CN115091725B - Intelligent blow molding machine for producing pesticide packing barrel and control method thereof - Google Patents
Intelligent blow molding machine for producing pesticide packing barrel and control method thereof Download PDFInfo
- Publication number
- CN115091725B CN115091725B CN202210780403.7A CN202210780403A CN115091725B CN 115091725 B CN115091725 B CN 115091725B CN 202210780403 A CN202210780403 A CN 202210780403A CN 115091725 B CN115091725 B CN 115091725B
- Authority
- CN
- China
- Prior art keywords
- feature
- vector
- uniformity
- intelligent
- blow molding
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 239000000575 pesticide Substances 0.000 title claims abstract description 158
- 238000000071 blow moulding Methods 0.000 title claims abstract description 149
- 238000000034 method Methods 0.000 title claims abstract description 74
- 238000012856 packing Methods 0.000 title description 2
- 238000007664 blowing Methods 0.000 claims abstract description 85
- 239000013598 vector Substances 0.000 claims description 272
- 238000004806 packaging method and process Methods 0.000 claims description 131
- 238000013527 convolutional neural network Methods 0.000 claims description 63
- 238000000465 moulding Methods 0.000 claims description 60
- 230000008569 process Effects 0.000 claims description 48
- 238000010586 diagram Methods 0.000 claims description 44
- 239000011159 matrix material Substances 0.000 claims description 39
- 238000012545 processing Methods 0.000 claims description 24
- 238000009826 distribution Methods 0.000 claims description 23
- 238000012546 transfer Methods 0.000 claims description 21
- 239000003905 agrochemical Substances 0.000 claims description 16
- 238000012937 correction Methods 0.000 claims description 16
- 230000003993 interaction Effects 0.000 claims description 15
- 238000004364 calculation method Methods 0.000 claims description 12
- 238000005259 measurement Methods 0.000 claims description 12
- 230000004913 activation Effects 0.000 claims description 11
- 238000013528 artificial neural network Methods 0.000 claims description 11
- 238000011176 pooling Methods 0.000 claims description 11
- 238000009512 pharmaceutical packaging Methods 0.000 claims description 5
- 230000005540 biological transmission Effects 0.000 claims description 3
- 238000010276 construction Methods 0.000 claims description 3
- 230000004927 fusion Effects 0.000 claims description 3
- 238000011156 evaluation Methods 0.000 claims description 2
- 238000003062 neural network model Methods 0.000 abstract description 5
- 238000007493 shaping process Methods 0.000 abstract description 5
- 238000009412 basement excavation Methods 0.000 abstract description 2
- 238000004519 manufacturing process Methods 0.000 abstract description 2
- 238000013473 artificial intelligence Methods 0.000 abstract 1
- 238000000605 extraction Methods 0.000 description 12
- 238000005065 mining Methods 0.000 description 9
- 230000008901 benefit Effects 0.000 description 5
- 238000005520 cutting process Methods 0.000 description 4
- 238000005516 engineering process Methods 0.000 description 4
- 238000013178 mathematical model Methods 0.000 description 4
- 230000000694 effects Effects 0.000 description 2
- 230000007246 mechanism Effects 0.000 description 2
- 238000002844 melting Methods 0.000 description 2
- 230000008018 melting Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000007789 sealing Methods 0.000 description 2
- 238000012360 testing method Methods 0.000 description 2
- 102100033620 Calponin-1 Human genes 0.000 description 1
- 102100033591 Calponin-2 Human genes 0.000 description 1
- 101000945318 Homo sapiens Calponin-1 Proteins 0.000 description 1
- 101000945403 Homo sapiens Calponin-2 Proteins 0.000 description 1
- 238000007792 addition Methods 0.000 description 1
- 230000004075 alteration Effects 0.000 description 1
- 238000000354 decomposition reaction Methods 0.000 description 1
- 230000002452 interceptive effect Effects 0.000 description 1
- 230000006798 recombination Effects 0.000 description 1
- 238000005215 recombination Methods 0.000 description 1
Images
Classifications
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B29—WORKING OF PLASTICS; WORKING OF SUBSTANCES IN A PLASTIC STATE IN GENERAL
- B29C—SHAPING OR JOINING OF PLASTICS; SHAPING OF MATERIAL IN A PLASTIC STATE, NOT OTHERWISE PROVIDED FOR; AFTER-TREATMENT OF THE SHAPED PRODUCTS, e.g. REPAIRING
- B29C49/00—Blow-moulding, i.e. blowing a preform or parison to a desired shape within a mould; Apparatus therefor
- B29C49/42—Component parts, details or accessories; Auxiliary operations
- B29C49/78—Measuring, controlling or regulating
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B29—WORKING OF PLASTICS; WORKING OF SUBSTANCES IN A PLASTIC STATE IN GENERAL
- B29C—SHAPING OR JOINING OF PLASTICS; SHAPING OF MATERIAL IN A PLASTIC STATE, NOT OTHERWISE PROVIDED FOR; AFTER-TREATMENT OF THE SHAPED PRODUCTS, e.g. REPAIRING
- B29C49/00—Blow-moulding, i.e. blowing a preform or parison to a desired shape within a mould; Apparatus therefor
- B29C49/42—Component parts, details or accessories; Auxiliary operations
- B29C49/78—Measuring, controlling or regulating
- B29C49/783—Measuring, controlling or regulating blowing pressure
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P70/00—Climate change mitigation technologies in the production process for final industrial or consumer products
- Y02P70/10—Greenhouse gas [GHG] capture, material saving, heat recovery or other energy efficient measures, e.g. motor control, characterised by manufacturing processes, e.g. for rolling metal or metal working
Landscapes
- Engineering & Computer Science (AREA)
- Manufacturing & Machinery (AREA)
- Mechanical Engineering (AREA)
- Blow-Moulding Or Thermoforming Of Plastics Or The Like (AREA)
- Shaping By String And By Release Of Stress In Plastics And The Like (AREA)
Abstract
The utility model relates to the field of intelligent manufacturing, it specifically discloses an intelligent blowing machine for producing pesticide package barrels and a control method thereof, and it utilizes artificial intelligence control technique to carry out deep implicit characteristic excavation to six views of pesticide package barrels and the power value of intelligent blowing machine's air-out equipment at a plurality of predetermined time points through the degree of depth neural network model, in order to excavate out the inside shaping condition in real time of pesticide package barrels, so as to classify and judge the air-out mode rationality of intelligent blowing machine's air-out equipment in the blow molding process, and then to blowing machine carries out blow molding control.
Description
Technical Field
The present invention relates to the field of intelligent manufacturing, and more particularly, to an intelligent blow molding machine for producing an agricultural chemical packaging barrel and a control method thereof.
Background
The blowing machine operates by first melting the plastic from the screw extruder and quantitatively extruding it from the blowing port and then placing it in a mold. Then, the cylinder is started so as to drive the cutting knife (including cold knife sealing and hot knife cutting). Next, the cylinder is activated to place the mold on the lower side of the blow port and blow-cool the mold to complete the molding.
In the blowing process, the air outlet control of the air blowing port is a key for controlling the forming precision of the pesticide packaging barrel. Ideally, the air speed of the air outlet should be controlled based on the real-time blow molding condition in the pesticide packaging barrel, but in the actual control process, on the one hand, the real-time internal molding condition of the pesticide packaging barrel is difficult to obtain, and on the other hand, the influence relationship between the air outlet mode and the uniformity of the molding wall thickness of the pesticide packaging barrel is difficult to model through a mathematical model.
Accordingly, there is a desire for an intelligent blow molding machine for producing pharmaceutical packaging barrels that is capable of fitting a relationship between the air out pattern and the uniformity of wall thickness of the pharmaceutical packaging barrel based on limited blow molding test data and performing blow molding control based on the air out pattern.
Disclosure of Invention
The present application has been made in order to solve the above technical problems. The embodiment of the application provides an intelligent blowing machine for producing an pesticide package barrel and a control method thereof, wherein an artificial intelligent control technology is utilized to conduct deep implicit feature mining on six views of the pesticide package barrel and power values of air outlet equipment of the intelligent blowing machine at a plurality of preset time points through a deep neural network model so as to mine out real-time internal forming conditions of the pesticide package barrel, so that air outlet mode rationality of the air outlet equipment of the intelligent blowing machine in a blowing forming process is judged in a classified mode, and then blowing control is conducted on the blowing machine.
According to one aspect of the present application, there is provided an intelligent blow molding machine for producing a packaging barrel for an agricultural chemical, comprising:
the device comprises an pesticide package barrel molding data acquisition module, a first image acquisition module, a second image acquisition module and a third image acquisition module, wherein the pesticide package barrel molding data acquisition module is used for acquiring first to sixth images of a blow-molded pesticide package barrel acquired by a camera, and the first to sixth images are six views of the pesticide package barrel;
the blowing process data acquisition module is used for acquiring power values of air outlet equipment of the intelligent blowing machine at a plurality of preset time points in the blowing molding process of the pesticide packaging barrel;
the molding data coding module is used for respectively passing the first image to the sixth image of the pesticide packaging barrel through a first convolutional neural network to obtain first feature images to sixth feature images;
the molding surface difference calculation module is used for calculating difference feature diagrams between every two feature diagrams in the first to sixth feature diagrams to obtain a plurality of difference feature diagrams, and arranging the difference feature diagrams according to sample dimensions to obtain uniformity feature diagrams;
the molding surface difference coding module is used for enabling the uniformity characteristic diagram to pass through a second convolution neural network serving as a filter so as to obtain uniformity characteristic vectors;
the blowing process data coding module is used for enabling power values of a plurality of preset time points of air outlet equipment of the intelligent blowing machine in the process of blowing the pesticide packaging barrel to pass through a time sequence coder comprising a one-dimensional convolution layer so as to obtain a power characteristic vector;
The molding surface difference characteristic distribution correction module is used for correcting the characteristic distribution of the uniformity characteristic vector based on the power characteristic vector to obtain a corrected uniformity characteristic vector; and
the control result generation module is used for enabling the corrected uniformity characteristic vector and the corrected power characteristic vector to pass through a classifier to obtain a classification result, wherein the classification result is used for indicating whether an air-out mode of air-out equipment of the intelligent blow molding machine in a blow molding process is reasonable or not.
In the above-mentioned intelligent blow molding machine for producing pesticide package barrels, the molding data encoding module is further configured to: and respectively carrying out convolution processing, pooling processing and activation processing on input data in forward transmission of layers by using each layer of the first convolutional neural network to respectively generate the first to sixth characteristic diagrams by the last layer of the first convolutional neural network, wherein the input of the first layer of the first convolutional neural network is respectively a first to sixth image of the pesticide packaging barrel.
In the above-mentioned intelligent blowing machine for producing pesticide package barrels, the molding surface difference calculation module includes: a difference unit configured to calculate a difference by position between each two of the first to sixth feature maps to obtain the plurality of difference feature maps; and the arrangement unit is used for arranging the plurality of differential feature maps according to the sample dimension so as to obtain the uniformity feature map.
In the above-mentioned intelligent blow molding machine for producing pesticide packaging barrels, the molding surface difference coding module is further configured to perform, in forward transfer of layers, input data by using each layer of the second convolutional neural network: carrying out convolution processing on input data to obtain a convolution characteristic diagram; carrying out mean pooling based on a local feature matrix on the convolution feature map to obtain a pooled feature map; performing nonlinear activation on the pooled feature map to obtain an activated feature map; the output of the last layer of the second convolutional neural network is the uniformity characteristic vector, and the input of the first layer of the second convolutional neural network is the uniformity characteristic map.
In the above-mentioned intelligent blowing machine for producing pesticide package barrels, blowing process data coding module includes: an input vector construction unit, configured to arrange power values of a plurality of predetermined time points of an air outlet device of the intelligent blow molding machine in a process of blow molding the pesticide packaging barrel into input vectors according to a time dimension; and the full-connection coding unit is used for carrying out full-connection coding on the input vector by using a full-connection layer of the time sequence coder according to the following formula to extract high-dimensional implicit characteristics of characteristic values of all positions in the input vector, wherein the formula is as follows: Wherein X is the input vector, Y is the output vector, W is the weight matrix, B is the bias vector, < >>Representing a matrix multiplication; and a one-dimensional convolution encoding unit, configured to perform one-dimensional convolution encoding on the input vector by using a one-dimensional convolution layer of the timing encoder to extract high-dimensional implicit correlation features between feature values of each position in the input vector, where the formula is:
where a is the width of the convolution kernel in the x direction, F is the vector of the convolution kernel parameters, G is the local vector matrix calculated with the convolution kernel function, and w is the size of the convolution kernel.
In the above-mentioned intelligent blowing machine for producing pesticide package barrels, the molding surface difference characteristic distribution correction module includes: the vector space interaction unit is used for calculating the position-based point multiplication between the uniformity characteristic vector and the power characteristic vector to obtain a space interaction characteristic vector; the data anisotropy evaluation unit is used for calculating the distance between the uniformity characteristic vector and the power characteristic vector; the measurement unit is used for dividing the characteristic value of each position in the space interaction characteristic vector by the distance between the uniformity characteristic vector and the power characteristic vector to obtain a measurement characteristic vector; an exponent operation unit for calculating a natural exponent function value with the eigenvalue of each position in the metric eigenvector as a power to obtain an exponent metric eigenvector; the first class probability calculation unit is used for passing the index measurement feature vector through the classifier to obtain first class probability; the second class probability calculation unit is used for passing the uniformity characteristic vector through the classifier to obtain second class probability; a class probability superposition unit, configured to calculate a product between the first class probability and the second class probability to obtain a weighted weight; and the weighting adjustment unit is used for weighting the characteristic values of each position in the uniformity characteristic vector by the weighting weight so as to obtain the corrected uniformity characteristic vector.
In the intelligent blow molding machine for producing the pesticide packaging barrel, the distance between the uniformity characteristic vector and the power characteristic vector is the square root of the Euclidean distance between the uniformity characteristic vector and the power characteristic vector.
In the above-mentioned intelligent blowing machine for producing pesticide package barrels, the control result generating module includes: the fusion unit is used for calculating a transfer matrix of the corrected uniformity characteristic vector relative to the power characteristic vector, wherein the power characteristic vector multiplied by the transfer matrix is equal to the corrected uniformity characteristic vector; and the classification unit is used for taking the transfer matrix as a classification characteristic matrix to pass through the classifier so as to obtain the classification result.
According to another aspect of the present application, a control method for an intelligent blow molding machine for producing an agricultural chemical packaging barrel, comprises:
acquiring first to sixth images of a blow-molded pesticide packaging barrel acquired by a camera, wherein the first to sixth images are six views of the pesticide packaging barrel;
acquiring power values of air outlet equipment of the intelligent blow molding machine at a plurality of preset time points in the process of blow molding the pesticide packaging barrel;
Respectively passing the first to sixth images of the pesticide packaging barrel through a first convolutional neural network to obtain first to sixth feature images;
calculating a difference feature map between every two feature maps in the first to sixth feature maps to obtain a plurality of difference feature maps, and arranging the difference feature maps according to sample dimensions to obtain a uniformity feature map;
the uniformity characteristic diagram is passed through a second convolution neural network serving as a filter to obtain uniformity characteristic vectors;
the power values of a plurality of preset time points of the air outlet equipment of the intelligent blowing machine in the process of blowing the pesticide packaging barrel are processed through a time sequence encoder comprising a one-dimensional convolution layer to obtain power characteristic vectors;
correcting the characteristic distribution of the uniformity characteristic vector based on the power characteristic vector to obtain a corrected uniformity characteristic vector; and
and passing the corrected uniformity characteristic vector and the power characteristic vector through a classifier to obtain a classification result, wherein the classification result is used for indicating whether an air outlet mode of air outlet equipment of the intelligent blow molding machine in a blow molding process is reasonable or not.
In the above control method for an intelligent blow molding machine for producing an agricultural chemical packaging barrel, the steps of passing the first to sixth images of the agricultural chemical packaging barrel through a first convolutional neural network to obtain first to sixth feature maps respectively include: and respectively carrying out convolution processing, pooling processing and activation processing on input data in forward transmission of layers by using each layer of the first convolutional neural network to respectively generate the first to sixth characteristic diagrams by the last layer of the first convolutional neural network, wherein the input of the first layer of the first convolutional neural network is respectively a first to sixth image of the pesticide packaging barrel.
In the above control method for an intelligent blow molding machine for producing an agricultural chemical packaging barrel, calculating a difference feature map between each two feature maps of the first to sixth feature maps to obtain a plurality of difference feature maps, and arranging the difference feature maps according to sample dimensions to obtain a uniformity feature map, including: calculating a difference by position between each two of the first to sixth feature maps to obtain the plurality of difference feature maps; and arranging the differential feature maps according to the sample dimension to obtain the uniformity feature map.
In the above control method for an intelligent blow molding machine for producing an pesticide packaging barrel, the step of passing the uniformity profile through a second convolutional neural network as a filter to obtain a uniformity profile vector comprises: input data are respectively carried out in forward transfer of layers by using each layer of the second convolutional neural network: carrying out convolution processing on input data to obtain a convolution characteristic diagram; carrying out mean pooling based on a local feature matrix on the convolution feature map to obtain a pooled feature map; performing nonlinear activation on the pooled feature map to obtain an activated feature map; the output of the last layer of the second convolutional neural network is the uniformity characteristic vector, and the input of the first layer of the second convolutional neural network is the uniformity characteristic map.
In the above control method for an intelligent blow molding machine for producing pesticide packaging barrels, the steps of passing power values of an air outlet device of the intelligent blow molding machine at a plurality of preset time points in the process of blow molding the pesticide packaging barrels through a time sequence encoder comprising a one-dimensional convolution layer to obtain power characteristic vectors include: arranging power values of a plurality of preset time points of air outlet equipment of the intelligent blow molding machine in the process of blow molding the pesticide packaging barrel into input vectors according to a time dimension; and performing full-connection coding on the input vector by using a full-connection layer of the time sequence coder to extract high-dimensional implicit characteristics of characteristic values of all positions in the input vector, wherein the formula is as follows: wherein X is the input vector, Y is the output vector, W is the weight matrix, B is the bias vector, < >>Representing a matrix multiplication; performing one-dimensional convolution encoding on the input vector by using a one-dimensional convolution layer of the time sequence encoder to extract high-dimensional implicit correlation features among feature values of each position in the input vector, wherein the formula is as follows:
where a is the width of the convolution kernel in the x direction, F is the vector of the convolution kernel parameters, G is the local vector matrix calculated with the convolution kernel function, and w is the size of the convolution kernel.
In the above control method for an intelligent blow molding machine for producing an agricultural chemical packaging barrel, correcting the characteristic distribution of the uniformity characteristic vector based on the power characteristic vector to obtain a corrected uniformity characteristic vector, including: calculating the position-based point multiplication between the uniformity characteristic vector and the power characteristic vector to obtain a space interaction characteristic vector; calculating the distance between the uniformity characteristic vector and the power characteristic vector; dividing the characteristic value of each position in the space interaction characteristic vector by the distance between the uniformity characteristic vector and the power characteristic vector to obtain a measurement characteristic vector; calculating natural exponential function values with the eigenvalues of all positions in the measurement eigenvector as powers to obtain an exponential measurement eigenvector; passing the exponential metric feature vector through the classifier to obtain a first class probability; passing the uniformity feature vector through the classifier to obtain a second class probability; calculating the product between the first type probability and the second type probability to obtain a weighted weight; and weighting the characteristic values of each position in the uniformity characteristic vector by the weighting weight to obtain the corrected uniformity characteristic vector.
In the above control method for an intelligent blow molding machine for producing an agricultural package barrel, the distance between the uniformity eigenvector and the power eigenvector is the square root of the euclidean distance between the uniformity eigenvector and the power eigenvector.
In the above control method for an intelligent blow molding machine for producing an agricultural chemical packaging barrel, the step of passing the corrected uniformity characteristic vector and the power characteristic vector through a classifier to obtain a classification result includes: calculating a transfer matrix of the corrected uniformity feature vector relative to the power feature vector, wherein the power feature vector multiplied by the transfer matrix is equal to the corrected uniformity feature vector; and taking the transfer matrix as a classification characteristic matrix to pass through the classifier so as to obtain the classification result.
Compared with the prior art, the intelligent blowing machine for producing the pesticide packaging barrel and the control method thereof utilize an artificial intelligent control technology to carry out deep implicit characteristic excavation on six views of the pesticide packaging barrel and power values of air outlet equipment of the intelligent blowing machine at a plurality of preset time points through a deep neural network model so as to excavate the real-time internal forming condition of the pesticide packaging barrel, so that the air outlet mode rationality of the air outlet equipment of the intelligent blowing machine in the blowing forming process is judged in a classified mode, and then the blowing machine is controlled.
Drawings
The foregoing and other objects, features and advantages of the present application will become more apparent from the following more particular description of embodiments of the present application, as illustrated in the accompanying drawings. The accompanying drawings are included to provide a further understanding of embodiments of the application and are incorporated in and constitute a part of this specification, illustrate the application and not constitute a limitation to the application. In the drawings, like reference numerals generally refer to like parts or steps.
Fig. 1 is an application scenario diagram of an intelligent blow molding machine for producing an pesticide packaging barrel according to an embodiment of the present application.
Fig. 2 is a block diagram of an intelligent blow molding machine for producing an agricultural package barrel according to an embodiment of the present application.
Fig. 3 is a block diagram of a molding surface difference profile correction module in an intelligent blow molding machine for producing an pesticide packaging barrel according to an embodiment of the present application.
Fig. 4 is a flow chart of a control method of an intelligent blow molding machine for producing an agricultural package barrel according to an embodiment of the present application.
Fig. 5 is a schematic diagram of the control method of the intelligent blow molding machine for producing the pesticide packaging barrel according to the embodiment of the present application.
Detailed Description
Hereinafter, example embodiments according to the present application will be described in detail with reference to the accompanying drawings. It should be apparent that the described embodiments are only some of the embodiments of the present application and not all of the embodiments of the present application, and it should be understood that the present application is not limited by the example embodiments described herein.
Scene overview
As previously mentioned, the blowing machine operates by first melting the plastic from the screw extruder and quantitatively extruding it from the blowing port and then placing it in a mold. Then, the cylinder is started so as to drive the cutting knife (including cold knife sealing and hot knife cutting). Next, the cylinder is activated to place the mold on the lower side of the blow port and blow-cool the mold to complete the molding.
In the blowing process, the air outlet control of the air blowing port is a key for controlling the forming precision of the pesticide packaging barrel. Ideally, the air speed of the air outlet should be controlled based on the real-time blow molding condition in the pesticide packaging barrel, but in the actual control process, on the one hand, the real-time internal molding condition of the pesticide packaging barrel is difficult to obtain, and on the other hand, the influence relationship between the air outlet mode and the uniformity of the molding wall thickness of the pesticide packaging barrel is difficult to model through a mathematical model.
Accordingly, there is a desire for an intelligent blow molding machine for producing pharmaceutical packaging barrels that is capable of fitting a relationship between the air out pattern and the uniformity of wall thickness of the pharmaceutical packaging barrel based on limited blow molding test data and performing blow molding control based on the air out pattern.
Based on the above, the inventor of the application indicates the uniformity of the formed wall thickness of the pesticide packaging barrel by collecting six views of the pesticide packaging barrel formed by blow molding, and uses a deep neural network to excavate the real-time internal forming condition of the pesticide packaging barrel, so as to classify and judge the rationality of the air outlet mode of the air outlet equipment of the intelligent blow molding machine in the blow molding process based on the deep characteristic information, and further carry out blow molding control on the blow molding machine.
Specifically, in the technical scheme of the application, first to sixth images of a blow-molded pesticide packaging barrel are acquired through a camera, wherein the first to sixth images are six views of the pesticide packaging barrel. Then, considering that the convolutional neural network has more excellent performance in terms of feature extraction of images, the first convolutional neural network is used for feature mining of the first to sixth images of the pesticide packaging barrel to extract local high-dimensional implicit features in the first to sixth images of the pesticide packaging barrel, thereby obtaining first to sixth feature maps.
It should be understood that, since in the six views of the pesticide packaging barrel, there will be areas where the respective features overlap, that is, overlapping feature information, in order to focus more on these implicit associated feature information in terms of high-dimensional feature extraction, the differential feature map between every two of the first to sixth feature maps is further calculated to obtain a plurality of differential feature maps. Accordingly, in one specific example, the difference by position between each two of the first to sixth feature maps is calculated to obtain the plurality of difference feature maps, that is, the following formula:
wherein ,Fi Characteristic values representing respective positions in the ith characteristic diagram, F j Characteristic value representing corresponding position in jth characteristic diagram, F k ' represents the eigenvalues of each location in the differential signature,representing the difference by location.
And then, arranging the differential feature graphs according to the dimension of the sample to integrate the feature information of the feature graphs in the dimension of the sample, thereby obtaining a uniformity feature graph. In this way, the uniformity feature map may be further subjected to local implicit feature mining by using a second convolutional neural network as a filter to obtain uniformity feature vectors.
In addition, when judging and controlling the rationality of the air outlet mode of the air outlet equipment of the intelligent blow molding machine in the blow molding process, the power value of each time point in the air outlet equipment needs to be considered. Therefore, it is also necessary to obtain power values of the air outlet device of the intelligent blowing machine at a plurality of preset time points in the process of blowing the pesticide packaging barrel, and code the power values in a time sequence encoder comprising a one-dimensional convolution layer, so as to extract characteristic distribution information of the power values of the air outlet device of the intelligent blowing machine in the process of blowing the pesticide packaging barrel in a time dimension, thereby obtaining a power characteristic vector.
It should be appreciated that due to uniformity feature vector V a Is characterized in that six-sided view as a data source is subjected to feature extraction through a first convolution neural network and a second convolution neural network which are cascaded, and the data density is higher than the power feature vector V in terms of the data volume of source data and the semantic density of feature extraction p This results in a quasi-probabilistic deviation between the uniformity feature vector and the power feature vector.
Thus, for uniformity eigenvector V a Data-intensive cluster-based corrections are made, namely:
wherein softmax (·) represents the probability value of the feature vector obtained by the classifier, exp (·) represents the exponential operation of the vector, performing the exponential operation on the vector represents the natural exponential function value which is a power of the feature value of each position in the vector, vector division represents dividing the feature value of each position in the molecular vector by the denominator, and as such, represents the multiplication of the vector.
Here, the data-dense cluster-based correction enables spatial interactions based on per-position feature correspondence through a self-attention mechanism between feature vectors, and measures data dissimilarity through vector distances, representing example similarity between data-dense objects and general objects. Therefore, the self-adaptive dependence of the classification probability value on different data-intensive clusters is determined by taking the product of the classification probability values as a weighting coefficient, the parameter self-adaptability of the corrected uniformity characteristic vector on the classification objective function is improved, the deviation of the class probability between the uniformity characteristic vector and the power characteristic vector is improved, and the classification accuracy is further improved. Therefore, the rationality of the air outlet mode of the air outlet equipment of the intelligent blow molding machine in the blow molding process can be accurately judged, so that blow molding control can be carried out on the blow molding machine.
Based on this, the application proposes an intelligent blow molding machine for producing a packaging barrel for an agricultural chemical, comprising: the device comprises an pesticide package barrel molding data acquisition module, a first image acquisition module, a second image acquisition module and a third image acquisition module, wherein the pesticide package barrel molding data acquisition module is used for acquiring first to sixth images of a blow-molded pesticide package barrel acquired by a camera, and the first to sixth images are six views of the pesticide package barrel; the blowing process data acquisition module is used for acquiring power values of air outlet equipment of the intelligent blowing machine at a plurality of preset time points in the blowing molding process of the pesticide packaging barrel; the molding data coding module is used for respectively passing the first image to the sixth image of the pesticide packaging barrel through a first convolutional neural network to obtain first feature images to sixth feature images; the molding surface difference calculation module is used for calculating difference feature diagrams between every two feature diagrams in the first to sixth feature diagrams to obtain a plurality of difference feature diagrams, and arranging the difference feature diagrams according to sample dimensions to obtain uniformity feature diagrams; the molding surface difference coding module is used for enabling the uniformity characteristic diagram to pass through a second convolution neural network serving as a filter so as to obtain uniformity characteristic vectors; the blowing process data coding module is used for enabling power values of a plurality of preset time points of air outlet equipment of the intelligent blowing machine in the process of blowing the pesticide packaging barrel to pass through a time sequence coder comprising a one-dimensional convolution layer so as to obtain a power characteristic vector; the molding surface difference characteristic distribution correction module is used for correcting the characteristic distribution of the uniformity characteristic vector based on the power characteristic vector to obtain a corrected uniformity characteristic vector; and the control result generation module is used for enabling the corrected uniformity characteristic vector and the corrected power characteristic vector to pass through a classifier to obtain a classification result, wherein the classification result is used for indicating whether an air-out mode of air-out equipment of the intelligent blow molding machine in a blow molding process is reasonable or not.
Fig. 1 illustrates an application scenario diagram of an intelligent blow molding machine for producing an pesticide packaging barrel according to an embodiment of the present application. As shown in fig. 1, in this application scenario, first to sixth images of a blow-molded pesticide packaging barrel (e.g., P as illustrated in fig. 1) are acquired by a camera (e.g., C as illustrated in fig. 1), which are six views of the pesticide packaging barrel, and power values of an air-out device (e.g., E as illustrated in fig. 1) of an intelligent blow-molding machine (e.g., B as illustrated in fig. 1) at a plurality of predetermined time points in blow-molding the pesticide packaging barrel are acquired. Then, the obtained power values of the first to sixth images and the plurality of predetermined time points are input into a server (for example, a cloud server S as illustrated in fig. 1) deployed with an intelligent blow molding machine algorithm for producing pesticide packages, wherein the server is capable of processing the power values of the first to sixth images and the plurality of predetermined time points with the intelligent blow molding machine algorithm for producing pesticide packages to generate a classification result for indicating whether an air-out mode of an air-out device of the intelligent blow molding machine in a blow molding process is reasonable.
Having described the basic principles of the present application, various non-limiting embodiments of the present application will now be described in detail with reference to the accompanying drawings.
Exemplary System
Fig. 2 illustrates a block diagram of an intelligent blow molding machine for producing a packaging barrel for an pesticide according to an embodiment of the present application. As shown in fig. 2, an intelligent blow molding machine 200 for producing a packaging barrel for an agricultural chemical according to an embodiment of the present application includes: an pesticide package barrel molding data acquisition module 210 for acquiring first to sixth images of a blow molded pesticide package barrel acquired by a camera, the first to sixth images being six views of the pesticide package barrel; a blowing process data obtaining module 220, configured to obtain power values of an air outlet device of an intelligent blowing machine at a plurality of predetermined time points in a process of performing blow molding on the pesticide packaging barrel; a molding data encoding module 230, configured to pass the first to sixth images of the pesticide packaging barrel through a first convolutional neural network to obtain first to sixth feature maps, respectively; a molding surface difference calculation module 240, configured to calculate a difference feature map between every two feature maps of the first to sixth feature maps to obtain a plurality of difference feature maps, and arrange the difference feature maps according to a sample dimension to obtain a uniformity feature map; the molding surface difference coding module 250 is configured to pass the uniformity feature map through a second convolutional neural network serving as a filter to obtain a uniformity feature vector; the blowing process data encoding module 260 is configured to pass power values of a plurality of predetermined time points of the air outlet device of the intelligent blowing machine in the process of performing blow molding on the pesticide packaging barrel through a time sequence encoder including a one-dimensional convolution layer to obtain a power feature vector; a molding surface difference feature distribution correction module 270, configured to correct a feature distribution of the uniformity feature vector based on the power feature vector to obtain a corrected uniformity feature vector; and a control result generating module 280, configured to pass the corrected uniformity feature vector and the power feature vector through a classifier to obtain a classification result, where the classification result is used to indicate whether an air-out mode of an air-out device of the intelligent blow molding machine in a blow molding process is reasonable.
Specifically, in this embodiment of the present application, the pesticide packaging barrel molding data obtaining module 210, the blow molding process data obtaining module 220 and the molding data encoding module 230 are configured to obtain first to sixth images of a blow molded pesticide packaging barrel collected by a camera, where the first to sixth images are six views of the pesticide packaging barrel, obtain power values of an air outlet device of an intelligent blow molding machine at a plurality of predetermined time points in a blow molding process of the pesticide packaging barrel, and then pass the first to sixth images of the pesticide packaging barrel through a first convolutional neural network to obtain first to sixth feature maps, respectively. . As previously mentioned, it should be understood that the control of the air outlet to the air blowing port is critical to control the accuracy of the molding of the package. Ideally, the wind speed of the air outlet should be controlled based on the real-time blow molding condition in the pesticide packaging barrel, but in the actual control process, on one hand, the real-time internal molding condition of the pesticide packaging barrel is difficult to obtain, and on the other hand, the influence relationship between the air outlet mode and the uniformity of the molding wall thickness of the pesticide packaging barrel is difficult to model through a mathematical model. Therefore, in the technical scheme of this application, six views through gathering blow molding's pesticide package barrel represent the shaping wall thickness degree of consistency of pesticide package barrel to use deep neural network to dig out the inside shaping condition in real time of pesticide package barrel, with based on this kind of deep characteristic information come to classify judgement in the air-out mode rationality of the air-out equipment of intelligent blowing machine in the blow molding process, and then to blow molding machine carries out blow molding control.
That is, specifically, in the technical solution of the present application, first to sixth images of a blow-molded pesticide package barrel are acquired by a camera, the first to sixth images being six views of the pesticide package barrel. In addition, when judging and controlling the rationality of the air outlet mode of the air outlet equipment of the intelligent blow molding machine in the blow molding process, the power value of each time point in the air outlet equipment needs to be considered. Therefore, it is also necessary to obtain power values of the air outlet device of the intelligent blowing machine at a plurality of preset time points in the process of blowing the pesticide packaging barrel.
Then, considering that the convolutional neural network has more excellent performance in terms of feature extraction of images, the first convolutional neural network is used for feature mining of the first to sixth images of the pesticide packaging barrel to extract local high-dimensional implicit features in the first to sixth images of the pesticide packaging barrel, thereby obtaining first to sixth feature maps. Accordingly, in one specific example, the input data is respectively subjected to convolution processing, pooling processing and activation processing in forward transfer of layers using the layers of the first convolutional neural network to respectively generate the first to sixth feature maps from the last layer of the first convolutional neural network, wherein the input of the first layer of the first convolutional neural network is the first to sixth images of the pesticide packaging barrel.
Specifically, in this embodiment of the present application, the molding surface difference calculating module 240 and the molding surface difference encoding module 250 are configured to calculate a difference feature map between every two feature maps of the first to sixth feature maps to obtain a plurality of difference feature maps, arrange the difference feature maps according to a sample dimension to obtain a uniformity feature map, and then pass the uniformity feature map through a second convolutional neural network serving as a filter to obtain a uniformity feature vector. It should be understood that, since in the six views of the pesticide packaging barrel, there will be areas where the respective features overlap, that is, overlapping feature information, in order to focus more on these implicit associated feature information in terms of high-dimensional feature extraction, the differential feature map between every two of the first to sixth feature maps is further calculated to obtain a plurality of differential feature maps.
And then, arranging the differential feature graphs according to the dimension of the sample to integrate the feature information of the feature graphs in the dimension of the sample, thereby obtaining a uniformity feature graph. In this way, the uniformity feature map may be further subjected to local implicit feature mining by using a second convolutional neural network as a filter to obtain uniformity feature vectors. Accordingly, in one specific example, the input data is performed separately in forward transfer of layers using each layer of the second convolutional neural network: carrying out convolution processing on input data to obtain a convolution characteristic diagram; carrying out mean pooling based on a local feature matrix on the convolution feature map to obtain a pooled feature map; performing nonlinear activation on the pooled feature map to obtain an activated feature map; the output of the last layer of the second convolutional neural network is the uniformity characteristic vector, and the input of the first layer of the second convolutional neural network is the uniformity characteristic map.
More specifically, in an embodiment of the present application, the molding surface difference calculating module includes: and a difference unit configured to calculate a difference by position between each two of the first to sixth feature maps to obtain the plurality of difference feature maps. Accordingly, in one specific example, the difference by position between each two of the first to sixth feature maps is calculated to obtain the plurality of difference feature maps, that is, the following formula:
wherein ,Fi Characteristic values representing respective positions in the ith characteristic diagram, F j Characteristic value representing corresponding position in jth characteristic diagram, F k ' represents the eigenvalues of each location in the differential signature,representing the difference by location. And the arrangement unit is used for arranging the plurality of differential feature maps according to the sample dimension so as to obtain the uniformity feature map.
Specifically, in the embodiment of the present application, the blowing process data encoding module 260 is configured to pass power values of a plurality of predetermined time points of the air outlet device of the intelligent blowing machine in the process of blowing the pesticide packaging barrel through a time sequence encoder including a one-dimensional convolution layer to obtain a power feature vector. It should be understood that when judging and controlling the rationality of the air outlet mode of the air outlet device of the intelligent blow molding machine in the blow molding process, the power value at each time point needs to be considered. Therefore, in the technical scheme of the application, it is further required to obtain power values of a plurality of preset time points of the air outlet device of the intelligent blow molding machine in the blow molding process of the pesticide packaging barrel, and code the power values in a time sequence encoder comprising a one-dimensional convolution layer, so as to extract characteristic distribution information of the power values of the air outlet device of the intelligent blow molding machine in the time dimension in the blow molding process of the pesticide packaging barrel, thereby obtaining a power characteristic vector.
More specifically, in an embodiment of the present application, the blow molding process data encoding module includes: an input vector construction unit for constructing the power of the air outlet device of the intelligent blowing machine at a plurality of preset time points in the process of blowing and molding the pesticide packaging barrelThe values are arranged as input vectors according to the time dimension; and the full-connection coding unit is used for carrying out full-connection coding on the input vector by using a full-connection layer of the time sequence coder according to the following formula to extract high-dimensional implicit characteristics of characteristic values of all positions in the input vector, wherein the formula is as follows:wherein X is the input vector, Y is the output vector, W is the weight matrix, B is the bias vector, < >>Representing a matrix multiplication; and a one-dimensional convolution encoding unit, configured to perform one-dimensional convolution encoding on the input vector by using a one-dimensional convolution layer of the timing encoder to extract high-dimensional implicit correlation features between feature values of each position in the input vector, where the formula is:
where a is the width of the convolution kernel in the x direction, F is the vector of the convolution kernel parameters, G is the local vector matrix calculated with the convolution kernel function, and w is the size of the convolution kernel.
Specifically, in the embodiment of the present application, the molding surface difference feature distribution correction module 270 is configured to correct the feature distribution of the uniformity feature vector based on the power feature vector to obtain a corrected uniformity feature vector. It should be appreciated that due to the uniformity feature vector V a Is characterized in that six-sided view as data source is subjected to feature extraction through the first convolutional neural network and the second convolutional neural network which are cascaded, and the data density is higher than the power feature vector V in terms of data volume of source data or semantic density of feature extraction p This results in a quasi-probabilistic deviation between the uniformity feature vector and the power feature vector. Therefore, in the technical scheme of the application,requiring the uniformity feature vector V a Data-intensive cluster-based corrections are made, namely:
wherein softmax (·) represents the probability value of the feature vector obtained by the classifier, exp (·) represents the exponential operation of the vector, performing the exponential operation on the vector represents the natural exponential function value which is a power of the feature value of each position in the vector, vector division represents dividing the feature value of each position in the molecular vector by the denominator, and as such, represents the multiplication of the vector.
More specifically, in an embodiment of the present application, the molding surface difference feature distribution correction module includes: firstly, calculating the position-by-position point multiplication between the uniformity characteristic vector and the power characteristic vector to obtain a space interaction characteristic vector. Then, a distance between the uniformity feature vector and the power feature vector is calculated. In particular, here, the distance between the uniformity feature vector and the power feature vector is the square root of the euclidean distance between the uniformity feature vector and the power feature vector. And dividing the characteristic value of each position in the space interaction characteristic vector by the distance between the uniformity characteristic vector and the power characteristic vector to obtain a measurement characteristic vector. Next, natural exponential function values, which are exponentiations of the eigenvalues of the individual positions in the metric eigenvector, are calculated to obtain an exponential metric eigenvector. The exponentially measured feature vectors are then passed through the classifier to obtain a first class probability. And then, the uniformity characteristic vector passes through the classifier to obtain second-class probability. Then, the product between the first type probability and the second type probability is calculated to obtain a weighted weight. And finally, weighting the characteristic values of each position in the uniformity characteristic vector by the weighting weight to obtain the corrected uniformity characteristic vector. It should be appreciated that the data-dense cluster-based correction enables spatial interactions based on per-location feature correspondence through a self-attention mechanism between feature vectors, and measures data dissimilarity through vector distance, representing example similarity between data-dense objects and general objects. Therefore, the self-adaptive dependence of the classification probability value on different data-intensive clusters is determined by taking the product of the classification probability values as a weighting coefficient, the parameter self-adaptability of the corrected uniformity characteristic vector on the classification objective function is improved, and the deviation of the class probability between the uniformity characteristic vector and the power characteristic vector is improved.
Fig. 3 illustrates a block diagram of a molding surface differential feature profile correction module in an intelligent blow molding machine for producing an pesticide packaging barrel according to an embodiment of the present application. As shown in fig. 3, the molding surface difference feature distribution correction module 270 includes: a vector space interaction unit 271, configured to calculate a space interaction feature vector by multiplying the position-wise points between the uniformity feature vector and the power feature vector; a data variability estimating unit 272 for calculating a distance between the uniformity characteristic vector and the power characteristic vector; a measurement unit 273, configured to divide the feature value of each position in the spatial interaction feature vector by the distance between the uniformity feature vector and the power feature vector to obtain a measurement feature vector; an exponent operation unit 274 for calculating a natural exponent function value exponentiated by the eigenvalue of each position in the metric eigenvector to obtain an exponent metric eigenvector; a first class probability calculation unit 275, configured to pass the exponential metric feature vector through the classifier to obtain a first class probability; a second probability calculation unit 276, configured to pass the uniformity feature vector through the classifier to obtain a second probability; a class probability superimposing unit 277 for calculating a product between the first class probability and the second class probability to obtain a weighted weight; and a weighting adjustment unit 278, configured to weight the feature values of each position in the uniformity feature vector with the weighting weight to obtain the corrected uniformity feature vector.
Specifically, in this embodiment of the present application, the control result generating module 280 is configured to pass the corrected uniformity feature vector and the power feature vector through a classifier to obtain a classification result, where the classification result is used to indicate whether an air-out mode of an air-out device of the intelligent blow molding machine in a blow molding process is reasonable. That is, in the technical solution of the present application, further, the corrected uniformity feature vector and the power feature vector may be passed through a classifier to obtain a classification result that is used to indicate whether the air-out mode of the air-out device of the intelligent blow molding machine in the blow molding process is reasonable.
More specifically, in an embodiment of the present application, the control result generating module includes: and the fusion unit is used for calculating a transfer matrix of the corrected uniformity characteristic vector relative to the power characteristic vector, wherein the power characteristic vector multiplied by the transfer matrix is equal to the corrected uniformity characteristic vector. Accordingly, in one specific example, the transfer matrix of the corrected uniformity feature vector relative to the power feature vector is calculated with the following formula: s=t×f, where F represents the power eigenvector, T represents the transfer matrix, and S represents the corrected uniformity eigenvector. And the classification unit is used for taking the transfer matrix as a classification characteristic matrix to pass through the classifier so as to obtain the classification result. In one specific example, the classifier processes the classification feature matrix to generate a classification result with the following formula: softmax { (W) n ,B n ):…:(W 1 ,B 1 ) Project (F), where Project (F) represents projecting the classification feature matrix as a vector, W 1 To W n Weight matrix for all the connection layers of each layer, B 1 To B n Representing the bias matrix for each fully connected layer.
To sum up, the intelligent blowing machine 200 for producing the pesticide packaging barrel according to the embodiment of the present application is illustrated, and uses an artificial intelligent control technology to perform deep implicit feature mining on six views of the pesticide packaging barrel and power values of air outlet devices of the intelligent blowing machine at a plurality of preset time points through a deep neural network model, so as to mine out real-time internal molding conditions of the pesticide packaging barrel, so as to perform classification judgment on air outlet mode rationality of the air outlet devices of the intelligent blowing machine in a blowing molding process, and further perform blowing control on the blowing machine.
As described above, the intelligent blow molding machine 200 for producing an agricultural chemical packaging barrel according to an embodiment of the present application may be implemented in various terminal devices, such as a server or the like for an intelligent blow molding machine algorithm for producing an agricultural chemical packaging barrel. In one example, the intelligent blow molding machine 200 for producing a packaging barrel of an pesticide according to embodiments of the present application may be integrated into the terminal device as a software module and/or hardware module. For example, the intelligent blow molding machine 200 for producing the package of pesticide may be a software module in the operating system of the terminal device or may be an application developed for the terminal device; of course, the intelligent blow molding machine 200 for producing the package barrels can also be one of the many hardware modules of the terminal device.
Alternatively, in another example, the intelligent blow molding machine 200 for producing the pesticide packaging barrel and the terminal device may be separate devices, and the intelligent blow molding machine 200 for producing the pesticide packaging barrel may be connected to the terminal device through a wired and/or wireless network and transmit the interactive information in a agreed data format.
Exemplary method
Fig. 4 illustrates a flow chart of a control method of an intelligent blow molding machine for producing a package of pesticide. As shown in fig. 4, a control method of an intelligent blow molding machine for producing an agricultural chemical packaging barrel according to an embodiment of the present application includes the steps of: s110, acquiring first to sixth images of a blow-molded pesticide packaging barrel acquired by a camera, wherein the first to sixth images are six views of the pesticide packaging barrel; s120, acquiring power values of air outlet equipment of an intelligent blow molding machine at a plurality of preset time points in the process of blow molding the pesticide packaging barrel; s130, respectively passing the first to sixth images of the pesticide packaging barrel through a first convolutional neural network to obtain first to sixth feature maps; s140, calculating a difference feature map between every two feature maps in the first to sixth feature maps to obtain a plurality of difference feature maps, and arranging the difference feature maps according to sample dimensions to obtain a uniformity feature map; s150, the uniformity characteristic map is passed through a second convolution neural network serving as a filter to obtain uniformity characteristic vectors; s160, enabling power values of a plurality of preset time points of air outlet equipment of the intelligent blowing machine in the process of blowing and molding the pesticide packaging barrel to pass through a time sequence encoder comprising a one-dimensional convolution layer to obtain a power characteristic vector; s170, correcting the characteristic distribution of the uniformity characteristic vector based on the power characteristic vector to obtain a corrected uniformity characteristic vector; and S180, passing the corrected uniformity characteristic vector and the power characteristic vector through a classifier to obtain a classification result, wherein the classification result is used for indicating whether an air-out mode of air-out equipment of the intelligent blow molding machine in a blow molding process is reasonable or not.
Fig. 5 illustrates a schematic architecture of a control method of an intelligent blow molding machine for producing an pesticide packaging barrel according to an embodiment of the present application. As shown in fig. 5, in the network architecture of the control method of the intelligent blow molding machine for producing the pesticide packaging barrel, first, the obtained first to sixth images (for example, P1 as illustrated in fig. 5) of the pesticide packaging barrel are respectively passed through a first convolutional neural network (for example, CNN1 as illustrated in fig. 5) to obtain first to sixth feature maps (for example, F1 to F6 as illustrated in fig. 5); next, calculating a differential feature map between each two of the first to sixth feature maps to obtain a plurality of differential feature maps (e.g., FD as illustrated in fig. 5), and arranging the differential feature maps in a sample dimension to obtain a uniformity feature map (e.g., FE as illustrated in fig. 5); the uniformity profile is then passed through a second convolutional neural network (e.g., CNN2 as illustrated in fig. 5) as a filter to obtain a uniformity profile vector (e.g., VF1 as illustrated in fig. 5); next, passing the obtained power values (e.g., P2 as illustrated in fig. 5) of the air outlet device of the intelligent blow molding machine at a plurality of predetermined time points in the process of blow molding the pesticide packaging barrel through a time sequence encoder (e.g., E as illustrated in fig. 5) comprising a one-dimensional convolution layer to obtain a power feature vector (e.g., VF2 as illustrated in fig. 5); then, based on the power feature vector, correcting a feature distribution of the uniformity feature vector to obtain a corrected uniformity feature vector (e.g., VF3 as illustrated in fig. 5); and finally, passing the corrected uniformity characteristic vector and the power characteristic vector through a classifier (for example, a circle S as illustrated in fig. 5) to obtain a classification result, wherein the classification result is used for indicating whether an air-out mode of an air-out device of the intelligent blow molding machine in a blow molding process is reasonable or not.
More specifically, in step S110, step S120 and step S130, first to sixth images of the blow-molded pesticide packaging barrel acquired by a camera are acquired, the first to sixth images are six views of the pesticide packaging barrel, power values of air outlet equipment of an intelligent blow molding machine at a plurality of preset time points in the process of blow molding the pesticide packaging barrel are acquired, and the first to sixth images of the pesticide packaging barrel are respectively passed through a first convolutional neural network to obtain first to sixth feature maps. It should be understood that the control of the air outlet to the air blowing port is critical to control the accuracy of the molding of the package barrel during the blow molding process. Ideally, the wind speed of the air outlet should be controlled based on the real-time blow molding condition in the pesticide packaging barrel, but in the actual control process, on one hand, the real-time internal molding condition of the pesticide packaging barrel is difficult to obtain, and on the other hand, the influence relationship between the air outlet mode and the uniformity of the molding wall thickness of the pesticide packaging barrel is difficult to model through a mathematical model. Therefore, in the technical scheme of this application, six views through gathering blow molding's pesticide package barrel represent the shaping wall thickness degree of consistency of pesticide package barrel to use deep neural network to dig out the inside shaping condition in real time of pesticide package barrel, with based on this kind of deep characteristic information come to classify judgement in the air-out mode rationality of the air-out equipment of intelligent blowing machine in the blow molding process, and then to blow molding machine carries out blow molding control.
That is, specifically, in the technical solution of the present application, first to sixth images of a blow-molded pesticide package barrel are acquired by a camera, the first to sixth images being six views of the pesticide package barrel. In addition, when judging and controlling the rationality of the air outlet mode of the air outlet equipment of the intelligent blow molding machine in the blow molding process, the power value of each time point in the air outlet equipment needs to be considered. Therefore, it is also necessary to obtain power values of the air outlet device of the intelligent blowing machine at a plurality of preset time points in the process of blowing the pesticide packaging barrel.
Then, considering that the convolutional neural network has more excellent performance in terms of feature extraction of images, the first convolutional neural network is used for feature mining of the first to sixth images of the pesticide packaging barrel to extract local high-dimensional implicit features in the first to sixth images of the pesticide packaging barrel, thereby obtaining first to sixth feature maps. Accordingly, in one specific example, the input data is respectively subjected to convolution processing, pooling processing and activation processing in forward transfer of layers using the layers of the first convolutional neural network to respectively generate the first to sixth feature maps from the last layer of the first convolutional neural network, wherein the input of the first layer of the first convolutional neural network is the first to sixth images of the pesticide packaging barrel.
More specifically, in step S140 and step S150, a difference feature map between each two of the first to sixth feature maps is calculated to obtain a plurality of difference feature maps, the difference feature maps are arranged according to sample dimensions to obtain a uniformity feature map, and the uniformity feature map is passed through a second convolutional neural network as a filter to obtain uniformity feature vectors. It should be understood that, since in the six views of the pesticide packaging barrel, there will be areas where the respective features overlap, that is, overlapping feature information, in order to focus more on these implicit associated feature information in terms of high-dimensional feature extraction, the differential feature map between every two of the first to sixth feature maps is further calculated to obtain a plurality of differential feature maps.
And then, arranging the differential feature graphs according to the dimension of the sample to integrate the feature information of the feature graphs in the dimension of the sample, thereby obtaining a uniformity feature graph. In this way, the uniformity feature map may be further subjected to local implicit feature mining by using a second convolutional neural network as a filter to obtain uniformity feature vectors. Accordingly, in one specific example, the input data is performed separately in forward transfer of layers using each layer of the second convolutional neural network: carrying out convolution processing on input data to obtain a convolution characteristic diagram; carrying out mean pooling based on a local feature matrix on the convolution feature map to obtain a pooled feature map; performing nonlinear activation on the pooled feature map to obtain an activated feature map; the output of the last layer of the second convolutional neural network is the uniformity characteristic vector, and the input of the first layer of the second convolutional neural network is the uniformity characteristic map.
More specifically, in step S160, power values of a plurality of predetermined time points of the air outlet device of the intelligent blow molding machine in the process of blow molding the pesticide packaging barrel are passed through a time sequence encoder comprising a one-dimensional convolution layer to obtain a power feature vector. It should be understood that when judging and controlling the rationality of the air outlet mode of the air outlet device of the intelligent blow molding machine in the blow molding process, the power value at each time point needs to be considered. Therefore, in the technical scheme of the application, it is further required to obtain power values of a plurality of preset time points of the air outlet device of the intelligent blow molding machine in the blow molding process of the pesticide packaging barrel, and code the power values in a time sequence encoder comprising a one-dimensional convolution layer, so as to extract characteristic distribution information of the power values of the air outlet device of the intelligent blow molding machine in the time dimension in the blow molding process of the pesticide packaging barrel, thereby obtaining a power characteristic vector.
More specifically, in step S170, based on the power feature vector, the feature distribution of the uniformity feature vector is corrected to obtain a corrected uniformity feature vector. It should be appreciated that due to the uniformity Feature vector V a Is characterized in that six-sided view as data source is subjected to feature extraction through the first convolutional neural network and the second convolutional neural network which are cascaded, and the data density is higher than the power feature vector V in terms of data volume of source data or semantic density of feature extraction p This results in a quasi-probabilistic deviation between the uniformity feature vector and the power feature vector. Therefore, in the technical solution of the present application, it is necessary to apply the uniformity characteristic vector V a Corrections based on data-intensive clusters are made.
More specifically, in step S180, the corrected uniformity feature vector and the power feature vector are passed through a classifier to obtain a classification result, where the classification result is used to indicate whether an air-out mode of an air-out device of the intelligent blow molding machine in a blow molding process is reasonable. That is, in the technical solution of the present application, further, the corrected uniformity feature vector and the power feature vector may be passed through a classifier to obtain a classification result that is used to indicate whether the air-out mode of the air-out device of the intelligent blow molding machine in the blow molding process is reasonable.
In summary, the control method of the intelligent blowing machine for producing the pesticide packaging barrel according to the embodiment of the application is clarified, and an artificial intelligent control technology is utilized to perform deep implicit feature mining on six views of the pesticide packaging barrel and power values of air outlet equipment of the intelligent blowing machine at a plurality of preset time points through a deep neural network model so as to mine out real-time internal forming conditions of the pesticide packaging barrel, so that air outlet mode rationality of the air outlet equipment of the intelligent blowing machine in a blowing forming process is judged in a classified mode, and further blowing control is performed on the blowing machine.
The basic principles of the present application have been described above in connection with specific embodiments, however, it should be noted that the advantages, benefits, effects, etc. mentioned in the present application are merely examples and not limiting, and these advantages, benefits, effects, etc. are not to be considered as necessarily possessed by the various embodiments of the present application. Furthermore, the specific details disclosed herein are for purposes of illustration and understanding only, and are not intended to be limiting, as the application is not intended to be limited to the details disclosed herein as such.
The block diagrams of the devices, apparatuses, devices, systems referred to in this application are only illustrative examples and are not intended to require or imply that the connections, arrangements, configurations must be made in the manner shown in the block diagrams. As will be appreciated by one of skill in the art, the devices, apparatuses, devices, systems may be connected, arranged, configured in any manner. Words such as "including," "comprising," "having," and the like are words of openness and mean "including but not limited to," and are used interchangeably therewith. The terms "or" and "as used herein refer to and are used interchangeably with the term" and/or "unless the context clearly indicates otherwise. The term "such as" as used herein refers to, and is used interchangeably with, the phrase "such as, but not limited to.
It is also noted that in the apparatus, devices and methods of the present application, the components or steps may be disassembled and/or assembled. Such decomposition and/or recombination should be considered as equivalent to the present application.
The previous description of the disclosed aspects is provided to enable any person skilled in the art to make or use the present application. Various modifications to these aspects will be readily apparent to those skilled in the art, and the generic principles defined herein may be applied to other aspects without departing from the scope of the application. Thus, the present application is not intended to be limited to the aspects shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.
The foregoing description has been presented for purposes of illustration and description. Furthermore, this description is not intended to limit the embodiments of the application to the form disclosed herein. Although a number of example aspects and embodiments have been discussed above, a person of ordinary skill in the art will recognize certain variations, modifications, alterations, additions, and subcombinations thereof.
Claims (8)
1. An intelligent blow molding machine for producing a packaging barrel for agricultural chemicals, comprising:
the device comprises an pesticide package barrel molding data acquisition module, a first image acquisition module, a second image acquisition module and a third image acquisition module, wherein the pesticide package barrel molding data acquisition module is used for acquiring first to sixth images of a blow-molded pesticide package barrel acquired by a camera, and the first to sixth images are six views of the pesticide package barrel;
The blowing process data acquisition module is used for acquiring power values of air outlet equipment of the intelligent blowing machine at a plurality of preset time points in the blowing molding process of the pesticide packaging barrel;
the molding data coding module is used for respectively passing the first image to the sixth image of the pesticide packaging barrel through a first convolutional neural network to obtain first feature images to sixth feature images;
the molding surface difference calculation module is used for calculating difference feature diagrams between every two feature diagrams in the first to sixth feature diagrams to obtain a plurality of difference feature diagrams, and arranging the difference feature diagrams according to sample dimensions to obtain uniformity feature diagrams;
the molding surface difference coding module is used for enabling the uniformity characteristic diagram to pass through a second convolution neural network serving as a filter so as to obtain uniformity characteristic vectors;
the blowing process data coding module is used for enabling power values of a plurality of preset time points of air outlet equipment of the intelligent blowing machine in the process of blowing the pesticide packaging barrel to pass through a time sequence coder comprising a one-dimensional convolution layer so as to obtain a power characteristic vector;
the molding surface difference characteristic distribution correction module is used for correcting the characteristic distribution of the uniformity characteristic vector based on the power characteristic vector to obtain a corrected uniformity characteristic vector; and
The control result generation module is used for enabling the corrected uniformity characteristic vector and the corrected power characteristic vector to pass through a classifier to obtain a classification result, wherein the classification result is used for indicating whether an air-out mode of air-out equipment of the intelligent blow molding machine in a blow molding process is reasonable or not;
wherein, the molding surface difference characteristic distribution correction module includes:
the vector space interaction unit is used for calculating the position-based point multiplication between the uniformity characteristic vector and the power characteristic vector to obtain a space interaction characteristic vector;
the data anisotropy evaluation unit is used for calculating the distance between the uniformity characteristic vector and the power characteristic vector;
the measurement unit is used for dividing the characteristic value of each position in the space interaction characteristic vector by the distance between the uniformity characteristic vector and the power characteristic vector to obtain a measurement characteristic vector;
an exponent operation unit for calculating a natural exponent function value with the eigenvalue of each position in the metric eigenvector as a power to obtain an exponent metric eigenvector;
the first class probability calculation unit is used for passing the index measurement feature vector through the classifier to obtain first class probability;
The second class probability calculation unit is used for passing the uniformity characteristic vector through the classifier to obtain second class probability;
a class probability superposition unit, configured to calculate a product between the first class probability and the second class probability to obtain a weighted weight; and
the weighting adjustment unit is used for weighting the characteristic values of each position in the uniformity characteristic vector by the weighting weight so as to obtain the corrected uniformity characteristic vector;
wherein the distance between the uniformity feature vector and the power feature vector is the square root of the Euclidean distance between the uniformity feature vector and the power feature vector.
2. The intelligent blow molding machine for producing an pesticide packaging barrel of claim 1, wherein the molding data encoding module is further configured to perform convolution processing, pooling processing, and activation processing, respectively, on input data in forward pass of layers using respective layers of the first convolutional neural network to generate the first through sixth feature maps, respectively, from a last layer of the first convolutional neural network, wherein inputs of a first layer of the first convolutional neural network are first through sixth images of the pesticide packaging barrel, respectively.
3. The intelligent blow molding machine for producing an agricultural package of claim 2, wherein the molding surface difference calculation module comprises:
a difference unit configured to calculate a difference by position between each two of the first to sixth feature maps to obtain the plurality of difference feature maps; and
and the arrangement unit is used for arranging the plurality of differential feature maps according to the sample dimension so as to obtain the uniformity feature map.
4. A smart blow-molding machine for producing pesticide packages according to claim 3, wherein the molding surface difference encoding module is further configured to use the layers of the second convolutional neural network to separately perform, in forward transfer of layers, input data:
carrying out convolution processing on input data to obtain a convolution characteristic diagram;
carrying out mean pooling based on a local feature matrix on the convolution feature map to obtain a pooled feature map; and
non-linear activation is carried out on the pooled feature map so as to obtain an activated feature map;
the output of the last layer of the second convolutional neural network is the uniformity characteristic vector, and the input of the first layer of the second convolutional neural network is the uniformity characteristic map.
5. The intelligent blow molding machine for producing an agricultural pharmaceutical packaging barrel of claim 4 wherein the blow molding process data encoding module comprises:
an input vector construction unit, configured to arrange power values of a plurality of predetermined time points of an air outlet device of the intelligent blow molding machine in a process of blow molding the pesticide packaging barrel into input vectors according to a time dimension;
full-connection coding unit for full-connection using the time sequence encoderThe connection layer carries out full-connection coding on the input vector by the following formula to extract high-dimensional implicit characteristics of characteristic values of all positions in the input vector, wherein the formula is as follows:, wherein />Is the input vector,/>Is the output vector, +.>Is a matrix of weights that are to be used,is a bias vector, ++>Representing a matrix multiplication; and a one-dimensional convolution encoding unit, configured to perform one-dimensional convolution encoding on the input vector by using a one-dimensional convolution layer of the timing encoder to extract high-dimensional implicit correlation features between feature values of each position in the input vector, where the formula is:
wherein , ais convolution kernel inxWidth in the direction,FIs a convolution kernel parameter vector, GFor a local vector matrix that operates with a convolution kernel,wis the size of the convolution kernel.
6. The intelligent blow molding machine for producing an agricultural chemical packaging barrel of claim 5, wherein the control result generation module comprises:
the fusion unit is used for calculating a transfer matrix of the corrected uniformity characteristic vector relative to the power characteristic vector, wherein the power characteristic vector multiplied by the transfer matrix is equal to the corrected uniformity characteristic vector; and
and the classification unit is used for taking the transfer matrix as a classification characteristic matrix to pass through the classifier so as to obtain the classification result.
7. A method of controlling an intelligent blow molding machine for producing a packaging barrel for an agricultural chemical according to claim 1, comprising:
acquiring first to sixth images of a blow-molded pesticide packaging barrel acquired by a camera, wherein the first to sixth images are six views of the pesticide packaging barrel;
acquiring power values of air outlet equipment of the intelligent blow molding machine at a plurality of preset time points in the process of blow molding the pesticide packaging barrel;
respectively passing the first to sixth images of the pesticide packaging barrel through a first convolutional neural network to obtain first to sixth feature images;
Calculating a difference feature map between every two feature maps in the first to sixth feature maps to obtain a plurality of difference feature maps, and arranging the difference feature maps according to sample dimensions to obtain a uniformity feature map;
the uniformity characteristic diagram is passed through a second convolution neural network serving as a filter to obtain uniformity characteristic vectors;
the power values of a plurality of preset time points of the air outlet equipment of the intelligent blowing machine in the process of blowing the pesticide packaging barrel are processed through a time sequence encoder comprising a one-dimensional convolution layer to obtain power characteristic vectors;
correcting the characteristic distribution of the uniformity characteristic vector based on the power characteristic vector to obtain a corrected uniformity characteristic vector; and
and passing the corrected uniformity characteristic vector and the power characteristic vector through a classifier to obtain a classification result, wherein the classification result is used for indicating whether an air outlet mode of air outlet equipment of the intelligent blow molding machine in a blow molding process is reasonable or not.
8. The control method for an intelligent blow molding machine for producing an agricultural package of claim 7, wherein passing the first through sixth images of the agricultural package through a first convolutional neural network, respectively, to obtain first through sixth feature maps, comprises:
And respectively carrying out convolution processing, pooling processing and activation processing on input data in forward transmission of layers by using each layer of the first convolutional neural network to respectively generate the first to sixth characteristic diagrams by the last layer of the first convolutional neural network, wherein the input of the first layer of the first convolutional neural network is respectively a first to sixth image of the pesticide packaging barrel.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210780403.7A CN115091725B (en) | 2022-07-04 | 2022-07-04 | Intelligent blow molding machine for producing pesticide packing barrel and control method thereof |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210780403.7A CN115091725B (en) | 2022-07-04 | 2022-07-04 | Intelligent blow molding machine for producing pesticide packing barrel and control method thereof |
Publications (2)
Publication Number | Publication Date |
---|---|
CN115091725A CN115091725A (en) | 2022-09-23 |
CN115091725B true CN115091725B (en) | 2023-05-05 |
Family
ID=83295866
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202210780403.7A Active CN115091725B (en) | 2022-07-04 | 2022-07-04 | Intelligent blow molding machine for producing pesticide packing barrel and control method thereof |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN115091725B (en) |
Families Citing this family (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN117011264A (en) * | 2023-08-03 | 2023-11-07 | 杭州高腾机电科技有限公司 | Underwear machine cluster production monitoring system and control method |
Family Cites Families (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101249721B (en) * | 2008-03-13 | 2010-09-08 | 华南理工大学 | Extrusion blow moulding wall-thickness on-line control system based on image recognition technology and method thereof |
DE102014010861A1 (en) * | 2014-07-25 | 2016-01-28 | Khs Corpoplast Gmbh | Volume controlled blast air supply |
US11836971B2 (en) * | 2018-08-23 | 2023-12-05 | Samsung Electronics Co., Ltd. | Method and device with convolution neural network processing |
DE102019203062A1 (en) * | 2019-03-06 | 2020-09-10 | Krones Ag | Method for product guidance in a stretch blow molding and / or filling system and stretch blow molding and / or filling system for bottles |
WO2021008328A1 (en) * | 2019-07-12 | 2021-01-21 | 腾讯科技(深圳)有限公司 | Image processing method and device, terminal, and storage medium |
CN114092972A (en) * | 2021-11-30 | 2022-02-25 | 深圳市辰星瑞腾科技有限公司 | Intelligent peep-proof electronic equipment display method and system and electronic equipment |
CN114475350A (en) * | 2022-02-14 | 2022-05-13 | 杭州鸽然科技有限公司 | Intelligent charging system and working method thereof |
-
2022
- 2022-07-04 CN CN202210780403.7A patent/CN115091725B/en active Active
Also Published As
Publication number | Publication date |
---|---|
CN115091725A (en) | 2022-09-23 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN105719188B (en) | The anti-method cheated of settlement of insurance claim and server are realized based on plurality of pictures uniformity | |
CN115586755B (en) | Production management control system and method for lithium hexafluorophosphate preparation | |
CN115239515B (en) | Precise intelligent processing and manufacturing system for mechanical parts and manufacturing method thereof | |
CN113313156A (en) | Internet of things equipment identification method and system based on time sequence load flow fingerprints | |
CN115013298B (en) | Real-time performance online monitoring system and monitoring method of sewage pump | |
CN115091725B (en) | Intelligent blow molding machine for producing pesticide packing barrel and control method thereof | |
CN105809672A (en) | Super pixels and structure constraint based image's multiple targets synchronous segmentation method | |
CN108537366B (en) | Reservoir scheduling method based on optimal convolution bidimensionalization | |
CN115100574A (en) | Action identification method and system based on fusion graph convolution network and Transformer network | |
CN111768000A (en) | Industrial process data modeling method for online adaptive fine-tuning deep learning | |
CN108763793A (en) | A kind of Weighted Fuzzy type D-S evidence theory frame | |
CN106709943A (en) | Point cloud registration method based on optimal transmission | |
CN107871103A (en) | Face authentication method and device | |
CN111931307B (en) | Part quality prediction and evaluation optimization method and system | |
CN112149896A (en) | Attention mechanism-based mechanical equipment multi-working-condition fault prediction method | |
CN116720438A (en) | Digital twin mine production system based on multi-source mine data and construction method | |
CN116151270A (en) | Parking test system and method | |
CN117407781B (en) | Equipment fault diagnosis method and device based on federal learning | |
CN114239384A (en) | Rolling bearing fault diagnosis method based on nonlinear measurement prototype network | |
CN116894180B (en) | Product manufacturing quality prediction method based on different composition attention network | |
CN116385365A (en) | Coal flow monitoring system and method based on scanning image | |
Galan-Marin et al. | Improving neural networks for mechanism kinematic chain isomorphism identification | |
US20230206063A1 (en) | Method for generating a trained convolutional neural network including an invariant integration layer for classifying objects | |
CN108734213A (en) | A kind of batch process modal identification method based on bayesian statistical analysis | |
CN114565051A (en) | Test method of product classification model based on neuron influence degree |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |