CN114994289B - Spinning product quality detection method and system - Google Patents

Spinning product quality detection method and system Download PDF

Info

Publication number
CN114994289B
CN114994289B CN202210916175.1A CN202210916175A CN114994289B CN 114994289 B CN114994289 B CN 114994289B CN 202210916175 A CN202210916175 A CN 202210916175A CN 114994289 B CN114994289 B CN 114994289B
Authority
CN
China
Prior art keywords
quality
spinning product
historical
quality detection
spinning
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
Application number
CN202210916175.1A
Other languages
Chinese (zh)
Other versions
CN114994289A (en
Inventor
曲柯宇
梁泽凯
李龙飞
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Jiangsu Zhuopeng Intelligent Mechanical And Electrical Co ltd
Original Assignee
Jiangsu Zhuopeng Intelligent Mechanical And Electrical Co ltd
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Jiangsu Zhuopeng Intelligent Mechanical And Electrical Co ltd filed Critical Jiangsu Zhuopeng Intelligent Mechanical And Electrical Co ltd
Priority to CN202210916175.1A priority Critical patent/CN114994289B/en
Publication of CN114994289A publication Critical patent/CN114994289A/en
Application granted granted Critical
Publication of CN114994289B publication Critical patent/CN114994289B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
    • G01N33/00—Investigating or analysing materials by specific methods not covered by groups G01N1/00 - G01N31/00
    • G01N33/36—Textiles
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
    • G06Q10/063—Operations research, analysis or management
    • G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
    • G06Q10/06395—Quality analysis or management
    • 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
    • Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
    • Y02P90/30—Computing systems specially adapted for manufacturing

Landscapes

  • Engineering & Computer Science (AREA)
  • Business, Economics & Management (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Human Resources & Organizations (AREA)
  • Health & Medical Sciences (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • General Physics & Mathematics (AREA)
  • General Health & Medical Sciences (AREA)
  • Biophysics (AREA)
  • Computational Linguistics (AREA)
  • Software Systems (AREA)
  • Mathematical Physics (AREA)
  • General Engineering & Computer Science (AREA)
  • Computing Systems (AREA)
  • Molecular Biology (AREA)
  • Development Economics (AREA)
  • Educational Administration (AREA)
  • Economics (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Evolutionary Computation (AREA)
  • Strategic Management (AREA)
  • Data Mining & Analysis (AREA)
  • Biomedical Technology (AREA)
  • Artificial Intelligence (AREA)
  • Chemical & Material Sciences (AREA)
  • Marketing (AREA)
  • General Business, Economics & Management (AREA)
  • Quality & Reliability (AREA)
  • Operations Research (AREA)
  • Textile Engineering (AREA)
  • Biochemistry (AREA)
  • Tourism & Hospitality (AREA)
  • Game Theory and Decision Science (AREA)
  • Pathology (AREA)
  • Medicinal Chemistry (AREA)
  • Food Science & Technology (AREA)
  • Immunology (AREA)
  • Analytical Chemistry (AREA)
  • General Factory Administration (AREA)

Abstract

The invention provides a spinning product quality detection method and a spinning product quality detection system, which belong to the technical field of spinning products, and comprise the following steps: acquiring multi-dimensional quality information of a first spinning product; constructing a spinning product quality analysis model; optimizing parameters of the spinning product quality analysis model to obtain a spinning product quality analysis model; inputting the multidimensional quality information into a spinning product quality analysis model to obtain a first quality prediction result and a second quality prediction result; according to the application scene of the first spinning product, carrying out weighted adjustment on the first quality prediction result and the second quality prediction result; and inputting the adjusted first quality prediction result and the adjusted second quality prediction result into a spinning product quality detection space to obtain a spinning product quality detection result. The invention solves the technical problems of low accuracy and low detection efficiency of the quality detection of the spinning product in the prior art, and achieves the technical effect of improving the accuracy and efficiency of the quality detection of the spinning product.

Description

Spinning product quality detection method and system
Technical Field
The invention relates to the technical field of spinning products, in particular to a spinning product quality detection method and system.
Background
The spinning product is made up by using various fibres as raw material through the processes of spinning and weaving, etc., and has the functions of insulating, covering and decorating. The quality of the spinning product is related to the comfort and the service life of the spinning product during the use process.
At present, the quality detection of spinning products is generally carried out by sampling detection, and whether the spinning products meet the corresponding quality requirements or not is judged by detecting main quality parameters of the spinning products and combining the operation standards with the subjective experience of technicians.
In the prior art, the manual participation degree of the quality detection of spinning products is high, and the technical problems of low quality detection accuracy, low detection efficiency and the like of the spinning products exist.
Disclosure of Invention
The application provides a method and a system for detecting quality of a spinning product, which are used for solving the technical problems of low accuracy and low detection efficiency of quality detection of the spinning product in the prior art.
In view of the above problems, the present application provides a method and a system for detecting quality of a spun yarn product.
In a first aspect of the present application, there is provided a spinning product quality detection method, the method comprising: acquiring and obtaining multi-dimensional quality information of a first spinning product to obtain a first quality information set and a second quality information set; constructing a spinning product quality analysis model; optimizing parameters of the spinning product quality analysis model to obtain the spinning product quality analysis model; inputting the first quality information set and the second quality information set into the spinning product quality analysis model to obtain a first quality prediction result and a second quality prediction result; according to the application scene of the first spinning product, carrying out weighting adjustment on the first quality prediction result and the second quality prediction result; and inputting the adjusted first quality prediction result and the adjusted second quality prediction result into a spinning product quality detection space to obtain a spinning product quality detection result.
In a second aspect of the application, there is provided a spinning product quality detection system, the system comprising: the first obtaining unit is used for acquiring and obtaining the multi-dimensional quality information of a first spinning product to obtain a first quality information set and a second quality information set; the first construction unit is used for constructing a spinning product quality analysis model; the first processing unit is used for optimizing parameters of the spinning product quality analysis model to obtain the spinning product quality analysis model; the second processing unit is used for inputting the first quality information set and the second quality information set into the spinning product quality analysis model to obtain a first quality prediction result and a second quality prediction result; the third processing unit is used for carrying out weighting adjustment on the first quality prediction result and the second quality prediction result according to the application scene of the first spinning product; and the fourth processing unit is used for inputting the adjusted first quality prediction result and the second quality prediction result into a spinning product quality detection space to obtain a spinning product quality detection result.
In a third aspect of the present application, there is provided an electronic device including: a processor coupled to a memory for storing a program that, when executed by the processor, causes an electronic device to perform the steps of the method according to the first aspect.
In a fourth aspect of the present application, a computer-readable storage medium is provided, on which a computer program is stored, which computer program, when being executed by a processor, carries out the steps of the method according to the first aspect.
One or more technical solutions provided in the present application have at least the following technical effects or advantages:
according to the technical scheme, the method comprises the steps of obtaining multi-dimensional quality information of a spinning product manufactured at present, obtaining a first quality information set and a second quality information set, then constructing a spinning product quality analysis model, optimizing parameters of the model by adopting an optimization method, obtaining the spinning product quality analysis model with high accuracy, further inputting the current first quality information set and the current second quality information set into the model, obtaining a first quality prediction result and a second quality prediction result, inputting the first quality prediction result and the second quality prediction result into a spinning product quality detection space after weighting and adjusting the first quality prediction result and the second quality prediction result based on an application scene of the current spinning product, and obtaining a final spinning product quality detection result. According to the method, the multi-dimensional quality information of the spinning product is acquired, the data dimension of the quality detection of the spinning product can be improved, the detection accuracy is improved, a spinning product quality analysis model is built, parameter optimization is carried out on the model, the accuracy and the convergence rate of model prediction can be improved, based on a first quality prediction result and a second quality prediction result output by the model, weighting adjustment is carried out by combining an application scene, the quality detection of the spinning product can be carried out more accurately based on the application scene, the universality is higher, the accuracy is higher, finally, a spinning product quality detection space is built, supervision adjustment is not needed in the building process, the adjusted first quality prediction result and second quality prediction result are input, the final spinning product quality detection result is obtained, the accuracy is higher, the quality detection method of the spinning product without manual participation evaluation judgment is built, the manual participation degree in the spinning product detection can be effectively reduced, the quantity of the spinning products for quality detection sampling can be reduced, the detection cost and the labor cost are reduced, and the technical effect of improving the accuracy and efficiency of the quality detection of the spinning product is achieved.
The foregoing description is only an overview of the technical solutions of the present application, and the present application can be implemented according to the content of the description in order to make the technical means of the present application more clearly understood, and the following detailed description of the present application is given in order to make the above and other objects, features, and advantages of the present application more clearly understandable.
Drawings
FIG. 1 is a schematic flow chart of a method for detecting quality of a spun yarn product provided by the present application;
FIG. 2 is a schematic flow chart of a spinning product quality analysis model constructed and obtained in the spinning product quality detection method provided by the application;
FIG. 3 is a schematic view of a flow chart of a spinning product quality detection result obtained in a spinning product quality detection method provided by the present application;
FIG. 4 provides a schematic structural view of a spinning product quality detection system according to the present application;
fig. 5 is a schematic structural diagram of an exemplary electronic device of the present application.
Description of reference numerals: the system comprises a first obtaining unit 11, a first constructing unit 12, a first processing unit 13, a second processing unit 14, a third processing unit 15, a fourth processing unit 16, an electronic device 300, a memory 301, a processor 302, a communication interface 303 and a bus architecture 304.
Detailed Description
The application provides a method and a system for detecting the quality of a spinning product, which are used for solving the technical problems of low accuracy and low detection efficiency of the quality detection of the spinning product in the prior art.
In view of the above technical problems, the technical solution provided by the present application has the following general idea:
the method comprises the steps of obtaining multi-dimensional quality information of a spinning product manufactured currently, obtaining a first quality information set and a second quality information set, then constructing a spinning product quality analysis model, optimizing parameters of the model by adopting an optimization method, obtaining the spinning product quality analysis model with high accuracy, further inputting the current first quality information set and the current second quality information set into the model, obtaining a first quality prediction result and a second quality prediction result, and inputting the first quality prediction result and the second quality prediction result into a spinning product quality detection space after weighting and adjusting the first quality prediction result and the second quality prediction result based on an application scene of the current spinning product, so as to obtain a final spinning product quality detection result.
Having described the basic principles of the present application, the technical solutions in the present application will be described clearly and completely with reference to the accompanying drawings, and it should be understood that the described embodiments are only a part of the embodiments of the present application, and not all of the embodiments of the present application, and the present application is not limited to the exemplary embodiments described herein. All other embodiments, which can be derived by a person skilled in the art from the embodiments of the present application without making any creative effort, shall fall within the protection scope of the present application. It should be further noted that, for the convenience of description, only some but not all of the elements relevant to the present application are shown in the drawings.
Example one
As shown in fig. 1, the present application provides a spinning product quality inspection method, including:
s100: acquiring multi-dimensional quality information of a first spinning product to obtain a first quality information set and a second quality information set;
in the embodiment of the present application, the first spun yarn product may be any product prepared based on a spinning process in the prior art, specifically, the first spun yarn product may be a finished product prepared by taking yarn as a raw material and performing spinning, weaving and finishing, and may also be a yarn product prepared by picking, carding, drawing, roving and spinning.
Illustratively, the first spun yarn product can be finished products such as underwear, gloves, bedding, handkerchiefs and the like, and can also be yarn products such as pure spun yarns and blended yarns.
Specifically, the multidimensional quality information of the first spun yarn product is acquired and acquired, and according to different classification rules, the multidimensional quality information can comprise parameter quality information and process quality information of the first spun yarn product, and can also comprise quality information and strength quality information of the first spun yarn product.
Illustratively, the parameter-based quality information may include quality information of the first spun yarn product, which may be obtained by detecting a micronaire value, a fiber length, a linear density, a trash content, a lint content, etc., and which has specific parameter values, and the process quality information may include parameters of whether a certain process is performed, such as whether combing is performed or not combing is performed.
Illustratively, the quality-type quality information may include quality information of the first spun product, such as color, hairiness, impurity content, foreign fiber content, and the like, which affect the quality of comfort in use of the product, and the strength-type quality information may include quality information of the first spun product, such as fiber length, breaking strength, micronaire value, linear density, slub, and cross-sectional morphology of the fibers, which affect the service life of the spun product.
Specifically, various kinds of quality information of the spun yarn products can be obtained by detecting the quality detection technology of the spun yarn products in the prior art, such as the image detection technology and the like.
Further, based on the collected multidimensional quality information of the first spinning product, the first spinning product is divided according to different classification rules, and a first quality information set and a second quality information set are obtained.
In the embodiment of the present application, preferably, the multidimensional quality information includes quality class quality information and intensity class quality information. Further, the first quality information set includes quality-type quality information, and the second quality information set includes intensity-type quality information.
S200: constructing a spinning product quality analysis model;
in the embodiment of the application, a spinning product quality analysis model is constructed based on an Artificial Neural Network (ANN) in machine learning, and an input layer, a hidden processing layer and an output layer of the model are specifically constructed, wherein the hidden processing layer comprises a plurality of simple units simulating neurons of human brains and is used for carrying out nonlinear logic operation according to multi-dimensional quality information of a spinning product and predicting the quality detection result of the spinning product, so that the quality detection of the spinning product without manual work is realized, and the detection accuracy and the detection efficiency are higher.
Specifically, the setting and construction of the network structure in the spinning product quality analysis model can be performed based on the dimension type number of the quality information of the first spinning product in the first quality information set and the second quality information set. Illustratively, according to the number of types of quality information, the number of types of model input parameters is confirmed, and then the network structure of the model is set, for example, the number of simple units is set.
S300: optimizing parameters of the spinning product quality analysis model to obtain the spinning product quality analysis model;
in the prior art, after a model is constructed based on an artificial neural network, the model is generally supervised and trained by acquiring training data and setting supervision information, and a nonlinear logic coupling relation between input parameters and output parameters of the model is found in a training process based on gradient descent, so as to form model parameters such as weights, thresholds and the like connected between simple units in the model, so that an output result of the model is converged or reaches a preset accuracy rate.
In the embodiment of the application, the model is not trained in a supervision training mode, but parameters in the model are optimized in an optimization-searching optimization mode, so that model parameters which can enable the output result of the model to be converged or reach the preset accuracy rate are obtained through an optimization searching method, and the convergence efficiency and the prediction accuracy rate of the model are improved.
For example, the optimization method can adopt a genetic algorithm and other methods in the prior art to optimize the parameters such as the weight, the threshold value and the like in the spinning product quality analysis model so as to obtain the spinning product quality analysis model which can lead the output result of the model to be converged.
S400: inputting the first quality information set and the second quality information set into the spinning product quality analysis model to obtain a first quality prediction result and a second quality prediction result;
preferably, in this embodiment of the application, the spinning product quality analysis model includes two network layers, and the two network layers may respectively use the first quality information set and the second quality information set as input data to perform analysis and prediction on different types of spinning product quality detection results to obtain a first quality prediction result and a second quality prediction result.
Specifically, the first quality prediction result and the second quality prediction result respectively include a quality detection result of the first spun yarn product obtained by preliminary prediction in two aspects as a data basis for performing subsequent quality detection. Preferably, the first quality prediction comprises a preliminary quality detection result in terms of use quality and the second quality prediction comprises a preliminary quality detection result in terms of use intensity.
Optionally, in the process of optimizing the model parameters, parameters in two network layers in the spinning product quality analysis model may be optimized respectively, so as to obtain network layer parameters that converge output results of the two network layers, and perform corresponding analysis and prediction of the detection result.
S500: according to the application scene of the first spinning product, carrying out weighting adjustment on the first quality prediction result and the second quality prediction result;
in the embodiment of the application, different application scenarios are provided for different types of first spinning products, for example, for the first spinning products as clothes and handkerchiefs, the requirements on the use quality such as comfort level are high. On the other hand, the first spun yarn used as a tablecloth is required to have a high demand for use quality such as strength and a long service life.
Therefore, after the first quality prediction result and the second quality prediction result are obtained through the spinning product quality analysis model prediction, the importance of the first quality prediction result and the second quality prediction result is assigned by weight based on the application scenario of the first spinning product, the corresponding weight assignment result is obtained, and the first quality prediction result and the second quality prediction result are adjusted by weight based on the weight assignment result, so that the first quality prediction result and the second quality prediction result of the first spinning product which are more suitable for the application scenario are obtained.
For example, the method of weight assignment may adopt any weight assignment method in the prior art, for example, methods such as AHP hierarchy analysis, expert weighting, and the like may be used.
After the weight distribution result is obtained through weight distribution, the weight values of the first quality prediction result and the second quality prediction result are included in the weight distribution result, and in an application scene of a first spinning product, if the importance of the quality prediction result of one aspect is larger, the corresponding weight value is larger.
The first quality prediction result and the second quality prediction result are weighted and adjusted after being recorded by adopting the weight distribution, so that the first quality prediction result and the second quality prediction result after weighted adjustment are obtained, the quality detection requirement of a first spinning product is more met, the accuracy is higher, and the judgment and analysis by technicians are not needed to be carried out by virtue of subjective experience.
S600: and inputting the adjusted first quality prediction result and the adjusted second quality prediction result into a spinning product quality detection space to obtain a spinning product quality detection result.
Specifically, in the embodiment of the present application, a spinning product quality detection space is constructed based on a large number of first quality detection results and second quality detection result data of a first spinning product obtained by previous detection. And after the adjusted first quality prediction result and the second quality prediction result are obtained, inputting the first quality prediction result and the second quality prediction result into the spinning product quality detection space, so that the corresponding spinning product quality detection result can be obtained, and the quality detection of the first spinning product is completed.
Preferably, the first quality detection result and the second quality detection result of the first spinning product in the history are input into the coordinate space by combining the visual coordinate space to form a plurality of coordinate points, and the plurality of coordinate points are clustered based on the clustering method to obtain a plurality of clustering results, wherein the plurality of clustering results can respectively correspond to quality detection results with high quality, qualified quality, unqualified quality and serious unqualified quality.
Further, according to the coordinate space, the current clustering result that the first quality prediction result and the second quality prediction result after adjustment are close to or belong to is judged, and the corresponding spinning product quality detection result is obtained, so that the visualization degree is high, the accuracy is high, and supervision adjustment is not needed.
According to the spinning product quality detection method, the multi-dimensional quality information of the spinning product is acquired through collection, the data dimension of the spinning product quality detection can be improved, the detection accuracy is improved, the spinning product quality analysis model is built and the parameter optimization is carried out on the model, the accuracy and the convergence rate of the model prediction can be improved, the first quality prediction result and the second quality prediction result output by the model are combined with the application scene for weighting adjustment, the quality detection of the spinning product can be carried out more accurately based on the application scene, the universality is higher, the accuracy is higher, the spinning product quality detection space is finally built, the construction process does not need supervision adjustment, the adjusted first quality prediction result and the adjusted second quality prediction result are input, the final spinning product quality detection result is obtained, the accuracy is higher, the spinning product quality detection method which does not need manual participation evaluation judgment is built, the manual participation degree in the spinning product detection can be effectively reduced, the number of the spinning products for quality detection can be reduced, the detection cost and the manual cost are reduced, and the technical effects of improving the accuracy and the spinning product quality detection efficiency are achieved.
Step S100 in the method provided in the embodiment of the present application includes:
s110: sampling and detecting the first textile product to obtain the multi-dimensional quality information;
s120: classifying the multi-dimensional quality information to obtain multi-dimensional quality information and multi-dimensional strength quality information;
s130: and taking the multi-dimensional quality information as the first quality information set, and taking the multi-dimensional strength quality information as the second quality information set.
Specifically, sampling detection is carried out on a first textile product which needs quality detection at present, and multi-dimensional quality information of the first textile product is detected.
The multidimensional quality information is classified to obtain multidimensional quality information and multidimensional intensity quality information which are respectively used as the first quality information set and the second quality information set.
According to the method provided by the embodiment of the application, the first spinning product is subjected to multi-dimensional quality information sampling detection, the first quality information set and the second quality information set are obtained in a classified mode, the quality of the spinning product can be detected based on different dimensions, and the accuracy of quality detection of the spinning product is improved.
As shown in fig. 2, step S200 in the method provided in the embodiment of the present application includes:
s210: constructing an input layer and an output layer of the spinning product quality analysis model based on an artificial neural network model;
s220: constructing a first implicit processing layer of the spinning product quality analysis model according to the first quality information set;
s230: according to the second quality information set, a second implicit processing layer of the spinning product quality analysis model is constructed, and the second implicit processing layer and the first implicit processing layer are arranged in parallel;
s240: and respectively connecting the input layer, the first hidden processing layer and the output layer, and the input layer, the second hidden processing layer and the output layer to obtain the spinning product quality analysis model.
Specifically, an input layer and an output layer of a spinning product quality analysis model are constructed based on an artificial neural network in machine learning and are respectively used for inputting parameters and outputting the output parameters.
And constructing a first implicit processing layer of the spinning product quality analysis model based on the first quality information set, wherein the first implicit processing layer is used for analyzing and predicting to obtain a first quality prediction result according to the first quality information set.
And constructing a second implicit processing layer of the spinning product quality analysis model based on the second quality information set, wherein the second implicit processing layer is used for analyzing and predicting according to the second quality information set to obtain a second quality prediction result.
And when the first quality information set and the second quality information set are input through the input layer, the first implicit processing layer and the second implicit processing layer can be respectively input into the first implicit processing layer and the second implicit processing layer through correspondingly connected channels for analysis and prediction.
And constructing the network structure in the network layer according to the quality information types in the first quality information set and the second quality information set respectively in the process of constructing the first implicit processing layer and the second implicit processing layer. For example, if the first quality information set includes 8 quality information, a corresponding network structure for analyzing and predicting 1 output parameter according to 8 input parameters may be correspondingly set.
In the process of constructing the spinning product quality analysis model, the two implicit processing layers are constructed based on the artificial neural network respectively and are used for performing different analysis and prediction according to different spinning product quality information respectively to obtain quality prediction results in two aspects, so that the data dimensionality of subsequent quality detection is improved, and the accuracy of quality detection is improved.
Step S300 in the method provided in the embodiment of the present application includes:
s310: acquiring a historical first quality information set and a historical second quality information set in the previous history of the first spinning product;
s320: acquiring a historical first quality detection result set and a historical second quality detection result set which are obtained by performing quality detection on the first spinning product in the previous history;
s330: respectively taking the historical first quality information set and the historical second quality information set as input parameters of the first implicit processing layer and the second implicit processing layer, and taking the historical first quality detection result set and the historical second quality detection result set as output parameters of the first implicit processing layer and the second implicit processing layer;
s340: optimizing and optimizing parameters in the first implicit processing layer and the second implicit processing layer respectively, wherein the parameters comprise weights and thresholds.
Specifically, a first quality information set and a second quality information set of a previous first spinning product in the quality detection process are collected and obtained and are respectively used as a historical first quality information set and a historical second quality information set.
And acquiring a first quality detection result and a second quality detection result which are obtained by detecting the previous first spinning product according to the historical first quality information set and the historical second quality information set in the quality detection process, and obtaining a historical first quality detection result set and a historical second quality detection result set.
Illustratively, historical quality information and historical quality detection results can be acquired according to quality detection log information collection in spinning product production units and obtained through classification.
And further, respectively taking the historical first quality information set and the historical second quality information set as input parameters of the first implicit processing layer and the second implicit processing layer, and taking the historical first quality detection result set and the historical second quality detection result set as output parameters of the first implicit processing layer and the second implicit processing layer.
In the prior art, corresponding training data is generally set according to input parameters and output parameters, and a neural network model is trained, so that the model obtains a complex logical relationship between the input parameters and the output parameters in a training process, and forms a weight and a threshold value for performing logical analysis processing by a simple unit, so that the model can predict the output parameters according to the input parameters to reach a convergence or preset accuracy.
In the embodiment of the application, parameters such as the weight and the threshold value for converging the model are not obtained through supervised training optimization, but the parameters such as the weight and the threshold value in the first implicit processing layer and the second implicit processing layer are obtained through an optimization algorithm, so that the model is prevented from falling into local optimum in the training process of gradient descent, and the convergence efficiency and the model accuracy are improved.
Taking the first implicit processing layer as an example, in the process of optimizing the parameters of the first implicit processing layer, step S340 includes:
s341: obtaining a first layer parameter scheme set according to the network structure of the first implicit processing layer;
s342: randomly selecting a parameter scheme from the first layer parameter scheme set as a first parameter scheme and as an optimization parameter scheme;
s343: under the first parameter scheme, the first implicit processing layer predicts the accuracy of the historical first quality detection result set according to the historical first quality information set to obtain a first accuracy;
s344: randomly selecting a parameter scheme from the first layer parameter scheme set again to serve as a second parameter scheme;
s345: under the second parameter scheme, the accuracy of the historical first quality detection result set is predicted by the first implicit processing layer according to the historical first quality information set, and a second accuracy is obtained;
s346: judging whether the second accuracy is greater than the first accuracy, if so, taking the second parameter scheme as the optimization parameter scheme, and if not, taking the second parameter scheme as the optimization parameter scheme according to a probability, wherein the probability is calculated by the following formula:
Figure 583553DEST_PATH_IMAGE001
wherein, the first and the second end of the pipe are connected with each other,
Figure 500694DEST_PATH_IMAGE002
in order to achieve the second accuracy, the first accuracy,
Figure 405065DEST_PATH_IMAGE003
k is an optimized speed factor for the first accuracy;
s347: continuing to randomly select a parameter scheme in the first layer parameter scheme set, and performing iterative optimization;
s348: when the optimization parameter scheme is not changed in the iteration of the preset times, outputting the parameter scheme corresponding to the optimization parameter scheme as an optimal parameter scheme;
s349: and adjusting parameters of the first implicit processing layer according to the parameters in the optimal parameter scheme to complete optimization.
Specifically, according to a network structure in a first hidden processing layer, random combination setting is carried out on weights and thresholds connected among simple units, a first layer parameter scheme set is obtained through combination, all possible combination schemes of a plurality of weights and a plurality of thresholds in the first hidden processing layer are included in the first layer parameter scheme set, the accuracy of the model is different under different schemes, and the purpose of supervision training is to find a parameter scheme capable of enabling the model to be converged in the training process.
In the embodiment of the present application, based on a specific optimization method, a parameter scheme is randomly selected and obtained from the first-layer parameter scheme set, and the parameter scheme is used as the first parameter scheme and is used as an optimization parameter scheme, that is, a currently optimal parameter scheme in an optimization process.
According to the first parameter scheme, parameter updating setting is carried out on the first implicit processing layer, the accuracy of the first implicit processing layer for predicting the first quality detection result set according to the first quality information set under the first parameter scheme is calculated, a first accuracy is obtained, the first accuracy is the accuracy of analysis and prediction carried out on the first implicit processing layer under the first parameter scheme, and the higher the accuracy is, the better the first parameter scheme is.
And randomly selecting a parameter scheme from the first layer parameter scheme set to be used as a second parameter scheme, and continuously calculating the accuracy of the first quality detection result set predicted by the first implicit processing layer according to the first quality information set under the second parameter scheme to obtain a second accuracy.
And comparing the first accuracy with the second accuracy, judging whether the second accuracy is greater than the first accuracy, if so, indicating that the second parameter scheme is superior to the first parameter scheme, and replacing the first parameter scheme with the second parameter scheme to serve as an optimization parameter scheme. And if the probability is less than the first probability, the second parameter scheme is inferior to the first parameter scheme, and at this time, the second parameter scheme is not directly abandoned, because the probability of the first parameter scheme is not the global optimum, the iterative optimization is prevented from being stopped at the first parameter scheme, and in order to improve the traversal optimization rate, the second parameter scheme replaces the first parameter scheme according to the probability to be used as the optimization parameter scheme.
The probability is calculated by:
Figure 862591DEST_PATH_IMAGE004
wherein, the first and the second end of the pipe are connected with each other,
Figure 739280DEST_PATH_IMAGE002
in order to achieve the second accuracy, the first accuracy,
Figure 561743DEST_PATH_IMAGE003
for the first accuracy, k is the optimized speed factor.
k is an optimization speed factor which gradually decreases with the number of optimization iterations, and optionally, can be a gradually decreasing constant. The decreasing mode of k can be any decreasing mode such as exponential decreasing mode or logarithmic decreasing mode. In the initial stage of the optimization iteration, k is larger so that P is larger, a second parameter scheme which is inferior can be accepted as the optimization parameter scheme with higher probability, the speed of the traversal iteration optimization is increased, and the optimization is prevented from being stopped at local optimum. And in the later stage of the optimizing iteration, k is smaller so that P is smaller, the probability of the optimizing parameter scheme in the later stage of optimizing is global optimal, the inferior parameter scheme is accepted as the optimizing parameter scheme with smaller probability, and the optimizing accuracy is improved.
Meanwhile, P is also related to the difference value between the second accuracy and the first accuracy, if the difference value is larger, the second parameter scheme is extremely inferior compared with the first parameter scheme, P is smaller, and the extremely inferior second parameter scheme is accepted as the optimization parameter scheme with smaller probability.
Therefore, based on the formula, optimization iteration is continuously carried out on the first-layer parameter scheme set to obtain a third parameter scheme, a fourth parameter scheme and the like, the accuracy is calculated, and judgment and analysis are carried out on the optimization parameter scheme.
In iterative optimization, if the optimization parameter scheme is not changed in the preset times of iterations, it indicates that the parameter scheme in the optimization parameter scheme is difficult to find a more optimal parameter scheme in the optimization, the k value is reduced, the P is also reduced, the optimization reaches the later stage, the probability of the optimization parameter scheme is globally optimal, and the optimal parameter scheme is output to obtain the optimal parameter scheme. Illustratively, the preset number of times may be set according to the number of parameter schemes in the first layer parameter scheme set, and may be 10 times, for example.
Based on parameters such as weight and threshold in the optimal parameter scheme, parameter adjustment is performed on the first hidden processing layer, parameter optimization of the current first hidden processing layer is completed, supervision training is not needed, and in a small data sample in the embodiment of the application, convergence efficiency is high and accuracy is high.
Further, the parameter optimization of the second hidden processing layer may be the same as the parameter optimization method of the first hidden processing layer, but the input parameters and the output parameters are different, and other optimization algorithms such as a genetic algorithm, a particle swarm optimization algorithm and the like may also be used for the optimization, which is not described herein again.
And obtaining a spinning product quality analysis model based on the first hidden processing layer and the second hidden processing layer which are optimized and completed by parameters. And inputting the first quality information set and the second quality information set of the current first spinning product into the model, and respectively performing analysis and prediction on the first hidden processing layer and the second hidden processing layer to obtain a first quality prediction result and a second quality prediction result output by the model, wherein the first quality prediction result and the second quality prediction result are used as quality detection results of two aspects of the current first spinning product.
According to the method and the device, the weight and the threshold parameter in the hidden processing layer in the model are optimized, the convergence rate and the accuracy of the model can be improved, the method and the device are suitable for the data dimensionality of the spinning product quality detection, and the quality detection result of the spinning product can be predicted more accurately.
As shown in fig. 3, step S600 in the method provided in the embodiment of the present application includes:
s610: acquiring a historical first quality detection result set and a historical second quality detection result set which are obtained by performing quality detection on the first spinning product in the previous history;
s620: acquiring a historical spinning product quality detection result obtained by performing quality detection on the first spinning product in the previous history;
s630: constructing a two-dimensional coordinate system based on the historical first quality detection result set and the historical second quality detection result set;
s640: inputting the historical first quality detection result set and the historical second quality detection result set into the two-dimensional coordinate system to obtain a plurality of coordinate points;
s650: clustering the coordinate points according to the quality detection result of the historical spinning product to obtain a plurality of clustering results;
s660: taking a plurality of clustering results and the two-dimensional coordinate system as a spinning product quality detection space;
s670: and inputting the first quality prediction result and the second quality prediction result into the spinning product quality detection space to obtain the spinning product quality detection result.
Specifically, a historical first quality detection result set and a historical second quality detection result set obtained by performing quality detection on a first spinning product before acquisition are collected. Optionally, in the collecting process, the historical first quality detection result set and the historical second quality detection result set are respectively subjected to weighting adjustment according to an application scenario of the first spinning product, so as to obtain the historical first quality detection result set and the historical second quality detection result set after weighting adjustment.
Further, a total quality detection result of the quality detection of the first spinning product in the previous history is collected, wherein the total quality detection result is a spinning product comprehensive quality detection result obtained by integrating the historical first quality detection result and the historical second quality detection result, and a historical spinning product quality detection result is obtained. Illustratively, the historical spinning product quality detection results comprise four results of high quality, qualified quality, unqualified quality and serious unqualified quality, and each result corresponds to a value range of a historical first quality detection result and a historical second quality detection result.
And respectively constructing an abscissa axis and an ordinate axis of a two-dimensional coordinate system based on the historical first quality detection result set and the historical second quality detection result set to obtain the two-dimensional coordinate system. And inputting the first quality detection result set and the second quality detection result set into the two-dimensional coordinate system, so that corresponding abscissa and ordinate values can be obtained according to the first quality detection result data and the second quality detection result data of each specific spinning product, and a plurality of coordinate points are obtained.
And clustering a plurality of coordinate points based on the ranges of the historical first quality detection results and the historical second quality detection results corresponding to different detection results in the historical spinning product quality detection results to obtain a plurality of clustering results, wherein the coordinate points in each clustering result correspond to one historical spinning product quality detection result, and can be qualified, unqualified and the like.
And taking the plurality of clustering results and the two-dimensional coordinate system as the spinning product quality detection space. And inputting a first quality prediction result and a second quality prediction result obtained by model prediction of the current first spinning product into the spinning product quality detection space based on the spinning product quality detection space, so that the quality detection result of the current first spinning product can be obtained.
Step S670 in the method provided in the embodiment of the present application includes:
s671: inputting the first quality prediction result and the second quality prediction result into the spinning product quality detection space to obtain a current quality detection coordinate point;
s672: calculating Euclidean distances between the current quality detection coordinate point and the centers of the clustering results to obtain a Euclidean distance set;
s673: according to the size of the Euclidean distances in the Euclidean distance set, carrying out weight distribution to obtain a weight distribution result;
s675: adopting the weight distribution result to carry out weighting adjustment on the Euclidean distances;
s676: and obtaining the adjusted minimum Euclidean distance, and taking the corresponding clustering result as the quality detection result of the spinning product.
Specifically, the current first quality prediction result and the current second quality prediction result are input into the spinning product quality detection space, corresponding abscissa and ordinate values are obtained, and then a corresponding current quality detection coordinate point is obtained.
And calculating Euclidean distances between the quality detection coordinate point and the centers of the plurality of clustering results to obtain an Euclidean distance set. Further, according to the size of each Euclidean distance in the Euclidean distance set, weight distribution is carried out, and a weight distribution result is obtained.
And performing weighting adjustment on the Euclidean distances in the Euclidean distance set by adopting the weight distribution result to obtain a plurality of adjusted Euclidean distances.
And selecting the minimum Euclidean distance in the adjusted plurality of Euclidean distances, and obtaining a spinning product quality detection result corresponding to the clustering result corresponding to the Euclidean distance to serve as the quality detection result of the current first spinning product.
The spinning product quality detection space is constructed based on the idea of the KNN algorithm, the coordinate points are formed according to the quality detection results of the strength and the quality, clustering is carried out according to the comprehensive quality detection results, and the final comprehensive quality detection result can be accurately and efficiently obtained according to the quality detection results of the current spinning product in two aspects.
To sum up, the embodiment of the application acquires the multi-dimensional quality information of the spinning product through collection, the data dimension of the quality detection of the spinning product can be improved, the detection accuracy is improved, the spinning product quality analysis model is built and the parameter optimization is carried out on the model, the accuracy and the convergence speed of the model prediction can be improved, the first quality prediction result and the second quality prediction result output by the model are combined with the application scene for weighting adjustment, the quality detection of the spinning product can be carried out more accurately based on the application scene, the universality is higher, the spinning product quality detection space is finally built, the construction process does not need supervision adjustment, the adjusted first quality prediction result and the adjusted second quality prediction result are input to obtain the final spinning product quality detection result, the accuracy is higher, the spinning product quality detection method without manual participation evaluation judgment is built, the manual participation degree in the spinning product detection can be effectively reduced, the number of the spinning products sampled by quality detection can be reduced, the detection cost and the manual cost are reduced, and the technical effects of improving the accuracy and the quality detection efficiency of the spinning product are achieved.
Example two
Based on the same inventive concept as the spinning product quality detection method in the previous embodiment, as shown in fig. 4, the present application provides a spinning product quality detection system, wherein the system comprises:
the first obtaining unit 11 is used for acquiring and obtaining multi-dimensional quality information of a first spinning product, and obtaining a first quality information set and a second quality information set;
a first construction unit 12 for constructing a spinning product quality analysis model;
a first processing unit 13, configured to optimize parameters of the spinning product quality analysis model to obtain the spinning product quality analysis model;
a second processing unit 14, configured to input the first quality information set and the second quality information set into the spinning product quality analysis model, so as to obtain a first quality prediction result and a second quality prediction result;
a third processing unit 15, configured to perform weighting adjustment on the first quality prediction result and the second quality prediction result according to an application scenario of the first spun yarn product;
and a fourth processing unit 16, configured to input the adjusted first quality prediction result and the adjusted second quality prediction result into a spinning product quality detection space, so as to obtain a spinning product quality detection result.
Further, the system further comprises:
the second obtaining unit is used for sampling and detecting the first textile product to obtain the multi-dimensional quality information;
a third obtaining unit, configured to classify the multidimensional quality information to obtain multidimensional quality information and multidimensional strength quality information;
a fourth obtaining unit, configured to use the multidimensional quality information as the first quality information set, and use the multidimensional strength quality information as the second quality information set.
Further, the system further comprises:
the second construction unit is used for constructing an input layer and an output layer of the spinning product quality analysis model based on an artificial neural network model;
the third construction unit is used for constructing a first implicit processing layer of the spinning product quality analysis model according to the first quality information set;
a fourth construction unit, configured to construct a second implicit processing layer of the spinning product quality analysis model according to the second quality information set, where the second implicit processing layer is arranged in parallel with the first implicit processing layer;
and the fifth construction unit is used for respectively connecting the input layer, the first hidden processing layer and the output layer, and the input layer, the second hidden processing layer and the output layer to obtain the spinning product quality analysis model.
Further, the system further comprises:
a fifth obtaining unit, which is used for acquiring a historical first quality information set and a historical second quality information set in the previous history of the first spinning product;
a sixth obtaining unit, configured to acquire a historical first quality detection result set and a historical second quality detection result set obtained by performing quality detection on the first spun yarn product in a previous history;
a fifth processing unit, configured to use the historical first quality information set and the historical second quality information set as input parameters of the first implicit processing layer and the second implicit processing layer, and use the historical first quality detection result set and the historical second quality detection result set as output parameters of the first implicit processing layer and the second implicit processing layer, respectively;
and a sixth processing unit, configured to perform optimization on parameters in the first implicit processing layer and the second implicit processing layer, where the parameters include a weight and a threshold.
Further, the system further comprises:
a seventh obtaining unit, configured to obtain a first layer parameter scheme set according to the network structure of the first implicit processing layer;
an eighth obtaining unit, configured to randomly select a parameter scheme from the first-layer parameter scheme set as the first parameter scheme and as an optimization parameter scheme;
a seventh processing unit, configured to calculate an accuracy of the historical first quality detection result set according to the historical first quality information set by the first implicit processing layer under the first parameter scheme, so as to obtain a first accuracy;
a ninth obtaining unit, configured to randomly select a parameter scheme again from the first layer parameter scheme set as a second parameter scheme;
the eighth processing unit is configured to calculate, under the second parameter scheme, an accuracy of the historical first quality detection result set, which is predicted by the first implicit processing layer according to the historical first quality information set, and obtain a second accuracy;
a first determining unit, configured to determine whether the second accuracy is greater than a first accuracy, if so, use the second parameter solution as the optimization parameter solution, and if not, use the second parameter solution as the optimization parameter solution according to a probability, where the probability is calculated by the following equation:
Figure 891093DEST_PATH_IMAGE001
wherein, the first and the second end of the pipe are connected with each other,
Figure 89993DEST_PATH_IMAGE002
in order to achieve the second accuracy, the first accuracy,
Figure 758872DEST_PATH_IMAGE003
k is an optimized speed factor for the first accuracy;
a ninth processing unit, configured to continue to perform random selection of parameter schemes in the first-layer parameter scheme set, and perform iterative optimization;
a tenth processing unit, configured to output a parameter scheme corresponding to the optimization parameter scheme as an optimal parameter scheme when the optimization parameter scheme does not change in a preset number of iterations;
and the eleventh processing unit is used for adjusting the parameters of the first implicit processing layer according to the parameters in the optimal parameter scheme to complete optimization.
Further, the system further comprises:
a tenth obtaining unit, configured to acquire a historical first quality detection result set and a historical second quality detection result set obtained by performing quality detection on the first spun yarn product in a previous history;
an eleventh obtaining unit, configured to acquire a quality detection result of a historical spinning product obtained by performing quality detection in a previous history of the first spinning product;
a sixth construction unit, configured to construct a two-dimensional coordinate system based on the historical first quality detection result set and the historical second quality detection result set;
a twelfth obtaining unit, configured to input the historical first quality detection result set and the historical second quality detection result set into the two-dimensional coordinate system, and obtain a plurality of coordinate points;
a twelfth processing unit, configured to cluster the multiple coordinate points according to the detection result of the quality of the historical spinning product, so as to obtain multiple clustering results;
a thirteenth obtaining unit configured to use the plurality of clustering results and the two-dimensional coordinate system as the spinning product quality detection space;
a fourteenth obtaining unit configured to input the first quality prediction result and the second quality prediction result into the spinning product quality detection space, and obtain the spinning product quality detection result.
Further, the system further comprises:
a fifteenth obtaining unit, configured to input the first quality prediction result and the second quality prediction result into the spinning product quality detection space, and obtain a current quality detection coordinate point;
a thirteenth processing unit, configured to calculate euclidean distances between the current quality detection coordinate point and centers of the multiple clustering results, to obtain a set of euclidean distances;
a fourteenth processing unit, configured to perform weight distribution according to the magnitudes of the euclidean distances in the euclidean distance set, to obtain a weight distribution result;
a fifteenth processing unit, configured to perform weighting adjustment on the magnitudes of the euclidean distances by using the weight distribution result;
and a sixteenth obtaining unit, configured to obtain the adjusted minimum euclidean distance, and use the corresponding clustering result as the spinning product quality detection result.
EXAMPLE III
Based on the same inventive concept as one of the spinning product quality detection methods in the foregoing embodiments, the present application also provides a computer-readable storage medium having stored thereon a computer program which, when executed by a processor, implements the method as in the first embodiment.
Exemplary electronic device
The electronic device of the present application is described below with reference to fig. 5.
Based on the same inventive concept as the method for detecting the quality of the spun yarn product in the previous embodiment, the present application also provides an electronic device, comprising: a processor coupled to a memory, the memory storing a program that, when executed by the processor, causes the electronic device to perform the steps of the method of embodiment one.
The electronic device 300 includes: processor 302, communication interface 303, memory 301. Optionally, the electronic device 300 may also include a bus architecture 304. Wherein, the communication interface 303, the processor 302 and the memory 301 may be connected to each other through a bus architecture 304; the bus architecture 304 may be a Peripheral Component Interconnect (PCI) bus, an Extended Industry Standard Architecture (EISA) bus, or the like. The bus architecture 304 may be divided into an address bus, a data bus, a control bus, and the like. For ease of illustration, only one thick line is shown in FIG. 5, but this is not intended to represent only one bus or type of bus.
Processor 302 may be a CPU, microprocessor, ASIC, or one or more integrated circuits for controlling the execution of programs in accordance with the teachings of the present application.
The communication interface 303 may be any device, such as a transceiver, for communicating with other devices or communication networks, such as an ethernet, a Radio Access Network (RAN), a Wireless Local Area Network (WLAN), a wired access network, and the like.
The memory 301 may be, but is not limited to, ROM or other type of static storage device that can store static information and instructions, RAM or other type of dynamic storage device that can store information and instructions, EEPROM, CD-ROM or other optical disk storage, optical disk storage (including compact disk, laser disk, optical disk, digital versatile disk, blu-ray disk, etc.), magnetic disk storage or other magnetic storage devices, or any other medium that can be used to carry or store desired program code in the form of instructions or data structures and that can be accessed by a computer. The memory may be self-contained and coupled to the processor through a bus architecture 304. The memory may also be integral to the processor.
The memory 301 is used for storing computer-executable instructions for implementing the present application, and is controlled by the processor 302 to execute. The processor 302 is used for executing computer-executable instructions stored in the memory 301, so as to realize the spinning product quality detection method provided by the above embodiment of the application.
In the above embodiments, the implementation may be wholly or partially realized by software, hardware, firmware, or any combination thereof. When implemented in software, it may be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the procedures or functions described in accordance with the present application are generated, in whole or in part. The computer may be a general purpose computer, a special purpose computer, a network of computers, or other programmable device. The computer instructions may be stored in a computer readable storage medium or transmitted from one computer readable storage medium to another, for example, from one website site, computer, server, or data center to another website site, computer, server, or data center via wired (e.g., coaxial cable, fiber optic, digital Subscriber Line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device including one or more available media integrated servers, data centers, and the like. The usable medium may be a magnetic medium (e.g., a floppy Disk, a hard Disk, a magnetic tape), an optical medium (e.g., a DVD), or a semiconductor medium (e.g., a Solid State Disk (SSD)), among others.
Although the present application has been described in conjunction with specific features and embodiments thereof, it will be evident that various modifications and combinations can be made thereto without departing from the spirit and scope of the application. Accordingly, the specification and figures are merely exemplary of the application and are intended to cover any and all modifications, variations, combinations, or equivalents within the scope of the application. It will be apparent to those skilled in the art that various changes and modifications may be made in the present application without departing from the scope of the application. Thus, if such modifications and variations of the present application fall within the scope of the present application and its equivalent technology, the present application is intended to include such modifications and variations.

Claims (7)

1. A method for detecting quality of a spun yarn product, the method comprising:
acquiring and obtaining multi-dimensional quality information of a first spinning product to obtain a first quality information set and a second quality information set;
constructing a spinning product quality analysis model;
optimizing parameters of the spinning product quality analysis model to obtain the spinning product quality analysis model;
inputting the first quality information set and the second quality information set into the spinning product quality analysis model to obtain a first quality prediction result and a second quality prediction result;
according to the application scene of the first spinning product, carrying out weighting adjustment on the first quality prediction result and the second quality prediction result;
inputting the adjusted first quality prediction result and the adjusted second quality prediction result into a spinning product quality detection space to obtain a spinning product quality detection result;
wherein, the step of constructing a spinning product quality analysis model comprises the following steps:
constructing an input layer and an output layer of the spinning product quality analysis model based on an artificial neural network model;
constructing a first implicit processing layer of the spinning product quality analysis model according to the first quality information set;
according to the second quality information set, a second implicit processing layer of the spinning product quality analysis model is constructed, and the second implicit processing layer and the first implicit processing layer are arranged in parallel;
respectively connecting the input layer, the first hidden processing layer and the output layer, and the input layer, the second hidden processing layer and the output layer to obtain a spinning product quality analysis model;
the optimization of the parameters of the spinning product quality analysis model comprises the following steps:
acquiring a historical first quality information set and a historical second quality information set in the previous history of the first spinning product;
acquiring a historical first quality detection result set and a historical second quality detection result set which are obtained by performing quality detection on the first spinning product in the previous history;
respectively taking the historical first quality information set and the historical second quality information set as input parameters of the first implicit processing layer and the second implicit processing layer, and taking the historical first quality detection result set and the historical second quality detection result set as output parameters of the first implicit processing layer and the second implicit processing layer;
optimizing and optimizing parameters in the first implicit processing layer and the second implicit processing layer respectively, wherein the parameters comprise weights and thresholds;
the inputting the adjusted first quality prediction result and the adjusted second quality prediction result into a spinning product quality detection space comprises:
acquiring a quality detection result of a historical spinning product obtained by performing quality detection in the previous history of the first spinning product;
constructing a two-dimensional coordinate system based on the historical first quality detection result set and the historical second quality detection result set;
inputting the historical first quality detection result set and the historical second quality detection result set into the two-dimensional coordinate system to obtain a plurality of coordinate points;
clustering the coordinate points according to the quality detection result of the historical spinning product to obtain a plurality of clustering results;
taking a plurality of clustering results and the two-dimensional coordinate system as a spinning product quality detection space;
and inputting the first quality prediction result and the second quality prediction result into the spinning product quality detection space to obtain the spinning product quality detection result.
2. The method of claim 1, wherein the acquiring multi-dimensional quality information of the first spun product comprises:
sampling and detecting a first textile product to obtain the multi-dimensional quality information;
classifying the multi-dimensional quality information to obtain multi-dimensional quality information and multi-dimensional strength quality information;
and taking the multi-dimensional quality information as the first quality information set, and taking the multi-dimensional strength quality information as the second quality information set.
3. The method of claim 1, wherein optimizing parameters within the first implicit processing layer comprises:
obtaining a first layer parameter scheme set according to the network structure of the first implicit processing layer;
randomly selecting a parameter scheme from the first layer parameter scheme set as a first parameter scheme and as an optimization parameter scheme;
under the first parameter scheme, the first implicit processing layer predicts the accuracy of the historical first quality detection result set according to the historical first quality information set to obtain a first accuracy;
randomly selecting a parameter scheme from the first layer parameter scheme set again to serve as a second parameter scheme;
under the second parameter scheme, the accuracy of the historical first quality detection result set is predicted by the first implicit processing layer according to the historical first quality information set, and a second accuracy is obtained;
judging whether the second accuracy is greater than the first accuracy, if so, taking the second parameter scheme as the optimization parameter scheme, and if not, taking the second parameter scheme as the optimization parameter scheme according to a probability, wherein the probability is calculated by the following formula:
Figure FDA0003854432580000031
wherein r is 2 To a second accuracy, r 1 K is an optimized speed factor for the first accuracy;
continuing to randomly select the parameter scheme in the first layer parameter scheme set, and performing iterative optimization;
when the optimization parameter scheme is not changed in the iteration of the preset times, outputting the parameter scheme corresponding to the optimization parameter scheme as an optimal parameter scheme;
and adjusting parameters of the first implicit processing layer according to the parameters in the optimal parameter scheme to complete optimization.
4. The method of claim 1, further comprising:
inputting the first quality prediction result and the second quality prediction result into the spinning product quality detection space to obtain a current quality detection coordinate point;
calculating Euclidean distances between the current quality detection coordinate point and the centers of the clustering results to obtain a Euclidean distance set;
according to the size of a plurality of Euclidean distances in the Euclidean distance set, carrying out weight distribution to obtain a weight distribution result;
weighting and adjusting the size of the Euclidean distances by adopting the weight distribution result;
and obtaining the adjusted minimum Euclidean distance, and taking the corresponding clustering result as the quality detection result of the spinning product.
5. A detecting system for a spinning product quality detecting method of claim 1, characterized in that the system comprises:
the first obtaining unit is used for acquiring and obtaining the multi-dimensional quality information of a first spinning product to obtain a first quality information set and a second quality information set;
the first construction unit is used for constructing a spinning product quality analysis model;
the first processing unit is used for optimizing parameters of the spinning product quality analysis model to obtain the spinning product quality analysis model;
the second processing unit is used for inputting the first quality information set and the second quality information set into the spinning product quality analysis model to obtain a first quality prediction result and a second quality prediction result;
the third processing unit is used for carrying out weighting adjustment on the first quality prediction result and the second quality prediction result according to the application scene of the first spinning product;
the fourth processing unit is used for inputting the adjusted first quality prediction result and the adjusted second quality prediction result into a spinning product quality detection space to obtain a spinning product quality detection result;
the second construction unit is used for constructing an input layer and an output layer of the spinning product quality analysis model based on an artificial neural network model;
the third construction unit is used for constructing a first implicit processing layer of the spinning product quality analysis model according to the first quality information set;
a fourth construction unit, configured to construct a second implicit processing layer of the spinning product quality analysis model according to the second quality information set, where the second implicit processing layer is arranged in parallel with the first implicit processing layer;
a fifth construction unit, configured to connect the input layer, the first hidden processing layer, and the output layer, and the input layer, the second hidden processing layer, and the output layer, respectively, to obtain the spinning product quality analysis model;
a fifth obtaining unit, configured to acquire a historical first quality information set and a historical second quality information set in a previous history of the first spinning product;
a sixth obtaining unit, configured to acquire a historical first quality detection result set and a historical second quality detection result set obtained by performing quality detection on the first spun yarn product in a previous history;
a fifth processing unit, configured to use the historical first quality information set and the historical second quality information set as input parameters of the first implicit processing layer and the second implicit processing layer, and use the historical first quality detection result set and the historical second quality detection result set as output parameters of the first implicit processing layer and the second implicit processing layer, respectively;
a sixth processing unit, configured to perform optimization on parameters in the first implicit processing layer and the second implicit processing layer, where the parameters include a weight and a threshold;
an eleventh obtaining unit, configured to acquire a quality detection result of a historical spinning product obtained by performing quality detection in a previous history of the first spinning product;
a sixth construction unit, configured to construct a two-dimensional coordinate system based on the historical first quality detection result set and the historical second quality detection result set;
a twelfth obtaining unit, configured to input the historical first quality detection result set and the historical second quality detection result set into the two-dimensional coordinate system, and obtain a plurality of coordinate points;
the twelfth processing unit is used for clustering the coordinate points according to the quality detection result of the historical spinning product to obtain a plurality of clustering results;
a thirteenth obtaining unit configured to use the plurality of clustering results and the two-dimensional coordinate system as the spinning product quality detection space;
a fourteenth obtaining unit, configured to input the first quality prediction result and the second quality prediction result into the spinning product quality detection space, and obtain the spinning product quality detection result.
6. An electronic device, comprising: a processor coupled with a memory, the memory for storing a program that, when executed by the processor, causes an electronic device to perform the steps of the method of any of claims 1 to 4.
7. A computer-readable storage medium, characterized in that the storage medium has stored thereon a computer program which, when being executed by a processor, carries out the steps of the method of any one of claims 1 to 4.
CN202210916175.1A 2022-08-01 2022-08-01 Spinning product quality detection method and system Active CN114994289B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202210916175.1A CN114994289B (en) 2022-08-01 2022-08-01 Spinning product quality detection method and system

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202210916175.1A CN114994289B (en) 2022-08-01 2022-08-01 Spinning product quality detection method and system

Publications (2)

Publication Number Publication Date
CN114994289A CN114994289A (en) 2022-09-02
CN114994289B true CN114994289B (en) 2022-10-28

Family

ID=83021116

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202210916175.1A Active CN114994289B (en) 2022-08-01 2022-08-01 Spinning product quality detection method and system

Country Status (1)

Country Link
CN (1) CN114994289B (en)

Families Citing this family (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115058807B (en) * 2022-08-17 2022-12-20 江苏卓鹏智能机电有限公司 Intelligent control method and system for spinning machine
CN115221736B (en) * 2022-09-20 2022-12-09 青岛宏大纺织机械有限责任公司 Method for constructing prediction model of yarn splicing strength
CN116131668A (en) * 2023-04-04 2023-05-16 山东盛日电力集团有限公司 Intelligent motor adjusting method, system, equipment and storage medium
CN116205543B (en) * 2023-05-04 2023-10-31 张家港广大特材股份有限公司 Method and system for detecting quality of metallurgical steel by combining feedback
CN116890222B (en) * 2023-08-17 2024-01-16 广州誉鑫精密部件有限公司 Intelligent positioning method and device of automatic assembling machine

Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107169565A (en) * 2017-04-27 2017-09-15 西安工程大学 Yarn quality prediction method based on fireworks algorithm improvement BP neural network
CN113408963A (en) * 2021-07-28 2021-09-17 上海致景信息科技有限公司 Textile yarn quality rating method and device, storage medium and processor

Patent Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107169565A (en) * 2017-04-27 2017-09-15 西安工程大学 Yarn quality prediction method based on fireworks algorithm improvement BP neural network
CN113408963A (en) * 2021-07-28 2021-09-17 上海致景信息科技有限公司 Textile yarn quality rating method and device, storage medium and processor

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
基于PSO-BP神经网络的纱线质量预测;熊经纬等;《东华大学学报(自然科学版)》;20150831;第41卷(第4期);第498-502页 *
基于遗传模拟退火神经网络的纱线强力预测;谷有众等;《上海纺织科技》;20160131;第44卷(第1期);第45-49页 *

Also Published As

Publication number Publication date
CN114994289A (en) 2022-09-02

Similar Documents

Publication Publication Date Title
CN114994289B (en) Spinning product quality detection method and system
CN107169565B (en) Spinning quality prediction method for improving BP neural network based on firework algorithm
CN109816031B (en) Transformer state evaluation clustering analysis method based on data imbalance measurement
CN112256739B (en) Method for screening data items in dynamic flow big data based on multi-arm gambling machine
CN113542241A (en) Intrusion detection method and device based on CNN-BiGRU mixed model
CN114742481A (en) Special steel performance evaluation method and system based on components
CN113592314A (en) Silk making process quality evaluation method based on sigma level
Chakraborty et al. Analysis of cotton fibre properties: a data mining approach
CN115874321B (en) Self-adaptive management method and system for improving yarn quality
CN111507824A (en) Wind control model mold-entering variable minimum entropy box separation method
CN109039797B (en) Strong learning based large flow detection method
CN116365519B (en) Power load prediction method, system, storage medium and equipment
CN110196797B (en) Automatic optimization method and system suitable for credit scoring card system
Das et al. Rough set-based decision tool for classification of cotton yarn neps
CN114597886A (en) Power distribution network operation state evaluation method based on interval type two fuzzy clustering analysis
Akgül et al. Optimization of the Murata Vortex Spinning machine parameters by the SMAA-MOORA approach
CN110913033A (en) IDCIP address allocation method based on CNN convolutional neural network learning
Beltran et al. Mill specific prediction of worsted yarn performance
CN117055508A (en) Intelligent control system and method for twisting production
CN115787160B (en) Static electricity removal control method and system for spinning machine group
CN107884362B (en) Method for rapidly detecting spandex content in cotton, polyester and spandex blended fabric
CN114706743B (en) Comprehensive evaluation method supporting real-time evaluation
CN113742216B (en) Method, device and storage medium for detecting efficiency of machine learning engine
Wu et al. An Algorithm for Establishing A Model of Optimal Cotton Blending
CN117112871B (en) Data real-time efficient fusion processing method based on FCM clustering algorithm model

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