CN115963397A - Rapid online detection method and device for surface defects of inner contour of motor stator - Google Patents
Rapid online detection method and device for surface defects of inner contour of motor stator Download PDFInfo
- Publication number
- CN115963397A CN115963397A CN202211530821.7A CN202211530821A CN115963397A CN 115963397 A CN115963397 A CN 115963397A CN 202211530821 A CN202211530821 A CN 202211530821A CN 115963397 A CN115963397 A CN 115963397A
- Authority
- CN
- China
- Prior art keywords
- motor stator
- inner contour
- detection
- image
- stator
- 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.)
- Granted
Links
Images
Classifications
-
- 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
- Manufacture Of Motors, Generators (AREA)
Abstract
The invention discloses a quick online detection method for surface defects of an inner contour of a motor stator, belonging to the technical field of defect detection and comprising the following steps: constructing a defect prediction model by adopting a YOLO v6 algorithm; calibrating the detection mechanism by adopting a calibration mould with a standard checkerboard in the inner contour; placing the motor stator on a placing platform; adjusting the position of the detection mechanism in the horizontal plane; collecting image parameters of the inner contour of the motor stator; the image parameters are input into a defect prediction model to judge the defects, and the invention also discloses a detection device of the corresponding method. The invention provides a method and a device for quickly detecting the surface defects of the inner contour of a motor stator on line, which can follow the production rhythm of a motor production line, quickly realize the quick on-line detection of the surface defects of the inner contour of the motor stator, have high automation degree, ensure the accuracy of a detection result, effectively improve the production efficiency of the motor stator, improve the product quality and reduce the detection cost in the production and preparation process.
Description
Technical Field
The invention belongs to the technical field of defect detection, and particularly relates to a method and a device for quickly detecting surface defects of an inner contour of a motor stator on line.
Background
The stator in the motor is one of two important core function units of the motor, the mass flow control of the stator is an extremely important link in the production and installation processes of the motor, and the stator has very important influence on the function and the service life of the motor.
At present, an automatic single machine or an automatic production line is basically adopted for the production of the motor, but the detection technology level of the surface defects of the stator is obviously lagged behind the technology level of the production, most enterprises still adopt a manual visual inspection method or a traditional image and manual method, the consumed time is long, the efficiency is low, the detection quality is greatly influenced by artificial subjective factors, and the real-time online detection in the scale production process cannot be met. The motor stator has various structures, specifications and models, the types of defects on the surface of a contour are pits, installation gaps, pollution of installation holes, rusting on the surface of a metal, foreign matters, scratches, scratch-off and rubbing, abnormal colors, pollution of welding beads, burrs and the like, and the detection requirements on the surface of the motor stator are also greatly different, so that the detection requirements of the motor stator with various differences are met by using the same device, and the detection compatible with the stators with different specifications as much as possible is a difficult problem which is urgently required to be solved by automatic surface defect detection of visual detection equipment in the production process.
Disclosure of Invention
Aiming at one or more of the above defects or improvement requirements in the prior art, the invention provides a method and a device for rapidly detecting the surface defects of the inner contour of the motor stator on line, which can follow the production rhythm of a motor production line, rapidly realize the rapid online detection of the surface defects of the inner contour of the motor stator, have high automation degree, ensure the accuracy of a detection structure, effectively improve the production efficiency of the motor stator, improve the product quality, reduce the cost in the production preparation process and improve the production efficiency.
In order to achieve the purpose, the invention provides a device for quickly detecting the surface defects of the inner contour of a motor stator on line, which comprises the following steps:
s1, constructing a defect prediction model by adopting a YOLO v6 algorithm;
s2, calibrating the detection mechanism by adopting a calibration mould with a standard checkerboard on the inner contour;
s3, placing the motor stator on a placing platform;
s4, adjusting the position of the detection mechanism in the horizontal plane to finish alignment of the detection mechanism and the axis of the motor stator;
s5, driving the detection mechanism to go deep into the motor stator, and collecting image parameters of the inner contour of the motor stator;
s6, inputting the image parameters into the defect prediction model, judging whether defects exist in the image parameters and defect positions and defect types when the defects exist through the defect prediction model, if so, entering a step S7, and if not, entering a step S8;
s7, sending out an alarm signal, informing an operator of the defect position and the defect type, and then entering the step S8;
s8, removing the motor stator from the placing platform, and repeating the steps S3-S6;
wherein, S1 comprises the following steps:
s11, constructing an image recognition training model;
s12, collecting a stator inner contour surface defect image set containing defect labels, wherein the stator inner contour surface defect image set comprises a training set and a testing set;
s13, putting the training set into the image recognition training model to obtain a defect prediction model;
s14, putting the test set into the defect prediction model, judging whether the undetected rate of the defect prediction model is 0 or not when the undetected rate does not exceed a threshold value, if the undetected rate is 0, entering a step S16, and if the undetected rate is not 0, entering a step S15;
s15, adjusting the training set and the test set, adjusting learning parameters in the image recognition training model for fine adjustment, and repeating the steps S12 to S14;
s16, completing the construction of a defect prediction model;
as a further preferred aspect of the present invention, the S13 includes the steps of:
s131, receiving the training set by an input end, and performing data enhancement, filtering and normalization on the pictures in the training set to obtain a preprocessed image with n × n channels;
s132, receiving the preprocessed image by using a backbone network, and extracting a feature map in the preprocessed image through depth separable convolution;
s133, receiving the characteristic diagram out2 by adopting a neck network, adding a scratch detection layer and a micro-recess detection layer to the tail end of the neck network, and mixing and combining the characteristics in the characteristic diagram out2 through the neck network to obtain a characteristic diagram out3;
and S134, the output layer receives the feature map out3, and determines the optimal regression solution and the confidence degree by adopting the judgment of the target frame and the multi-step screening.
As a further preferred of the present invention, said depth separable convolution comprises the steps of:
s1321, obtaining and determining channels to be convolved from the preprocessed image, setting convolution kernels corresponding to different channels one by one, and convolving different channels channel by adopting independent convolution kernels to obtain feature maps out1 with the same number as the convolution kernels;
s1322, determining the size of a convolution kernel corresponding to the channel according to the number of the channels in the channel-by-channel convolution, and performing weighted combination on the feature diagram out1 in the channel direction to obtain a feature diagram out2 containing the surface defect features of the inner contour of the motor stator.
As a further preferred aspect of the present invention, the data enhancement includes flipping, rotating, cropping, superimposing and mosaic algorithms of the training set images at random.
As a further preferred aspect of the present invention, S2 includes the steps of:
s21, preparing a calibration mould with the same inner diameter as the stator of the motor to be tested, and arranging a standard checkerboard on the inner contour of the calibration mould;
s22, placing the calibration mould on a placing platform, and adjusting the axis of the detection mechanism to be coincident with the axis of the calibration mould;
s23, acquiring images of a standard checkerboard in a calibration mold through a detection mechanism to primarily adjust the posture of the detection mechanism;
s24, acquiring calibration image information of the marked checkerboards in the inner contour of the calibration mould again by using the detection mechanism after the initial adjustment;
and S25, obtaining a correction matrix of the detection mechanism through the calibration image information, and using the correction matrix to correct the graphic parameters acquired by the detection mechanism.
As a further preferred aspect of the present invention, S4 includes the steps of:
s41, collecting contour information of an inner hole circle of a motor stator;
s42, adopting a first calculation formula to calculate the edge pixel position in the contour information;
s43, repeatedly iterating by adopting a least square method to obtain the accurate circle center and radius of the fitted inner hole circle;
and S44, adjusting the position of the detection mechanism in the horizontal plane according to the circle center and the radius to finish the alignment of the detection mechanism and the motor stator.
As a further preferred aspect of the present invention, the first calculation formula is:
f=∑((X i -X C ) 2 +(Y i -Y C ) 2 -R 2 ) 2
wherein (X) i ,Y i ) Fitting the positions of the pixels at the edge of the inner hole circle; (X) C ,Y C ) The position of the center of the fitting inner hole circle is obtained; r is the radius of the fitting inner hole circle; f is the objective function.
The invention also discloses a device for rapidly detecting the surface defects of the inner contour of the motor stator on line, which adopts the method for rapidly detecting the surface defects of the inner contour of the motor stator on line as claimed in any one of claims 1 to 7 to complete the detection of the motor stator, and comprises the following steps:
a carrying mechanism; the motor stator placing platform is horizontally arranged and used for bearing a motor stator;
a detection mechanism; the device comprises a detection motion component and an image acquisition component, wherein the detection motion component is fixedly connected with the image acquisition component and is used for driving the image acquisition component to reciprocate in space; the image acquisition component comprises a plurality of image acquisition cameras and at least one centering camera, and the plurality of image acquisition cameras are horizontally and circumferentially arranged and are used for acquiring image parameters of the inner contour of the motor stator; the centering camera is opposite to the end face of the top end of the placing platform and is used for acquiring the inner hole circular outline information of the motor stator;
a support; the bearing mechanism and the detection mechanism are arranged oppositely along the vertical direction and are fixedly installed on the bracket.
As a further preferable aspect of the present invention, the carrying mechanism further includes a carrying moving member disposed between the placing platform and the support, and configured to drive the placing platform to move in a horizontal plane.
As a further preferable aspect of the present invention, a positioning assembly is disposed on the placing platform, and is configured to fix the motor stator on the placing platform when the motor stator is detected.
The above-described improved technical features may be combined with each other as long as they do not conflict with each other.
Generally, compared with the prior art, the technical scheme conceived by the invention has the following beneficial effects:
(1) According to the method and the device for quickly detecting the surface defects of the inner contour of the motor stator on line, the image information of the inner contour of the motor stator acquired by an analysis and detection mechanism is quickly acquired by adopting a YOLO v6 algorithm component defect prediction model, so that the defects of the inner contour of the motor stator are quickly determined, the production rhythm of a motor production line is kept up with, and the production efficiency is improved; and accurate calibration of a certain detection mechanism is adopted, so that the accuracy of image information acquisition can be ensured, the time consumed by calibration can be simplified, the detection speed of the surface defects of the inner contour of the motor stator is increased, the requirement on operation beats in the production process of a production line is met, and the production efficiency of the whole motor production line is increased.
(2) According to the method and the device for quickly detecting the surface defects of the inner contour of the motor stator on line, the preprocessed image is convolved in a depth separable convolution mode, convolution kernels corresponding to different channels are arranged corresponding to the preprocessed image, and weighted combination is performed in the channel direction according to the sizes of the convolution kernels, so that the preprocessed image subjected to the depth separable convolution can be ensured to contain a characteristic diagram of the surface defects of the inner contour of the motor stator, the operation cost of a network model can be reduced, the operation speed can be improved, the detection precision of the surface defects can be ensured, and the accuracy of quickly detecting the surface defects of the inner contour of the motor stator on line can be further ensured.
(3) According to the method and the device for quickly detecting the surface defects of the inner profile of the motor stator on line, the forms of the scratch detection layer aiming at tiny scratches and pits and the micro pit detection layer are added in the neck network in a specific manner, so that the data of splicing multiple characteristic diagrams, scratch and micro pits are copied and enhanced, the attention of the neck network to the detection layers and the micro pits is improved, the defects of pits and scratches are effectively prevented from being submerged in massive characteristic information, further, the accurate detection of the tiny scratches and pits by on line detection is greatly improved with low operation cost, and the accuracy of quickly detecting the surface defects of the inner profile of the motor stator on line is further improved.
(4) According to the method and the device for quickly detecting the surface defects of the inner contour of the motor stator on line, the defect prediction model for quickly detecting the motor stator is constructed by adopting the mode of adding the scratch detection layer and the micro-depression detection layer by the deep convolution and the neck network, so that the stable detection of the inner contour defects of the motor stator can be greatly increased, the detection speed of the inner contour of the motor stator can be ensured, the quick online detection of the inner contour of the motor stator can keep up with the production beat of a motor production line, and the production efficiency of the motor is improved. And the calibration mold with the checkerboards on the inner wall effectively realizes accurate calibration of the image acquisition cameras arranged in the annular direction in the detection mechanism, so that the accuracy of acquisition of the contour image parameters in the motor stator is ensured. Meanwhile, the motor stator structures with different diameters are accurately positioned by arranging the positioning assembly on the placing platform, so that the detection precision of the motor stator is ensured, and the motor stator positioning device has good economic benefits and popularization prospects.
Drawings
FIG. 1 is a flow chart of a method for rapidly detecting surface defects of an inner contour of a stator of a motor on line according to the invention;
FIG. 2 is a flow chart of step S1 of a method for rapidly detecting surface defects of an inner contour of a stator of a motor on line according to the present invention;
FIG. 3 is a flow chart of step S13 of the method for rapidly detecting surface defects of an inner contour of a motor stator on line according to the present invention;
FIG. 4 is a flow chart of step S2 of the method for rapidly detecting surface defects of an inner contour of a stator of a motor on line according to the present invention;
FIG. 5 is a flow chart of step S4 of the method for rapidly detecting surface defects of an inner contour of a stator of a motor on line according to the present invention;
FIG. 6 is a perspective view of a device for rapidly detecting surface defects of an inner contour of a stator of an electric motor on line according to the present invention;
FIG. 7 is a perspective view of the backside of a device for rapidly detecting surface defects of an inner contour of a stator of an electric motor on-line according to the present invention;
FIG. 8 is a left side view of the perspective of FIG. 6 of an apparatus for rapid on-line detection of surface defects of an inner contour of a stator of an electric machine according to the present invention;
FIG. 9 is a rear view of the device for rapidly detecting surface defects of inner contours of a stator of an electric motor on line according to the present invention;
FIG. 10 is a perspective view of an image capturing component of a device for rapidly detecting surface defects of an inner contour of a stator of a motor on line.
In all the figures, the same reference numerals denote the same features, in particular:
1. a carrying mechanism; 11. placing a platform; 12. a carrier motion member;
2. a detection mechanism; 21. an X-axis motion assembly; 22. a Y-axis motion assembly; 23. a Z-axis motion assembly; 24. a camera fixing platform; 25. an image capture camera; 26. centering the camera;
3. a support; 4. and a Z-axis adjusting component.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and do not limit the invention. In addition, the technical features involved in the embodiments of the present invention described below may be combined with each other as long as they do not conflict with each other.
In the description of the present invention, it is to be understood that the terms "central," "longitudinal," "lateral," "length," "width," "thickness," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," "outer," "clockwise," "counterclockwise," "axial," "radial," "circumferential," and the like are used in the orientations and positional relationships indicated in the drawings for convenience in describing the invention and to simplify the description, and are not intended to indicate or imply that the referenced device or element must have a particular orientation, be constructed and operated in a particular orientation, and are not to be considered limiting of the invention.
Furthermore, the terms "first", "second" and "first" are used for descriptive purposes only and are not to be construed as indicating or implying relative importance or to implicitly indicate the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one such feature. In the description of the present invention, "a plurality" means at least two, e.g., two, three, etc., unless specifically limited otherwise.
In the present invention, unless otherwise expressly stated or limited, the terms "mounted," "connected," "secured," and the like are to be construed broadly and can, for example, be fixedly connected, detachably connected, or integrally formed; can be mechanically or electrically connected; they may be directly connected or indirectly connected through intervening media, or they may be connected internally or in any other suitable relationship, unless expressly stated otherwise. The specific meanings of the above terms in the present invention can be understood according to specific situations by those of ordinary skill in the art.
In the present invention, unless otherwise expressly stated or limited, the first feature "on" or "under" the second feature may be directly contacting the first and second features or indirectly contacting the first and second features through an intermediate. Also, a first feature "on," "over," and "above" a second feature may be directly or diagonally above the second feature, or may simply indicate that the first feature is at a higher level than the second feature. A first feature being "under," "below," and "beneath" a second feature may be directly under or obliquely under the first feature, or may simply mean that the first feature is at a lesser elevation than the second feature.
Example (b):
as shown in fig. 1 to 10, the method for rapidly detecting surface defects of an inner contour of a motor stator on line in the preferred embodiment of the present application can keep up with the production tact of a motor production line, rapidly implement rapid online detection of surface defects of an inner contour of a motor stator, and at the same time, not only has high automation degree, but also can ensure accuracy of detection structure, effectively improve production efficiency of the motor stator, improve product quality, and reduce production and preparation costs.
Specifically, as shown in fig. 1, in a preferred embodiment of the present application, the method for rapidly detecting surface defects of an inner contour of a stator of an electric machine on line comprises the following steps:
s1, constructing a defect prediction model by adopting a YOLO v6 algorithm;
s2, calibrating the detection mechanism 2 by adopting a calibration mould with a standard checkerboard in the inner contour;
s3, placing the motor stator on the placing platform 11, and fixing the motor stator through a positioning component;
s4, adjusting the position of the detection mechanism 2 in the horizontal plane to finish the alignment of the detection mechanism 2 and the axis of the motor stator;
s5, driving the detection mechanism 2 to penetrate into the motor stator along the Z direction, and collecting image parameters of the inner contour of the motor stator;
s6, inputting the image parameters into a defect prediction model, judging whether defects exist in the image parameters and the positions and types of the defects when the defects exist through the defect prediction model, if so, entering a step S7, and if not, entering a step S8;
s7, sending out an alarm signal, informing an operator of the defect position and the defect type, and then entering the step S8;
and S8, removing the motor stator from the placing platform 11, and repeating the steps S3 to S6.
Further, as shown in fig. 2, in the preferred embodiment of the present application, step S1 includes the following steps
S11, constructing an image recognition training model consisting of an input end, a backbone network, a neck network and an output layer;
preferably, the image recognition training model predicts a plurality of bounding boxes and confidence levels through a single convolutional neural network;
s12, collecting a stator inner contour surface defect image set containing defect labels, wherein the stator inner contour surface defect image set comprises a training set and a testing set;
preferably, the defect label contains a plurality of defect types and positions of the defect types in the image.
Further preferably, the ratio between training set and test set is 8.
S13, putting the training set into an image recognition training model to obtain a defect prediction model;
s14, putting the test set into a defect prediction model, judging whether the undetected rate of the defect prediction model is 0 or not when the undetected rate does not exceed a threshold, if the undetected rate is 0, entering a step S16, and if the undetected rate is not 0, entering a step S15;
s15, adjusting the training set and the test set, adjusting learning parameters in the image recognition training model for fine adjustment, and repeating the steps S12 to S14;
preferably, the learning parameters include a learning rate (learning rate) and a decay (decay).
And S16, completing the construction of a defect prediction model.
More specifically, as shown in fig. 3, in the preferred embodiment of the present application, step S13 includes the following steps:
s131, receiving a training set by an input end, and performing data enhancement, filtering and normalization on pictures in the training set to obtain a preprocessed image with n × n channels;
preferably, the data enhancement includes flipping, rotating, cropping, superimposing, and mosaic algorithms of the training set images at random.
S132, receiving the preprocessed image by using a backbone network, and extracting a feature map in the preprocessed image through depth separable convolution;
preferably, the depth separable convolution comprises the steps of:
s1321, obtaining and determining channels to be convolved from the preprocessed image, setting convolution kernels corresponding to different channels one by one, and convolving different channels channel by adopting independent convolution kernels to obtain feature maps out1 with the same number as the convolution kernels;
s1322, determining the size of a convolution kernel corresponding to a channel according to the number of the channels in the channel-by-channel convolution, and performing weighted combination on the feature diagram out1 in the channel direction to obtain a feature diagram out2 containing the surface defect features of the inner contour of the motor stator;
s133, receiving the characteristic diagram out2 by adopting a neck network, adding a scratch detection layer and a micro-recess detection layer to the tail end of the neck network, and mixing and combining the characteristics in the characteristic diagram out2 by the neck network to obtain a characteristic diagram out3;
preferably, a scratch detection layer and a micro-recess detection layer are specifically added at the end of the neck network, and the attention of the network to the scratch detection layer and the micro-recess detection layer is improved by introducing a specific channel, performing Concat fusion on the multi-layer features of the backbone network and copying and enhancing the data of the scratch detection layer and the micro-recess detection layer, so that the recesses and scratches are effectively prevented from being submerged in massive feature information. In actual use, under the condition of increasing 19% of calculation amount cost, the scratch detection rate of 7% and the dent detection rate of 3% on the surface of the rotor are improved approximately.
And S134, the output layer receives the feature diagram out3, and determines the optimal regression solution and the confidence coefficient by adopting the judgment of the target frame and the multi-step screening.
Further, as shown in fig. 4, in the preferred embodiment of the present application, the step S2 includes the following steps:
s21, preparing a calibration mould with the same inner diameter as the stator of the motor to be measured, and arranging a standard checkerboard in the inner contour of the calibration mould;
preferably, the standard checkerboard is a long strip, the length of the standard checkerboard can be the same as the perimeter of the inner contour of the calibration mold, and the height of the standard checkerboard is the same as the axial height of the calibration mold.
S22, placing the calibration mould on the placing platform 11, and adjusting the axis of the detection mechanism 2 to be coincident with the axis of the calibration mould;
s23, acquiring an image of a standard checkerboard in the calibration mold through the detection mechanism 2, and primarily adjusting the posture of the detection mechanism 2;
preferably, the detection mechanism 2 includes four image capturing cameras 25 arranged in the horizontal circumferential direction, and the postures of the four image capturing cameras 25 are preliminarily adjusted by comparing the images captured in the image capturing cameras 25.
S24, acquiring calibration image information of the marked checkerboards in the inner contour of the calibration mould again by using the detection mechanism 2 after the initial adjustment;
preferably, four image acquisition cameras 25 uniformly arranged along the circumferential direction in the detection mechanism 2 acquire calibration image information of the annular checkerboard, so that multiple times of calibration required when the multiple cameras are calibrated by adopting a calibration plate can be avoided, errors existing in the process of multiple times of calibration and splicing can be avoided, the time required for calibration can be reduced, the detection rate of the whole rapid online detection method can be further improved, and the production beat of a motor production line can be adapted.
And S25, obtaining a correction matrix of the detection mechanism 2 through the calibration image information, and using the correction matrix to correct the graphic parameters acquired by the detection mechanism 2.
Preferably, the correction matrix includes distortion coefficients, internal and external parameters, and data matching parameters.
Further preferably, because the inner diameter of the calibration mold is the same as the inner diameter of the motor to be measured, the acquired calibration image information is the linear position of the detection mechanism 2 on the motor stator, which can ensure that the focal length and the object distance of the image acquisition camera 25 are both values suitable for acquiring the inner contour image of the motor, so that the correction matrix solved reversely according to the image calibration data acquired by the image acquisition camera 25 is the real distortion reaction of the inner contour of the stator, and further the accurate correction matrix of the inner contour of the motor stator is obtained, and the image parameters revised by the correction matrix are the image information accurately solving the actual state of the inner contour of the motor stator.
Further, as shown in fig. 5, the step S4 includes the steps of:
s41, collecting contour information of an inner hole circle of a motor stator;
s42, adopting a first calculation formula to calculate the edge pixel position in the contour information;
the first calculation formula is:
f=∑((X i -X C ) 2 +(Y i -Y C ) 2 -R 2 ) 2
wherein (X) i ,Y i ) Fitting the positions of the pixels at the edge of the inner hole circle; (X) C ,Y C ) The position of the center of the fitting inner hole circle is obtained; r is the radius of the fitting inner hole circle; f is the objective function.
S43, repeatedly iterating by adopting a least square method to obtain the accurate circle center and radius of the fitted inner hole circle;
and S44, adjusting the position of the detection mechanism 2 in the horizontal plane according to the circle center and the radius, and completing alignment of the detection mechanism 2 and the motor stator.
Further, in a preferred embodiment of the present application, the image parameters in step S6 are subjected to a defect prediction model to obtain an optimal regression solution and confidence corresponding to the image parameters, and whether a defect, a defect position, and a defect type exist is determined by judging the optimal regression solution and confidence.
In addition, in another preferred embodiment of the present application, a device for fast online detecting surface defects of an inner contour of a motor stator is disclosed, which is used for implementing online detection of the motor stator by using the method for fast online detecting surface defects of an inner contour of a motor stator.
Specifically, as shown in fig. 6 to 10, the device for rapidly detecting surface defects of an inner contour of a motor stator on line comprises a carrying mechanism 1, a detection mechanism 2, a bracket 3 and a data processing module; the support 3 is a main supporting structure of the online detection device, and is integrally in a rectangular frame structure and used for carrying the object mechanism 1 and the detection mechanism 2. The object carrying mechanism 1 is arranged inside the support 3 and comprises a placing platform 11 horizontally arranged at the top end of the object carrying mechanism 1, and the placing platform is used for stably bearing a motor stator or calibrating a mold, so that the detection mechanism 2 can accurately acquire image parameters of the inner contour of the motor stator. The detection mechanism 2 is arranged opposite to the carrying mechanism 1, is also fixedly arranged on the bracket 3 and comprises a detection movement component and an image acquisition component. Wherein, detect moving member fixed mounting on support 3, image acquisition component fixed mounting is on detecting moving member, through detecting moving member drive image acquisition component and move in the space to when carrying out profile acquisition in the motor stator through the image acquisition component, the image parameter of profile in the motor stator is gathered to the accuracy.
Further, in the preferred embodiment of the present application, a positioning assembly is disposed on the placing platform 11, and is used for completing quick and accurate positioning of the motor stator when the motor stator is placed on the placing platform 11, for example, the positioning assembly is a positioning fixture, a positioning slot, a magnetic attraction point, and the like disposed on the placing platform 11. Preferably, in this application preferred embodiment, be provided with a plurality of magnetism suction points along the hoop distribution in the place table 11, the diameter that this magnetism suction point is arranged along the hoop is the same with motor stator internal diameter, and when in actual use, place motor stator on place table 11 the back, to the magnetic suction point circular telegram production magnetic field of inhaling, realize the location to motor stator through magnetic field. Further preferably, a plurality of magnetic attraction points are circumferentially arranged on concentric circles with different diameters, so that when motor stators with different inner diameters are produced, magnetic fields can be generated through the magnetic attraction points on the corresponding diameters, and the motor stators with different inner diameters are fixedly arranged on the placing platform 11.
More in detail, in a preferred embodiment of the present application, a stage moving member 12 is provided between the placing platform 11 and the stand 3 for driving the movement of the placing platform 11 in the horizontal plane. In the actual use process, the placing platform 11 is driven to move in the horizontal plane through the object carrying moving component 12, so that part or all of the placing platform 11 extends out of the bracket 3, a motor stator can be accurately and quickly placed on the placing platform 11, the detection speed of the online inspection device is improved, and the production rhythm of a motor production line is adapted. Preferably, the portion of the resting platform 11 that protrudes beyond the support 3 does not exceed 10% of its total area. Further preferably, the object movement member 12 is communicatively connected to a data processing module for controlling movement of the object movement member 12 according to predetermined instructions.
Further, in the preferred embodiment of the present application, the object moving member 12 is a corresponding matching guide slot and a sliding block, wherein the guide slot is fixedly installed on the bracket 3, the sliding block is correspondingly matched in the guide slot while the placing platform 11 is fixedly connected, and the sliding block is driven by a corresponding driving member to move, so that the placing platform 11 can realize reciprocating motion along the direction of the guide slot. Preferably, the guide groove may be a triangular frame as shown in fig. 1, a rectangular frame-like structure, or even a plate-like structure in which the guide groove is opened.
Of course, the object movement member 12 is not limited to the guide groove and the slider, and may also adopt a driving structure similar to a tank chain, or adopt a ball screw, so long as the structure that can horizontally drive the placing platform 11 can implement the solution in this embodiment, and is not described herein again.
In more detail, as shown in fig. 7, in the preferred embodiment of the present application, the object moving member 12 further includes a Z-axis adjusting assembly 4, preferably at least one screw motor extending along the Z-axis direction, for driving the object moving member 12 to reciprocate in the Z-axis direction, thereby adjusting the position of the placing platform 11 in the space. Further preferably, at least one guide mechanism is provided between the carrier moving member 12 and the frame 3 in the Z-axis direction for guiding the carrier moving member when moving in the Z-axis direction.
Further, in a preferred embodiment of the present application, the detection motion component is configured to drive the image capturing component to move stably in the space, and therefore, in order to achieve the self-movement of the image capturing component in the space, the detection motion component includes an X-axis motion component 21, a Y-axis motion component 22, and a Z-axis motion component 23, which are respectively configured to achieve the stable movement of the image capturing mechanism in the three-axis directions. Preferably, the Y-axis moving assembly 22 is symmetrically disposed on two first guide rods on the top truss of the support 3, and a first driving part is also disposed on the guide rods, and the two ends of the X-axis moving assembly 21 are fixedly mounted on the first driving part, and the first driving part drives the X-axis moving assembly 21 to move on the first guide rods. Correspondingly, the X-axis moving assembly 21 also includes a second guiding rod and a second moving portion, two ends of the second guiding rod are fixed on the first driving part of the Y-axis moving assembly 22, the second moving portion is disposed on the second guiding rod, and the second moving portion can reciprocate on the second guiding rod to drive the Z-axis moving assembly 23 disposed on the second moving portion to reciprocate in the X-axis direction. One end of the Z-axis moving assembly 23 is fixedly mounted on the end surface of the second driving part facing the placing platform 11, the other end of the Z-axis moving assembly is provided with a telescopic part which can stretch along the Z-axis direction, and an image collecting mechanism is fixedly mounted at the end part of the telescopic part facing the placing platform 11 and used for driving the image collecting mechanism to reciprocate along the Z-axis direction. Preferably, the X-axis motion assembly 21, the Y-axis motion assembly 22 and the Z-axis motion assembly 23 are all in communication connection with the data processing module to realize accurate transmission of control commands of the data processing module.
Further preferably, in the preferred embodiment of the present application, a ball screw or a roller structure is used between the guide rod and the moving portion, so as to realize the reciprocating motion of the moving portion on the guide rod.
Further, in the preferred embodiment of the present application, the image capturing means comprises a camera holding platform 24, an image capturing camera 25 and a centering camera 26. Wherein, camera fixed platform 24 is the basic installation component of this image acquisition component, deviates from place the terminal surface and the 23 pars contractilis fixed connection of Z axle motion subassembly of place the platform 11 to evenly be provided with the first mounting groove of a plurality of along circumference on the camera fixed platform 24 side wall face for image acquisition camera 25's stable erection, simultaneously, seted up the second mounting groove on this camera fixed platform 24 towards place the terminal surface of platform 11, be used for centering camera 26's fixed mounting. The image acquisition camera 25 and the centering camera 26 are respectively used for acquiring image parameters of the inner contour of the motor stator, acquiring contour information of an inner hole circle of the motor stator, and aligning the axis of the image acquisition component with the axis of the motor stator through the contour information. Preferably, four image capturing cameras 25 are uniformly arranged on the side wall of the camera fixing platform 24 along the circumferential direction, so as to simultaneously capture image information of 360 degrees of the circumferential direction of the inner contour surface of the motor stator. Further preferably, the camera fixing platform 24 is a rectangular parallelepiped structure.
According to the method and the device for quickly detecting the surface defects of the inner contour of the motor stator on line, the defect prediction model for quickly detecting the motor stator is constructed by adopting the mode of adding the scratch detection layer and the micro-depression detection layer by the deep convolution and the neck network, so that the stable detection of the inner contour defects of the motor stator can be greatly increased, the detection speed of the inner contour of the motor stator can be ensured, the quick online detection of the inner contour of the motor stator can keep up with the production rhythm of a motor production line, and the production efficiency of the motor is improved. And the calibration mould with the checkerboard on the inner wall effectively realizes the accurate calibration of the image acquisition camera 25 which is arranged in the detection mechanism 2 in the radial direction, thereby ensuring the accuracy of the acquisition of the contour image parameters in the motor stator. Meanwhile, the motor stator structures with different diameters are accurately positioned by arranging the positioning assembly on the placing platform 11, so that the detection precision of the motor stator is guaranteed, and the motor stator positioning device has good economic benefits and popularization prospects.
It will be understood by those skilled in the art that the foregoing is only a preferred embodiment of the present invention, and is not intended to limit the invention, and that any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the scope of the present invention.
Claims (10)
1. A quick online detection method for surface defects of an inner contour of a motor stator is characterized by comprising the following steps:
s1, constructing a defect prediction model by adopting a YOLO v6 algorithm;
s2, calibrating the detection mechanism by adopting a calibration mould with a standard checkerboard on the inner contour;
s3, placing the motor stator on a placing platform;
s4, adjusting the position of the detection mechanism in the horizontal plane to finish the alignment of the detection mechanism and the axis of the motor stator;
s5, driving the detection mechanism to go deep into the motor stator, and collecting image parameters of the inner contour of the motor stator;
s6, inputting the image parameters into the defect prediction model, judging whether defects exist in the image parameters and the positions and types of the defects when the defects exist through the defect prediction model, if so, entering a step S7, and if not, entering a step S8;
s7, sending out an alarm signal, informing an operator of the defect position and the defect type, and then entering the step S8;
s8, removing the motor stator from the placing platform, and repeating the steps S3-S6;
wherein, S1 includes the following steps:
s11, constructing an image recognition training model;
s12, collecting a stator inner contour surface defect image set containing defect labels, wherein the stator inner contour surface defect image set comprises a training set and a testing set;
s13, putting the training set into the image recognition training model to obtain a defect prediction model;
s14, putting the test set into the defect prediction model, judging whether the undetected rate of the defect prediction model is 0 or not when the undetected rate does not exceed a threshold, if the undetected rate is 0, entering a step S16, and if the undetected rate is not 0, entering a step S15;
s15, adjusting the training set and the test set, adjusting learning parameters in the image recognition training model for fine adjustment, and repeating the steps S12-S14;
and S16, completing the construction of a defect prediction model.
2. The method for rapidly detecting the surface defect of the inner contour of the stator of the motor as claimed in claim 1, wherein the S13 comprises the steps of:
s131, receiving the training set by an input end, and performing data enhancement, filtering and normalization on the pictures in the training set to obtain a preprocessed image with n × n channels;
s132, receiving the preprocessed image by using a backbone network, and extracting a feature map in the preprocessed image through depth separable convolution;
s133, receiving the characteristic diagram out2 by adopting a neck network, adding a scratch detection layer and a micro-recess detection layer to the tail end of the neck network, and mixing and combining the characteristics in the characteristic diagram out2 through the neck network to obtain a characteristic diagram out3;
and S134, the output layer receives the feature map out3, and determines the optimal regression solution and the confidence degree by adopting the judgment of the target frame and the multi-step screening.
3. The method for rapid on-line detection of surface defects in an electric machine stator contour of claim 2 wherein said depth separable convolution comprises the steps of:
s1321, obtaining and determining channels to be convolved from the preprocessed image, setting convolution kernels corresponding to different channels one by one, and convolving different channels channel by adopting independent convolution kernels to obtain feature maps out1 with the same number as the convolution kernels;
s1322, determining the size of a convolution kernel corresponding to the channel according to the number of the channels in the channel-by-channel convolution, and performing weighted combination on the feature diagram out1 in the channel direction to obtain a feature diagram out2 containing the surface defect features of the inner contour of the motor stator.
4. The method of claim 2, wherein the data enhancement includes random inversion, rotation, cropping, superimposing and mosaic algorithms on the training set images.
5. The method for rapidly detecting the surface defect of the inner contour of the stator of the motor in an online manner as claimed in any one of claims 1 to 4, wherein the step S2 comprises the steps of:
s21, preparing a calibration mould with the same inner diameter as the stator of the motor to be tested, and arranging a standard checkerboard on the inner contour of the calibration mould;
s22, placing the calibration mould on a placing platform, and adjusting the axis of the detection mechanism to be coincident with the axis of the calibration mould;
s23, acquiring an image of a standard checkerboard in a calibration mold through a detection mechanism to preliminarily adjust the posture of the detection mechanism;
s24, acquiring calibration image information of the marked checkerboards in the inner contour of the calibration mould again by using the detection mechanism after the initial adjustment;
and S25, obtaining a correction matrix of the detection mechanism through the calibration image information, and using the correction matrix to correct the graphic parameters acquired by the detection mechanism.
6. The method for rapidly detecting the surface defect of the inner contour of the stator of the motor in an online manner as claimed in any one of claims 1 to 4, wherein the step S4 comprises the steps of:
s41, collecting contour information of an inner hole circle of a motor stator;
s42, adopting a first calculation formula to calculate the edge pixel position in the contour information;
s43, repeatedly iterating by adopting a least square method to obtain the accurate circle center and radius of the fitted inner hole circle;
and S44, adjusting the position of the detection mechanism in the horizontal plane according to the circle center and the radius, and completing alignment of the detection mechanism and the motor stator.
7. The method for rapidly detecting surface defects of inner contours of stators of electric machines according to claim 6, wherein the first calculation formula is:
f=∑((X i -X C ) 2 +(Y i -Y C ) 2 -R 2 ) 2
wherein (X) i ,Y i ) Fitting the positions of the pixels at the edge of the inner hole circle; (X) C ,Y C ) The position of the center of the fitting inner hole circle is obtained; r is the radius of the fitting inner hole circle; f is the objective function.
8. An apparatus for rapidly detecting surface defects of inner contour of motor stator on line, which is characterized in that the detection of the motor stator is completed by the method for rapidly detecting surface defects of inner contour of motor stator on line according to any one of claims 1 to 7, and the method comprises the following steps:
a carrying mechanism; the motor stator placing platform comprises a placing platform which is horizontally arranged and used for bearing a motor stator;
a detection mechanism; the device comprises a detection motion component and an image acquisition component, wherein the detection motion component is fixedly connected with the image acquisition component and is used for driving the image acquisition component to reciprocate in space; the image acquisition component comprises a plurality of image acquisition cameras and at least one centering camera, and the plurality of image acquisition cameras are horizontally and circumferentially arranged and are used for acquiring image parameters of the inner contour of the motor stator; the centering camera is opposite to the end face of the top end of the placing platform and is used for acquiring inner hole circular contour information of the motor stator;
a support; the bearing mechanism and the detection mechanism are arranged oppositely along the vertical direction and are fixedly installed on the bracket.
9. The device for rapidly detecting the defects on the inner contour surface of the motor stator as claimed in claim 8, wherein the carrying mechanism further comprises a carrier moving member disposed between the placing platform and the bracket for driving the placing platform to move in a horizontal plane.
10. The device for rapidly detecting the defects of the inner contour surface of the motor stator on the line as claimed in claim 8 or 9, wherein a positioning assembly is arranged on the placing platform for fixing the motor stator on the placing platform during the detection of the motor stator.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202211530821.7A CN115963397B (en) | 2022-12-01 | 2022-12-01 | Rapid online detection method and device for surface defects of inner contour of motor stator |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202211530821.7A CN115963397B (en) | 2022-12-01 | 2022-12-01 | Rapid online detection method and device for surface defects of inner contour of motor stator |
Publications (2)
Publication Number | Publication Date |
---|---|
CN115963397A true CN115963397A (en) | 2023-04-14 |
CN115963397B CN115963397B (en) | 2023-07-25 |
Family
ID=85898567
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202211530821.7A Active CN115963397B (en) | 2022-12-01 | 2022-12-01 | Rapid online detection method and device for surface defects of inner contour of motor stator |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN115963397B (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN117309889A (en) * | 2023-11-27 | 2023-12-29 | 宁德时代新能源科技股份有限公司 | Pole piece edge detection device and method |
Citations (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
KR101458426B1 (en) * | 2013-07-16 | 2014-11-07 | 동명대학교산학협력단 | Auto defects detection apparatus and method for testing inner side of cylinder tube |
CN111598846A (en) * | 2020-04-26 | 2020-08-28 | 鲲鹏通讯(昆山)有限公司 | Rail defect detection method in tunnel based on YOLO |
WO2021135331A1 (en) * | 2019-12-30 | 2021-07-08 | 歌尔股份有限公司 | Product defect detection method, apparatus and system |
US20210407062A1 (en) * | 2020-06-30 | 2021-12-30 | Beijing Baidu Netcom Science And Technology Co., Ltd. | Product defect detection method and apparatus, electronic device and storage medium |
CN114067197A (en) * | 2021-11-17 | 2022-02-18 | 河南大学 | Pipeline defect identification and positioning method based on target detection and binocular vision |
CN114283205A (en) * | 2021-12-21 | 2022-04-05 | 众信方智(苏州)智能技术有限公司 | Panoramic multi-view camera position calibration device and method |
CN114894808A (en) * | 2022-05-27 | 2022-08-12 | 华南理工大学 | Device and method for detecting defects of heat pipe orifice based on machine vision |
CN115201220A (en) * | 2022-08-17 | 2022-10-18 | 泰州市健新智造科技有限公司 | Wheel hub processing surface defect detecting system based on machine vision |
CN115272271A (en) * | 2022-08-09 | 2022-11-01 | 郑州芯视道机器人技术有限公司 | Pipeline defect detecting and positioning ranging system based on binocular stereo vision |
-
2022
- 2022-12-01 CN CN202211530821.7A patent/CN115963397B/en active Active
Patent Citations (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
KR101458426B1 (en) * | 2013-07-16 | 2014-11-07 | 동명대학교산학협력단 | Auto defects detection apparatus and method for testing inner side of cylinder tube |
WO2021135331A1 (en) * | 2019-12-30 | 2021-07-08 | 歌尔股份有限公司 | Product defect detection method, apparatus and system |
CN111598846A (en) * | 2020-04-26 | 2020-08-28 | 鲲鹏通讯(昆山)有限公司 | Rail defect detection method in tunnel based on YOLO |
US20210407062A1 (en) * | 2020-06-30 | 2021-12-30 | Beijing Baidu Netcom Science And Technology Co., Ltd. | Product defect detection method and apparatus, electronic device and storage medium |
CN114067197A (en) * | 2021-11-17 | 2022-02-18 | 河南大学 | Pipeline defect identification and positioning method based on target detection and binocular vision |
CN114283205A (en) * | 2021-12-21 | 2022-04-05 | 众信方智(苏州)智能技术有限公司 | Panoramic multi-view camera position calibration device and method |
CN114894808A (en) * | 2022-05-27 | 2022-08-12 | 华南理工大学 | Device and method for detecting defects of heat pipe orifice based on machine vision |
CN115272271A (en) * | 2022-08-09 | 2022-11-01 | 郑州芯视道机器人技术有限公司 | Pipeline defect detecting and positioning ranging system based on binocular stereo vision |
CN115201220A (en) * | 2022-08-17 | 2022-10-18 | 泰州市健新智造科技有限公司 | Wheel hub processing surface defect detecting system based on machine vision |
Non-Patent Citations (2)
Title |
---|
张凯 等: "基于轻量化YOLOv4的发电机定子表面缺陷检测算法", 《计算机与数字工程》, vol. 49, no. 4, pages 686 - 691 * |
王紫媛: "基于深度学习的敞车厢体缺陷识别与定位研究", 《万方学位论文》, pages 2 - 5 * |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN117309889A (en) * | 2023-11-27 | 2023-12-29 | 宁德时代新能源科技股份有限公司 | Pole piece edge detection device and method |
CN117309889B (en) * | 2023-11-27 | 2024-05-14 | 宁德时代新能源科技股份有限公司 | Pole piece edge detection device and method |
Also Published As
Publication number | Publication date |
---|---|
CN115963397B (en) | 2023-07-25 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN106949848B (en) | A kind of high-precision laser 3D profile phone structural detection method | |
CN106248686A (en) | Glass surface defects based on machine vision detection device and method | |
CN110501342B (en) | Cheese yarn rod positioning visual detection method | |
CN106767518A (en) | A kind of phone housing quality on-line detecting device and method based on line laser structured light | |
CN102353349B (en) | Machine vision based micro-sound film concentricity online testing system and testing method | |
CN108288288A (en) | Accurate shaft size measurement method, the device and system of view-based access control model identification | |
CN109211110A (en) | A kind of flat-type workpiece size detection system and method based on machine vision | |
CN110470247B (en) | Device and method for detecting coaxiality of inner and outer circular surfaces of part | |
CN109840900A (en) | A kind of line detection system for failure and detection method applied to intelligence manufacture workshop | |
CN115963397A (en) | Rapid online detection method and device for surface defects of inner contour of motor stator | |
CN111220544A (en) | Lens quality detection device and detection method | |
CN110953993A (en) | Detection device and method for sag and distance limit of power transmission line | |
CN212301356U (en) | Wheel hub welding seam visual detection device | |
CN116465315B (en) | Automatic screen quality detection method and system | |
CN116105604A (en) | Steel pipe quality detection system and detection method | |
CN115014248B (en) | Laser projection line identification and flatness judgment method | |
CN115931871A (en) | Device and method for detecting outer contour defects of permanent magnet motor rotor | |
CN109945792A (en) | A kind of diameter measuring method, device and the application method of high reflective cylindrical body | |
CN112881424A (en) | AI + fluorescence permeation small-sized pipe surface defect detection and quality grading method and system | |
CN109990711A (en) | A kind of appearance quality detection method of punched nickel-plated steel band | |
CN115984177A (en) | Machine vision detection device, control method thereof, control device, and storage medium | |
CN114858817A (en) | Detection device and method for protective lens of galvanometer | |
CN117788400A (en) | Binocular vision-based photovoltaic module defect detection method and system | |
CN110470250B (en) | Detection device and detection method for surface flatness of part | |
CN112378797A (en) | Full-automatic indentation measuring device and method |
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 |