Motor stator defect detection method based on three-dimensional point cloud

Abstract

The invention discloses a motor stator defect detection method based on a three-dimensional point cloud, and the method comprises the steps: carrying out the scanning collection of a motor stator from different angles through an image collection device, so as to obtain a plurality of pieces of image data of the motor stator and corresponding depth information; reconstructing three-dimensional point cloud data of the motor stator based on the image data and the depth information of the plurality of motor stators; and inputting the three-dimensional point cloud data of the motor stator into a pre-trained defect detection model, and judging the type and position of the defect through the defect prediction model. Through the method, automatic detection of motor stator defects can be realized, and the detection accuracy is high.

Description

Technical Field

[0001] The present invention relates to the technical field of motor detection, and in particular to a motor stator defect detection method and system based on three-dimensional point cloud. Background Art

[0002] With the widespread popularity of new energy vehicles, research on new energy vehicles has been highly valued by scientific research institutions and companies in various countries. As one of the three core systems of new energy vehicles, the drive motor provides the main driving force for the vehicle. Its characteristics determine the main performance indicators of the vehicle and directly affect the vehicle's power, economy and user driving experience.

[0003] The motor stator and rotor are important components of the motor, and they work together to realize the operation of the motor. Among them, the motor stator is a fixed part of the motor, usually composed of an iron core, windings, and a frame, and its main function is to generate a magnetic field. The iron core in the motor stator is made of laminated silicon steel sheets with good magnetic conductivity. Its function is to form the main magnetic circuit of the motor and fix the windings at the same time. The iron core is generally composed of two layers of inner iron and outer iron, with the windings on the inner iron and the end caps on the outer iron. The winding is another important component of the motor stator, which is wound with insulated wires in a certain shape and manner. The function of the winding is to generate a magnetic field after power is turned on, realizing the conversion of electrical energy into mechanical energy. According to the number of phases of the power supply, the winding can be divided into single-phase windings and three-phase windings.

[0004] As an important component of the motor, the quality of the motor stator directly affects the performance and life of the motor. However, traditional motor stator defect detection methods mostly rely on manual inspection, which is not only inefficient, but also easily affected by the subjective factors of the inspector himself, resulting in low accuracy of the detection results, which seriously affects the motor quality inspection process. Therefore, it is necessary to provide an efficient and accurate motor stator defect detection method. Summary of the invention

[0005] The purpose of the present invention is to provide a motor stator defect detection method based on three-dimensional point cloud to solve the problems of low efficiency and poor accuracy of traditional detection methods with high degree of automation.

[0006] To achieve the above object, the present invention adopts the following technical solutions:

[0007] The motor stator is scanned and collected from different angles by an image acquisition device to obtain multiple image data of the motor stator and corresponding depth information; based on the multiple image data of the motor stator and the depth information, the three-dimensional point cloud data of the motor stator is reconstructed; the three-dimensional point cloud data of the motor stator is input into a pre-trained defect detection model, and the type and location of the defect are determined by the defect prediction model; and the detection is ended.

[0008] Furthermore, the multiple images of the motor stator may be multiple images acquired at the same scanning angle from multiple scanning positions around the motor stator to be tested, or multiple images acquired at different scanning angles from multiple scanning positions around the motor stator to be tested.

[0009] Furthermore, the image acquisition device includes an array camera, the overall angle of the array camera is adjustable, and each camera has a different shooting angle, so that multiple appearance images of the motor stator are obtained in one shooting.

[0010] Furthermore, the defect detection model is obtained by training a neural network model.

[0011] Furthermore, the defect type may be cracks, scratches, bubbles, stains, delamination, etc.

[0012] Furthermore, the acquired image data is preprocessed, and the resolution of the preprocessed image data is enhanced.

[0013] Furthermore, the resolution enhanced image is convolved to obtain a first image feature, and a plurality of cascaded feature extraction units are set. The feature extraction units perform convolution processing, and the input of the first-level feature extraction unit is the first image feature. The input of each subsequent level of feature extraction unit is the first image feature and the output of each previous level of feature extraction unit. The output of the last level of feature extraction unit is fused with the first image feature to obtain image data with higher resolution.

[0014] Furthermore, the three-dimensional point cloud data is input into the first defect detection model to determine whether a defect exists; if a defect exists, the three-dimensional point cloud data is further input into the second defect detection model to determine the type and location of the defect. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0016] Figure 1 It is a flow chart of an embodiment of a motor stator defect detection method based on three-dimensional point cloud of the present application;

[0017] Figure 2 It is a flowchart of an embodiment of processing image data before reconstructing a three-dimensional point cloud in the present application;

[0018] Figure 3This is a schematic diagram of a process of processing data collected by an array camera according to an embodiment of the present application;

[0019] Figure 4 It is a flow chart of an embodiment of the defect detection method of the present application. DETAILED DESCRIPTION

[0020] The technical scheme in the embodiment of the present invention will be clearly described below in conjunction with the accompanying drawings in the embodiment of the present invention. Obviously, the described embodiment is only a part of the embodiment of the present invention, rather than all the embodiments. Based on the embodiment of the present invention, all other embodiments obtained by those skilled in the art without making creative work belong to the scope of protection of this application.

[0021] See also Figure 1 , Figure 1 1 is a flow chart of an embodiment of a motor stator defect detection method based on three-dimensional point cloud of the present application. As shown in the figure, the method includes:

[0022] Step S1: Scan and capture the motor stator from different angles using an image acquisition device to obtain multiple images of the motor stator and corresponding depth information.

[0023] Specifically, the above-mentioned multiple images of the motor stator image data can be multiple images obtained from multiple scanning positions around the motor stator to be tested at the same scanning angle, or can be multiple images obtained from multiple scanning positions around the motor stator to be tested at different scanning angles, so as to completely cover the appearance details of the motor stator to be tested.

[0024] Preferably, the image acquisition device may include an array camera, the overall angle of the array camera is adjustable, and each camera has a different shooting angle. Multiple appearance images of the motor stator can be obtained in one shooting, which can not only improve the acquisition efficiency but also obtain image data from more angles, which is beneficial to fully reproduce the three-dimensional appearance of the motor stator to be tested when reconstructing the three-dimensional point cloud.

[0025] Step S2: reconstructing the three-dimensional point cloud data of the motor stator based on the multiple images of the motor stator and the depth information.

[0026] The image data provides the two-dimensional coordinates of the pixels, namely the x and y values ​​in the XY coordinate system, while the depth information provides the Z coordinate in the camera coordinate system. Based on the intrinsic and extrinsic parameters of the camera, the coordinates of any pixel point in the world coordinate system can be calculated, thereby reconstructing the three-dimensional point cloud data model of the motor stator.

[0027] Step S3: input the three-dimensional point cloud data of the motor stator into a pre-trained defect detection model, and determine the type and location of the defect through the defect prediction model.

[0028] The defect detection model is obtained by training a neural network model, and its input is the three-dimensional point cloud data of the motor stator. By directly processing the three-dimensional point cloud data, the complex operations of projecting, registering, and performing 2D-3D mapping of the three-dimensional point cloud data, as well as the data errors and information loss caused by the projection and mapping process, are avoided.

[0029] Preferably, the defect types can be divided into cracks, scratches, bubbles, stains, delamination, etc.

[0030] Step S4: End the detection.

[0031] Since image data affects the accuracy of subsequent 3D point cloud data reconstruction, and the acquisition of image data may be affected by environmental interference and equipment, resulting in noise or unclear image data, the image can also be processed before 3D point cloud data reconstruction.

[0032] Please refer to Figure 2 , Figure 2 It is a flowchart of an embodiment of processing image data before reconstructing a three-dimensional point cloud in the present application.

[0033] Specifically, in some embodiments, step S2 may include:

[0034] Step S21: pre-processing the acquired image.

[0035] Preprocessing operations can be denoising, cropping, contrast enhancement, histogram equalization, etc.

[0036] Step S22: Perform resolution enhancement on the pre-processed image.

[0037] Specifically, the preprocessed image is convolved to obtain a first image feature, a plurality of cascaded feature extraction units are set, and the feature extraction units perform convolution processing, the input of the first-level feature extraction unit is the first image feature, and the input of each subsequent level of feature extraction unit is the first image feature and the output of each previous level of feature extraction unit, and the output of the last level of feature extraction unit is fused with the first image feature, so as to obtain image data with higher resolution. Each level of feature extraction unit includes sequentially connected convolution layers and residual layers.

[0038] The specific calculation process of resolution enhancement is as follows: assuming that the convolution calculation is recorded as f, the activation function is expressed as t, R i Represents the residual block of the i-th level, recorded as:

[0039]

[0040] Among them, H irepresents the output of the i-th level residual block, Represents a parameter set, represents the parameters of the jth convolutional layer in the i-th level residual block. The output feature of the final residual block is represented as H u , expressed as:

[0041]

[0042] For the first-level residual block, its input is f(X; W c ), where X is the original image data, W c is the convolution parameter for convolving the original image.

[0043] The j-th residual block of the i-th level feature extraction unit is defined as represents the parameter set of the local cascade block at level i, then the local cascade block at level i is represented as:

[0044]

[0045] Among them, B i,U Defined as recursive callable, that is:

[0046] B i,0 =H i-1

[0047]

[0048] Through the above method, the final cascade block H can be obtained u Output features:

[0049] H 0 =f(X;W c )

[0050]

[0051] Among them, B is the number of cascaded feature extraction units.

[0052] Step S23: Reconstructing a three-dimensional point cloud based on the image data after resolution enhancement.

[0053] Specifically, the 3D point cloud reconstruction process is:

[0054] Feature points are extracted from each image, and then feature point matching is performed between the collected images based on the feature points to determine the position of the same feature point in each image, and the pixel value of the feature point and the corresponding position coordinates are recorded.

[0055] Based on the camera intrinsic parameter matrix and pixel depth, the pixel coordinates in the two-dimensional image are converted into point cloud coordinates. The calculation formula is as follows:

[0056]

[0057] Among them, P(u,v) is the point cloud coordinate of pixel point P, u and v are the pixel coordinates of pixel point P in the two-dimensional image, D(u,v) is the depth information of the corresponding pixel point P, K is the intrinsic parameter matrix of the camera, is the coordinate matrix.

[0058] The three-dimensional space coordinate information of the feature points is obtained from the recorded pixel values ​​and corresponding position coordinates.

[0059] Based on the feature points, the existing PCL point cloud fusion technology is used to fuse the point cloud of multi-angle data, eliminate redundant and repeated point cloud data, and reconstruct the final point cloud model.

[0060] Preferably, when the image acquisition device is an array camera, step S23 may further include:

[0061] S231: Identify feature points of each of n images, select an image as a reference image, and perform feature point matching based on the reference image, wherein the n images are acquired by n cameras in the array camera respectively, and the feature points are pixel points that can indicate a certain feature of the motor stator.

[0062] S232: Calculate the distance differences between the feature points of the remaining n-1 images and the reference image respectively, and when the distance difference is greater than a threshold, remove the corresponding image.

[0063] When the distance difference between the matched feature points is large, it indicates that the image is very different from the reference image due to noise or other interference factors and is not suitable as data for point cloud reconstruction.

[0064] In a preferred embodiment, step S3 may include the following steps: Figure 4 As shown:

[0065] S31: inputting the three-dimensional point cloud data into the first defect detection model to determine whether there is a defect;

[0066] S32: If there is a defect, continue to input the three-dimensional point cloud data into the second defect detection model to determine the type and location of the defect.

[0067] In this embodiment, the defect detection model is composed of a first defect detection model and a second defect detection model. The first defect detection model is a simple two-classification detection model that can quickly screen out defective data. Only point cloud data with defects is input into the second defect detection model to determine the defect type and location. Since the complexity of the convolutional neural network is closely related to computational efficiency, the binary classification model has a simpler network structure and faster calculation speed than the multi-classification model. If there is only one defect detection model, data without defects also needs to be judged by multiple classifications, which affects the detection efficiency. By setting a cascaded first defect detection model and a second defect detection model, this embodiment can quickly screen out defective point cloud data, and only perform multi-classification judgments on defective point cloud data, which greatly improves the calculation speed.

[0068] The specific training process of the defect detection model is as follows:

[0069] Step 1: Obtain a labeled training set, where the label of the first defect detection model training set is whether there is a defect, and the label of the second defect detection model training set is the defect type and location information. In order to improve the comprehensiveness and accuracy of the defect detection model, the training set can include training data of multiple types of defects and all possible locations where defects may appear.

[0070] Step 2: Input the training set into the defect detection model to be trained for training. For the first defect detection model, set the false detection rate threshold. When the false detection rate does not exceed the threshold, determine whether the missed detection rate of the training model is 0. If the missed detection rate is 0, end the training of the first defect detection model. If the missed detection rate is not 0, continue to adjust the parameters in the training model. For the second defect detection model, determine whether the model training has converged by calculating whether the loss function is less than the threshold. The loss function can be any loss function in the prior art, such as the cross entropy loss function, the logarithmic loss function, etc.

[0071] Among them, the false detection rate refers to the proportion of samples that are mistakenly judged as defective among samples that are actually not defective. A too high false detection rate means that the model will mistakenly judge normal motor stators as defective. The missed detection rate refers to the proportion of samples that are mistakenly judged as non-defective among samples that actually have defects. By setting the false detection rate not to exceed a certain threshold, the proportion of normal motor stators judged as defective stators can be reduced, and setting the missed detection rate to 0 can ensure that all defective motor stators can be detected. Through the above settings, the effectiveness and accuracy of the defect detection model can be improved.

[0072] The motor stator defect detection method based on three-dimensional point cloud in the present invention scans and collects the motor stator from different angles through an image acquisition device to obtain multiple images of the motor stator and corresponding depth information. Based on the multiple images and depth information of the motor stator, the three-dimensional point cloud data of the motor stator is reconstructed, and the three-dimensional point cloud data of the motor stator is input into a pre-trained defect detection model, and the type and position of the defect are judged by the defect prediction model. Through this method, the automatic detection of motor stator defects can be realized, and the detection accuracy is high. In addition, by processing the collected stator image to improve the imaging quality, it is beneficial to the reconstruction quality of the three-dimensional point cloud. At the same time, a cascaded secondary defect detection model is adopted to first determine whether the stator has defects, and only when there are defects, the defect type and location are judged and detected, so that the point cloud data with defects can be quickly screened out, and only the point cloud data with defects are multi-classified, which greatly improves the calculation speed and has good economic benefits and promotion prospects.

[0073] Finally, it should be noted that those skilled in the art will easily understand that the embodiments described above are only preferred embodiments of the present invention, which are used to illustrate the technical solutions of the present application rather than to limit them. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A motor stator defect detection method based on three-dimensional point cloud, characterized in that: The method comprises: Scanning and collecting the motor stator from different angles by using an image acquisition device to obtain multiple images of the motor stator and corresponding depth information; Reconstruct the three-dimensional point cloud data of the motor stator based on multiple images and depth information of the motor stator; Inputting the three-dimensional point cloud data of the motor stator into a pre-trained defect detection model, and determining the type and location of the defect through the defect prediction model; End detection.

2. The method according to claim 1, characterized in that The multiple motor stator image data are multiple images acquired from multiple scanning positions around the motor stator to be tested at the same scanning angle, or multiple images acquired from multiple scanning positions around the motor stator to be tested at different scanning angles.

3. The method according to claim 2, characterized in that The image acquisition device includes an array camera, the overall angle of the array camera is adjustable, and each camera has a different shooting angle, so that multiple appearance images of the motor stator are obtained in one shooting.

4. The method according to claim 1, characterized in that: The defect detection model is obtained by training the neural network model.

5. The method according to claim 1, characterized in that Defect types include cracks, scratches, bubbles, stains, delamination, etc.

6. The method according to claim 1, characterized in that The acquired image data is preprocessed, and the resolution of the preprocessed image data is enhanced.

7. The method according to claim 5, characterized in that The resolution-enhanced image data is convolved to obtain a first image feature. A plurality of cascaded feature extraction units are set. The feature extraction units perform convolution processing. The input of the first-level feature extraction unit is the first image feature. The input of each subsequent level of feature extraction unit is the first image feature and the output of each previous level of feature extraction unit. The output of the last level of feature extraction unit is fused with the first image feature to obtain image data with higher resolution.

8. The method according to claim 1, characterized in that The three-dimensional point cloud data is input into the first defect detection model to determine whether there is a defect; if there is a defect, the three-dimensional point cloud data is further input into the second defect detection model to determine the type and location of the defect.

Images