CN-121998887-A - Method for detecting defects of thick edges of battery pole piece coating

Abstract

The invention provides a method for detecting a coating thick edge defect of a battery pole piece, which belongs to the technical field of power battery manufacturing and comprises the steps of obtaining three-dimensional point cloud data of the battery pole piece after a coating drying process, preprocessing original point cloud data, extracting a coating edge region, calculating and extracting point cloud characteristics of the point cloud data of the coating edge region, inputting the point cloud characteristics into a pre-trained coating thick edge defect detection model, and judging whether the region has the coating thick edge defect. Compared with the traditional two-dimensional defect detection method, the method can detect the abnormal thickening defect of the coating edge area of the polar plate, and can solve the problem that the two-dimensional detection cannot effectively identify the coating thick edge defect. The method is simple and quick, has higher accuracy compared with the traditional method, and can be suitable for different lithium ion battery production lines.

Inventors

• ZHENG YUEJIU
• CHEN TIANXIN
• LAI XIN
• ZHAO WENZHENG
• LIU YINHUA
• CHEN FEI

Assignees

• 上海理工大学

Dates

Publication Date

20260508

Application Date

20241101

Claims (8)

1. 1. The method for detecting the thick edge defect of the battery pole piece coating is characterized by comprising the following steps: S1, acquiring three-dimensional point cloud data of a battery pole piece after a coating and drying process; s2, preprocessing the three-dimensional point cloud data, and extracting a coating edge area; s3, calculating point cloud data of the coating edge area, and extracting point cloud characteristics; S4, inputting the extracted point cloud characteristics into a pre-trained coating thick edge defect detection model, and judging whether the coating thick edge defect exists in the area according to output.
2. 2. The method for detecting the thick edge defect of the battery pole piece coating according to claim 1, wherein in the step S1, three-dimensional point cloud data of the battery pole piece after the coating and drying is obtained through scanning by a distance measuring and depth acquiring device, wherein the data comprises x, y and z coordinates of a pole piece current collector area and a coating area, and the advancing direction of the pole piece is parallel to a y axis.
3. 3. The method for detecting the thick edge defect of the battery pole piece coating according to claim 1, wherein in S2, the preprocessing of the three-dimensional point cloud data includes a point cloud space transformation and a point cloud filtering operation for removing scanning errors of the ranging and depth acquisition device.
4. 4. The method for detecting the thick edge defect of the battery pole piece coating according to claim 1, wherein in S2, the step of extracting the coating edge region specifically comprises the following steps: S21, acquiring a contour curve of a cross section at any position through pole piece point cloud data; S22, distinguishing a current collector region and a coating region through a contour curve, and marking the junction between the current collector region and the coating region as x 0 ; And S23, setting the width of the coating edge in the x direction as w, and taking the point cloud data in the x coordinate range of the coating edge area as { x 0 ,x 0 +w } or { x 0 -w,x 0 }, as the coating edge area.
5. 5. The method for detecting a thick edge defect of a battery pole piece coating according to claim 1, wherein in S3, the point cloud features include a point cloud coordinate, a point cloud normal vector, a point cloud curvature, a distance from a fitted surface to an actual surface, the number of points outside the fitted surface, and an angle between a normal vector of a fitted plane and an actual normal vector, and each region extracts a plurality of features as a set of data representing the region.
6. 6. The method for detecting a coated bead defect of a battery pole piece according to claim 1, wherein in S4, the training of the coated bead defect detection model comprises the steps of: S41, acquiring point cloud characteristic data of a plurality of groups of coating edge areas, and labeling areas represented by each group of data; S42, preprocessing the obtained data set, and dividing the data set into a training set, a verification set and a test set; s43, designing a neural network model, taking the point cloud characteristics of each group of data as the input of the model, taking the numerical value corresponding to the label as the output of the model, and continuously training the neural network model to finally obtain the coating thick edge defect detection model.
7. 7. The method for detecting thick edge defects of battery pole piece coating according to claim 6, wherein in the step S41, the label is specifically marked as 0 for normal coating and 1 for thick edge defects.
8. 8. The method for detecting a thick edge defect of a battery pole piece coating according to claim 6, wherein in S42, the training set, the verification set and the test set are divided according to a ratio of 6:2:2.

Description

Method for detecting defects of thick edges of battery pole piece coating Technical Field The invention belongs to the technical field of power battery manufacturing, and particularly relates to a method for detecting a thick edge defect of a battery pole piece coating. Background With the rapid development of new energy automobiles and electrochemical energy storage technologies, the market demand of lithium ion batteries has increased dramatically. In order to meet the market demand for high-performance lithium ion batteries, battery manufacturers continue to expand production lines and improve production efficiency and product quality. In the production process of lithium ion batteries, the coating process is one of the key steps, which directly affects the performance and life of the battery. The bead defects in the coating process are common quality problems and need to be identified and controlled by effective detection means. The coating thick edge defect refers to a phenomenon that the coating thickness of the edge part of the pole piece is abnormally thickened in the coating process. Such defects may not only affect the external appearance quality of the battery, but also may cause non-uniform electrochemical reactions inside the battery, affecting the capacity and cycle life of the battery. The traditional detection method, such as using an area density meter to perform Z-type scanning, has the problems of incomplete coverage and radiation hazard. While 2D machine vision detection based on a CCD camera can provide intuitive image information, the detection capability of the coating bead is limited, especially in the case of small gray scale differences on the coating surface. Therefore, the current lithium ion battery production line lacks a method capable of effectively detecting the coating bead defects, which limits the improvement of the quality and the safety of the battery. With the increase of market demands, the requirements on battery performance are also higher and higher, and the conventional detection method cannot meet the requirements of high-precision detection. Therefore, in order to improve the reliability and safety of the battery, development of a new detection technology is urgently needed to overcome the limitations of the prior art and realize rapid and accurate detection of the coating bead defects. Disclosure of Invention The invention aims to provide a method for detecting a thick edge defect of a battery pole piece coating, which is characterized by comprising the following steps: And 1, acquiring three-dimensional point cloud data of the battery pole piece after a coating and drying process. Three-dimensional point cloud data of the battery pole piece after coating and drying is obtained through scanning of ranging and depth acquisition equipment such as a line laser camera, a structured light camera and a time flight camera, and the data comprises, but is not limited to, x, y and z coordinates of a pole piece current collector area and a coating area. Wherein, the direction of pole piece advance is parallel with the Y axle. And 2, preprocessing the three-dimensional point cloud data, and extracting a coating edge area. Preprocessing operations on three-dimensional point cloud data include, but are not limited to, point cloud filtering, point cloud cropping, point cloud space transformation, and the like. Extracting the coated edge region comprises the steps of: S21, acquiring a contour curve of a cross section at any position through pole piece point cloud data; Step S22, fluid and coating areas are separated through a contour curve area, and the junction between the two areas is marked as x 0; in step S23, if the width of the coating edge in the x direction is set to be w, the x coordinate range of the coating edge region is { x 0,x0 +w } or { x 0-w,x0 }, and the point cloud data in this range is taken as the coating edge region. And 3, calculating the point cloud data of the coating edge area, and extracting the point cloud characteristics. The point cloud features include, but are not limited to, point cloud coordinates, point cloud normal vectors, point cloud curvature, distance of the fitted surface to the actual surface, number of points outside the fitted surface, angles between the fitted plane normal vector and the actual surface normal vector, and the like. Each region extracts features as a set of data representing the region. And 4, inputting the extracted point cloud characteristics into a pre-trained coating thick edge defect detection model, and judging whether the coating thick edge defect exists in the area according to output. The training process of the coating thick edge defect detection model comprises the following steps: Step 41, acquiring a plurality of groups of point cloud characteristic data of the coating edge areas, and labeling the areas represented by each group of data; step 42, preprocessing the data set, and dividing the data set into a