Machine vision collaborative detection system for welding defects of energy storage system

Abstract

This paper discloses a machine vision collaborative inspection system for welding defects in energy storage systems, including a multi-view image acquisition module, an image preprocessing module, a defect identification and collaborative analysis module, a collaborative control module, and a result output and feedback module. The multi-view image acquisition module uses distributed industrial cameras to achieve omnidirectional, blind-spot-free image acquisition of the welded parts of the energy storage system. The image preprocessing module performs noise reduction, enhancement, stitching, and calibration on the acquired images to improve image quality. The defect identification and collaborative analysis module uses a combination of traditional machine vision algorithms and deep learning algorithms to accurately identify various welding defects and uses multi-view image fusion to locate and classify the severity of defects. The result output and feedback module outputs a defect detection report and feeds back abnormal signals to the production control system. This invention achieves automated, high-precision, and omnidirectional detection of welding defects in energy storage systems, improving detection efficiency and accuracy.

Description

Technical Field

[0001] This invention relates to a machine vision collaborative inspection system for welding defects in energy storage systems. Background Technology

[0002] Welding is a critical process in the production of energy storage systems, and its quality directly affects the structural strength, sealing performance, conductivity, and operational safety of these systems. During the welding process, various welding defects can easily occur due to factors such as welding process parameters, welding materials, and the on-site environment. These defects mainly include incomplete fusion, weak welds, shrinkage cavities, weld slag, spatter, overheating, and burrs. Among these, hidden defects such as weak welds and incomplete fusion can easily lead to open circuits, overheating, cracking, and even leaks and explosions during long-term operation of the energy storage system due to vibration and temperature changes. Defects such as weld slag and spatter can puncture the cell separator, causing short circuit risks and seriously threatening the operational safety of the energy storage system.

[0003] Currently, welding defect detection in energy storage systems mainly employs two methods: manual inspection and single machine vision inspection. Manual inspection relies on the experience of inspectors, resulting in low efficiency, high labor intensity, large inspection errors, and a high rate of missed defects. Furthermore, it is difficult to adapt to the demands of large-scale mass production and cannot accurately identify minute defects. In addition, existing inspection systems mostly operate independently and cannot coordinate with the process parameters of the energy storage welding equipment. This makes it difficult to optimize welding process parameters based on inspection results, leading to recurring welding defects and failing to improve welding quality from the source.

[0004] Meanwhile, the existing defect classification standards of the detection system are unclear, making it difficult to accurately assess the severity of defects, which is detrimental to subsequent defect handling and quality control. Therefore, developing a machine vision collaborative detection system for welding defects in energy storage systems that can achieve comprehensive, high-precision, and automated detection and can be coordinated with the welding process has become an urgent technical problem to be solved in the energy storage industry. Summary of the Invention

[0005] The purpose of this invention is to overcome the above-mentioned defects of the prior art and provide a machine vision collaborative inspection system for welding defects in energy storage systems. This system solves the technical problems of low detection efficiency, limited field of view, insufficient recognition accuracy, and inability to coordinate with the welding process. It enables automated, comprehensive, and high-precision inspection of welding defects in energy storage systems, thereby improving welding quality and the operational safety of energy storage systems.

[0006] A machine vision collaborative detection system for welding defects in an energy storage system includes a multi-view image acquisition module, an image preprocessing module, a defect identification and collaborative analysis module, a collaborative control module, and a result output and feedback module.

[0007] The multi-view image acquisition module is used to acquire multi-view images of the welding parts of the energy storage system, including the energy storage cell terminal, the battery module housing, and the tab connection part.

[0008] The image preprocessing module is used to preprocess the acquired multi-view images. The preprocessing steps include image denoising, contrast enhancement, geometric calibration, and image stitching.

[0009] The defect identification and collaborative analysis module is used to identify and collaboratively analyze defects in the preprocessed image to classify the severity of defects.

[0010] The collaborative control module is used to realize the synchronous linkage of each module, control the acquisition angle, frame rate and supplementary light intensity of multiple cameras, adjust the preprocessing algorithm parameters and defect identification algorithm threshold, and communicate with the control system of the welding equipment to obtain welding process parameters.

[0011] The result output and feedback module is used to receive the analysis results of the defect identification and collaborative analysis module, output the defect type, defect location, severity and detection time data, and issue an alarm signal for unqualified defects.

[0012] Furthermore, the multi-view image acquisition module uses a distributed industrial CCD camera, equipped with a supplementary lighting unit and a protective housing, and achieves synchronous exposure of multiple cameras through hardware triggering to ensure the consistency of image acquisition.

[0013] Furthermore, the image preprocessing module employs an algorithm combining Gaussian filtering and median filtering for image denoising to remove image noise caused by dust and spatter at the welding site; contrast enhancement uses an adaptive histogram equalization algorithm to improve the grayscale difference between welding defects and the background.

[0014] Furthermore, the collaborative control module adopts a PLC controller and an industrial control computer for collaborative control. The PLC controller is used to control the synchronous triggering of multiple cameras and the adjustment of the supplementary lighting unit, while the industrial control computer is used to run image preprocessing and defect recognition algorithms, realizing real-time data interaction and dynamic parameter adjustment of each module, with a response time of no more than 10ms.

[0015] Furthermore, the result output and feedback module supports multiple output methods, including real-time display on the display terminal, printing of inspection reports by printer, and uploading of inspection data to the MES system via Ethernet. The alarm signal adopts both audible and visual alarms and system pop-up alarms, and different alarm levels are set according to the severity of the defects.

[0016] Furthermore, a calibration module is included, which communicates with the collaborative control module to periodically calibrate the camera of the multi-view image acquisition module, including intrinsic and extrinsic parameter calibration, to ensure the accuracy of image acquisition. The calibration cycle can be set to once every 2000 inspection stations, depending on the field usage. Periodic intrinsic and extrinsic parameter calibration can eliminate acquisition errors caused by camera installation deviations and operational vibrations. Fixed-cycle calibration balances inspection accuracy and production efficiency, continuously ensuring the system's long-term stable high-precision image acquisition capability.

[0017] Furthermore, the severity of the defects is graded into four levels: minor defects, general defects, severe defects, and fatal defects. The grading criteria are determined based on the correlation between defect area, defect depth, defect location, and the operational safety of the energy storage system. This quantitative grading standard can objectively assess the defect risk level, providing a clear basis for product repair, isolation, and disposal, enabling differentiated defect handling, and reducing potential safety hazards in the energy storage system.

[0018] Beneficial effects:

[0019] By using multi-view collaborative acquisition, precise image preprocessing, intelligent defect recognition, and process linkage control, we can achieve comprehensive, automated, and high-precision detection of welding defects in energy storage systems, significantly reducing the missed detection rate and improving detection efficiency and recognition accuracy. We can also optimize welding processes and implement graded defect management to enhance the system's environmental adaptability and operational stability, ensuring the welding quality and operational safety of energy storage systems from the source. Detailed Implementation

[0020] To enhance understanding of the present invention, the present invention will be further described in detail below with reference to embodiments and accompanying drawings. These embodiments are only used to explain the present invention and do not constitute a limitation on the scope of protection of the present invention.

[0021] Example 1: A machine vision collaborative inspection system for welding defects in energy storage systems, comprising a multi-view image acquisition module, an image preprocessing module, a defect identification and collaborative analysis module, a collaborative control module, a result output and feedback module, and a calibration module. These modules work together to complete the entire automated inspection process for welding defects in energy storage systems.

[0022] The system's multi-view image acquisition module enables multi-view image acquisition of energy storage cell terminals, battery module housings, and electrode connection points. It employs distributed industrial CCD cameras paired with supplementary lighting units and protective housings, and uses hardware triggering to achieve simultaneous exposure of multiple cameras to ensure consistent acquisition.

[0023] The image preprocessing module sequentially performs image denoising, contrast enhancement, geometric calibration, and image stitching on the acquired images. Denoising uses a combination of Gaussian filtering and median filtering to remove dust and splash noise. Contrast enhancement uses an adaptive histogram equalization algorithm to improve the grayscale difference between defects and the background.

[0024] The defect identification and collaborative analysis module can complete defect identification, collaborative analysis and severity classification of preprocessed images. Defects are classified into four levels: minor, general, severe and fatal. The classification criteria are determined by combining the defect area, depth, location and the degree of correlation with the operational safety of the energy storage system.

[0025] The collaborative control module adopts a collaborative architecture of PLC controller and industrial control computer to realize synchronous linkage of various modules, control camera acquisition parameters, preprocessing and recognition algorithm parameters, and communicate with the welding equipment control system to obtain process parameters. The PLC is responsible for camera triggering and supplementary lighting adjustment, and the industrial control computer is responsible for algorithm operation. The overall response time is no more than 10ms, which can realize real-time data interaction and dynamic parameter adjustment.

[0026] The results output and feedback module receives defect analysis results and outputs defect type, location, severity, and detection time data. It supports multiple output methods, including terminal display, report printing, and MES system upload. It employs dual alarms of sound and light and system pop-up windows, and sets alarm prompts according to defect level. The calibration module communicates with the collaborative control module to periodically calibrate the camera's internal and external parameters to ensure acquisition accuracy. The calibration cycle can be set to once every 2000 detection stations, depending on the site conditions.

[0027] System initialization and calibration

[0028] After the system is powered on, the collaborative control module first retrieves the calibration data from the calibration module. If the current detection station count reaches 2000, the calibration module automatically starts and performs comprehensive calibration of the internal and external parameters of the industrial CCD camera in the multi-view image acquisition module. After calibration, the parameters are synchronized to the collaborative control module to ensure that the camera acquisition accuracy meets the standard. If the calibration cycle has not been reached, the system directly enters the waiting-for-inspection state.

[0029] Multi-view image synchronous acquisition

[0030] Once the energy storage system components (cell terminals / module housings / tab connection points) are transported to the testing station and positioned via the production line, the PLC controller of the collaborative control module issues a trigger command. This triggers the distributed industrial CCD camera to synchronously expose via hardware. The supplementary lighting unit automatically activates to adapt to the light intensity, and the protective casing isolates the welding area from dust and splashes. This process enables multi-angle, blind-spot-free image acquisition of the welding area and transmits the original images to the image preprocessing module in real time.

[0031] The process is as follows:

[0032] S1. After receiving the original multi-view images, the image preprocessing module performs the following preprocessing operations in sequence: first, it uses a combination of Gaussian filtering and median filtering to remove dust and splash noise; then, it uses an adaptive histogram equalization algorithm to improve the grayscale difference between defects and the background; subsequently, it performs geometric calibration to correct distortion; and finally, it completes the multi-view image stitching to generate a high-quality, distortion-free, and feature-clear preprocessed image, which is then transmitted to the defect recognition and collaborative analysis module.

[0033] S2, the defect identification and collaborative analysis module performs intelligent identification and multi-view collaborative analysis on the pre-processed image to accurately determine the defect type; at the same time, based on the defect area, depth, location and the degree of impact on the safety of the energy storage system, the defect is divided into four levels: minor, general, serious and fatal, and the analysis results such as defect type, location, severity and detection time are synchronously transmitted to the collaborative control module and the result output and feedback module.

[0034] After receiving the defect analysis results, the S3 collaborative control module uses an industrial control computer to retrieve the process parameters of the welding equipment in real time. Based on the defect type and level, it dynamically adjusts the image preprocessing algorithm parameters and the defect recognition algorithm threshold. The PLC controller synchronously optimizes the camera acquisition angle, frame rate, and supplementary light intensity to achieve real-time linkage between hardware and algorithms. The entire control process has a response time of ≤10ms, ensuring that the system is always in the optimal detection state and providing data support for welding process optimization.

[0035] S4. After receiving the defect analysis data, the result output and feedback module displays the test results in real time through the display terminal. Test reports can be printed as needed, and the data can be uploaded to the MES system via Ethernet to achieve full-process traceability of quality data. If unqualified defects are detected, the system immediately triggers the corresponding alarm according to the defect level: minor defects are only recorded and prompted, general defects are reminded by pop-up window, and serious / fatal defects trigger a dual alarm of sound and light + pop-up window to prevent defective workpieces from flowing into the next process and ensure production quality.

[0036] S5. After completing the inspection of a single workpiece, the system automatically resets the station count and enters the next round of inspection process; after every 2000 workpieces are inspected, the calibration module automatically performs camera calibration to continuously ensure the long-term, stable and high-precision operation of the system.

[0037] This embodiment achieves comprehensive acquisition, high-precision identification, automated detection, real-time control, and visual feedback of welding defects in energy storage systems through the coordination and operation of the above modules.

[0038] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A machine vision collaborative inspection system for welding defects in energy storage systems, characterized in that, It includes a multi-view image acquisition module, an image preprocessing module, a defect identification and collaborative analysis module, a collaborative control module, and a result output and feedback module; The multi-view image acquisition module is used to acquire multi-view images of the welding parts of the energy storage system, including the energy storage cell terminal, the battery module housing, and the tab connection part. The image preprocessing module is used to preprocess the acquired multi-view images. The preprocessing steps include image denoising, contrast enhancement, geometric calibration, and image stitching. The defect identification and collaborative analysis module is used to identify and collaboratively analyze defects in the preprocessed image to classify the severity of defects. The collaborative control module is used to realize the synchronous linkage of each module, control the acquisition angle, frame rate and supplementary light intensity of multiple cameras, adjust the preprocessing algorithm parameters and defect identification algorithm threshold, and communicate with the control system of the welding equipment to obtain welding process parameters. The result output and feedback module is used to receive the analysis results of the defect identification and collaborative analysis module, output the defect type, defect location, severity and detection time data, and issue an alarm signal for unqualified defects.

2. The machine vision collaborative detection system for welding defects in energy storage systems according to claim 1, characterized in that, The multi-view image acquisition module uses distributed industrial CCD cameras, equipped with a supplementary lighting unit and a protective shell. It achieves synchronous exposure of multiple cameras through hardware triggering to ensure the consistency of image acquisition.

3. The machine vision collaborative detection system for welding defects in energy storage systems according to claim 2, characterized in that, The image preprocessing module employs an algorithm combining Gaussian filtering and median filtering for image denoising to remove image noise caused by dust and spatter at the welding site; contrast enhancement uses an adaptive histogram equalization algorithm to improve the grayscale difference between welding defects and the background.

4. The machine vision collaborative inspection system for welding defects in energy storage systems according to claim 1, characterized in that, The collaborative control module adopts a PLC controller and an industrial control computer for collaborative control. The PLC controller is used to control the synchronous triggering of multiple cameras and the adjustment of the supplementary lighting unit, while the industrial control computer is used to run image preprocessing and defect recognition algorithms, realizing real-time data interaction and dynamic parameter adjustment of each module, with a response time of no more than 10ms.

5. The machine vision collaborative inspection system for welding defects in energy storage systems according to claim 1, characterized in that, The result output and feedback module supports multiple output methods, including real-time display on the display terminal, printing of test reports by printer, and uploading of test data to the MES system via Ethernet. The alarm signal adopts both audible and visual alarm and system pop-up alarm, and different alarm levels can be set according to the severity of the defect.

6. The machine vision collaborative detection system for welding defects in energy storage systems according to claim 1, characterized in that, It also includes a calibration module, which is connected in communication with the collaborative control module, and is used to periodically calibrate the camera of the multi-view image acquisition module, including internal parameter calibration and external parameter calibration, to ensure the accuracy of image acquisition. The calibration cycle can be set to once every 2000 test stations according to the field usage.

7. The machine vision collaborative inspection system for welding defects in energy storage systems according to claim 1, characterized in that, The severity of the defects is classified into four levels: minor defects, general defects, severe defects, and fatal defects. The classification criteria are determined based on the correlation between the defect area, defect depth, defect location, and the operational safety of the energy storage system.