A car body welding condition detection system and detection method based on artificial intelligence

Abstract

The present invention aims to provide a body welding condition detection system and detection method based on artificial intelligence, the system includes the following modules: camera module, configuration expansion module, production line PLC, information collection module, image acquisition module, intelligent detection module, control module. The present invention can quickly build a detection model according to different production models, can meet the automatic switching recognition model of multiple models of production lines, and identify the phenomenon of missing welding of parts in the process of automatic welding of the body through a deep learning model, thereby improving the reliability of welding production and avoiding negligence and waste of manpower caused by manual detection.

Description

Technical Field

[0001] The present invention relates to the field of automobile parts welding detection, and in particular to an artificial intelligence-based vehicle body welding condition detection system and detection method. Background Art

[0002] Different parts in automobile manufacturing require different welding methods. Among them, automobile body components are mainly made of cold-rolled steel stamping parts with good stamping and welding processability, which are assembled and welded. The welding production is large in batches, fast in production speed, and requires high assembly and welding precision of the welded parts. It mostly adopts the production method of assembly line operation, with high mechanization and automation, and a large number of special equipment and fixtures are used. More and more welding work is done by automatic welding machines and welding robots.

[0003] When the robot is automatically welding, if a fault occurs, such as wrong welding and missing welding, if it is not discovered in time, it will cause a large number of products to be defective, requiring rework or even scrapping, which will bring huge economic losses. Therefore, online production welding quality monitoring of the production line is of great significance.

[0004] The automated welding production line is flexible and can produce multiple products on the same line. In the current existing technologies, the detection and identification technology of body welding parts generally uses the professional image recognition technology based on OPenCV, which detects and identifies parts through filtering, binarization, grayscale, edge operator, expansion corrosion and other technologies. These traditional technologies have high thresholds for use and poor applicability. When the model and parts change, new programs need to be written to adapt. These have brought inconvenience to actual use and increased costs. Summary of the invention

[0005] The present invention aims to provide a body welding condition detection system and method based on artificial intelligence. The system and method can quickly build a detection model according to different production models, can meet the automatic switching recognition model of multiple models of production lines, and can identify the welding leakage and lack of parts in the process of automatic welding of the body through a deep learning model, thereby improving the reliability of welding production and avoiding negligence and waste of manpower caused by manual detection.

[0006] The technical solution of the present invention is as follows:

[0007] The vehicle body welding condition detection system based on artificial intelligence includes the following modules:

[0008] Camera module: cameras are set up at each process point of each production line. Each camera is connected to the control module through the network to take pictures of the production conditions of each process of the production line;

[0009] Configure the expansion module to set the ID of each production line, the IP address of the production line PLC, the IP address of each camera of the production line, set the production model and its corresponding timing action information and the camera that needs to be started for welding detection;

[0010] The production line PLC is used to read the vehicle model information in the configuration expansion module, as well as the timing action signals of each robot, fixture, etc. in the production line, control the production line, and exchange information with the system host computer;

[0011] The information collection module reads the IP address of the production line PLC in the configuration module, connects to the data block of the production line PLC with the IP address, obtains the vehicle model information being produced on the production line and the timing action information of the welding robot, and transmits it to the image acquisition module;

[0012] The image acquisition module reads the IP addresses of the cameras required for detection, determines the welding station that needs to be photographed based on the timing action information of the welding robot in the production line obtained by the information collection module, calls the camera with the corresponding IP address at the station, takes a photo when welding is completed, obtains the collected image, and transmits it to the intelligent detection module;

[0013] The intelligent detection module uses a recognition model trained by a deep neural network to identify the welds of the body parts in the captured image, identify the number of welds and the type of each weld, and transmit it to the control module;

[0014] The control module receives the detection result of the intelligent detection module and compares it with the welding information in the pre-stored vehicle model information, so as to detect the welding position.

[0015] The recognition model of the neural network deep training is a trained yolo-V3 convolutional neural network. The yolo-V3 convolutional neural network is trained with hundreds of labeled pictures to obtain a recognition model of the neural network deep training.

[0016] The artificial intelligence-based body welding condition detection system also includes a real-time alarm module. When the intelligent detection module detects a missed weld or a wrong weld, the real-time alarm module writes an offset to the production line PLC. After receiving the offset, the production line PLC stops the production line operation and issues an alarm through a display screen and an alarm light.

[0017] The artificial intelligence-based body welding condition detection system also includes a real-time display module for displaying the monitoring images of all cameras on the production line in real time.

[0018] The artificial intelligence-based body welding condition detection system also includes an information sending module, which is electrically connected to the control module and is used to send the detection information and detection results calculated by the control module to the enterprise WeChat and / or database in real time.

[0019] The control module is a computer.

[0020] The present invention also provides a method for detecting the welding condition of a vehicle body based on artificial intelligence, which applies the above-mentioned vehicle body welding condition detection system and comprises the following steps:

[0021] A. Cameras are set up at each process point of each production line, and each camera is connected to the control module through the network;

[0022] B. Use the configuration expansion module to set the ID of each production line, the IP address of the production line PLC, the IP address of each camera of the production line, determine the production model and its corresponding timing action information and the camera that needs to be started for welding detection;

[0023] C. The production line PLC reads the vehicle model information in the configuration expansion module, as well as the timing action signals of each robot, fixture, etc. in the production line, and controls the production line for production;

[0024] D. The information collection module reads the IP address of the production line PLC in the configuration module, connects to the data block of the production line PLC of the IP address, obtains the vehicle model information being produced by the production line and the timing action information of the welding robot, and transmits them to the image acquisition module;

[0025] E. The image acquisition module reads the IP addresses of the cameras required for the reading and detection, determines the welding station that needs to be photographed according to the timing action information of the welding robot in the production line obtained by the information collection module, calls the camera with the corresponding IP address at the station, takes a photo when the welding is completed, obtains the collected image, and transmits it to the intelligent detection module;

[0026] F. The intelligent detection module uses a recognition model with deep neural network training to collect images of various welding points of the body parts, identify the number of welding points and the type of each welding point, and transmit them to the control module. The control module compares the detection results with the welding information in the pre-stored vehicle model information to detect the welding position.

[0027] The vehicle body welding condition detection method based on artificial intelligence further includes the following steps:

[0028] G. When missed welding and wrong welding are found, the real-time alarm module writes the offset to the production line PLC. After receiving the offset, the production line PLC stops the production line operation and alarms through the display screen and alarm lights to notify the on-site personnel to handle it. After the processing is completed, reset it through the reset button and restart the operation.

[0029] H. The detection information and detection results calculated by the control module are sent to the enterprise WeChat and / or database in real time through the information sending module.

[0030] The present invention can quickly build a detection model according to different production vehicle models by configuring the expansion module, which can meet the automatic switching of recognition models for multiple vehicle models in the production line. During operation, it reads the signal of the production line PLC data block to drive detection and recognition, and uses the vehicle model signal of the production line to realize the switching of the artificial intelligence detection and recognition model of each vehicle model, which can meet the needs of flexible production and detection of the production line.

[0031] The configuration expansion module of the present invention can conveniently add detection points, can optimize the configuration according to the corresponding hardware and model, and can instantly adjust the camera installation position, so as to facilitate and quickly layout new detection points.

[0032] The present invention uses a neural network deep learning model to identify the phenomenon of weld leakage and lack of welding of parts in the automatic welding process of the car body, thereby improving the reliability of the welding process and avoiding negligence and waste of manpower caused by manual inspection.

[0033] The present invention can quickly build detection models according to different production models, can meet the needs of automatic switching recognition models for multi-model production on the production line, and can identify the phenomenon of weld leakage and lack of parts in the process of automatic welding of the car body through a deep learning model, thereby improving the reliability of welding production and avoiding negligence and waste of manpower caused by manual inspection. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] Figure 1 is a module diagram of the system of Example 1;

[0035] Figure 2 is a composition diagram of the system of Example 1;

[0036] Figure 3 This is a configuration file structure diagram of the configuration extension module of Example 1;

[0037] Figure 4 This is the solder joint diagram of the parts actually identified in Example 2. DETAILED DESCRIPTION

[0038] The present invention will be described in detail below with reference to the accompanying drawings and embodiments. Example 1

[0039] like Figure 1-4As shown in the figure, the body welding condition detection system based on artificial intelligence includes the following modules:

[0040] Camera module: cameras are set up at each process point of each production line. Each camera is connected to the control module through the network to take pictures of the production conditions of each process of the production line;

[0041] Configure the expansion module to set the ID of each production line, the IP address of the production line PLC, the IP address of each camera of the production line, set the production model and its corresponding timing action information and the camera that needs to be started for welding detection;

[0042] The production line PLC is used to read the vehicle model information in the configuration expansion module, as well as the timing action signals of each robot, fixture, etc. in the production line, control the production line, and exchange information with the system host computer;

[0043] The information collection module reads the IP address of the production line PLC in the configuration module, connects to the data block of the production line PLC with the IP address, obtains the vehicle model information being produced on the production line and the timing action information of the welding robot, and transmits it to the image acquisition module;

[0044] The image acquisition module reads the IP addresses of the cameras required for detection, determines the welding station that needs to be photographed based on the timing action information of the welding robot in the production line obtained by the information collection module, calls the camera with the corresponding IP address at the station, takes a photo when welding is completed, obtains the collected image, and transmits it to the intelligent detection module;

[0045] The intelligent detection module uses the trained yolo-V3 convolutional neural network to identify the welding points of the body parts in the collected images, identify the number of welding points and the type of each welding point, and transmit them to the control module;

[0046] The control module receives the detection result of the intelligent detection module and compares it with the welding information in the pre-stored vehicle model information, thereby detecting the welding position;

[0047] Real-time alarm module: when the intelligent detection module detects missed welding or wrong welding, the real-time alarm module writes the offset to the production line PLC. After receiving the offset, the production line PLC stops the production line operation and issues an alarm through the display screen and alarm light.

[0048] Real-time display module, used to display the monitoring images of all cameras on the production line in real time;

[0049] The information sending module is electrically connected to the control module and is used to send the detection information and detection results calculated by the control module to the enterprise WeChat and / or database in real time.

[0050] The control module is a computer. Example 2

[0051] The vehicle body welding condition detection method based on artificial intelligence, using the vehicle body welding condition detection system of embodiment 1, comprises the following steps:

[0052] A. Cameras are set up at each process point of each production line, and each camera is connected to the control module through the network;

[0053] B. Use the configuration expansion module to set the ID of each production line, the IP address of the production line PLC, the IP address of each camera of the production line, determine the production model and its corresponding timing action information and the camera that needs to be started for welding detection;

[0054] C. The production line PLC reads the vehicle model information in the configuration expansion module, as well as the timing action signals of each robot, fixture, etc. in the production line, and controls the production line for production;

[0055] D. The information collection module reads the IP address of the production line PLC in the configuration module, connects to the data block of the production line PLC of the IP address, obtains the vehicle model information being produced by the production line and the timing action information of the welding robot, and transmits them to the image acquisition module;

[0056] E. The image acquisition module reads the IP addresses of the cameras required for the reading and detection, determines the welding station that needs to be photographed according to the timing action information of the welding robot in the production line obtained by the information collection module, calls the camera with the corresponding IP address at the station, takes a photo when the welding is completed, obtains the collected image, and transmits it to the intelligent detection module;

[0057] F. The intelligent detection module uses a recognition model with deep neural network training to collect images of various welding points of the body parts, identify the number of welding points and the type of each welding point, and transmit them to the control module. The control module compares the detection results with the welding information in the pre-stored vehicle model information to detect the welding position.

[0058] G. When missed welding and wrong welding are found, the real-time alarm module writes the offset to the production line PLC. After receiving the offset, the production line PLC stops the production line operation and alarms through the display screen and alarm lights to notify the on-site personnel to handle it. After the processing is completed, reset it through the reset button and restart the operation.

[0059] H. The detection information and detection results calculated by the control module are sent to the enterprise WeChat and / or database in real time through the information sending module.

[0060] Example of solder joint recognition results Figure 4 .

Claims

1. A car body welding condition detection system based on artificial intelligence, characterized in that: Includes the following modules: Camera module: cameras are set up at each process point of each production line. Each camera is connected to the control module through the network to take pictures of the production conditions of each process of the production line; Configure the expansion module to set the ID of each production line, the IP address of the production line PLC, the IP address of each camera of the production line, set the production model and its corresponding timing action information and the camera that needs to be started for welding detection; The production line PLC is used to read the vehicle model information in the configuration expansion module, as well as the timing action signals of each robot, fixture, etc. in the production line, control the production line, and exchange information with the system host computer; The information collection module reads the IP address of the production line PLC in the configuration module, connects to the data block of the production line PLC with the IP address, obtains the vehicle model information being produced on the production line and the timing action information of the welding robot, and transmits it to the image acquisition module; The image acquisition module reads the IP addresses of the cameras required for detection, determines the welding station that needs to be photographed based on the timing action information of the welding robot in the production line obtained by the information collection module, calls the camera with the corresponding IP address at the station, takes a photo when welding is completed, obtains the collected image, and transmits it to the intelligent detection module; The intelligent detection module uses a recognition model trained by a deep neural network to identify the welds of the body parts in the captured image, identify the number of welds and the type of each weld, and transmit it to the control module; The control module receives the detection result of the intelligent detection module and compares it with the welding information in the pre-stored vehicle model information, so as to detect the welding position.

2. The vehicle body welding condition detection system based on artificial intelligence as claimed in claim 1, characterized in that: The neural network deep training recognition model is a trained yolo-V3 convolutional neural network model.

3. The vehicle body welding condition detection system based on artificial intelligence as claimed in claim 1, characterized in that: It also includes a real-time alarm module. When the intelligent detection module detects missing or wrong welding, the real-time alarm module writes the offset to the production line PLC. After receiving the offset, the production line PLC stops the production line operation and issues an alarm through the display screen and alarm light.

4. The vehicle body welding condition detection system based on artificial intelligence as claimed in claim 1, characterized in that: It also includes a real-time display module, which is used to display the monitoring images of all cameras on the production line in real time.

5. The vehicle body welding condition detection system based on artificial intelligence as claimed in claim 1, characterized in that: It also includes an information sending module, which is electrically connected to the control module and is used to send the detection information and detection results calculated by the control module to the enterprise WeChat and / or database in real time.

6. The vehicle body welding condition detection system based on artificial intelligence as claimed in claim 1, characterized in that: The control module is a computer.

7. A method for detecting body welding conditions based on artificial intelligence, using a body welding condition detection system as claimed in any one of claims 1 to 6, characterized in that: The following steps are involved: A. Cameras are set up at each process point of each production line, and each camera is connected to the control module through the network; B. Use the configuration expansion module to set the ID of each production line, the IP address of the production line PLC, the IP address of each camera of the production line, determine the production model and its corresponding timing action information and the camera that needs to be started for welding detection; C. The production line PLC reads the vehicle model information in the configuration expansion module, as well as the timing action signals of each robot, fixture, etc. in the production line, and controls the production line for production; D. The information collection module reads the IP address of the production line PLC in the configuration module, connects to the data block of the production line PLC of the IP address, obtains the vehicle model information being produced by the production line and the timing action information of the welding robot, and transmits them to the image acquisition module; D. The image acquisition module reads the IP addresses of the cameras required for the reading and detection, determines the welding station that needs to be photographed according to the timing action information of the welding robot in the production line obtained by the information collection module, calls the camera with the corresponding IP address at the station, takes a photo when the welding is completed, obtains the collected image, and transmits it to the intelligent detection module; E. The intelligent detection module uses a recognition model deeply trained by a neural network to identify each welding point of the body parts in the collected image, identify the number of welding points and the type of each welding point, and transmit them to the control module. The control module compares the detection results with the welding information in the pre-stored vehicle model information to detect the welding position.

8. The method for detecting the welding condition of a vehicle body based on artificial intelligence as claimed in claim 7 further comprises the following steps: F. When missed welding and wrong welding are found, the real-time alarm module writes the offset to the production line PLC. After receiving the offset, the production line PLC stops the production line operation and alarms through the display screen and alarm lights to notify the on-site personnel to handle it. After the processing is completed, reset it through the reset button and restart the operation.

9. The method for detecting the welding condition of a vehicle body based on artificial intelligence as claimed in claim 7 further comprises the following steps: G. The detection information and detection results calculated by the control module are sent to the enterprise WeChat and / or database in real time through the information sending module.

Images