In mechanical manufacturing, stamping is a very important plastic forming method, widely used in the automotive, aerospace, and electrical appliance industries. As is well known, most of the body panels and structural components of automobiles are thin-sheet stamped parts, making the level of stamping technology and the quality of stamping crucial for automobile manufacturers.
The collision between the "physical world" (represented by manufacturing equipment) and the "digital world" (represented by technologies such as artificial intelligence and sensors) has spurred a massive transformation in manufacturing. The fusion of these two worlds will inject new momentum into the next round of economic development. New technologies, represented by artificial intelligence, are having a profound impact on production processes, production models, and supply chain systems. The application value of artificial intelligence in manufacturing process diagnostics is becoming increasingly apparent, especially in the quality inspection and process optimization of stamped parts, where it is demonstrating advantages unmatched by human intervention. In short, artificial intelligence technologies can replace human eyes in performing functions such as identification, measurement, positioning, and judgment of stamped parts. Furthermore, artificial intelligence possesses "learning" capabilities, and through sample accumulation and model training and optimization, it can accurately predict the cracking risk of stamped parts, thereby achieving precise control and optimization of stamped product quality. The following are application cases of artificial intelligence technology in automotive manufacturing stamping workshops.
Project Background
In mechanical manufacturing, stamping is a very important plastic forming method, widely used in the automotive, aerospace, and electrical appliance industries. As is well known, most of the body panels and structural components of automobiles are thin-sheet stamped parts, making the level of stamping technology and the quality of stamping crucial for automobile manufacturers.
The stamping workshop of a certain automobile manufacturing plant has three stamping production lines, mainly producing passenger vehicle body panels with large outline dimensions and spatial curved shapes, such as side panels, fenders, doors, and hoods. During the stamping production process, some side panels are prone to localized cracking during the stretching process, requiring repeated parameter adjustments and trial production; at the end of the production line, a large number of quality inspectors are needed to manually inspect the surface defects of the stamped parts.
Problems and Challenges
1. The current inspection method at the end of the stamping production line is manual inspection. Within a limited production cycle time, stamped parts with surface defects such as cracks, scratches, slip lines, and bumps need to be quickly sorted out. The inspection standards are not uniform, the stability is not high, and the quality inspection data is difficult to quantify and store effectively, which is not conducive to the company's data resource collection, quality problem analysis and traceability.
2. During the stamping production trial, many factors can affect the local cracking of the side panel during the stretching process, such as equipment parameters, mold condition, and sheet material properties. Adjusting parameters and repeated trials are somewhat blind, costly, and inefficient.
3. The data is influenced by many factors, varies greatly in form, and is distributed across different business systems in the workshop. It includes both real-time equipment data and unstructured image data, which places extremely high demands on data acquisition, management, and storage.
Solution
Based on the above, Merrill Data builds a big data platform for enterprises to achieve the integration, storage and unified management of equipment, molds, materials, manufacturing process data and quality inspection data in the factory's stamping workshop. With the help of data mining based on machine learning and intelligent detection technology based on machine vision, it can predict side stamping cracks and intelligently identify surface defects of product parts.
Based on stamping equipment processing parameters, sheet metal parameters, die performance parameters, and maintenance records, an intelligent prediction model for the stamping process is established using data mining and machine learning algorithms. Through sample accumulation and model training and optimization, the cracking risk of stamped parts is accurately predicted. Finally, the correlation between influencing factors in the manufacturing process is determined, and a control strategy for the combination of production process parameters is formulated to support the optimization of stamping manufacturing processes and quality control.
◎ Intelligent identification and detection of stamped parts defects based on machine vision: Utilizing existing production line conditions, an image acquisition system is designed to quickly identify surface defects in stamped parts through real-time image acquisition and intelligent analysis. All detected images and process data are automatically stored in a big data platform. By correlating quality inspection data, production process parameters, and product design parameters, and leveraging big data analytics, a closed-loop system for analyzing and managing stamped product quality issues is formed, achieving precise control and optimization of stamped product quality.
Application value
1. By predicting the cracking risk of stamped parts, the efficiency of designing processing parameters for stamped parts of new car models can be improved, reducing the number of trial productions and trial production costs.
2. By rapidly and intelligently detecting surface defects in stamped parts, the stability and reliability of production line inspection are improved, while reducing the labor intensity and costs for quality inspectors. Simultaneously, product quality inspection data is effectively stored, providing crucial data support for achieving closed-loop quality analysis and traceability.
3. It has explored a practical and feasible demonstration path for the intelligent manufacturing transformation of enterprises, and accumulated valuable experience for the promotion and application of technologies such as industrial big data and artificial intelligence in peer companies.
Applicable industries
Industries such as automobile manufacturing, aerospace, and home appliance production that employ stamping and spraying processes and have high requirements for product surface quality.
