关键词: 自动螺钉机, 图像识别, 最短路径, 遗传算法, 自适应变异, Tecnomatix

Abstract: Aiming at the difficulty during the process design of the automatic screw machines, such as low efficiency and poor quality, a screw machine process design and verification system is developed. Supported by image recognition technology, the system would automatically obtain the quantity and the locations of the screwing holes from the top view. An optimization model will be established to optimize the tightening sequence of the screw according to RSA (Radar Spiral Algorithm) and improved genetic algorithm. An integration is conducted into Tecnomatix to simulate those optimized sequence in case of collision. The results indicate that the error range between the coordinate obtained by image recognition technology and the real one is tiny enough for simulation; applying adapted mutation to generic algorithm helps to speed up the convergence in early stage and escape the local optimal solution in late stage simultaneously; optimizing with RSA and improved genetic algorithm can not only improve the efficiency of searching the solution, but also shorten the screwing path based on the process requirements.

Key words: automatic screwing machine, image recognition technology, shortest path, genetic algorithm, adapted mutation, Tecnomatix