Domain knowledge integrated CAM system based on multi-objective path optimal planning and deep convolutional neural network

In the era of intelligent manufacturing, increasing consumers’ demands for customized products necessitates innovative approaches to design and processing efficiency. This research proposes an intelligent computer-aided manufacturing (CAM) system integrating domain knowledge, multi-objective optimization and deep convolutional neural networks (DCNNs). Using wire electrical discharge machining (WEDM) process as a case study, a multi-objective optimization model was developed to enhance machining quality, accuracy, and efficiency. A dataset of optimal machining paths and corresponding surface models was utilized to train the DCNN, enabling predictive path generation and real-time application. Comparative experiments between the optimized paths and traditional equidistant interpolation paths were conducted on a six-axis WEDM machine tool with constant RC power supply parameters. The machining efficiency finds an improvement of 18.83% to 40.7% on five randomly generated non-uniform rational B-splines (NURBS) free ruled surfaces. These findings underscore the high efficiency and practicality of the proposed system, advancing intelligent CAM solutions for complex manufacturing scenarios.