Knowledge-guided particle swarm optimization of multi-link systems for cold and warm forging presses

Abstract. The performance of the multi-link system for mechanical press directly affects the performance of the press and the quality of the forgings. In order to improve the quality of forgings and enhance the design efficiency, a knowledge-guided particle swarm optimization design framework for multi-link systems is proposed, taking the cold and warm forging press as the research object. A knowledge database consisting of historical cases, rule knowledge and other knowledge was created. A reasoning machine consisting of retrieval and adoption was then built to determine the configuration of the multi-link system through case similarity calculations. A multi-link system optimization model was established for the selected configuration. To solve the optimization model, a knowledge-guided improved particle swarm optimization algorithm is developed, from which we obtain the optimized dimensions of the multi-link system. Comparing the results before and after optimization, we found that all the performance indexes have been improved to a certain extent, which proves the effectiveness of the knowledge-guided particle swarm optimization design framework for multi-link systems. Moreover, the framework facilitates the storage and reuse of design knowledge, which improves design efficiency and promotes design automation.