Applying particle swarm optimization algorithm and support vector machine for optimizing metal flow in extrusion

Abstract

Aluminum alloy profiles constitute a critical material for aviation structural components, and their die quality significantly impacts the reliability of these parts. In response to the current problems of cracking and twisting in profile quality, this study analyzes the flow characteristics of metal billets during the extrusion forming process from the perspective of energy consumption, combined with improved support vector machine and particle swarm algorithms. A multi-objective metal extrusion process parameter optimization model has been established. The results showed that the model’s predictive accuracy was above 0.9 on both the training and validation sets. The improved support vector machine and particle swarm optimization algorithms had better recall and average accuracy than the compared algorithms, and could be applied to research on multi-objective metal extrusion process parameter optimization. Meanwhile, compared with the initial plan, the optimized plan resulted in more uniform metal flow in the product thickness and width directions for the extended forming of aluminum alloy. The results show that the improved velocity field is obtained under the research method, which can promote the improvement of product quality, reduce the energy consumption of extrusion forming, and provide a reference for the setting of process parameters for the production of new types of profiles.