Optimization of Structural Parameters of Air Pump Spring based on Machine Learning

Abstract

Spring leaves usually need to bear huge load and impact, thus wear, fracture, fatigue and other forms of failure easily take place. In this paper, the fatigue behavior of the air pump spring is numerically analyzed by finite element simulation technology. Based on the finite element result data set and the improved particle swarm BP neural network, the mapping relationship between spring structural parameters and fatigue damage is established. With the combination of neural network and genetic algorithm, a spring structure optimization method is proposed, and on this basis, the structural parameters of air pump springs are optimized by multi-objective, to improve its fatigue resistance. This study provides a reference for the health management of air pump spring, a new idea for its structure optimization.