Research on Prediction Method of Profile Grinding Surface Roughness Based on PSO-LS SVM Algorithm

Abstract

In order to improve the prediction accuracy of the grinding surface roughness, this paper presents a prediction algorithm based on Least Squares Support Vector Machine with Particle Swarm Optimization (PSO-LS SVM). Chose the radial basis function as the kernel function of this new algorithm in where the regularization parameter and kernel function parameter of LS SVM were optimized by PSO, then the grinding surface roughness prediction model was established with the parameters optimized before. Through experimental analysis, it can be realized that compared to the conventional prediction methods, the prediction results of the proposed algorithm are closer to the actual roughness and the relative errors are smaller. As a result, the proposed algorithm is more acceptable.