Rolling force prediction based on PSO optimized support vector regression

Rolling force prediction is very important in hot bar rolling process. Aiming at the problem of predicting the bar rolling force accurately, an optimal approach of support vector regression based on improved particle swarm optimization (PSO) is proposed. A mathematic model based on the support vector regression optimized by particle swarm optimization is established, and the optimal parameter of which is searched by PSO. The experiment results shows that the proposed prediction model has better prediction results than the support vector regression algorithm and BP-NN algorithm, increasing the average prediction accuracy from 78.5% to 94.1%.