Temperature prediction of nano-iron powder decomposing furnace based on SAPSO-ELM

Abstract

A temperature prediction model based on Self-adaption Particle Swarm Optimization (SAPSO) and Extreme Learning Machine (ELM) is proposed in this paper. The nano-iron powder decomposing furnace temperature prediction model is established based on ELM. ELM, a neural network, is developed rapidly in recent years, but it requires a lot of hidden layer neurons to achieve ideal prediction accuracy. In order to solve this problem, SAPSO is used to optimize ELM input weight matrix and the hidden layer threshold value in this paper, which simplifies the ELM. The accuracy of this method is verified by choosing the related factors, using field data and simulation. Different numbers of neurons are set to compare with ordinary ELM. And the test shows that SAPSO-ELM temperature prediction model has better generalization.