Coal Moisture Intelligent Modeling and Optimization Based on Resampling by Half-Mean

Abstract

Coal moisture automatic online control has important practical significance on the actual production, which is realized by analyzing and modelling the existing coal moisture control system. In this research, experimental training data are used RHM (Resembling by Half-Mean) to exclude abnormal values. The study adopts RBF (Radical Basis Function) neural network for coal moisture control system to model, then PSO (Particle Swarm Optimization) algorithm is applied to RBF model parameter identification and optimization. The rolling optimization in this algorithm can modify target function, and improve the accuracy of model prediction. Experimental results show that the model based on the method of PSO-RBF using RHM is obviously better than the one which do not. When the model is applied to coal moisture control system, it could enhance the accuracy of the forecasts and the model significantly.