Comparison on prediction wood moisture content using ARIMA and improved neural networks

Wood moisture content (MC) is one of the key parameters which influenced on wood product cost, qualities and efficiency, etc. The fiber saturation point (FSP) cannot be measured directly based the principle of electrical method. In this paper, two prediction measuring algorithms based the autoregressive integrated moving average (ARIMA) and functional link artificial neural network models are considered along with various combinations of these models for predicting wood moisture content (MC) around the fiber saturation point. The predicting principle and procedure of these methods are presented in detail. Measurement experiments are performed to get the time series data of wood moisture content. Simulation comparison of predicting performances shows that the improved neural network models with functional link ANN give a better performance in solving the wood moisture content prediction problem.