A Novel Neural Network for Sensor Fusion Applied to Wood Growth Ring Moisture Measurement

Abstract

Wood growth ring moisture content is considered as an important measured quantity for the analysis of wood drying stress. However, wood moisture sensors working in the complicated drying conditions are intensively jammed by ambient factor parameters. In this paper, a novel functional-link neural network (FLNN) is presented for the implementation of wood moisture sensor fusion. FLNNs are single layer networks that are able to handle linearly nonseparable classes due to the dimensions of the inputs being increased by using nonlinear combinations of the input features. Simulation results have demonstrated that FLNN has a less computational complexity and a better performance compared with conventional multilayer neural networks.