Applying General Probabilistic Neural Network to Adaptive Measurement Fusion

Abstract

A neural-network-based adaptive state estimation is presented to measurement fusion for use of a multisensor system tracking a maneuvering target. The proposed approach consists of a group of parallel alpha-beta-gamma filters and a general probabilistic neural network (GPNN). By incorporating a general probabilistic formulation and Markov chain into a general regression neural network, GPNN is developed as a decision logic algorithm for online classification. Each activation function of GPNN is defined as Gaussian basis function whose smooth factor is a constant selected from filter’s innovation covariance matrix by utilizing the parametric method. Based upon fused outputs of alpha-beta-gamma filters and a GPNN-based classifier, an adaptive alpha-beta-gamma filter is developed to improve tracking accuracy. The simulation results are presented to demonstrate the effectiveness of the proposed method.