RBF Neural Network for Threat Sequencing

The air combat situation of fighter can be formulated as a vector, an element of which is a decision factor. Properly determining the weights of all factors is crucial for threat assessment. In this paper, radial basis function neural network was employed to approximate the nonlinear complex relationship of all factors. Analytic hierarchy process was used to form initial training patterns. Then an adjustment was given to adjust unreasonable training patterns among them. The resulting training patterns were supplied to RBF neural network. The experimental results show that RBF neural network can successfully approximate the weights of all factors.