Electrostatic target characteristic predictor based on fuzzy neural network

To overcome the disadvantages of the electrostatic target characteristic signal predictors based on Adaline and Elman neural network, such as big error and slow speed, a prediction model based on a four-layer fuzzy neural network (FNN) was proposed for electrostatic target detection. In this model, the membership functions were Gauss functions and the back propagation (BP) algorithm was applied to train the network parameters. The simulation results show that the performance of FNN prediction model is superior to that of Adaline and Elman neural network. It is characterized by smaller error, smaller calculation amount and faster speed which are crucial in real-time system. The proposed prediction method can provide an effective way to compensate the delay of circuits and improve the real-time of the electrostatic detection system.