Radar fault diagnosis based on adaptive genetic algorithm and neural network

Abstract

A scheme for radar fault diagnosis is proposed. It is based on artificial neural networks whose learning sample comes from the Pspice simulation. For single circuit, single layer artificial neural networks are used. Aiming at the problems of slow rate of convergence and falling easily into part minimums in BP algorithm, a new improved genetic BP algorithm was put forward. To determine whether the network fall into part minimum point, a discriminator of part minimum was put forth in the training process of neural network. Genetic algorithm was used to revise the weights of the neural network if the BP algorithm fell into minimums. For large and complex system multiplayer artificial neural networks are used, as well as preprocessing layer and post processing layer. It can raise diagnosis precision, shorten to train time, raise fault tolerance and the development of fault diagnosis, effective carry out fault diagnosis. Circuit faults of certain radar are diagnosed using the simulation and neural network scheme put forward in this paper, which testified the validity of this diagnosis process.