Advanced BP Neural Network Algorithm on Radar Jamming Effectiveness Evaluation

Aimed at the problems of initial weights and thresholds of standard back propagation neural network effecting the network training in the presence of radar jamming effectiveness evaluation model to a great extent, first, this paper puts forward the idea of using GA according to selection, crossover and mutation among the random initialization and regarding the error function between predicted and expected values as the fitness function to optimize initial weights and thresholds of BP neural network to predict data sample better. Then train and predict the BP neural network at last. Compared to the algorithm without GA, the simulation results show that the using genetic algorithm to optimize the BP network has smaller error in evaluation, and the advantages and disadvantages of the two evaluation algorithms are verified.