改进GA优化BP神经网络的雷达信号识别

Radar signal recognition based on BP neural network optimized by improved GA

为了提高雷达辐射源信号的识别率,提出了一种改进遗传算法优化反向传播(back propagation,BP)神经网络的雷达信号识别算法。该算法先提取差分近似熵、调和平均盒维数和信息维数等3个特征作为雷达信号识别的特征,通过对遗传算法的适应度函数和交叉算子进行改进,并用改进的遗传算法优化BP神经网络,得到最优的初始权值和阈值,从而对雷达信号进行识别。仿真结果表明,与BP神经网络和标准遗传算法优化的BP神经网络的相比,改进雷达信号识别算法能够提高雷达辐射源信号的识别率,并提高了算法的收敛速度,证实了算法的有效性。

In order to improve the recognition rate of radar emitter signal,an improved genetic algorithm is proposed to optimize the back propagation(BP)neural network for radar signal recognition.The algorithm first extracts three features of radar signal recognition,such as difference approximate entropy,harmonic mean box dimension and information dimension.By improving the fitness function and crossover operator of genetic algorithm,and Optimizing BP neural network with improved genetic algorithm,the optimal initial weight and threshold value are obtained,so as to identify radar signal.The simulation results show that compared with BP neural network and BP neural network optimized by standard genetic algorithm,the improved radar signal recognition algorithm can improve the recognition rate of radar emitter signal,and improve the convergence speed of the algorithm,which proves the effectiveness of the algorithm.