反向传播(BP)神经网络算法能准确有效地对电子对抗干扰效能进行评估,并已获得理论和实践验证,但其存在训练时间长、收敛速度慢的缺陷。文中基于计算机BP神经网络,主要针对如何快速、准确计算出雷达自适应优化数值进行了大量分析,发现唯有将神经网络计算方式引入自适应优化计算方法才可以得出结果,最终按照实际情况再对该值进行修改,以此减少过程时间,提高收敛速度,增强评估时效性。在计算机大环境相同的运行程序条件下,就时间代价而言,BP神经网络跟踪的处理耗时比波形捷变提升了一个数量级,比波形固定的计算方法提升了两个数量级。通过计算神经网络映射数值,能够进一步对雷达抗干扰性能进行计算和测评。研究表明:倘若设定评分指标不变,BP神经网络法的雷达抗干扰效能指标要优于传统加权评估法。

Back propagation (BP) neural network algorithm can accurately and effectively evaluate the jamming effectiveness of electronic countermeasure, which has been verified in practice and theory, but it has the defects of long training time and slow convergence speed. Based on computer BP neural network, a large number of analysis on how to quickly and accurately calculate radar adaptive optimization numbers is carried out, and the results can be gotten only by introducing the BP neural network algorithm into adaptive optimization algorithm. Then the results are modified according to actual situations so as to reduce the training time, improve the convergence speed, and enhance the timeliness of evaluation. Under the operating condition with the same computer environment, in terms of time cost, compared with waveform agility, the processing time of BP neural network tracking is increased by one order of magnitude; compared with waveform fixed method, the processing time is increased by two orders of magnitude. Using mapping ability of BP neural network, the radar anti-jamming effectiveness is calcuated and evaluated. The results show that the radar anti-jamming effectiveness index of BP neural network method is better than the traditional weighted evaluation method under the same scoring index.

TN973

内蒙古自治区科技项目