基于遗传算法的空空导弹消耗规律神经网络预测方法

Neural Network Prediction Method for the Law of Air to Air Missile Consumption Based on Genetic Algorithm

摘要:

针对弹药预测问题,介绍了基于遗传算法的 BP神经网络的相关原理及理论,研究了基于遗传算法的空空导弹消耗量 BP神经网络预测方法,弥补了传统的 BP神经网络法在弹药预测方面存在许多缺点,采用遗传算法求得影响弹药消耗各因素的权值及阈值,具有较高的准确性；同时,优化过程是对神经网络算法的权值及阈值进行优化,由适应度函数计算出染色体的适应度值,再经过选择、交叉及复制等操作,得到适应度最高的个体。对历次空战空空导弹消耗量数据进行归一化处理,将处理结果带入传统 BP模型和遗传算法优化后的 BP算法模型进行预测,得到预测结果并对比分析,预测结果显示了遗传算法优化的有效性,避免了传统 BP算法局部性强的缺点,预测结果较优化前较大提升,验证了改进算法的有效性和先进性。

Abstract:

The problem of ammunition prediction was discussed in this paper. The principle and theory of GA-BP was in.troduced and the BP neural network prediction method of air to air missile consumption based on genetic algorithm wasstudied for remedying the shortcomings of traditional BP neural network method in the prediction of ammunition. Theweights and thresholds of all factors affecting ammunition consumption were obtained by genetic algorithm with high accu.racy. At the same time, the optimization process was that optimizing the weights and thresholds of the neural network algo.rithm, and the fitness values of chromosomes were calculated from fitness function. After selection, crossover and duplica.tion, the highest fitness individuals were obtained. Finally, normalization of air to air missile consumption data of previousair combat was carried out, the resultsed were brought into the traditional BP model and the BP algorithm optimized by ge.netic algorithm to make predictions and comparative analysis. The prediction results show the effectiveness of genetic algo.rithm optimization that avoids the disadvantages of the traditional BP algorithm with strong locality, and the effectivenessand advance of the improved algorithm are verified.