结合BP神经网络的遗传算法优选PEE制备参数

为方便快捷地寻找粉状乳化炸药PEE（powery emulsion explosives）的最佳制备参数，根据PEE制备数据建立了BP神经网络模型。以建立的非线性神经网络模型作为个体的适应度函数。采用Matlab编程，编写了制备参数的优选遗传算法迭代过程，并寻求到了最佳制备参数。依据灰关联分析方法，解析了PEE制备各因素对衡量指标的影响。结果表明，搅拌速度对PEE的制备也较为重要，而添加剂和喷雾压力影响较小，说明优化参数结果与原始数据的误差对衡量指标影响不大。理论分析结果与实际相符合，为PEE制备提供了一种新的优选方法。

Genetic algorithm with BP neural network of optimizing selecting parameters for PEE preparation

In order to find the optimizing parameters of the PEE. The BP neural network was found based on the data of the process of the PEE preparation. The nonlinear network was acted as the individual fitness function. The process of GA was programmed by Matlab language to find the optimizing parameters of the PEE and the optimizing parameters were found. The influencing factors on the scaling index of the PEE preparation was studied with the grey relationship analysis methods. It shows that the whisking velocity is very important to the PEE preparation. But the additive and the spraying pressure are very important. It accounts for the influence of the error between the optimizing parameters and the original data on the index of the PEE. The theoretical analysis results are in accordance with the fact. At the same time,it proposes a new optimizing way for explosives preparation.