基于遗传BP神经网络的装备研制风险分析方法

简要介绍了GA和BP算法,利用遗传算法全局性搜索的特点,改变BP算法依赖梯度信息的指导来调整网络权值的方法,寻找最为合适的网络连接权和网络结构,提出了遗传算法优化BP神经网络的思路及其数学模型.最后,结合对某型航空装备的风险源的分析,利用此优化模型进行了验证.结果表明了该方法的可行性,为装备研制风险分析提供了一种新思路.

An Analysis Method of Weapon Research and Development Risk based on the Fusion of GA and BP Neural Network

Genetic algorithms(GA) and back-propagation(BP) algorithms are introduced briefly. BP algorithm adjusts network weight according to the instruction of grads information. Using the globally search capability of GA to modify the method of looking for the most suitable network structure and link weight. Putting forward the way of BP optimized by GA. Bringing forward its mathematics model. Finally, verification of the model is carried on by the risk source analysis of the aviation equipment. Result expresses the possibility of this method which provides a new way for the risk analyzing of weapon research and development .