基于BP和Hopfield神经网络的航空肼燃料保障安全评价

针对航空肼燃料保障安全评价的复杂性和非线性,提出并建立了基于BP和Hopfield神经网络的动态安全评价模型。在综合分析国内外肼燃料保障安全评价的基础上,针对航空肼燃料保障过程中出现的问题,构建并优化了指标体系,选取前馈神经网络中的BP网络和反馈神经网络中的Hopfield网络建立评价模型。在详细说明了BP和Hopfield神经网络的构建方法后,进行实例验证,并对预测效果进行了比较分析。仿真表明,两种模型都能正确评价安全保障状态。但在收敛速度、联想记忆功能方面Hopfield神经网络优于BP神经网络。将BP和Hopfield神经网络用于肼燃料保障安全评价过程中,具有适用性和可行性,对于航空肼燃料保障的安全建设与安全管理研究具有重要意义。

Research of Aerial Hydrazine Fuel Safety Assessment Based on BP and Hopfield Neural Network

Objective A kind of aerial hydrazine fuel dynamic safety assessment models based on BP and Hopfield neural network were raised and constructed in the light of the complexity and nonlinear characteristic of aerial hydrazine fuel guarantee. Methods Based on synthetically analyzing research data at home and abroad on aerial hydrazine fuel safety assessment,aiming at the problems of aerial hydrazine fuel guarantee,an indexes system was advanced and has been optimized,then BP network in feedforward neural networks and Hopfield network in feedback neural networks were chosen. After elaborating the method of establishing network,the models were examined by practical cases. Finally the outputs got analyzed and compared. Results The study show that the two models can evaluate the safety status correctly. Hopfield network is superior to BP network in convergence and associational memory. Conclusion The usage of BP and Hopfield neural network in aerial hydrazine fuel safety assessment is applicable and feasible. It is significant for safety construction and safety management of aerial hydrazine fuel guarantee.