基于逆向思维的系统可靠性精确分配模型  

Accurate allocation model of system reliability based on converse thinking

作　　者：张宏斌[1,2] 贾志新[1] 

机构地区：[1]北京科技大学机械工程学院,北京10008 [2]陆军航空兵学院机载设备系,北京101114

出　　处：《计算机应用》2009年第9期2468-2470,共3页journal of Computer Applications

基　　金：国家自然科学基金资助项目(50875021)

摘　　要：为了实现系统可靠性设计后期可靠度的精确分配,提出一种采用逆向思维,利用神经网络能够通过学习逼近任意非线性映射的能力,对系统可靠度进行精确分配的方法。以前期可靠性试验数据为基础,分析各子系统可靠度及自身约束条件变化过程中,系统可靠度的变化程度,从而获得子系统可靠度变化对系统整体可靠性影响的程度,作为可靠性分配的依据。将系统可靠度及两两对应子系统自身约束条件作为输入,同时以相对的子系统可靠度比值作为输出,对误差反向传播(BP)神经网络和径向基函数(RBF)神经网络进行训练,并对比较了测试结果,得出了系统可靠度精确分配的神经网络模型。For realizing accurate reliability allocation in the late stage of system reliability design, a new reliability allocation method was proposed. The basic idea of the method was using converse thinking and the ability of neural network that it could approach to any non-liner mapping. Based on the early reliability data, the influence of the system reliability changing by the sub-system reliability and the sub-system serf-constrained conditions changing could be achieved. It could be taken as the base of the reliability allocation. The system reliability and the sub-system self-constrained conditions were taken as the input of the neural network; the ratio of the corresponding sub-system reliability was taken as the output of the neural network. Then, the Back Propagation （BP） neural network and Radial Basis Function （RBF） neural network were trained. After comparing the test results, the neural network model of system reliability allocation could be achieved.

关 键 词：可靠性分配 逆向思维 误差反向传播神经网络 径向基函数神经网络 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]