基于神经网络的机械系统可靠性评价模型

机械系统相关失效的存在,严重地削弱了冗余结构的安全作用。从系统层的应力强度—干涉模型出发,通过Monte-Carlo仿真不同冗余系统的各阶失效概率,获得失效数据。利用BP神经网络算法的函数逼近功能、非线性映射功能和容错能力对系统相关失效概率模型进行离散化处理,建立系统失效概率与零件失效数据之间的非线性关系,进而构建基于神经网络的机械系统可靠性评价的参数化模型。利用该模型可以预测系统中的任意阶相关失效概率。实例给出其应用方法并与实验结果进行对比,表明该方法准确可靠。

Reliability Evaluation Model of Mechanical System Based on Neural Network

The security role of redundant mechanical system was seriously weakened by its relevant failure.The failure probability of different redundant structure was simulated at various mode by the Monte-Carlo based on the point of stress-strength and interference model,and the failure data was obtained.The relevant failure probability model was discretized by the powerful function approximation function,non-linear mapping function and tolerance of the BP neural network.The non-linear relationship between the system failure probability and the components failure data was built,and then the parametric model of mechanical system reliability evaluation based on the neural network was constructed.Any multiplicity relevant failure probability could be predicted by the model.The application of the method was described by a example.The comparison between the simulation and the experimental results show that the method is high accuracy and reliability.