Abstract: For variable amplitude hydraulic system complexity, uncertainty, ambiguity, it proposes fuzzy neural network based on fault tree as a method for variable amplitude hydraulic system fault diagnosis. The method extracts fault diagnosis input and output variables of variable amplitude hydraulic system using fault tree knowledge, introducing the concept of fuzzy logic, fuzzy membership functions to describe the extent to which these failures, using fuzzy membership functions to describe the extent of these failures, using Levenberg-Marquardt algorithm to train the neural network system, getting a better performance in inference speed and fault-tolerant, and analyzing and verifying the effectiveness of variable amplitude hydraulic system fuzzy neural network fault diagnosis through case.

Key words: fault diagnosis, fault tree, fuzzy neural network

摘要： 针对变幅液压系统复杂性、不确定性、模糊性的特点，提出基于故障树的模糊神经网络作为变幅液压系统故障诊断的方法。该方法利用故障树知识提取变幅液压系统故障诊断的输入变量和输出变量，引入模糊逻辑的概念，采用模糊隶属函数来描述这些故障的程度，利用Levenberg-Marquardt优化算法对神经网络进行训练，系统推理速度快，容错能力强，并通过实例分析验证了变幅液压系统模糊神经网络故障诊断的有效性。

关键词: 故障诊断, 故障树, 模糊神经网络