基于SOM的真空断路器机械故障诊断

给出了适用于小样本训练的自组织映射(SOM)网络的基本概念和突出特点,分析了真空断路器的机械特性与对应机械故障的关系。在此基础上,提出以真空断路器的机械特性作为训练与识别样本并基于SOM的真空断路器机械故障诊断方法。重点介绍了应用该方法进行断路器机械故障分类的全过程:通过提取正常与故障状态下断路器的机械特性并输入至SOM网络中进行故障区分。实验分析表明,该故障诊断方案可有效对真空断路器常见机械故障进行分类。

Mechanical Fault Diagnosis of Vacuum Circuit Breaker Based on SOM

The basic concepts and highlights of self-organization map (SOM) neural network which is suitable for small samples are given.The relationship between mechanical characteristics of vacuum circuit breaker and the corresponding mechanical failure was analyzed.On this basis, a method of mechanical fault diagnosis of vacuum circuit breaker that mechanical characteristics parameters were taken as training and identification samples of SOM neural network is proposed.The whole process of applying this method to conduct the mechanical fault classification of circuit breakers is emphasized: the fault differentiation is carried out by extracting the mechanical characteristics of circuit breaker under the normal and fault states as inputs of SOM network.The experimental results show that the fault diagnosis scheme achieves high accuracy of mechanical fault classification for vacuum circuit breakers.