隐式半马尔科夫模型下的变压器故障诊断方法

Transformer Fault Diagnosis Method Based on Hidden Semi-Markov Model

摘要:

变压器运行过程中存在多种状态,能够正确划分运行状态,对变压器的维修和故障诊断有着重要的意义。首先,详细分析了马尔科夫链的衍生模型,并构造了隐式半马尔科夫模型(Hidden Semi-Markov Models, HSMM)；然后,通过引入“微状态-宏状态”的对应关系,用于在HSMM中描述变压器运行过程中的状态；最后,建立了涵盖变压器历史状态信息,并包含特征提取、状态分类和故障识别过程的HSMM故障诊断流程。通过变压器DGA故障诊断的算例分析,结果表明所述方法的有效性。

Abstract:

There are many states in the operation of the transformer, which can correctly divide the operating state, which is of great significance for the maintenance and fault diagnosis of the transformer. Firstly, the derivative models of Markov chain are analyzed in detail and Hidden Semi-Markov Models (HSMM) are constructed. And then, the state of the transformer operation process is described in HSMM by introducing the correspondence of "micro-state macro-state". Finally, the HSMM fault diagnosis process covering the historical state of the transformer and including feature extraction, state classification and fault identification process is established. Through the analysis of the transformer DGA fault diagnosis, the results show the effectiveness of the proposed method.