基于贝叶斯网络的变压器故障推理模型

Fault Reasoning Model of Transformer Bsed on Bayesian Network

变压器是电力网络的重要设备。为保障变压器安全运行,针对以往故障推理过程中征兆特征过于复杂和没有加入不良工况问题,建立基于贝叶斯网络的变压器故障推理模型。该模型通过对变压器的在线监测,在对以往变压器的不良工况、故障类型、征兆特征进行数据分析的基础上,对变压器监测数据进行推理,得到变压器故障类型的可能性排序。该模型选取征兆特征的一个优化组合进行分析,使推理过程更加简洁,提高了推理结果的准确性。以某种典型的油浸式变压器故障为例进行推理仿真,仿真结果证明了该模型的可靠性和准确性。

In order to ensure the safe operation of the transformer which is an important equipment in electric power network,a transformer fault reasoning model based on Bayesian Networks is built to solve the problems that the symptom characteristics are too complex and not adding the bad conditions in previous fault reasoning process.Based on the data analysis of the bad conditions,the model could rank the possibility of the transformer faults by reasoning the on-line monitoring data,types of faults and symptom characteristics of former transformers.The model selects an optimized combination of symptom characteristics for analysis which makes the reasoning process more concise and improves the accuracy of the reasoning results.Some typical oilimmersed transformer faults are selected for reasoning simulation which shows the reliability and accuracy of the model.