基于贝叶斯网格的变压器故障诊断方法优化研究

Bayesian Mesh-based Optimization of Transformer Fault Diagnosis

变压器内部运行状态参数存在复合性,导致直接进行故障诊断时,对具体故障位置的诊断结果存在较大误差,为此提出基于贝叶斯网格的变压器故障诊断方法优化研究。以变压器节点导纳矩阵为基础,构建以电力系统变压器串联部分漏磁导纳阵为核心的模型,将构成复杂的变压器转换为由若干个节点导纳矩阵构成的模型。在诊断阶段,引入贝叶斯网格,根据感知区域内随机均匀散布在贝叶斯网格内的稀疏位置矢量,确定具体的故障位置。测试结果表明,所设计诊断方法对具体故障位置的诊断结果误差不仅表现出了较高的稳定性,而且具体误差均处于较低水平。

Due to the compounding of the internal operating state parameters of the transformer,there is a large error in the diagnosis result of the specific fault location when the fault diagnosis is carried out directly,so the optimization of the transformer fault diagnosis based on Bayesian mesh is proposed.Based on the transformer node admittance matrix,a model with the magnetic flux leakage admittance array of the transformer in series of the power system was constructed,and the complex transformer was converted into a model composed of several node admittance matrices.In the diagnosis stage,a Bayesian mesh is introduced to determine the specific fault location according to the sparse position vectors randomly and evenly scattered in the Bayesian mesh in the perception area.The test results show that the diagnostic method for the specific fault location not only shows high stability,but also achieves small errors.