基于油中溶解气体特征量筛选的变压器故障诊断方法

油中溶解气体分析对变压器故障预警及诊断具有重要意义。针对油中溶解气体特征量种类众多、故障关联特征分析不足等问题，文中以油浸式变压器为研究对象，提出了基于油中溶解气体特征量筛选的变压器故障诊断方法。首先，对油中溶解气体的原始特征量进行特征衍生，通过随机森林(random forest, RF)计算特征量对故障诊断的重要度，筛选得到最佳特征组合。其次，采用树结构概率密度估计(tree-structured parzen estimator, TPE)实现RF模型的参数寻优，并形成TPE-RF诊断模型。同时，结合多种评价指标，证明所提方法能够对变压器作出准确的故障诊断。最后，提出TreeSAHP模型分析特征量对各故障的重要度，优选出各故障关联的主要特征量，并根据变压器运行案例，探讨了该方法在电力行业现场应用中的适用性，验证了该方法的有效性。

Fault diagnosis method for transformers based on feature selection of dissolved gas in oil

Dissolved gas analysis is important to the early waring and diagnosis of transformer fault. Considering that the large numbers of features of dissolved gas in oil and the insufficient analysis of fault associated features, the paper proposes a new fault diagnosis method for transformers based on feature selection of dissolved gas in oil. Firstly, the paper com-pletes derivation of original features for dissolved gas in the oil, and the optimal combination of feature is selected by calculating the importance of feature for fault diagnosis based on random forest (RF). Then, the tree-structured parzen estimator (TPE) is used to realize the parameter optimization of RF model and obtain the TPE-RF diagnostic model. Combined with the various evaluated indicators, it shows that the proposed method can accurately diagnose the transformer faults. Finally, the TreeSHAP model is introduced to analyze the importance of the new features corre-sponding to each fault and select the specialized features for each fault. In case study, according to the case of trans-former operation, the applicability of the method in the power system is discussed, and the effectiveness of the method is verified.