WRSR与改进朴素贝叶斯融合的变压器故障诊断技术研究

电力变压器的运行状态评估及其故障准确定位,一直是制约电网运行安全和设备运维效率的技术瓶颈。建立一种基于加权秩和比(Weighted Rank Sum Ratio, WRSR)并结合改进朴素贝叶斯网络的诊断模型,用以评估电力变压器整体运行状态,确定故障位置及具体故障类型。首先从多个变电站收集变压器的历年故障数据,并将其作为训练集,在改进朴素贝叶斯网络中建立起特征参量与故障位置、故障类型之间的非线性映射关系。结合某电网的具体变压器运行状态信息与检测数据,利用WRSR模型对具体变压器整体运行状态进行评价,然后将状态性能较差的变压器故障检测数据作为测试集代入至改进朴素贝叶斯网络中来预测故障位置。最终结果表明,所提模型能够实现对电力变压器状态的合理评价,又可在预测故障部位及故障类型时保持较高的准确率。

Transformer fault diagnosis technology based on the fusion of WRSR and improved naive Bayes

Evaluation of the operational status of power transformers and accurate location of faults have always been technical bottlenecks affecting the safety of power grid operation and the efficiency of equipment operation and maintenance. This paper establishes a diagnosis based on the Weighted Rank Sum Ratio (Weighted Rank Sum Ratio, WRSR) combined with improved naive Bayesian networks. The model is used to evaluate the overall operational status of the power transformer, determine the fault location and specific fault types. The paper first collects historical fault data of transformers from multiple substations and uses it as a training set to establish a nonlinear mapping relationship between characteristic parameters and fault locations and fault types in an improved naive Bayesian network. Combined with specific transformer operating status information and detection data of a power grid, it first uses the WRSR model to evaluate the overall operating status of the specific transformer, and then substitutes the transformer fault detection data with poor performance as a test set into the improved naive Bayes network to predict fault location. The final results show that the model proposed can realize a reasonable evaluation of the state of power transformers, and can maintain a high accuracy rate in predicting fault locations and fault types. This work is supported by the National Natural Science Foundation of China (No. 51907109) and Shandong Electric Power Science and Technology Project (No. SGSDZBOOJXJ2000375).