基于时效评分函数和贝叶斯概率的电力变压器状态实时评估

Real-time Condition Assessment of Power Transformers Based on Time-effect Score Function and Bayesian Probability

为了有助于设备安全的动态管控,通过融合设备更新周期不同指标数据,提出电力变压器状态实时评估方法。在构建评估指标体系基础上,提出实时监测和例行试验2类指标评分值函数;同时基于设备样本和该变压器评估时间,提出结合贝叶斯后验概率与时变函数的指标权重计算方法。就综合指标劣化不均衡问题,引入变权重公式来调整计算其权重系数。样本实例计算结果显示采用该方法的评估正确率达93.5%。通过充分利用变压器本身数据信息,辅以同类型设备样本统计规律挖掘,该方法适于对电力变压器状态进行实时评估。

In order to facilitate the dynamic monitoring of the safety of equipment,we proposed a method for real-time assessment of power transformers by fusing different data of equipment index in update periods. Based on the construction of the evaluation index system,two kinds of score functions on data were proposed for real-time monitoring indices and preventive test indices,respectively. At the same time,based on the equipment samples and the assessment time of the transformer,a method for calculating the index weights combining Bayesian posterior probability and time-varying function was proposed. To solve the imbalance problem of the comprehensive indices,we introduced the variable weight formula to adjust and calculate the weight coefficients. Sample calculation results show that the correct rate of evaluation is up to 93.5%. By making full use of the data of the equipment itself,combined with mining the statistical laws from the same type of equipment,the method is suitable for real-time condition assessment of power transformers.