基于BP网络改进模糊评判隶属度函数的变压器状态评估

Improved Fuzzy Evaluation Based on BP Network Membership Function of the Transformer Condition Assessment

通过对当前国内电力变压器状态检修现状的分析,以模糊理论为基础,论文提出一种采用模糊综合评判结合BP神经网络算法对隶属度函数的训练优化来评估电力变压器工作状态的方法。此方法根据影响变压器健康状态的因素与其内部器件的对应关系,利用BP神经网络训练样本,得出训练好的最优隶属度函数,以此判断出相应内部器件的健康状态。验证了结合BP神经网络算法的模糊方法的评估结果比仅仅采用模糊理论评估的结果更准确。

Through analyzing the current situation of the present domestic state overhaul of electric power transformer,based on the fuzzy theory,a method is presented that is used for training the membership of fuzzy comprehensive evaluation combined with BP neural network algorithm to optimize membership function to evaluate the working state of power transformer.The method according to the influencing factors of transformer health status and its internal components,the relation between the parameters provided by the transformer,the training samples of BP neural network are used to train good optimal membership functions,which can be used for judging the health status of the corresponding internal device.It is verified that the results of fuzzy method of combining the BP neural network algorithm evaluation have more accuracy than the results of fuzzy theory to evaluate.