基于最大熵隐马尔科夫模型的电网故障诊断方法

A Power Grid Fault Diagnostic Method Based on Maximum Entropy Hidden Markov Model

随着电网调控一体化的全面推进,针对低价值密度故障数据的有效诊断成为实现电网自愈化的关键。该文提出了一种基于最大熵隐马尔科夫模型(maximum entropy hidden Markov model,ME-HMM)的电网故障诊断方法,该方法首先对调度中心所接收到的遥信信息进行去噪解析,并基于保护-断路器关联关系定义了待诊断信息类型以及异常信息模式,然后结合电气量信息和开关量信息构建特征函数向量,并通过训练ME-HMM模型对故障数据所隐藏的异常模式进行挖掘。通过实例分析证明该方法能够实现对原始故障数据的精简,有效识别包括信息畸变、保护断路器不正确动作在内的异常信息,从而提高电网故障诊断效率。

With comprehensive promotion of power grid regulation and integration,effective diagnosis of low-valuedensity fault data becomes the key to power grid self-healing.In this paper,a fault diagnostic method for power grid based on maximum entropy hidden Markov model(ME-HMM)was proposed.Firstly the remote signal information received by dispatching center was denoising analyzed.Based on protection-circuit breaker association relationship,information type and abnormal information mode were defined.According to electrical quantity information and switch quantity information,feature function vectors were built.The abnormal patterns hidden in fault data were excavated with the model.Results demonstrated that the method can realize simplification of original fault data,effectively identifying the abnormal information including distortion of information and incorrect action of circuit breaker,so as to improve fault diagnosis efficiency of power grid.