基于混沌理论和K-means聚类的有载分接开关机械状态监测

Mechanical Condition Monitoring of On-load Tap Changers Based on Chaos Theory and K-means Clustering Method

变压器有载分接开关的可靠性对保障电力系统的安全稳定运行意义重大。考虑到有载分接开关操作过程中的振动信号与其机械性能密切相关,从振动信号的混沌动力学特性出发,通过计算嵌入维数和延迟时间,对实测到的振动信号进行了相空间重构。根据振动信号的相空间分布特性,使用K-means聚类分析方法对重构后的振动信号模式进行识别,据此对有载分接开关的机械性能进行监测。现场实测数据的计算结果表明:使用改进Wolf法计算得到的最大李雅普诺夫指数为正,进一步证实了有载分接开关切换过程中振动信号的混沌特性,基于K-means聚类分析法得到的簇中心位移矢量的变化可清晰地识别出有载分接开关的故障隐患。研究结果为从混沌动力学角度监测有载分接开关的运行状态提供了理论依据。

Reliability of on-load tap changer （OLTC） is essential to the secure and stable operation of power system. Since the vibration signals during the operation of OLTC were closely related to its mechanical performance, the measured vibration signals were reconstructed in the phase space with the calculated embedding dimension and delay time based on the chaotic feature of vibration signal. Then the K-means clustering analysis method was applied to recognize the signal feature according to the distribution pattern of the reconstructed vibration signal. Consequently, the mechanical performance of OLTC could be analyzed successfully. Calculated results of measured vibration signals have shown that the chaotic feature of vibration signals during the operation of OLTC is further verified due to the positive of the largest Lyapunov exponent obtained by the improved Wolf algorithm. The fault of OLTCs can be found clearly according to the variations of displacement vector of cluster centers calculated by K-means method. The obtained results can provide theoretical basis for the condition monitoring of OLTC in the viewpoint of chaos dynamics.