为了更加有效地对变压器绕组松动故障进行监测与识别，提出了一种变压器绕组松动故障的混沌特征分析方法。首先，针对振动信号的混沌动力学特性，采用互信息量法和G-P算法分别确定延迟时间和嵌入维数，对变压器振动信号进行相空间重构；其次，通过判断最大Lyapunov指数是否为正，进而证明变压器振动信号的混沌特性，在此基础上分析不同程度的绕组松动故障对相空间轨迹变化的影响；最后，将关联维数、Kolmogorov熵和最大Lyapunov指数作为一组混沌特征用以量化变压器绕组发生松动故障前后振动信号的混沌特性。结果表明：变压器振动信号的最大Lyapunov指数均大于0，证实了其具有混沌特性，所得到的混沌特征能够有效反映变压器绕组松动故障。研究结果为变压器绕组松动状态监测提供了一种理论依据。

In order to monitor and identify transformer winding looseness fault more effectively, a chaotic feature analysis method for transformer winding looseness fault is proposed. Firstly, according to the chaotic dynamic characteristics of vibration signals, mutual information method and G-P algorithm are used to determine the delay time and embedding dimension respectively to reconstruct the phase space of transformer vibration signals; Secondly, the chaotic characteristic of transformer vibration signal is proved by judging whether the maximum Lyapunov exponent is positive. On this basis, the influence of different degrees of winding looseness fault on the change of phase space trajectory is analyzed; Finally, correlation dimension, Kolmogorov entropy and maximum Lyapunov exponent are used as a set of chaotic features to quantify the chaotic characteristics of vibration signals before and after the happening of transformer winding looseness fault. The results show that the maximum Lyapunov exponents of transformer vibration signals are all greater than 0, which proves that they have chaotic characteristics, and the obtained chaotic characteristics can effectively reflect the looseness fault of transformer windings. The research results provide a theoretical basis for monitoring the loosing state of transformer windings.

TM41

国家自然科学基金项目(51577050)；国网江苏省电力有限公司科技项目(J2021053)