提取和分析了电能扰动信号的非线性动力学参数———信息熵、Kolmogorov 熵和最大 Lyapunov 指数，结果表明单一参数对不同的电能扰动信号有较好的区分能力；然后添加幅度熵、相位熵，总共5维特征作为电能扰动信号的特征矢量集，应用高斯混合模型对暂降、中断、谐波、振荡、切痕、尖峰、暂升、波动8种单一电能扰动信号，以及暂升加谐波、暂降加谐波、中断加谐波、波动加谐波4种复合扰动信号进行建模与识别。结果表明：非线性动力学参数能较好地分辨这12种扰动信号，当5种特征进行组合后获得了96.42%的识别率，识别效果较好。%This paper extracts and analyzes nonlinear dynamic parameters of power disturbance signals including information entropy,Kolmogorov entropy and maximum Lyapunov exponent and results indicate that single parameter has better distin-guishing capacity on different power disturbance signals. Total five dimension features such as amplitude entropy and phase entropy are added as characteristic vector set,and Gaussian mixed model (GMM)is used for modeling and identification for eight kinds of single power disturbancesignals including transient sag,interruption,harmonic,oscillation,voltage notches, swell and fluctuation,as well as for mixed disturbance signals including swell plus harmonic,transient sag plus harmonic, fluctuation plus harmonic and interruption plus harmonic. Results indicate that nonlinear dynamic parameters can well iden-tify these 12 kinds of disturbance signals and when five characteristics are mixed,it is able to acquire identification rate of 96.42% which means better identification effect.
展开▼
