中文摘要：

借鉴Shannon小波熵在电力系统故障检测中的应用，将非广延Tsallis熵与小波分析相结合，构造Tsallis小波熵算法并提出一种基于Tsallis小波熵电力系统暂态信号特征提取的方法。从广延特性对Shannon小波熵展开讨论，指出其对复杂暂态信号进行特征提取时存在局限性；通过对Tsallis小波熵和Shannon小波熵区别与联系的分析，指出Tsallis小波熵更适合对具有非广延特性的小波系数（或各尺度重构信号）进行二次数据挖掘；利用数理统计法揭示Tsallis小波时间熵和能量熵在暂态信号特征提取及其复杂度分析方面的应用机制，并论证其理论可行性；仿真及实验结果证明：较Shannon小波熵，Tsallis小波熵应用更为灵活，特别是其小波能量熵在对电力系统信号复杂度正确表征的同时，能准确提取暂态信号特征。

英文摘要：

Based upon the application of Shannon wavelet entropy （WE） to fault detection of power system, Tsallis WE algorithm was proposed and implemented with the combination of the Tsallis entropy theory and the wavelet analysis to provide a novel approach for feature extraction of transient signals in power system. In terms of the extension of Shannon entropy, the limitations of Shannon WE applied to feature extraction of transient signals were discussed. Studying on the relationships and differences between Tsallis WE and Shannon WE, it can be concluded that Tsallis WE is more suitable for secondary data mining based on non-extensive wavelet coefficients （or reconstruction signals） under different scales. Taking wavelet time entropy （WTE） and wavelet energy entropy （WEE） as the illustration respectively, the application mechanism of Tsallis WE to signal feature extraction and complexity estimation was presented through mathematical statistics of data dispersion. The simulation and experiment results indicated that Tsallis WE is more valid and flexible than Shannon WE for this application. In addition, Tsallis WEE can extract features of transient signals precisely and estimate signal complexity properly in power system.