Identifying Early Stator Fault Severity in DFIGs Based on Adaptive Feature Mode Decomposition and Multiscale Complex Component Current Trajectories
Shouwang Zhao,Yu Chen,Feng Liang,Sichao Zhang,Nadeem Shahbaz,Shuang Wang,Yong Zhao,Wei Deng,Yonghong Cheng
DOI: https://doi.org/10.1109/tim.2024.3443347
IF: 5.6
2024-09-03
IEEE Transactions on Instrumentation and Measurement
Abstract:Early fault detection is critical for ensuring reliable operation in wind power generation systems employing doubly fed induction generators (DFIGs). Although the widespread use of motor current signature analysis (MCSA) for noninvasive fault detection, early-stage faults in DFIGs often present with weak characteristic fault information, posing challenges for detection amidst noise and interference signals. This article proposes a novel method to assess the severity of early stator interturn short circuits (ITSCs) in DFIGs by combining adaptive feature mode decomposition (AFMD) with multiscale analysis of complex component current trajectories. AFMD is applied to perform quadratic time-frequency decomposition of current signals, resulting in a series of modal components with diverse frequency contents. By selecting the low-frequency current modal components with the fundamental frequency and low-order harmonics, high-frequency interferences, and noise are effectively filtered out. The extracted low-frequency current modal components undergo multiscale Park vector trajectory analysis, thereby enhancing the expression of weak fault characteristics associated with early ITSCs. Further analysis involves the extraction of negative sequence current (NSC) and zero-sequence current (ZSC) components from the low-frequency modal components. The residual ZSC signal is constructed by filtering out the fundamental frequency and low-order harmonic components from the ZSC component of the low-frequency modal. Analysis of the symmetrized dot pattern (SDP) of the residual ZSC signal captures the evolution process of early-stage ITSCs with a low number of turns. In addition, the NSC of the low-frequency modal components is qualitatively and quantitatively evaluated for early low-turn insulation short circuits of varying severity levels. Experimental validation on a 100-kW DFIG test platform demonstrates the efficacy of the proposed method in enhancing the detection and diagnosis of early-stage ITSC faults.
engineering, electrical & electronic,instruments & instrumentation