复杂电网行波全景特征挖掘与故障定位新原理研究

电网故障产生宽广频带行波，可用于挖掘丰富故障信息。传统行波定位方法仅单一利用局部波头信息，面向交直流混联电网，存在波头准确提取困难、故障特征混叠严重且呈现非线性分布等难题，导致定位可靠性不高；而多点检测行波具有相关性和冗余性，可通过挖掘时空故障特征提高行波定位的精度和可靠性。为此，项目拟研究：1）考虑时频域行波波形形成机理，给出行波全景波形的理论诠释和数学描述，实现对其全面定性认识；2）探讨最优时频特征联合分布表达的时频分析方法，研究利用深度学习的深层抽象特征挖掘方法，凸显故障敏感特征，实现时空多层次全景特征准确可靠提取；3）分析故障产生机制变化和演化过程对行波全景特征的影响机理，探索时空全景特征与故障位置的映射关系，揭示行波全景波形唯一性理论，提出时空多层次全景特征融合的行波定位新原理。项目旨在构筑一套准确可靠的复杂电网行波全景特征挖掘和故障定位新方法，为电网安全稳定运行提供技术保障。

When a fault occurs, a wide-band traveling wave is produced, which can be used to excavate the abundant fault information. The traditional traveling wave-based fault location method is just based on the traveling wave-fronts. It exists the following difficulties, such as: accurate extraction for traveling wave-front, serious aliasing of complicated fault features, nonlinear distribution of fault features in time-space domain. The traditional traveling wave-based fault location method is with low reliability. While multi-point detection traveling wave has correlation and redundancy, it can increase the accuracy and reliability of the fault location by mining temporal and spatial multi-level fault features. Thus, there are three main contents. First, considering the formation mechanism of traveling wave in the time and frequency domains, the panoramic waveform theoretical interpretation and mathematical description are proposed, which realizes the comprehensive and qualitative cognition for traveling wave panoramic waveform. Second, a novel time-frequency analysis method used for the optimal distribution representation of time-frequency features is deeply analyzed, and deep level features is mined by using deep leaning. The temporal and spatial multi-level fault features can be accurately and reliably extracted, which can prominently reflect the fault sensitive features. Finally, the panoramic waveform influenced by fault production mechanism and its evolution process is systematically analyzed. The mapping relation between temporal and spatial panoramic features and fault location is explored. The uniqueness theory of traveling wave panoramic waveform is revealed. On the basis, a novel traveling wave-based fault location method by fusing temporal and spatial multi-level fault features is proposed. The project aims to proposed a set of accurate and reliable panoramic features mining method and a novel fault location principle, which can provide technical support for safe and stable operation of the power grid.