基于参数化时频分析的电网故障行波信号精确识别方法研究

项目名称： 基于参数化时频分析的电网故障行波信号精确识别方法研究

项目编号： No.51507015

项目类型： 青年科学基金项目

立项/批准年度： 2016

项目学科： 电工技术

项目作者： 席燕辉

作者单位： 长沙理工大学

项目金额： 20万元

中文摘要： 电网故障行波信号的可靠准确识别是故障行波保护与定位的基础。现有行波识别一般采用非参数化时频分析方法，难以实现对强电磁干扰条件下轻微故障行波信号和复杂折反射行波信号的可靠辨识。为此，项目提出并研究基于参数化时频分析的行波识别方法：1）研究电网故障行波信号特征，构建Alpha稳定分布-时变自回归模型，准确描述行波信号的时域动态特性，并研究基于该参数化模型的噪声干扰消除方法；2）针对模型参数突变的延迟性，研究以参数噪声统计特性估计为核心、带时变遗忘因子的自适应粒子滤波算法，实现行波参数的准确估计；3）利用精确的参数信息从时域和时频域提取关联特征量，特别是与故障信息密切相关的Alpha稳定分布参数和瞬时频率，时频联合分析准确可靠地感知强电磁干扰下微弱故障行波信号和多次折反射复杂行波信号。本项目提出行波信号的新型识别参数，实现受扰行波信号的准确识别，为构建新型故障行波保护与定位原理提供参数支持。

中文关键词： 行波识别；参数化时频分析；故障特征提取；时变自回归模型；Alpha稳定分布

英文摘要： The reliable and accurate identification of traveling wave lays a foundation for traveling wave protection and location. The existing traveling wave identification methods are usually based on non-parameterized time-frequency analysis. It is difficult to achieve reliable identification of some minor and complex traveling wave signals, which are influenced by strong electromagnetic interference and complex refraction and reflection. Therefore, the project proposes and investigates the identification method of traveling wave signals based on parameterized time-frequency analysis. 1) The characteristics of traveling wave signals are analyzed. Time-varying autoregressive (TVAR) model with alpha stable distribution is constructed to describe temporal dynamic characteristics of traveling wave signals. And then, the project explores the noise interference cancellation method. 2) For the delay of the parameters’ abrupt changes, the project investigates the adaptive particle filtering methods with variable forgetting factor, which focus on estimating the statistical characteristics of parameter noise. The accurate parameters are obtained by use of the proposed optimization algorithms. 3) The correlation feature values are extracted from time domain and time-frequency domain by using a wealth of precise parameters information. Especially the feature values, such as alpha stable distribution parameters and instantaneous frequency, are closely related to the fault information. On the basis, the minor and complex traveling wave signals are perceived reliably and accurately by joint time-frequency analysis of feature values. Some novel identification parameters are proposed, and the disturbed traveling wave signals are accurately identified in the project, which will provide parameters support for construction of novel traveling wave protection and fault location principles.

英文关键词： traveling wave identification;parameterized time-frequency analysis;fault feature extraction;time-varying autoregressive model;alpha stable distribution