基于稀疏表示的电力线路故障特征分析和继电保护研究

Fault features are the basis for fault detection and protective relaying. The existing methods for power systems to obtain fault features rely on fault analysis. Classical fault analysis is processed by building physical models and solving the mathematical equations, which satisfy electric circuit and electromagnetic field constraints. However, as for large interconnected power gird, existing mathematical methods can hardly obtain the analytical solution, for example, for the fault in the power system with power lines based on distributed RLC parameters model, and the fault in AC/DC hybrid power system including converters. This project proposes to obtain fault features by avoiding directly solving the mathematical equations. Instead, our proposed method exploits fault features by utilizing the correlations between fault signals. The research is planned to use limited known fault signals to reconstruct unknown fault signals, and to represent fault features based on known signals, and to present a new intelligent non-setting protective relaying algorithm. This detailed research contents contain to analyze the relationship between fault signals and non-faulty signals, the correlation between fault signals, the sparse representation of fault signals, the generation of over-complete fault signals basis, the fault detection and protection via sparse representation. The research goal is to overcome the shortcoming of existing fault analysis, and lays foundations for the new generation of intelligent protection technology.

故障特征是故障检测和继电保护的基础。现有的电力系统故障特征获取依赖故障分析。经典的电力系统故障分析通过建立物理模型，列写并求解满足电路和电磁场理论约束的数学方程来实现。理想的故障特征描述是数学方程的解析解。然而，针对目前的互联大电网，现有的数学工具难以获得电力系统故障特征的解析表达，如基于分布参数RLC模型的电力线路故障和含换流器的混联电力系统故障。本项目拟避免直接求解故障方程，而利用故障信号之间的相关性，用少量的故障信号重构全部的故障信号，实现基于部分已知故障信号的故障特征描述，并在此基础上，构造智能的免整定的新型继电保护方法。项目拟具体研究故障信号与非故障信号之间的关系以及故障信号之间的相关性，故障信号的稀疏表示，过完备基矩阵的生成，基于稀疏表示的故障检测和继电保护算法。项目成果有望突破现有故障分析的桎梏，为新一代智能继电保护技术的实现奠定基础。

项目聚焦基于稀疏表示的电力系统故障特征提取，突破了传统电力系统故障特征依赖物理模型数学方程求解的思路，解决了电力系统复杂故障以及交直流互联电网故障特征难以解析的问题；提出了基于结构化稀疏表示和抗噪声干扰能力强的故障选相方法；提出了用动态变形算法量化故障波形和参考波形的相似度，利用核函数技巧的稀疏表示故障区段识别方法，解决了目前配电网单端行波测距法无法确定故障分支的问题；提出了单端行波测量条件下的辐射型配电网故障测距方法，解决了配电网单端行波测距法无法消除干扰折反射波头的问题； 提出了基于稀疏表示的中性点非有效接地系统故障选线方法，拓展了选线技术方案；发现了行波在故障点会发生折反射而在雷击点不会发生折反射的规律，提出了基于单端行波信息的雷击干扰识别方法。构造了故障定位特征量并形成了故障定位判据，提出了直流配电线路单端量故障定位方法，提高了故障检测抗干扰能力；提出了利用配电线路网络拓扑和重合闸行波的单端故障测距方法，解决了单端行波法受配电线路分支影响的故障点反射波识别问题，消除了线路分支对故障点反射波的影响，实现了单端行波精确测距。研制了配电线路故障检测和继电保护样机，设计了基于稀疏表示的配电线路故障数据采集处理和故障检测硬件系统，开发了基于稀疏表示的配电线路故障检测系统软件。项目研究拓展了电力系统故障数据特征提取的方法，提出了免整定的故障检测和继电保护技术，减少了继电保护受现场运行整定影响的不确定性，保证了继电保护理论与实践的一致性。