现有电网故障实时诊断与分析中存在2个主要问题：一是数据源的问题，即保护和开关动作信息上传不及时、时标不一致和信息缺失；二是误动拒动分析中存在2个重要盲点。同时为实现智能输电网的自愈性特征，又必须实现对电网故障的实时准确诊断和精确分析。针对上述问题，提出基于广域量测系统(wide area measurement system，WAMS)的电网故障快速诊断与分析的解决方案。同时还设计以多智能体技术为基础的面向数据服务的体系结构，并详细分析各智能体的主要结构和实现方法。最后，通过实际扰动试验的测试结果表明，该方法具有故障诊断速度快、故障诊断与分析结果准确等特点，尤其在误动拒动问题上可达到实时精确地分析，如在智能输电网中应用可有效降低大停电事故发生的概率。

Two main problems exist in current online fault diagnosis and analysis system for power grid. One is data source problem due to time delay for transferring information and asynchronous time of accepting data. The other is two key blind spots which exist in false-operation and fail-operation diagnosis. Meanwhile, real-time fault diagnosis and exact fault analysis are key self-healing characters of smart grid. A novel fast fault diagnosis and analysis approach of power grid based on WAMS was presented in this paper for solving problems mentioned above. Firstly, the architecture was designed for data service oriented and developed from multi-agent technology. Then the structure and methods of each agent were analysed in detail. Finally, it is shown by test results of disturbance experiment in a real power grid that this approach is fast and correct in fault diagnosis and analysis result. It will be very useful to reduce the risk of system blackouts when used in practical smart grid.

国家重点基础研究发展计划项目(973项目)(2004CB217904)。