电力系统动态安全分析的ANN动态安全域

ANN BASED DYNAMIC SECURITY REGIONS OF DYNAMIC SECURITY ANALYSIS OF POWER SYSTEM

采用人工神经网络(ANN)多维空间的强拟合特性,能够较精确地拟合动态安全域,减少了临界点以及稳定点和非稳定点的误判,克服了起平面理论近似拟合电力系统动态安全域误差较大的不足。同时,对于某个具体的故障,采用临界切除时间t_(cr)建立ANN稳定裕度拟合器,确定电力系统中运行点的稳定裕度。通过PSASP中的3机9节点模型验证了此理论的正确性,而且准确性优于采用超平面理论拟合的动态安全域。

Using the strong imitation characteristics of high-dimensions spaces based on ANN, this paper can imitate the dynamic security regions (DSR) more accurately, reduce mistakes in judging border points, the stable points or the unstable points, and overcome the bigger error brought by the hyper-plane theory approximately imitating power system DSR. At the same time, to a given fault, this paper builds the ANN stable remain imitation machine used by the critical cutting-time (CCT) and defines the stable remains of running points on power system. The test on the 3-machine 9-bus model of PS ASP proves the correctness of the theory and the higher accuracy of DSR than that imitated by the hyper-plane theory.