基于神经网络和模糊理论的电力系统动态安全评估

在实用动态安全域的基础上提出了实用动态安全域的神经网络和模糊理论方法，构建了用于拟合动态安全域的ANN拟和器和用模糊理论在安全边界上构造一个模糊中间带的模糊识别器，并对该方法中的系统灵敏度和超平面度进行了分析。最后对4机11节点测试系统进行了仿真计算。实验表明，用神经网络拟合的动态安全域其拟和合度和精度均高于传统最小二乘法的拟合，其模糊识别器进一步减少了临界点的误判和在安全域边界上稳定和非稳定点的误判。

Dynamic Security Assessment of Power System Based on ANN and Fuzzy Theory

The paper proposed the ANN and fuzzy theory method for the assessment of practical dynamic security regions(DSR) of power systems, and built the imitator for dynamic security regions and the fuzzy identifier based on fuzzy theory. The sensitivity and hyper-plane parameter of the method were analyzed. The simulation on the four-machine and eleven-bus test system proves that the DSR imitated by ANN is more accurate than the traditional least square method, and the fuzzy identifier can availably reduce the wrong judges of the border points and the stable or unstable points in the vicinity of border.