应用人工神经网络进行电力系统动态安全评价的新方法

A NEW METHOD FOR POWER DYNAMIC SECURITY ASSESSMENT USING ARTI FI CIAL NEURAL NETWORKS

本文提出了一种应用人工神经网进行电力系统动态安全评价的新方法,把故障前系统的稳态运行参数作为特征量,摇摆过程中发电机转子间的最大相对摇摆角作为系统稳定性的量度,并证明了它们之间呈连续映射关系。随后表明了用三层前向网络实现这类映射的可行性。为减少神经网络的训练时间,本文提出了网络“分解- 集结”方法和映射与分类相结合的观点。6机、22节点系统的试验结果证明了本文提出方法的有效性。

This paper presents a new method for power system dynamic security assessment using artificial neural networks with the operating conditions of pre-fault system taken as features, and the maximum relative swing angle of generator rotors in swinging process as the measurement of the system,s stability. The continuous mapping significance between them is proved. This followed by showing the feasibility of realizing the mapping with three-layer forward networks. In order to reduce training time of neural networks,a network 'Decomposition-aggregation' technque and an idea to combine mapping and classifying are proposed. The test result of 6-generators, 22-buses system shows the effectiveness of the proposed method.