基于人工智能的大系统分层在线暂态稳定评估

在对比分析电力系统已有暂态稳定评估（TSA）方法的基础上,提出了一种以人工智能为主体的层次型大系统在线暂态稳定评估方案。以一个7机24节点系统为例,详细介绍了基于启发式推理和人工神经网络的方案实现过程及仿真测试结果,并重点讨论了基于样本的输入特征向量筛选方法。该方案无需时域积分运算,能提供故障筛选、稳定评级、主导失稳机组估计以及危险事件的故障临界切除时间（CCT）估计等多级稳定信息,较好地满足了在线TSA的要求。

COMBINATION OF HEURISTIC REASONING AND ANN TO REALIZE ON-LINE TRANSIENT STABILITY ASSESSMENT IN LARGE SCALE POWER SYSTEMS

With the comparison of multiple power system transient stability assessment (TSA) methods, a new hierarchicalTSA scheme based on artificial intelligence (Al) techniques is proposed. Realization of the scheme with the application ofboth rule-based reasoning and artificial neural networks, and digital simulation test results are presented with an example of7--machine 24--bus power system. The sample--based input features screening is also discussed in detail. The new schemeavoids time--consuming digital simulation and can provide multi-level stability information, including the stability levels, theleading instability generator or generator group and the fault critical clearing time (CCT) for dangerous cases.This project is jointly supported by "artificial intelligence based power system stability margin on-line assessment andpreventive control' (Guangdong Ph. D Foundation, No. 974112) and "intelligent power system stability assessment andenhancement' (Croucher Foundation).