基于自校正方法的多机电力系统稳定器设计  

Multi-machine Power System Stabilizers Based on Adaptive Critic Design

作　　者：乔嘉赓[1] 闵勇[1] 吴小林[2] 

机构地区：[1]清华大学电机系电力系统国家重点实验室,北京100084 [2]北京电力公司,北京100073

出　　处：《现代电力》2006年第6期1-5,共5页Modern Electric Power

基　　金：国家重点基础研究发展计划项目(973项目)(2004CB217904);国家自然科学基金资助项目(50323002)。

摘　　要：低频振荡危害了电力系统的安全和稳定,电力系统稳定器(PSS)是抑止低频振荡的有效方法。但是在复杂系统中,配置和整定电力系统稳定器是一个难题。本文利用自校正方法(ACD)设计了一类新型非线性电力系统稳定器。该稳定器通过在线训练进行设计,摆脱了对系统数学模型的依赖,同时能够对系统中不确定性予以充分考虑,适用于复杂多机环境中PSS的优化设计。训练成功的PSS能够在各种工况下有效抑制系统振荡,同传统稳定器相比具有很强的鲁棒性。文中以四机二区域系统为例,证实了这种新型电力系统稳定器用于多机系统时的优良特性。Low frequency oscillations always take place in large-scale power system with huge-flow and long-distance tielines, which seriously threatens the system operation with stability and safety. Power System Stabilizer （PSS） is a kind of effective equipment in restraining such oscillations, but in multiple power system, it is difficult to programming PSSs at different positions. Adaptive Critic Design （ACD） theory is introduced to design a kind of novel PSS to meet the need of multi-system. The PSS is designed by training on line, and system mathematical model is not required in the training process. Well-trained PSS copes with the uncertainty of system well, so it can work more effectively in restraining all kinds of oscillations rather than conventional PSS. The simulation results in four-generator two-area system confirm the excellence of new PSS.

关 键 词：电力系统稳定器 自校正设计 动态规划 人工神经网络 强化学习 

分 类 号：TM712[电气工程—电力系统及自动化] TM761