机构地区: 浙江师范大学计算机科学与工程学院计算机科学与技术系
出 处: 《电力系统自动化》 1998年第7期35-39,共5页
摘 要: 分析了现有的基于人工神经网络的电力系统实时控制器设计与训练中存在的问题,在此基础上提出了一种新的具有训练样本在线自动生成能力的电力系统实时控制器设计方案。从理论上论证了该方案的可行性,并依据这一方案实现了神经网络电力系统稳定器的设计。在线训练与控制的仿真结果均充分显示出该方案的优越性。 The problerns existing in designing and training neural network based power system real-time controllers(NNPSRC) are discussed in this paper. For the NNPSRC. a new scheme that can automatically generate on-line trainingsamples and realize self-learning is therefore proposed. The feasibility of the new scheme is proved theoretically. Based onthis scheme. a novel neural network based power systern stabilizer (NNPSS) is then designed. Simulations on the on-lineself-learning and control performance of the designed NNPSS show the unique advantages of the proposed scheme.