水压频率的递归网络灵敏度预测控制研究  

On Sensitivity Predictive Control of Water Hammer and Frequency with Memory Recurrent Network

作　　者：曾杰[1] 梁年生[1] 刘育骐[1] 郭创新[1] 

机构地区：[1]华中理工大学电力工程系

出　　处：《华中理工大学学报》1996年第9期91-93,共3页Journal of Huazhong University of Science and Technology

摘　　要：提出一种对水轮发电机组水压频率进行综合调节的记忆递归网络灵敏度预测控制器．控制器由预测网络和控制网络组成，预测网络向控制网络提供控制灵敏度信号．两个网络均根据学习和控制的收敛与稳定性定理自适应学习．仿真结果表明控制器能在线实时控制具有非线性、时变、多变量特性的水轮发电机组，有效控制水压．A sensitivity predictive controller with memory recurrent network for use in the comprehensive control of the water hammer and the frequency is proposed. The controller consists of a predictive and a control network. The predictive network provides the sensitivity signal to the control network. According to the theory of convergence and stability on learning and control, the two networks are adaptive to the learning rate. Simulation results show that teh controller is good in the control of nonlinear, timevarying, multivariable hydrogenerator unit and the water hammer.

关 键 词：水轮发电机组 水压频率 递归网络 预测控制算法 

分 类 号：TM312[电气工程—电机]