论文部分内容阅读
一种水轮发电机组分层神经网络自适应预测控制器，采用多层前向神经网络，由两个分别用于预测和控制的子神经网络组成．提出一种误差反向传播在线实时自变学习率算法．仿真结果表明，该控制器对具有非线性、时变、多变量特性的水轮发电机组，具有良好性能．
A stratified neural network adaptive predictive controller for a hydroelectric generating set is composed of a multi-layer forward neural network and two sub-neural networks respectively used for prediction and control. This paper presents a real-time self-learning learning algorithm of error back propagation. Simulation results show that the controller has good performance for hydro-generator with nonlinear, time-varying and multivariable characteristics.