RBF Neural Network And Its Application In The Superheated Steam Temperature Control System

With their powerful ability of approximating nonlinear functions and the characteristics of adaptive learning, parallel and distributed processing, strong robustness and fault tolerance, neural networks have been an effective approach to model and control the unknown and uncertain nonlinear systems. The applications of common multi-layer feed-forward neural networks are limited because of a large amount of computation time, slow convergence speed and being easy to plunge into local minimal. Radial basis function (RBF) neural networks have attracted extensive attention these years mainly because of its ability to model arbitrary nonlinear mapping, simple network structure, linear relationship between the network weights and the output such that a linear optimal algorithm can be employed for weight updating.Based on existing learning algorithms for RBF neural networks, novel learning algorithms are proposed in this paper, the applications of RBF neural networks with proposed algorithms in processes modeling and controlling are studied. Based on clonal selection algorithm and immune network theory, the memory, learning and self-organization abilities of artificial immune system are introduced into the selecting of the number and position of hidden layer radial basis function centers, the output layer weights are decided with the recursive least squares algorithm.There have been considerable interests in dynamic neural networks these years. M-RAN algorithm based RBF neural network is a dynamic neural network and suitable for process online modeling. Dynamic RBF neural network is applied to super-heated steam temperature system control, Simulations for power plant super-heated steam temperature control system using presented control system schemes are carried out.