Application Of Neural Network In Modeling Of Power Plant

Aim to the lack of forward neural network, this paper put out a kind of Elman recurrent neural network. This Elman neural network include the feedback memory links in its inner part, and overcome the shortcoming of forward neural network that can only make static mapping. Secondly, the paper detailed the advantages and disadvantages of two kinds of modeling methods (physical model and neural network model) concerned with thermal process of power plant, and put forward a new kind of physical neural network modeling method. This kind of method overcomes the weaknesses of purely physical model (complicacy, too many parameters, lower accuracy, etc) by adding neural network to its inner part; on the other hand, the neural network makes modeling process close the actual thermal system, and guarantees the forecast accuracy. Finally, this paper simulated the drum's water level and pressure signals of NO. 2 unit of Tianjin's PanShan power plant using this kind of physical neural network method, and the result was acceptable. Dai Yue (Control Theory and Control Engineering)Directed by prof. YuGuang Niu...