Abstract: With the efforts intensified in energy conservation and emission reduction policies, great importance has been attached to the development and research of energy-saving and consumption-reducing technologies for thermal power plants from state level. Aiming at current unreasonable way of boiler blowing on the heating surface of the boiler, the pollution rate (FF) is used to characterize and represent the impacts of the cleanliness of the heated surface on the heat transfer of the boiler heating surface. The forecasting model of the heating surface cleanliness based on PSO-Elman neural network is established. Using the particle swarm optimization (PSO) algorithm in combination with Elman dynamic neural network, the Elman neural network structure is first determined according to the number of input and output parameters. Then by taking advantage of PSO algorithm the weights and thresholds of the neural network are optimized. Finally the derived optimal values and thresholds are assigned to the Elman neural network as the initial value for network training such that the heating surface cleanliness forecast model is established based on PSO-Elman neural network. Through simulations of specific examples, satisfactory forecasting accuracy is obtained and hence the effectiveness of the proposed method is verified.

Key words: intelligent generation, heating surface, clean state, PSO-Elman, prediction