This paper describes a systematic approach to predict the water level in the drum of a steam boiler with the help of artificial neural networks (ANN). The parameters of the model can be obtained from the physical dimensions and characteristics of the boiler. The frequency of deviations and the degree of deviation of the water level in the drum can be significantly reduced by the ANN modeling of the water tube boiler water feed system to the drum. The ANN model to be applied for the boiler feed system in the power plant will not only increases the efficiency of the system but shall considerably reduce the tripping of the power plant. The model so developed can be used for synthesis of model-based control algorithms of boiler system.
