Modeling and prediction of boiler drum in a thermal power plant

Abstract

The boiler drum is non-linear complex industrial equipment due to its phase change and inverse response behavior. One of the crucial control parameter of the industrial process is the drum level, reflected mainly by the load and feed water indirectly. As a result, the prediction of boiler drum level in thermal power plant which needs the modeling of boiler drum using data-driven approach is chosen in this study. This paper proposes a heuristic technique to develop Artificial Neural Network (ANN) model to predict the level of water in boiler drum. Experimental data are obtained from an operational power plant (210MW Thermal Power Station Expansion-1, NLC India Ltd., Neyveli). The data obtained from the plant is used to predict the drum level using static and dynamic model neural network. The static model is developed using feed forward architecture and the dynamic model is developed using NARX neural network. The results of static model as well as dynamic model have been compared and presented.