Modeling of a Fossil Fuel-Fired Power Plant Process and Analysis of Prediction Success of Main Steam Pressure Parameter

Although fossil fuel fired power plants, which are frequently used nowadays, have critical importance due to electricty and steam production for industry and manufacturing works, they are under criticism due to their negative impacts on the environment. In this study, an operation and process analysis of a 135 MW fossil fuel fired power plant located at Turkey is carried out and the plant is modeled with Artificial Neural Networks (ANN), which is one of contemporary artificial intelligence method, selecting 19 critical input parameters. Previous process values of the parameters are obtained from the power plant, each contains 1440 process data, blended with data mining techniques, and main steam pressure parameter, which is one of the most important output parameter of the plant, is estimated by various approaches and experiments. The results are compared with the Multiple Linear Regression (MLR) approach, one of the frequently used statistical forecasting method in literature. In this estimation study, root mean square error and determination coefficient terms are used for comparison, and the outputs are founded as 0.0994 and 0.0039 for ANN model, 0.970 and 0.0172 for MLR model respectively. Additionally, a comparison benchmarking is made with a similar study in literature, and it is shown that the parameters of ANN model and selection of input variables are found very successful