Superiority of Data Mining Techniques to Predict the Amount of Power Generated by Thermal Power Plants

Abstract

This paper presents the superiority of data mining techniques in predicting the amount of power generated by thermal power plants, over the traditional approaches that use thermodynamic laws or the power plant manufacturer’s guides. The paper first compares between amount of power calculated using thermodynamic laws, and the amount of power predicted using manufacturers’ guides with the actual power generated. Then prediction model was built to predict the amount of generated power using the controllable parameters at turbine inlet. Models were evaluated using separate test sets, or cross validation in case of small sets. The values predicted by this model is then compared with actual and other predicted values to prove that data mining tool is most accurate than traditional methods.