Prediction of Electric Power Generation in Turbine Systems

This research utilizes a secondary data set of 100 samples for exploring predictive modelling of electric power generation in turbine system. The prediction models will be trained utilizing this data set. There are four attributes associated with each sample - fuel type (coal, natural gas, nuclear energy), turbine efficiency, steam pressure, and electric power generated. The basic goal of this work is to establish and compare the two prediction models which are linear regression and generalized regression neural network (GRNN) for anticipating generated electric power based on three inputs- fuel type, turbine efficiency and the steam pressure. MATLAB has been used to develop linear aggression and GRNN models. Their prediction accuracy is assessed in terms of mean square value of prediction error. The Linear aggression model is based on a linear relationship between inputs and output. On the other hand, the GRNN model is based on nonlinear approach using radial basis function network for identifying complex patterns in data. This technique potentially offers more accurate predictions.