Machine learning model for hourly power forecast in combined cycle plants using MARS and whale optimisation

• The net hourly output electrical power (PE) of CCPPs is analysed in detail. • The developed WOA/MARS model has predicted satisfactorily the PE. • The MARS model was compared with Ridge, Lasso and Elastic-net techniques. • The correlation coefficient of the WOA/MARS - relied model is about 0.97. • The physico-chemical operation input variables are studied in depth. The objective of this investigation was to develop a model capable of accurately and efficiently identifying and calculating the net hourly load output of a Combined Cycle Power Plant (CCPP), utilising a dataset comprising 9,568 actual CCPP samples. This study introduces the population-inspired Whale Optimisation Algorithm (WOA) in conjunction with Multivariate Adaptive Regression Splines (MARS), an effective machine learning technique. For comparative purposes, Lasso, Elastic-net, and Ridge regressions were also applied to the same experimental dataset with the WOA optimiser. The principal operational input variables—ambient temperature (AT), atmospheric pressure (AP), relative humidity (RH), and exhaust steam pressure (vacuum)—were monitored throughout the study. The findings demonstrate that the WOA/MARS model effectively identifies and quantifies the relative importance of independent variables influencing the full load output electrical power of the CCPP. By employing the WOA/MARS regression approach, determination and correlation coefficients of 0.9394 and 0.9692, respectively, were achieved, with a Root Mean Square Error (RMSE) of 4.1970, indicating a high degree of predictive accuracy and reliability. Comparative analysis revealed that the predictive performance of the Ridge, Lasso, and Elastic-net models was inferior to that of the WOA/MARS model. The strong agreement between the WOA/MARS model predictions and the experimental data underscores its superior efficiency and potential for practical application in optimising CCPP performance and the operation of energy systems.