A hybrid genetic algorithm and linear regression for prediction of NOx emission in power generation plant

• Main Authors: Bunyamin M.A., Yap K.S., Aziz N.L.A.A., Tiong S.K., Wong S.Y., Kamal M.F.
• Other Authors: 55812855600
• Format: Conference paper
• Published: Institute of Physics Publishing 2023
• Subjects: Estimation; Gas emissions; Genetic algorithms; Nitrogen oxides; Optimal systems; Accurate prediction; Emission estimation; Explanatory variables; Hybrid genetic algorithms; Model parameters; Optimal solutions; Power generation plants; Prediction errors; comparative study; electricity generation; error analysis; estimation method; genetic algorithm; industrial emission; numerical model; optimization; parameterization; regression analysis; Forecasting

This paper presents a new approach of gas emission estimation in power generation plant using a hybrid Genetic Algorithm (GA) and Linear Regression (LR) (denoted as GA-LR). The LR is one of the approaches that model the relationship between an output dependant variable, y, with one or more explanatory variables or inputs which denoted as x. It is able to estimate unknown model parameters from inputs data. On the other hand, GA is used to search for the optimal solution until specific criteria is met causing termination. These results include providing good solutions as compared to one optimal solution for complex problems. Thus, GA is widely used as feature selection. By combining the LR and GA (GA-LR), this new technique is able to select the most important input features as well as giving more accurate prediction by minimizing the prediction errors. This new technique is able to produce more consistent of gas emission estimation, which may help in reducing population to the environment. In this paper, the study's interest is focused on nitrous oxides (NOx) prediction. The results of the experiment are encouraging. � Published under licence by IOP Publishing Ltd.