Load forecasting for air conditioning systems using linear regression and artificial neural networks

The increasing demand for energy efficiency in industrial sectors necessitates innovative approaches to optimize energy consumption. This research addresses the challenge of accurately forecasting energy loads in air conditioning systems within the metal printing industry. Traditional forecasting methods often fail to capture industrial settings' complex, dynamic energy demands. This study aims to develop a precise load forecasting model by integrating Linear Regression (LR) and Artificial Neural Networks (ANN). Using real-world data from Kian Joo Can Factory Berhad, the ANN model demonstrated superior performance with a Mean Absolute Percentage Error (MAPE) of 11.44% and a Coefficient of Variation of the Root Mean Square Error (CVRMSE) of 4.214%. These findings suggest significant potential for reducing energy consumption, lowering operational costs, and improving equipment maintenance. Implementing machine learning algorithms in this context underscores their value in enhancing the efficiency, reliability, and cost-effectiveness of Air Handling Units (AHU) in industrial air conditioning systems.