Improving Energy Baseline Models Using Artificial Neural Networks in Industrial Facilities: a Case Study of Small and Medium-Sized Company

Abstract

This study presents an innovative approach for developing energy-efficient reference models specifically designed for industrial environments, with a particular focus on a plastic injection moulding plant. The research analyses three advanced modelling techniques, linear and non-linear regression, Artificial Neural Networks (ANN), and Adaptive Neuro-Fuzzy Inference Systems (ANFIS), in order to address challenges such as production mix data, unidentified losses, and variable environmental conditions. A set of performance metrics has been established to evaluate the models, and computational experiments have been conducted to validate the results. Despite uncertainties in the input variables, the ANNs have demonstrated high accuracy and simplicity in estimating baseline energy consumption, making them an effective tool for improving energy performance in the industrial sector. Furthermore, the study highlights the importance of integrating sustainable practices that not only improve operational efficiency by reducing wastes and conserving resources, but also deliver long-term financial benefits and contribute to pollution reduction through decreased energy demand.