A comparative study of classical static and AI models predicting energy consumption in machining for Learning Factories

One of the main targets of Learning Factories (LF) is to minimise the impact on the natural environment. By following this trend, sustainability has been widely referred in smart factories, and the energy consumption is identified as a key indicator to exam the green machining processes. Hence, acquiring robust and accurate energy predictive models is a decisive step to enable sustainable manufacturing in a learning factory environment. There is a big controversy regarding the choice of using classical or artificial intelligence (AI) techniques to form up those models. Thus, this paper carries out a comparative study of regression and AI energy predictive models of milling processes and provides a comprehensive analysis of their performance based on the requirements of smart factories. Three modelling methodologies: response surface, artificial neural networks, and adaptive neuro-fuzzy systems, are employed to construct the models and a new criterion set is adapted for the assessment. The results disclose the best model and method for various application requirements for sustainable Learning Factories.