Prediction of energy efficiency, power factor and associated carbon emissions of machine tools using soft computing techniques

Abstract

Due to growing environmental issues and strict carbon emission (CEM) guidelines imposed throughout the globe, low-carbon emission of machine tools, which aims to reduce carbon intensity and improve process efficiency, has evolved as an emerging issue that has encouraged a lot of research into accurate prediction of energy-related performance characteristics such as energy efficiency (EE), power factor (PF) and associated CEM of machine tools. In practice, EE and PF are the two significant indicators of a machine tool's effective electrical energy utilization. In the present study, three soft computing techniques, multi-gene genetic programming (MGGP), least square-support vector machine (LS-SVM) and fuzzy logic, are employed to model a machine tool's EE, PF and associated CEM. The experiments were performed on a CNC lathe machine tool to capture the data required for development of models using soft computing techniques. The performance of the models was evaluated on six statistical indicators i.e. coefficient of determination (R2), mean absolute error, root mean square error, mean square error, sum of squared error and relative percentage error. In all cases, R2 values were found in the range of 94–99% in training and 84–94% in testing, signifying a strong relationship between the experimental and predicted values. The models' comparative performance evaluation reveals that the LS-SVM consistently outperforms the MGGP and fuzzy logic. Further, hypothesis testing i.e. mean paired t-test and variance of F-test, were conducted to validate the goodness of fit of the developed models. The developed models can be used to eliminate the need for advanced costly laboratory set-up and time-consuming measurement procedures required for performing experiments.