An economical model for predicting machine tool energy consumption using machine learning

Developing an accurate and reliable prediction model for machine tool energy consumption is crucial for effective energy management and process optimization in machining processes. However, such models typically require large amounts of energy consumption data and additional sensor signals for training due to the complexity of machining parameters, which substantially increases the time and economic costs associated with developing and deploying energy prediction models. To address these challenges, this study proposes a Fully Connected-Radial Basis Function (FC-RBF) model and a Machining Processes-based Virtual Sample Generation (MP-VSG) method for predicting machine tool energy consumption from small and cost-effective datasets. The FC-RBF model achieves high-precision predictions under limited training data, while the MP-VSG method effectively generates representative virtual samples to mitigate information scarcity in small datasets. Experimental results based on machine tool energy consumption data demonstrate the effectiveness of the proposed approach. The integrated FC-RBF and MP-VSG framework achieves a Mean Absolute Percentage Error (MAPE) of 6.67% when trained on only 10% of the experimental dataset, outperforming commonly used models and demonstrating both high prediction accuracy and cost-effectiveness.