Prediction of cutting forces in turning Ti6Al4V in different cooling environments using a multiple signal data fusion approach

This study investigates the relationship between cutting forces and real-time energy consumption signals during the turning of titanium alloys, as well as the impact of various cooling methods on these forces. A hybrid approach combining energy signal analysis and forward regression modeling was applied to model the cutting forces. This work aims to model and predict cutting forces using individual and multiple signal data fusion approaches using data obtained from energy signals, such as reactive power, power factor, apparent power, and active power. A random forest regressor was preferred for model set-up, and the Scopt Optimizer was preferred for fine-tuning the hyperparameters. Six methods were used to score the features based on their importance level. The results showed that multi-energy signal data fusion outperforms the models based on single signals in predicting cutting forces. This method offers a new approach for optimizing metrics, such as the main cutting force (Fc) and feed force (Ff). The MAPE value of 4.91% obtained in the modeling estimation of the Ff emphasizes the power of the proposed method. The presented framework has the potential to increase processing performance while reducing energy consumption. The findings highlight the effectiveness of combining energy signals and machine learning in developing sustainable manufacturing processes. The proposed model can be adapted to different processing scenarios and provide a basis for industrial applications. © Indian Academy of Sciences 2026.