Grinding parameters prediction under different cooling environments using machine learning techniques

Selection of optimum process parameters is vital for performing a sound grinding operation on Inconel 751 alloy. This paper co-relates the relationship between the most influential input parameters like cutting velocity, depth of cut, feed rate, and environmental conditions to the output parameters, namely, tangential grinding forces, normal grinding forces, temperature, and surface roughness. Three types of machine-learning (ML) algorithms such as support vector machine (SVM), Gaussian process regression (GPR), and boosted tree ensemble techniques are employed to develop a ML model for predicting the output variables during grinding operation of Inconel 751. In order to develop a better ML model, K-fold technique is employed on a total of 81 datasets which are extracted from experimental studies. ML models developed from different algorithms are compared based on performance metrics like R2 score and root-mean-square error (RMSE). GPR algorithm exhibits best results with relatively better R2 score and RMSE value in predicting grinding forces and temperature at wheel work interface. From analyzing the ML models, it is found that cooling environments determined the output grinding parameters to a greater extent when compared with the input grinding parameters.