Machine learning-based predictive modelling of dry electric discharge machining process

The conventional methods like linear or polynomial regression, despite their overwhelming accuracy on training data, often fail to achieve the same accuracy on independent test data. In this research, a comparative study of three different machine learning techniques (linear regression, random forest regression, and AdaBoost) is carried out to build predictive models for dry electric discharge machining process. Six different process parameters namely voltage gap, discharge current, pulse-on-time, duty factor, air inlet pressure, and spindle speed are considered to predict the material removal rate. Statistical tests on independent test data show that despite linear regression's considerable accuracy on training data, it fails to achieve the same on independent test data. Random forest regression is seen to have the best performance among the three predictive models.