An innovative design support system for industry 4.0 based on machine learning approaches

Electric machines together with power electronic converters are the major components in industrial and automotive applications. The frequent situation in the engineering practice is that designers, final or intermediate users have to roughly estimate some basic performance data or specification data or other metrics related to the specific task they have, on the basis of few data available at a particular instant of time or at the time of use. This paper addresses this problem in the Industry 4.0 scenario by introducing innovative Design support system (DesSS), originated from the Decision Support System (DSS), for the prediction and estimation of machine specification data such as machine geometry and machine design on the basis of other heterogeneous parameters (i.e. motor performance, field of application, geographic market, and range of cost). For the development of the DesSS different machine learning techniques were compared such as Decision/Regression Tree (DT/RT), Nearest Neighbors (NN), and Neighborhood Component Features Selection (NCFS). Experimental results obtained on the real use case demonstrated the appropriateness of the application of the machine learning approaches as the main core of the DesSS used for the estimation of the machine parameters. In particular, the results show high reliability in terms of accuracy and macro-F1 score of the 1-NN+NCFS and RT for solving respectively the classification and regression task. This approach can viably replace the model-based tools used for the parameters prediction, being it more accurate and with higher computational speed.