Predictive modelling and Pareto optimization for energy efficient grinding based on aANN-embedded NSGA II algorithm

Abstract

A large amount of global power consumption and environmental pollution problems are attributed to manufacturing industries. Grinding is one of the most energy intensive precision machining processes. Stimulating energy-saving potential in grinding to improve energy efficiency will be great helpful to accelerate energy and environmental sustainability. However, the energy utilization and efficiency problem has not been fully considered in traditional grinding strategies for production benefits. Therefore, a Pareto optimal design method is proposed to obtain optimal grinding parameters under the dual drive of energy management and machining performance. A nonparametric model based on the improved adaptive artificial neutral network (aANN) has been built to predict the surface quality, machining time, total and active power consumption. Parametric studies of the aANN model have been performed to achieve more accurate predictions. Accordingly, the multi-objective optimization problem searching a trade-off among energy and time efficient criteria, as well as product quality, has been solved using non-dominated sorting genetic algorithm II (NSGA II). Moreover, an energy efficiency benchmark has been suggested to indicate eco-friendly ability of grinding. Experimental results of the AISI 1045 steel have demonstrated that the optimal solution could improve energy efficiency by 89.52% and reduce machining time by 174.36% keeping the same product quality. Similar procedures will be performed to build the energy efficient grinding database towards sustainable and intelligent manufacturing.