Improvement of dry EDM process characteristics using artificial soft computing methodologies

Dry electrical discharge machining (EDM) is an environmentally-friendly alternative of die-sinking EDM process, which it uses gaseous medium instead of liquid as a dielectric. Due to contribution of too many parameters in this process, selection of optimal parameters to increase the process performances is a really crucial concern. In this work, a predictive model based on back-propagation neural network has been applied to correlate the inputs and outputs of dry EDM process. Herein, the inputs were gap voltage, pulse current, pulse on time, duty factor, air intake pressure and rotational speed of tool, and also the main outputs were material removal rate (MRR) and surface roughness (SR). Firstly a back-propagation (BP) and radial basis function neural network have been developed based on data generated from literature [Saha and Choudhary Int J Mach Tools Manuf 49:297–308 (2009)]. Then, the accuracy of proposed models has been checked by their values of error percent via testing data. Hereafter, the most accurate model was served as an objective function to optimize the process using artificial bee colony (ABC) algorithm. In optimization stage, firstly a single objective optimization was fulfilled to determine the optimal factors related to each output separately. Then a multi-objective optimization was implemented to calculate the best solutions in the case of higher MRR and lower SR simultaneously. Results indicated that the predictive model can estimate the dry EDM process precisely, and also the ABC algorithm could find the optimal solution sets logically.