Neural-network-based modeling and optimization of the electro-discharge machining process

In this research, a new integrated neural-network-based approach is presented for the prediction and optimal selection of process parameters in die sinking electro-discharge machining (EDM) with a flat electrode (planing mode). A 3–6–4–2-size back-propagation neural network is developed to establish the process model. The current (I), period of pulses (T), and source voltage (V) are selected as network inputs. The material removal rate (MRR) and surface roughness (Ra) are the output parameters of the model. Experimental data were used for training and testing the network. The results indicate that the neural model can predict process performance with reasonable accuracy, under varying machining conditions. The effects of variations of the input machining parameters on process performance are then investigated and analyzed through the network model. Having established the process model, a second network, which parallelizes the augmented Lagrange multiplier (ALM) algorithm, determines the corresponding optimum machining conditions by maximizing the MRR subject to appropriate operating and prescribed Ra constraints. The optimization procedure is carried out in each level of the machining regimes, such as finishing (Ra≤2 μm), semi-finishing (Ra≤4.5 μm), and roughing (Ra≤7 μm), from which, the optimal machining parameter settings are obtained. The optimization results have also been discussed, verified experimentally, and the amounts of relative errors calculated. The errors are all in acceptable ranges, which, again, confirm the feasibility and effectiveness of the adopted approach.