Optimization and Modeling of Plasma Cutting of AISI 309 Stainless Steel by Using Neural Network-Genetic Algorithm Hybrid Model

In plasma cutting, a noble gas at high speed is blown from the nozzle and ionized with the help of a frequency spark at high voltage and an electric arc is created which cause the gas changes to the plasma state. Plasma cutting is an ideal process for cutting of the hard metals. In this research, the effect of the input parameters and their optimization in plasma cutting of AISI 309 stainless steel were studied. By conducting the different experimental tests, the effect of input parameters including amperage, gas pressure and the cutting speed of torch on the three output parameters of the width of cut (Kerf), heat-affected zone (HAZ) and surface roughness (Ra) were investigated. Analysis of the results showed that the amperage, cutting speed and gas pressure have the highest impact on the output parameters, respectively. The artificial neural network (ANN)-genetic algorithm was used to predict and optimize the output parameters. The results indicate that the artificial neural networks model trained by the genetic algorithm are able to predict the output parameters accurately. Finally, the optimization of output parameters to achieve the best cutting conditions was carried out using the genetic algorithm. The artificial neural network models were considered as the objective function and also, the parameters of the heat-affected zone, surface roughness, and the width of cut were introduced as inputs of the algorithm. According to results, a combination of the neural network and genetic algorithm is an efficient method for optimization of the plasma cutting process. This method can be easily modified and utilized for other advanced cutting methods.