Modeling and optimization of the effective parameters in the pickling operation of the titanium alloy by the artificial neural network and genetic algorithm

Alfa layer due to the hot forging of the titanium alloy is removed using the acid pickling operation. Because of the interaction between the effective parameters, investigation and optimization of the effective parameters needs experimentation and modeling methods. In this research effect of the temperature, operation time and percent of the hydrofluoric and nitric acids on the surface roughness and depth of cutting are modeled and optimized. First the experiments are designed using taguchi method and implemented. Then the process is modeled by an artificial neural network and the effect of the parameters on the surface roughness and depth of cutting are investigated. Next by combining the artificial neural network with genetic algorithm the process is optimized. Results show that a Multilayer Feed forward network with Levenberg-Marquardt backpropagation learning algorithm and ten nods in the hidden layer could model the operation precisely. Increasing the temperature and hydrofluoric acid percent cause to increase the depth of cutting. Nitric acid percent and operation temperature have interaction in affecting on the amount of depth of cutting. Operation time have no affect on the roughness. Increasing the temperature at low percent of hydrofluoric acid cause to decrease the roughness however it increases at high percent of hydrofluoric acid.