Study On Cutting Database Of ASJ Based On Neural Network

Due to the different mechanisms of acceleration of abrasive particles, the speedof abrasive particles in ASJ jet is far greater than the speed of abrasive particles inAWJ, which cutting ability is up to5-10times that of AWJ. ASJ, suitable for all kindsof metal materials and non-metal materials of high speed precision machining, hasbeen successfully used in semiconductor chip manufacturing industry.In the paper, the material removal mechanism, mechanism of remaining linesforming on surface of ASJ technology is studied in details. According to the results ofthe experiment, there is obvious quality gap between the upper and lower two separate,so we separately research the two sections. We can get the influence ranking of cuttingparameters in cutting quality, and the highly nonlinear in the twothroughorthogonal analysis. So it is difficult to create the cutting model using the traditionalmathematical model. Therefore, combined with the current research status at home andabroad abrasive jet cutting model, the issues that creating a pre-mixed abrasive waterjet cutting model based on artificial neural network are proposed.In the paper, under the MATLAB environment, the realization of the roughnessforecast and the rate predicted models of aluminum alloy1060, which adopted thesymbol and extrapolating simulation, confirmed that the network model is fitted with ahigh degree of ability and generalization ability. To avoid being confined to a singlematerial, we create the roughness predict and velocity predictions neural network modelabout material304stainless steel, and the model can also meet the requirements ofmodeling errors.