Research on Penetration Detection in EDM Small Hole Machining Based on BP Neural Algorithm

In this article, a penetration detection technology for small hole machining in electrical discharge machining (EDM) based on Backpropagation (BP) neural network classification algorithm is proposed to improve the machining efficiency of gas film holes in aviation turbine discs. During the penetration period of small hole machining, due to the drastic variations in machining voltage, five parameters, including the number of machining voltage variations(p), pulse width(ON_all), pulse interval(OFF), servo distance(SV), and machining current(PS), are determined as input variables for the BP model features. Based on the above analysis, using the data obtained from model training, the BP neural network classification algorithm model was established in MATLAB. As a result, a small hole machining penetration detection model was obtained in a field programmable gate array (FPGA). At the same time, considering factors such as FPGA on-chip resources and algorithm time consumption, the FPGA program was transplanted on the experimental platform through table lookup, and the feasibility was verified through machining experiments on this method.