Abstract: Turbine blades are crucial components in aero engine and electrochemical machining(ECM) is an important processing method to machine blade. Accurate cathode design and modification of turbine blade in ECM is the key technology for improving the blade precision. Digital cathode modification model, using back propagation (BP) neural network together with Levenberg-Marquardt(L-M) optical algorithm, is developed on the basis of the linear cathode modification method and its experimental data. In order to test the validity of the network model, experimental investigations are carried out by using a self-designed blade ECM tool system. The machining samples are detected by the three coordinates measuring machine, and the results show that the precision is enhanced obviously after cathode modification. The accuracy of blade root, blade tip of blade basin is increased by 0.09 mm and the modification efficiency is improved. It proves that the correction of cathode modification can be finished accurately. This network model can be used to carry out digital cathode modification to those electrodes that have complex surfaces. By this cathode digital modification, the modification cycle can be shortened, and the accuracy of turbine blade in ECM can be enhanced.

Key words: Blade, Cathode modification, Electrochemical machining, Experimental investigation, Neural network