Multi-objective Parameter Optimization And Process Modeling Of EDM For Superalloy GH4169

Because high-temperature alloys have high hardness and strength,good thermal stability,corrosion resistance and fatigue resistance,they are widely used to manufacture key components for aero-engines and steam turbines.Among all superalloys,GH4169 has the largest production and use,so GH4169 is the most representative.However,when GH4169 was cut by the conventional style of cutting,many problems was encountered,such as large cutting force and easy wear of the tool.Since EDM removes the material by the thermal energy generated by spark,and it is not affected by the strength and hardness of the workpiece material with high processing precision,it is very suitable to use EDM to cut GH4169.Based on the relevant theories,this paper studies the EDM parameters of GH4169 by the aspects of single factor test,orthogonal experiment,multi-objective parameter optimization and process modeling.It has certain guidance on the parameter selection in actual production.significance.The effects of peak current,pulse width,interpulse and gap voltage on material removal rate and surface roughness is studied by single factor experiment.The hardness of the processed surface of GH4169 is measured by Rockwell hardness tester.The cracks on the processed surface of GH4169 is observed by scanning electron microscopy and analyzed accordingly.Through the orthogonal experiment,the primary and secondary order of influence of peak current,pulse width,interpulse and gap voltage on material removal rate and surface roughness is further explored.Based on the orthogonal experiment and the gray correlation analysis method,the multi-objective parameters of EDM GH4169 is optimized,and the optimal parameter combination is obtained: the peak current is 3A,the pulse width is 60?s,and the pulse is 40?s,the gap voltage is 27 steps.It is tested and verified finally.Comparing the verification results with the optimal combination in the orthogonal table,it is found that although the roughness value is slightly increased,the material removal rate is increased by about 2.9 times.On the basis of the orthogonal experiment,other 35 sets of experiments performed is combined to form sample data with 16 sets of data of the orthogonal test.6 sets ofdata is randomly selected to constitute test samples,and the remaining 45 sets of data constitutes training samples.By reasonably selecting the relevant functions and parameters,the peak current,pulse width,interpulse and gap voltage are used as the input of the network,and the surface roughness and material removal rate are the output of the network,then the prediction of GH4169 based on BP neural network is established.The model is tested and verified.The results show that although the model has certain errors in the actual parameter prediction,these errors are within a reasonable range.Therefore,the prediction model is still effective.The genetic algorithm is used to optimize the initial weight and threshold of the established prediction model,and the experimental verification is carried out.Finally,the prediction models before and after optimization are compared.The results show that the prediction error of the optimized model is generally much smaller than the prediction error before optimization for both of surface roughness and material removal rate.The analysis of variance shows that the optimized model not only has higher prediction accuracy,but also better stability.