Modeling And Simulation Of Precise WEDM And Optimization Of Its Process

Wire electrical discharge machining (WEDM), instead of direct contact, is one of the fastest growing and accepted manufacturing technologies in non-traditional machining processes; a series of discharges, which occure between a wire electrode and a workpiece, remove the workpiece in the machining process. In WEDM process, material removal is an interaction of many effects of heat, electricity, magnetism, etc. Hence, there are difficulties in explaining the complexity of the process by theoretical mode. In addition, WEDM machining results are closely related with the cutting parameters, and then process optimization has become a highlight issue. Therefore, this thesis focuses on the above two problem in precise WEDM.In this thesis, based on molecular dynamics theory, Cu and W models of molecular dynamics are constructed for Micro-EDM, which can directly reflect the changes of the material during processing in dynamic microscopic view and can record the various physical quantities in the process. Comparative study of processing Cu and W is done to investigage the similarities and differences by computer simulation, and its inherent processing regularity is described from the multi-perspective views. These models can provide the necessary guidance for the thermal physics model in EDM or WEDM.To enhance the prediction accuracy in simulation model, one new model of single-spark EDM process is constructed in this thesis based on FEM method, considering variable heat distribution coefficient of cathode (fc), the latent heat and plasma flushing effciency (PFE), to predict material removal rate (MRR) and surface roughness (Ra). This model is validated using reported analytical and experimental results. Then, this thesis proposes a hybrid intelligent process model, based on FEM and Gaussian process regression (GPR), to investigate the optimized regularity in EDM. The average relative error of the regression model to predict MRR and Ra, confirmed by verification experiment, are20.74%and14.75%, respectively. Further, the precise WEDM-LS optimization model of process is also established using this method.To study the cutting process optimization in the precise WEDM-LS (WEDM-Low Speed) process, this thesis analyzes the problem based on test data of both rough and finish cutting by the signal noise ratio (SNR) analysis.Then a mathematical regression model and a single objective process optimization in both rough and finish cutting are obtaineed, respectively. The confirmation experiments show that the MRR is increased by2.3times in rough cutting and Ra is decreased by1.41times in finish cutting, which has a more significant improvement. This thesis also studies the multi-objective cutting process optimization in precise WEDM. On the one hand, two goals of process optimization in precise WEDM-MS (WEDM-Middle Speed) machining for SKD11has done, comparativing study of NSGA-Ⅱ algorithm parameters to get robust algorithm, thereby obtaining the Pareto frontier of MRR and Ra in two-dimensional morphology; On the other hand, multi-objective optimization process in MRR and three-dimensional morphology (Sq and Sz) have also been done for WEDM-LS. To determine whether the algorithm has the problem of premature convergence, a method of calculating the diversity of the population is put forward. Then, Pareto frontier can be obtained for the multi-objective optimization process. Two aspects of these studies can provide guidance for the practical application in engineering.To solve the problem in recognition of discharge state for online process optimization, this thesis presents an algorithm for calculating multi-view eigenvalues based on image moments, fractal coefficient and geometric features and build image correlation algorithm of discharge waveform. By signal acquisition, image reconstruction, computing eigenvalues and fuzzy neural network (FNN) classifier, the discharged state can be drawn. The experiments show that the result of recognition accuracy rate is98.7%.Based on the research mentioned above, the application is studied in the precise WEDM-MS, and it is proposed for the corresponding integrated approach and design ideas. Then, one CNC of WEDM-MS is developed.