Research Of Process Modeling And Simulation System For High Speed Wire EDM

After lots of investigation of the internal and external methods on modeling of wire EDM process, the dissertation presents an attempt at modeling the process through different ways on different high speed WEDM machins, a predictive system software is developed. Machining effectiveness prediction and parameters optimization is considered in this software. Moreover it can be used on different wire EDM machine tools through models training module. It is verified that accurate results has been provided through Artificial Neural Networks on different high speed WEDM machines. This method can be used to guide the proceeding of production.The dissertation includes six chapters and the contents are as follow:In chapter 1, through the comparison between high speed WEDM and low speed WEDM, problems and the trend of development of high speed WEDM is brought forward. Then the modeling methods on modeling of EDM domestically and abroad is given. The idea of establish a simulation system to predict technological performance and optimize machining parameters is given.In chapter 2, based on the samples of TP7725 high speed WEDM machine from ShangHai wireless apparatus factory and DK7725e high speed WEDM machine from SuZhou SanGuang electromechanical corporation, the virtues and defects of known process databases are given. For the TP7725 machine which has complete database liner regression model and nonlinear regression model are brought out. The two models are compared for goodness of fit and prediction. And the process regularity is analyzed.In chapter 3, the application of Artificial Net Works in function fitting is discussed, a process modd based on TP7725 process database using BP Artificial Neural Netwroks is brought out. During the modeling procedure, how to determine the structure of net and accelerate rate of convergence is discussed. Then another model using RBF Artificial Neural Networks is also discussed. The two models are compared for goodness of fit and prediction.The appraisal of different models is given.In chapter 4, for the DK7725e which has no complete database, though survey of experiments schemes , a central composite rotatable experimental design is brought out. Then based on test data nonlinear regression model and RBF Neural Networks model are brought out. Verification experiments have been carried out to check the validity of the developed models.In chapter 5, the makeup and function of the process simulation system software is introduced firstly, then the particular design of the master module, database module, models training module and simulating module is discussed.In chapter 6, the work of the dissertation is summarized and the further research work is looked ahead.