Multi-parameter Optimization Of Magnetic Field Assisted Micro-ecm With BP Neural Network

With the development of aviation,automobile,biomedicine and other industries,some micro-structures made of high-performance difficult-to-process materials,such as micro-pits,micro-grooves and micro-pores,have been widely used.In the field of micro-electrochemical machining,the workpiece material itself does not affect the machining process,and the future application prospect is very broad.Magnetic field composite technology has further improved the machining quality of micro-electrochemical machining technology,but the current processing quality of this technology still needs to be improved.The selection of process parameters is generally obtained by experimental iteration,and this method can only get close to the optimal solution through iteration,which has low efficiency and high cost.In this thesis,the purpose of efficiently optimizing process parameters is to establish a prediction model using BP neural network,and the optimization of process parameters of magnetic field composite micro-electrochemical machining is studied.The main research contents are as follows:(1)The effects of machining parameters on the experimental results were studied through the basic experiments of magnetic field compound micro-electrochemical machining.The basic experiment aims to obtain better machining effect,and constantly adjusts the machining parameter values according to the machining results.By analyzing the basic experimental results,the ideal range of each machining parameter was determined.In order to achieve better results in BP neural network model training,orthogonal experiments were carried out based on basic experiments,and 25 groups of sample data with uniform distribution and strong representation were obtained.Together with 18 groups of effective basic experiments,a total of43 groups of sample data were obtained.Compound micro-electrochemical machining were studied.BP neural network theory;The theory of multi-parameter optimization using BP neural network model.(2)Based on the data of basic experiment and orthogonal experiment,the structure of the BP neural network processing result prediction model was determined.According to the model structure,the BP neural network calculation program was developed under the framework of Tensor Flow using Python language,including the forward propagation calculation program and the back propagation calculation program.The correctness of the BP neural network program is verified by the existing data.The experimental data was made into a data set and input into the BP neural network for training.By comparing the training effect,the optimal value of the model parameters was determined.The number of hidden layer nodes was 12,the training times was 2000,the learning rate was 0.09,and the weight parameter values of the model were obtained.(3)The process parameters of the magnetic field composite micro-electrochemical machining process were optimized,and two optimization methods were proposed.The first method was to use the prediction model of the processing results and get the better process parameters by constantly substituting the process parameters to compare the experimental results.After optimization,the comprehensive error was reduced by about 25%.The second method is to establish the machining parameter prediction model of magnetic field compound micro-electrochemical machining,input the expected machining results,and get the optimized process parameters directly.After optimization,the comprehensive error is reduced by about31%.(4)Development of multi-parameter optimization system for magnetic field composite micro-electrolysis.Based on Python Tkinter library,a multi-parameter optimization system for magnetic field composite micro-electrochemical machining was designed and built.Based on the process parameter prediction model,the system can directly obtain the optimal process parameters to achieve the expected machining results by inputting the expected machining results.The optimization of process parameters improves the machining accuracy of the magnetic field composite micro-electrochemical machining process,and also provides convenience for other experimental personnel to obtain the optimized process parameters.By constructing the BP neural network processing result prediction model,the optimization of the process parameters is realized,and the optimization effect is good.The processing experiment is carried out with the optimized process parameters,and the processing effect is obviously improved,which proves the effectiveness of the optimization method.Through the establishment of BP neural network process parameter prediction model,the efficiency of process parameter acquisition is greatly improved.Through the experimental verification,the optimization effect is better than the processing result prediction model.Finally,based on the BP neural network process parameter optimization model,the parameter optimization system is developed,so that relevant personnel can quickly obtain the optimized process parameters through the parameter optimization system.