Research On Modeling And Process Of Grinding Material Removal For GH4169 Nickel Based Superalloy

With great mechanical properties of high strength and hardness,nickel based superalloy can be used in harsh environment for a long time.It keeps stable properties even in extreme conditions like high temperature and used as one of the common aviation materials.Due to high dimensional accuracy of aviation materials,grinding is deployed as one of the main process methods to achieve the final fishing accuracy.Since the shapes of aerospace parts are often complex,the flexibility of process equipment is needed.Therefore,intelligent robotic grinding platform has become more and more popular these years.Developing accurate material removal model in consideration of grinding parameters and tool wear becomes the key to achieving desired finishing accuracy of 3D surface.This thesis analyzes parameters in grinding process and examines which parameters are related to material removal to simplify model through intelligent robotic grinding system.In this paper,the abrasive ability index of abrasive belt is defined and the time-domain variable ? is introduced to measure the effective working time of the belt under certain state.The time-domain variable can be divided by a certain time step as the grinding order ? from the initial state till the working state,based on the intermittent feature of grinding process.Analyzing experimental data,the mathematical relationship between the abrasive ability index and the grinding order of the abrasive belt is established,which makes it convenient to introduce the abrasive ability index to the model to quantify wear situation.The proposed method solves the problem of modeling abrasive tool wear due to the difficulty of quantification.Based on the experimental data,the relationship between the grinding parameters and the material removal is analyzed.The correlation of each grinding parameter and the material removal is obtained.The correlation of the normal grinding force is the most significant,followed by grinding belt linear speed and the abrasive ability index.For different belts,the influence of abrasive ability index is different.For a normal belt,the influence of the abrasive ability of the abrasive belt is even nearly as strong as belt linear speed.Therefore,it is necessary to consider the influence of the wear of belts for the nickel-based superalloy and similar difficult-to-machine materials.The mathematic model of the grinding process is established through the method of multiple regression analysis.A modified empirical model based on Preston's correlation equation is proposed.The experimental data are used to derive the grinding model of the difficult-to-machine materials.The quantitative relationship between the material removal and the belt linear speed,the normal grinding force and the grinding order was obtained.The average percentage error of the model is shown to meeting the needs of grinding superalloys.In view of the defects and shortcomings of the traditional BP neural network in modeling,the generalized regression neural network model and the cross-validation optimization algorithm are introduced to establish the regression model of the grinding process.The average error percentage of the model is decreased to a large extent.The accuracy of the model is analyzed compared with the traditional BP neural network network and the existing model.Finally,the advantages and disadvantages of the mathematical modeling method and the neural network modeling method of regression analysis are analyzed from the aspects of modeling difficulty and model visualization.The validity of the time-domain variable is analyzed,and the rationalization suggestions are given for the development of modeling grinding process.