Optimization Of Machining Parameters Affecting Surface Roughness Using Surrogate Techniques

In metal cutting process, machining parameters are not only determine the production efficiency of the cutting process, but also determine the quality of the surface roughness for cutting force and cutting parts, in fact, cutting force and surface quality of the machined parts is strongly correlated, the cutting force is also affecting the surface quality. The optimization of machining parameters is very important for improving the cutting force stability and surface roughness quality in the machining process. In this paper, the machining parameters are optimized by using the surrogate model based on response surface method and the artificial neural network model techniques.This paper mainly carries the following aspects of the work:(1) The main factors that affect the cutting force are analyzed based on the model of cutting force. In the research work of this thesis, the cutter tooth number is introduced as an important machining parameter besides cutting speed, cutting depth and feed rate as the key machining parameters. Based on the analysis of the exits model for the machining roughness of response surface, the range and level of the four key parameters are determined by experiment design.(2) The experimental study and sensitivity analysis of the model based on the response surface is studied. First according to the level of the four key processing parameters, by cutting experiment the sample data set of relevant machining parameters are collected, then quadratic response surface method of processing parameters and surface roughness of surrogate model are established. Secondly, the sensitivity analysis of the model is further simplified. Finally, the machining parameters are optimized based on surrogate model, and the cutting experiments show that the error of the surface roughness model based on response surface method is within the acceptable range.(3) The optimization of machining parameters based on response surface method is carried out which affect cutting force and surface roughness. According to the levels of four key machining parameters through the experiment acquisition of roughness degree and cutting force, surface roughness and cutting force for linear and second-order response surface agent model was built, and verified on agent model based on the cutting experiments, to verify the reliability of the agent model. With the application of models, machining parameters for single and multi-objective machining parameters are optimized. Similarly, cutting experiments show the surrogate model for the surface roughness and cutting force in single and multiple objective optimizations is reliable.(4) The research on the single objective and multi-objective optimization of the machining parameters based on the ANN model. On the basis of the study of artificial neural network model structure and principle, application of artificial neural network was established based on cutting force and surface roughness of the processing parameters of single and multi-objective. Optimization model and cutting experiments show that based on the neural network model of the cutting force and surface roughness model can well approximate the real cutting model.The work is carried out based on experimental design, through a variety of agent model of machining parameters on influence of cutting force and surface roughness degree, as well as the application of agent model of machining parameters in multi-objective optimization method. Cutting test results show that the established various agent model and optimization method is effective and feasible.There are three techniques used in this study. These techniques are Response Surface Method (RSM), Taguchi and Artificial Neural Networks (ANN). In response surface, sensitivity analysis is carried out in order to evaluate the significant degree for each cutting parameter. Single response shows that tool flute has higher significant level among other machining parameters used in this study. In multi response (dual response) surface roughness is play as independent response while cutting force is known as corresponding responses.Taguchi techniques are applied in this study with the same machining parameters and same levels, the results carried out gives better results compared with full factorial design. The number of data used in Taguchi techniques is equal to one third of the data used in full factorial method. Neural networks method carried out for machining parameters optimization, optimization results obtained in this application are very good compared with the others.From Taguchi and RSM techniques, tool flutes number is significant parameter affecting both surface roughness and cutting force. Tool flutes number interacts with cutting speed are classified with significant parameters, which give strong indication to the vibrations due to the high spindle speed in milling machine.