Tool Tip Frequency Response Function Prediction For Non-constant Connection Properties

Chatter in machining usually results in poor surface finish,damage to cutters,and reduces machine tool life.Tool tip frequency response function(FRF)is an important basis for chatter suppression.The receptance coupling substructure analysis method(RCSA)is widely used to predict the tool tip FRF.However,the existing RCSA methods make a lot of assumptions,simplifications and approximations to model the connection properties between different substructure,which seriously affects the accuracy of the methods.For this question,the relationship between the connection properties and the dynamics of the machine tool substructure is studied in this thesis,and the phenomenon that the holder-tool connection properties change with the position of the machine tool spindle and the tool overhang is found.Then,a tool tip FRF prediction method for non-constant connection properties is proposed and experimentally verified on a five-axis machine tool.The main work of the paper is as follows:(1)The influence of the spindle position on the holder-tool connection properties is studied.An IRCSA method considering the change of spindle position is established.The correlation between the spindle-holder system FRF and the holder-tool connection properties at different spindle positions is analyzed.It is found that the holder-tool connection properties change with the spindle position,which is then verified experimentally on a five-axis machine tool.(2)The influence of tool overhang on the holder-tool connection properties is studied.Based on the Euler-Bernoulli beam theory,Timoshenko beam theory,and the parameterisation to the connection properties,it is found that the holder-tool connection properties change with the tool overhang,which is experimentally verified later on a five-axis machine tool.(3)A tool tip FRF prediction method for non-constant connection properties is proposed.For the initial holder-tool,the connection properties are identified to connection parameters at different spindle positions,and are further modeled through KNN.Then this model is transferred to a target holder-tool through Tr Ada Boost.R2 algorithm to acquire the connection parameters of the target holder-tool.Finally,the tool tip FRF of the target holder-tool at different spindle positions could be predicted by coupling the substructure FRFs.Experiments are carried out for verification and compared with traditional RCSA method,the proposed method can get a smaller error between the predicted tool tip FRFs and the measured ones.