Thesis Type: Postgraduate
Institution Of The Thesis: Orta Doğu Teknik Üniversitesi, Faculty of Engineering, Department of Mechanical Engineering, Turkey
Approval Date: 2008
Student: ORKUN ÖZŞAHİN
Supervisor: HASAN NEVZAT ÖZGÜVENAbstract:
In machining centers, with the increasing trends in high precision machining, chatter has become an important problem which results in poor surface finish and low material removal rate. Chatter can be avoided with stability diagrams which provide the stable regions in the machining process for the depth of cut and spindle speed combinations. In order to obtain stability diagrams, tool point frequency response function (FRF) of the system should be obtained. Throughout this study, contact parameters which are the most critical part of the analytical modeling of spindle-holder-tool assembly in order to obtain tool point FRF, are examined. For the accurate identification of the contact parameters, a recently suggested closed form approach based on measured FRFs is improved and applied to real structures by solving several application problems. In addition to the identification of contact parameters from experimental results, in order to eliminate the dependency on experiments, artificial neural networks are used to predict contact parameters for cases for which no experiments were carried out. By using trained neural network, contact parameters are predicted for the first seen combination of tool gauge length and diameter with a high accuracy. Such an application will have an important contribution to the machining stability studies since elimination of dependency on experiments will make it possible to predict stability diagrams for different combinations of spindle, holder and tool without performing any experiments. Additionally, since accurate identification of contact parameters, thus tool point FRFs and stability diagrams are highly dependent on accuracy of the performed experiments, possible errors due the mass of the accelerometers are also investigated. In order to compensate the mass effect of the accelerometers, a structural modification with matrix inversion method is applied to the accelerometer based results.