Akademik Veri Yönetim Sistemi
Tezin Türü: Doktora
Tezin Yürütüldüğü Kurum: Orta Doğu Teknik Üniversitesi, Mühendislik Fakültesi, Makina Mühendisliği Bölümü, Türkiye
Tezin Onay Tarihi: 2018
Öğrenci: MASOUD LATIFI NAVID
Danışman: ERHAN İLHAN KONUKSEVEN
Özet:Automatic grinding using robot manipulators, requires simultaneous control of the robot endpoint and force interaction between the robot and the constraint surface. In robotic grinding, surface quality can be increased by accurate estimation of grinding forces where significant tool and workpiece deflection occurs. Tool deflection during robotic grinding operation causes geometrical errors in the workpiece cross-section. Also, it makes controlling the grinding cutting depth difficult. Moreover small diameter of the tool in robotic grinding causes different behavior in the grinding process in comparison with the tools that are used by universal grinding machines. In this study, a robotic surface grinding force model is developed in order to predict the normal and tangential grinding forces. A physical model is used based on chip formation energy and sliding energy. To improve the model for robotic grinding operations, a refining term is added. In order to include the stiffness of the tool and setup in the force model, penetration tests are implemented and their results are used in refining term of the force model. The model coefficients are estimated using a linear regression technique. The proposed model is validated by comparing model outputs with experimentally obtained data. Evaluation of the test results demonstrates the effectiveness of the proposed model in predicting surface grinding forces. In this thesis, a method is proposed for calculation of the tool deflection in normal and tangential directions based on grinding force feedback in these directions. Based on calculated values, a real-time tool deflection compensation algorithm is developed and implemented. Implementing surface grinding with constant normal force is a well-known approach for improving surface quality. Tool deflection in the robotic grinding causes orientation between the force sensor reference frame and tool reference frame. This means that the measured normal and tangential forces by the sensor are not actual normal and tangential interaction forces between the tool and workpiece. In order to eliminate this problem, a resultant grinding force control strategy is designed and implemented for a parallel hexapod-robotic light abrasive surface grinding operation. Due to the nonlinear nature of the grinding operation, a supervised fuzzy controller is designed where the reference input is identified by the proposed grinding force model. Evaluation of the experimental results demonstrates significant improvement in grinding operation accuracy using the proposed resultant force control strategy in parallel with a real-time tool deflection compensation algorithm. The final aim of this thesis is to develop a posture optimization strategy for robotic grinding operation using 12 DOF hybrid redundant manipulator. The 12 DOF redundant hybrid manipulator of present study is composed of a 6 DOF serial ABB IRB2000 robot and a 6 DOF PI H-824 hexapod where the parallel hexapod is connected to the end of the serial ABB manipulator. Here the fifth joint (wrist) of the ABB serial manipulator is the weakest joint in the robot, so the computed torque of this joint is selected as the cost function. The aim is to minimize this factor by finding the best configuration of the hybrid manipulator using genetic algorithm approach. For such a purpose, a complete kinematic and dynamic model of the 12 DOF manipulator is developed where the output of the grinding force model is fed into the dynamic model as external reaction forces. The computed torque of the wrist joint is given to the optimization module and new configuration is generated by the module and is given to the dynamic model. This process continues until converge to the minimum computed torque value. Then the optimal configuration is chosen for the grinding operation. The evaluation of this posture optimization approach shows its great ability to decrease the necessary actuating torques of the redundant manipulator joints.