Experimental Verification of Nonlinear System Model Parameter Identification Based On Trajectory Pattern Method

Abstract

The purpose of this thesis was to verify the Trajectory Pattern Method (TPM) based nonlinear system model parameter identification method for a two degrees of freedom closed loop planar robotic manipulator. The systematic procedure and the mathematical proof of convergence of the method were introduced first. A model based feedforward control system was constructed for the control of the planar manipulator. The software of the control system was implemented. Experiments were performed to demonstrate the effectiveness of the TPM based parameter identification method. The experimental results, which can be extended to a large class of nonlinear mechanical systems whose models are linear with respect to their model parameters, showed that the estimated parameters were getting closer to their nominal values during the parameter updating process. As the kinematics and dynamics parameters of the system approached their nominal values, the model based feedforward control system would control the system better. The integral of the squared errors of the outputs were calculated to show that the overall control performance was improved after each parameter updating step. The principle conclusion drawn from the experiment revealed that the developed parameter identification method was very effective on the highly nonlinear dynamic system.