A Systematic Method for Model Parameter Identification of Nonlinear Dynamics Systems Using Trajectory Pattern Method
In this paper, a new method is presented for model parameter identification of a large class of fully controlled nonlinear dynamics systems such as robot manipulators. The method uses trajectory patterns with feed-forward controls to identify model parameters of the system. The developed method ensures full system stability, does not require close initial estimated values for the parameters to be identified, and provides a systematic method of emphasizing on the estimation of the parameters associated with lower order terms of the system dynamics model and gradually upgrading the accuracy with which the model parameters, particularly those associated with the higher order terms of the system dynamics, are estimated. The developed method is based on Trajectory Pattern Method (TPM). In this method, for a pattern of motion the inverse dynamics model of the system is derived in algebraic form in terms of the trajectory pattern parameters. The structure of the feedback error with feedforward signal calculated with the estimated model parameters will then be fixed, and its measurement can be used to systematically upgrade the model parameter estimation. The mathematical proof of convergence of the developed method and results of its implementation on a robot manipulator with highly non-linear dynamics are provided.