Adaptive Iterative Learning Control of a Single-Link Flexible Manipulator Based on an Identified Adaptive NARX Model

Abstract

Flexible-link robots are more widely used in areas such as microsurgery, defense and space vehicles. Compared to rigid robot manipulator because of light weight, low inertia, higher payload carrying capacity and faster executable motion. Flexible robots consist of manipu- lators that are made of flexible and lightweight materials. In addition to these benefits they are associated with serious control problem of vibration. As the structure is flexible when it is actuated and it vibrates with low frequency and after some time the vibrations gets reduced. Hence the control problem for the flexible robot becomes much more complex than rigid link robots. So to overcome this difficulty many control strategies have been proposed in the past, but in most of the works they have considered linear model and the actuator dynamics are not taken into account. In this project work, a non linear model of a single link flexible robot manipulator obtained using Assumed Mode Method (AMM) is considered. The actuator dynamics has also been considered in the modelling of the single link flexible robot. The model obtained by using AMM method is quite complex and require complete knowledge of flexible manipulator dy- namics. As the system is nonlinear and time-varying so to meet the demands of the control system design, an adaptive nonlinear autoregressive with exogenous input (NARX) model is identified using the input/output experimental data. An adaptive iterative learning con- troller (AILC) is designed based on both the existing mathematical model and identified model of a single-link flexible manipulator (SLFM). Tuning of the AILC controller is carried out using least square method. Simulation results demonstrate that the AILC controller designed using the identified adaptive NARX model gives better results than the AILC controller designed using an existing mathematical model.