Backstepping adaptive iterative learning control for strict-feedback nonlinear systems under varying iteration lengths

This paper presents a neural network-based adaptive iterative learning control for strict-feedback nonlinear systems subject to varying iteration lengths. A command filter backstepping approach is adopted to address the derivative explosion caused by continuous differentiation. Radial basis function neural networks are developed to approximate nonlinear functions of each subsystem. Updating laws of the controller are constructed in the piecewise framework to estimate unknown parameters with the consideration of varying iteration lengths. The convergence of the resultant tracking error is guaranteed by means of the composite energy function. Effectiveness of the proposed algorithm is verified on a third-order robot manipulator.