ADP-Based Fault Tolerant Control for Flexible Joint Manipulator System

This research proposes a solution to fault-tolerant method in manipulator systems with flexible joints experiencing nonaffine actuator failure through the application of adaptive dynamic programming (ADP) methodology. Leveraging Bellman’s optimality criterion, the system’s cost function is estimated via a constructed critic neural network featuring dynamically adjusted weight vectors, simultaneously addressing the fault and enhancing system performance. Control input for trajectory tracking is derived through numerical resolution of the Hamilton-Jacobi-Bellman (HJB) equation, accompanied by a designed fault estimation mechanism for identifying unknown nonaffine actuator failure. Lyapunov stability analysis confirms uniform ultimate boundedness (UUB) of all the system variables. Finally, numerical simulations ultimately validate the applicability and operational efficacy of the developed control framework.