Event-Triggered Incremental Adaptive Dynamic Programming Based Model-Free Robust Control for Zero-Sum Game Guidance

This study proposes a model-free robust control method designed by Event-Triggered Incremental Adaptive Dynamic Programming (ET-IADP) scheme for zero-sum game guidance, which reduces the computational burden for high sampling frequency and guarantee the closed-loop system stability with the completely unknown system dynamic. First, to reduce the dependence on explicit dynamic model, the timescale separation (TSS) principle based approximated incremental linear dynamic (ILD) system is constructed to replace the original nonlinear system, where only the input-state data with its time-delayed one is exploited to derive the approximated ILD system under the model uncertainties. Subsequently, the state-related ET condition for reducing the computational burden is designed into the IADP-based bilateral optimal control policy solving process of zero-sum game with the completely unknown dynamic, where the ET based critic network is exploited to obtain the approximate solution of Hamilton-Jacobi-Issac (HJI) equation. The ET-IADP method updating control policy only at event triggering moment not only overcomes the shortcoming of the time-triggered method suffering from the high sampling frequency but also enhances the robustness against system uncertainties. The incremental error is incorporated into the cost function to attenuate its influence on control performance. And then, the uniformly ultimately bounded (UUB) of the closed-loop system is proven regardless of whether the triggering condition is satisfied. Finally, A series of simulations have been conducted to verify the effectiveness of the proposed method.