Neuroadaptive learning algorithm for constrained nonlinear systems with disturbance rejection
Guichao Yang,Jianyong Yao,Zhenle Dong
DOI: https://doi.org/10.1002/rnc.6143
IF: 3.8973
2022-01-01
International Journal of Robust and Nonlinear Control
Abstract:For practical plants, there are not only inevitable endogenous and exogenous disturbances, but also constraint demands for control input amplitude, output performance, and system states. The existing controllers are difficult to handle the aforementioned control issues simultaneously and may be affected by "explosion of complexity." In this work, we will develop a neuroadaptive learning algorithm for constrained nonlinear systems with disturbance rejection. In detail, the performance prescribed function and time-varying barrier Lyapunov functions will be constructed to achieve the prescribed output performance and time-varying state constraints, respectively. Furthermore, the neural network adaptive control and the extended state observers will be combined to respectively estimate the endogenous uncertainties and exogenous disturbances on-line and meanwhile compensate for them feedforwardly. In particular, a filter will be introduced to estimate the virtual control law at each procedure, and this filtered value will replace its actual value in synthesizing the controller. In addition, the filtering errors and input saturation nonlinearity will be compensated by introducing an auxiliary system. Eventually, the whole closed-loop stability is strictly guaranteed and the achievable control performance is verified by the application results.