19 March 2019 in IEEE Transactions on Neural Networks and Learning Systems
In this paper, a learning-based robust tracking control scheme is proposed for a quadrotor unmanned aerial vehicle system. The quadrotor dynamics are modeled including time-varying and coupling uncertainties. By designing position and attitude tracking error subsystems, the robust tracking control strategy is conducted by involving the approximately optimal control of associated nominal error subsystems. Furthermore, an improved weight updating rule is adopted, and neural networks are applied in the learning-based control scheme to get the approximately optimal control laws of the nominal error subsystems. The stability of tracking error subsystems with time-varying and coupling uncertainties is provided as the theoretical guarantee of learning-based robust tracking control scheme. Finally, considering the variable disturbances in the actual environment, three simulation cases are presented based on linear and nonlinear models of quadrotor with competitive results to demonstrate the effectiveness of the proposed control scheme.
20 September 2019 in IEEE Transactions on Smart Grid
With the increasing proportion of renewable power generations, the frequency control of microgrid becomes more challenging due to stochastic power generations and dynamic uncertainties. The energy storage system (ESS) is usually used in microgrid since it can provide flexible options to store or release power energy. In this paper, an intelligent control strategy completely based on the adaptive dynamic programming (ADP) is developed for the frequency stability, which is designed to adjust the power outputs of micro-turbine and ESS when photovoltaic (PV) power generation is connected into the microgrid. Further, considering the changes of PV power and load demand in a day, the full utilization of PV power and the recycling of energy storage are realized through the proposed regulation strategy. Numerical simulation results validate the energy-storage-based intelligent frequency control strategy for the microgird with stochastic model uncertainties, and comparative studies based on PID, LQR and fuzzy logic control illustrate the superiority of the proposed control strategy.
18 February 2019 in IEEE Transactions on Systems, Man, and Cybernetics: Systems
In this paper, an approximately optimal control strategy is developed for the tracking control of a class of continuous-time nonlinear systems with unmatched uncertainties. By transforming the unmatched uncertain term, the auxiliary system associated with the uncertain nonlinear system is established. The auxiliary system is divided into steady and transient parts, and the related controllers are separately solved, meanwhile the transient tracking error system is also obtained by introducing the reference system. A neural network-based adaptive dynamic programming method is used to get the approximately optimal tracking control law of uncertain nonlinear systems with a predefined cost function. Furthermore, the ultimately uniform boundedness of neural network weights and the stability of tracking error systems are both proved through Lyapunov theory. Two cases of nonlinear systems with unmatched uncertainties are investigated to illustrate the effectiveness of the proposed robust tracking control strategy.
24 May 2022 in IEEE Transactions on Cybernetics
In this article, a data-based feedback relearning (FR) algorithm is developed for the uncertain nonlinear systems with control channel disturbances and actuator faults. Uncertain problems will influence the accuracy of collected data episodes, and in turn affect the convergence and optimality of the data-based reinforcement learning (RL) algorithm. The proposed FR algorithm can update the strategy online by relearning from the empirical data. The strategy can continuously approach the optimal solution, which improves the convergence and optimality of the algorithm. Moreover, based on the experience replay technology, a data processing method is designed to further improve the data utilization efficiency and the algorithm convergence. A neural network (NN)-based fault observer is used to achieve the model-free fault compensation. The polynomial activation function is redesigned by using the sigmoid function/hyperbolic tangent activation function, to reduce the difficulty of NNs design for an unknown nonlinear system and improve the generalization. In the face of disturbances and actuator faults, the control performance, algorithm convergence, and optimality of the proposed strategy can be well guaranteed through comparative simulation.
24 July 2018 in IEEE Access
In this paper, a novel observer-based control strategy is proposed for a class of uncertain continuous-time nonlinear systems based on the Hamilton-Jacobi-Bellman (HJB) equation. Due to the complexity of nonlinear systems, the approximately optimal control for affine uncertain continuous-time nonlinear systems is pursued. Considering that only the output variables can be measured in the control practice, the state observer is designed to reconstruct all system states by using the output variables. The observer-based policy iteration algorithm can solve the HJB equation within the adaptive dynamic programming framework for the unknown-state uncertain nonlinear systems, where a critic neural network is constructed to approximate the optimal cost function, and then, the approximate expression of the optimal control policy can be directly derived from solving the HJB equation. In addition, the stability of the whole closed-loop system is provided based on the Lyapunov analysis.
4 September 2023 in IEEE Transactions on Emerging Topics in Computational Intelligence
Reinforcement learning (RL) algorithms require continuous interaction with the controlled plant to optimize the objective function and control policy. However, the security risk and poor real-time performance limit the application of RL algorithms. In this article, a data generation feedback relearning (DGFR) control algorithm is developed to avoid these limitations. The DGFR control algorithm improves the control performance by interacting with a proposed delayed neural network-based data generation model. Considering the time-delay characteristics of industrial production, the data generation model is different from the identifier of the system model, but it is used to generate the target data required by the DGFR control algorithm. The interaction between the algorithm and the data generation model avoids the risk of trial-and-error and improves real-time performance. Finally, the superiority of the proposed control algorithm is verified by giving comparative experiments in conjunction with a power grid system and a permanent magnet synchronous motor system.