A dynamic event-triggered network control algorithm combined with gradient-sharing asynchronous advantage actor-critic strategy

In the existing memory event-triggered control (METC) algorithms, the threshold parameters and memory weights are fixed, reducing the system's adaptability. In this paper, a new intelligent dynamic METC algorithm is proposed to reduce the amount of transmission data and decrease the communication burden in networked control systems. The proposed dynamic METC mechanism applies the gradient-sharing asynchronous advantage actor-critic (A3C-GS) learning algorithm to optimized memory event-triggered function of memory weights and threshold parameters. A time-varying delay system model for dynamic METC is developed by incorporating the A3C-GS for the networked control system with communication delays. Then, by solving two linear matrix inequalities (LMIs), the controller gain parameters of the networked control systems are derived. Finally, the performance of the proposed new METC algorithm is compared with the current three event-triggered control methods. Simulation results show that the proposed intelligent event-triggered algorithm reduces the number of triggers by about 40% compared with the traditional event-triggered algorithm under the pregiven simulation time. Thus, the effectiveness of the main results is verified.

Automated summary

In this paper, a new intelligent dynamic METC algorithm is proposed to reduce the amount of transmission data and decrease the communication burden in networked control systems. The proposed algorithm optimizes the memory event-triggered function by applying the A3C-GS learning algorithm. Simulation results show that the proposed algorithm reduces the number of triggers by about 40% compared with traditional event-triggered algorithms.