Networked predictive control is an effective method to handle time delays and data dropouts in networked control systems, but it is difficult to apply the control strategy to networked multi-agent systems. First, the computational burden will increase exceedingly when the control strategy is applied to networked multi-agent systems. Second, restrictions have to be imposed on the communication topology to implement the control strategy; Third, most existing networked predictive controllers lack robustness to model uncertainties. In this paper, these problems are solved by introducing predicting error to the control scheme. The proposed control strategy not only guarantees consensus and stability of the overall system, but also reduces the computational burden exceedingly. In addition, the proposed algorithm is robust to model uncertainties. At last, the proposed control strategy can be executed in distributed processors, therefore, flexibility and robustness are preserved for the networked multi-agent systems. These features make the control strategy applicable in practice. Digital simulations and experiments based on spacecraft simulators are carried out to verify the effectiveness of the proposed control strategy.

网络预测控制是处理网络控制系统中的时间延迟和数据丢失的有效方法，但是很难将控制策略应用于网络多智能体系统。首先，当控制策略应用于网络化多智能体系统时，计算负担将大大增加。其次，必须对通信拓扑施加限制以实施控制策略。第三，大多数现有的网络预测控制器缺乏对不确定性进行建模的鲁棒性。本文通过将预测误差引入控制方案来解决这些问题。所提出的控制策略不仅保证了整个系统的一致性和稳定性，而且极大地减轻了计算负担。此外，所提出的算法对于不确定性建模具有鲁棒性。最后，所提出的控制策略可以在分布式处理器中执行，因此，为网络化多智能体系统保留了灵活性和鲁棒性。这些功能使控制策略在实践中适用。进行了基于航天器模拟器的数字仿真和实验，以验证所提出的控制策略的有效性。