Distributed Predictive Control for Multiagent Systems With Communication Delays via a Hybrid Data-Driven and Model-Based Approach
针对多智能体系统在通信延迟下的一致性和跟踪问题，提出一种基于观测器的分布式预测控制方案，通过辅助变量和数据驱动优化主动补偿延迟，数值和工业案例验证了有效性。
This article focuses on the consensus and tracking problem of multiagent systems under communication delays, and proposes an observer-based distributed predictive control scheme to actively compensate for communication delays. First, an auxiliary variable and its corresponding estimator are innovatively designed to provide an observation of a globally consensus value that converges faster than the actual system output. Then, a cost function is formulated for this observer, and the gain of the correction term is optimized in a data-driven way to enhance the accuracy of the estimation. Based on the auxiliary variables, a distributed predictive controller is presented. Unlike existing control schemes, the proposed controller compensates for the delayed data in an active way, and the rolling prediction process is carried out in a distributed manner, facilitated by the designed auxiliary variables. Case studies conducted in both numerical and industrial scenarios demonstrate the effectiveness of the proposed method.