.png){kind=icon}
.png){kind=icon}
Return
Cloud-Based Computational Data-Enabled Predictive Control
.png){kind=icon}
PRE
AI
DOI：10.1109/JIOT.2022.3194945.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
This article considers the data-driven optimal control problem for unknown linear systems subject to system constraints from a computation efficiency improvement perspective. We propose a novel Computational Data-enabled Predictive Control (CDeePC) algorithm in a cloud environment, built on a recent work DeePC [1]. First, massive input-output data samples are precollected to describe the system input/output behavior through a behavioral systems theory approach, which might result in a large-scale online optimization problem with high dimensional decision variables. To solve it, in CDeePC, a multiblock alternating direction method of multipliers (ADMMs) algorithm is then employed, in which a solution to the original large problem can be calculated by solving iteratively a set of small subproblems. Next, to further improve the robustness of the algorithm, an online feedback CDeePC (OCDeePC) algorithm is proposed by utilizing real-time data samples to better capture the system characterization. Two analytic inversion formulas are derived for fast computation of time-varying controller parameters based on the result at the previous time. After that, we discuss the computational complexity and provide the convergence proof of proposed algorithms. Finally, after designing a cloud-based control architecture and formulating the iterative scheme to compute control actions as a workflow, we construct and deploy the proposed controllers as a service in the cloud. A case study of tracking control of wheeled mobile robots is provided to illustrate the efficacy of the proposed algorithms.
Keywords:
Alternating direction method of multipliers (ADMM)
cloud computing
convergence analysis
data-driven control
model predictive control (MPC)
Journal
IF:
8.9
Papers:
1.4W
Citations:
7.8W