Distributed Time-Varying Quadratic Optimal Resource Allocation Subject to Nonidentical Time-Varying Hessians With Application to Multiquadrotor Hose Transportation
Distributed Time-Varying Quadratic Optimal Resource Allocation Subject to Nonidentical Time-Varying Hessians With Application to Multiquadrotor Hose Transportation
复制标题
DOI：
10.1109/tsmc.2021.3137814
复制
发表时间：
2022-10
期刊：
影响因子：
--
通讯作者：
B. Wang;Shan Sun;W. Ren
中科院分区：
文献类型：
--
作者：
B. Wang;Shan Sun;W. Ren
研究了多智能体系统中具有时变二次代价函数和时变耦合等式约束的分布式时变最优资源分配问题。我们的目标是设计一个分布式算法的代理与单积分动态合作，以满足耦合等式约束和最小化所有的本地成本函数的总和。这里，耦合等式约束和成本函数都显式地依赖于时间。成本函数是二次型的，并且可以具有不同的时变海森。为了以分布式方式解决问题，首先为每个代理开发基于分布式平均跟踪方法的估计器，以估计某些全局信息。通过利用估计的全局信息和自适应增益方案，提出了一种分布式连续时间算法，该算法确保智能体发现和跟踪时变最优轨迹，且误差消失。我们说明了所提出的方法在使用多个四旋翼的最优软管运输问题的适用性。
This article considers the distributed time-varying optimal resource allocation problem with time-varying quadratic cost functions and a time-varying coupled equality constraint for multiagent systems. The objective is to design a distributed algorithm for agents with single-integrator dynamics to cooperatively satisfy the coupled equality constraint and minimize the sum of all local cost functions. Here, both the coupled equality constraint and cost functions depend explicitly on time. The cost functions are in quadratic form and may have nonidentical time-varying Hessians. To solve the problem in a distributed manner, an estimator based on the distributed average tracking method is first developed for each agent to estimate certain global information. By leveraging the estimated global information and an adaptive gain scheme, a distributed continuous-time algorithm is proposed, which ensures the agents to find and track the time-varying optimal trajectories with vanishing errors. We illustrate the applicability of the proposed method in the optimal hose transportation problem using multiple quadrotors.