Distributed Control Agents Research Articles

Analysis of coupling dynamics for power systems with iterative discrete decision making architectures

Iterative “learning” by distributed control agents has been proposed for power system decision making. Such decision making can achieve agreement among control agents while preserving privacy. The iterative decision making process may interact with power system dynamics. In such cases, coupled dynamics are expected. The objective of this paper is to propose a modeling approach that can conduct stability analysis for these hybrid systems. In the proposed approach, the discrete decision making process is approximated by continuous dynamics. As a result, the entire hybrid system can be represented by a continuous dynamic system. Conventional stability analysis tools are then used to check system stability and identify key impacting factors. An example power system with multiple control agents is used to demonstrate the proposed modeling and analysis. The analysis results are then validated by nonlinear time-domain simulation.The continuous dynamics models developed in this paper sheds insights into the control nature of each distributed optimization algorithm. An important finding is documented in this paper: a consensus algorithm based decision making may act as an integrator of frequency deviation. It can bring the frequency back to nominal while the primal-dual based decision making cannot.

Cooperative distributed MPC of linear systems with coupled constraints

This paper develops a cooperative, distributed form of MPC for linear systems subject to persistent, bounded disturbances. The distributed control agents make decisions locally and communicate plans with each other. Cooperation is promoted by consideration of a greater portion of the system-wide objective by each local agent; specifically, a local agent designs hypothetical plans for other agents, sacrificing local performance for the benefit of system-wide performance. These hypothetical plans are never communicated and no negotiation takes place. The method guarantees robust feasibility by permitting only one agent to optimize per time step, while ‘freezing’ the plans of others, and sufficient conditions are given for robust stability. These properties hold for all structures of cooperation between agents. Thus, a key feature is that coupled constraint satisfaction is compatible with inter-agent cooperation.

Distributed model predictive control of linear systems with persistent disturbances

This article presents a new form of robust distributed model predictive control (MPC) for multiple dynamically decoupled subsystems, in which distributed control agents exchange plans to achieve satisfaction of coupling constraints. The new method offers greater flexibility in communications than existing robust methods, and relaxes restrictions on the order in which distributed computations are performed. The local controllers use the concept of tube MPC – in which an optimisation designs a tube for the system to follow rather than a trajectory – to achieve robust feasibility and stability despite the presence of persistent, bounded disturbances. A methodical exploration of the trades between performance and communication is provided by numerical simulations of an example scenario. It is shown that at low levels of inter-agent communication, distributed MPC can obtain a lower closed-loop cost than that obtained by a centralised implementation. A further example shows that the flexibility in communications means the new algorithm has a relatively low susceptibility to the adverse effects of delays in computation and communication.

Designing Communication Networks to Decompose Network Control Problems

The pressure from today’s economic and energy markets demands further distribution of decision making in large, dynamic networks. To this end, distributed model predictive control (MPC) divides the task of operating a dynamic network into a set of small, localized subtasks, one for each distributed control agent. To guide the decomposition of the overall task, which plays a central role in the quality of the operation yielded by the distributed agents, this paper proposes a model for problem decomposition based on the design of agent communication networks. The model gives rise to the communication-network design problem: A bicriteria optimization problem whose objectives are the maximization of the influence perceived by the agents and the minimization of the communication cost. The benefit of the model is the potential to concentrate the effort of implementing distributed MPC, and measuring its actual performance, on a reduced number of decompositions, namely those that are Pareto efficient for the model. The paper gives an account of related work, including the computational complexity of finding Pareto efficient solutions, an integer programming formulation, and families of valid inequalities. Its main contribution is the demonstration that the model can be effective, meaning that Pareto efficient solutions to the model tend to induce efficient problem decompositions. The experimental evidence was gathered by applying the model to decompose control problems of two representative networks, namely arrays of pendulums and electric power networks, and measuring the quality of the operation delivered by the distributed control agents.

Designing communication networks for distributed control agents

This paper is concerned with a bicriteria combinatorial optimization problem and its applications to the design of communication networks for distributed controllers. In a bipartite graph, a subset of the edges yields a communication network for which we pay a price but, on the other hand, it induces another network that encompasses the former and for which we receive a reward. The problem is then to find the efficient solutions. The paper delivers an IP formulation for the problem of finding supported solutions, which is shown to be NP-Hard, along with families of strong valid inequalities, which are shown to draw the LP bounds closer to the Pareto optimal frontier. To approximate the frontier, heuristics are integrated in a problem-solving architecture called asynchronous team and then put to the test in two prototypical scenarios.