Multi-agent Control Research Articles

Fault-tolerant Control of a Cyber-physical System

Cyber-physical systems represent a new emerging field in automatic control. The fault system is a key component, because modern, large scale processes must meet high standards of performance, reliability and safety. Fault propagation in large scale chemical processes can lead to loss of production, energy, raw materials and even environmental hazard. The present paper develops a multi-agent fault-tolerant control architecture using robust fractional order controllers for a (13C) cryogenic separation column cascade. The JADE (Java Agent DEvelopment Framework) platform was used to implement the multi-agent fault tolerant control system while the operational model of the process was implemented in Matlab/SIMULINK environment. MACSimJX (Multiagent Control Using Simulink with Jade Extension) toolbox was used to link the control system and the process model. In order to verify the performance and to prove the feasibility of the proposed control architecture several fault simulation scenarios were performed.

METHOD FOR ANTIFAULT CONTROL OF COMPLEX TECHNICAL SYSTEMS

A method for antifault control of complex technical systems (CTS) is proposed using hybrid expert systems (HES), which are built on the basis of neural networks (NN) and fuzzy logic that uses databases (DB) and automated knowledge bases (KB). This method uses monitoring, diagnostics and forecas-ting of the CTS data, calculating its operability.
 Then the reliability of the system under consideration is determined from the obtained data. Also, the system is capable of self-learning. Proceeding from this, with the help of a multi-agent control system, the system is influenced to avoid emergencies.
 As a result, the CTS emergency control system was developed that combines the advantages of the previously known HES and antifault control methods, which complement each other's advantages and compensate each other's shortcomings.

Multi-agent oriented solution for forecasting-based control strategy with load priority of microgrids in an island mode – Case study: Tunisian petroleum platform

To improve the power supply availability in an island microgrid, this paper proposes a new approach that integrates distributed energy sources economically, reliably and efficiently. In an island mode, a microgrid must ensure its self-sufficiency of energy production since it cannot make an energetic exchange with a main grid. However, in this mode, the random behavior of the resources affected by meteorological factors presents a major constraint. The challenge related to the power availability in microgrids is to find a solution that faces the operation of intermittent power sources. The microgrid should guarantee a useful power management in order to achieve a high availability of energy. In this paper, we present a mathematical model to describe the influence of the meteorological factors on the sources production. We propose a multi-agent control strategy based on the production forecasting and load shedding for a high availability of the microgrid power supply. The proposed multi-agent system uses the master-slave model in which the communication and negotiation between the defined agents are performed by a concept of tokens. The developed control system is implemented on Spartan 6 FPGA-Board. The paper's contribution is applied to a Tunisian petroleum platform where several blackouts are recorded between 2012 and 2014. Simulation and experimental results show clearly a high availability as a performance of the proposed control strategy.

A robotic system for researching social integration in honeybees.

In this paper, we present a novel robotic system developed for researching collective social mechanisms in a biohybrid society of robots and honeybees. The potential for distributed coordination, as observed in nature in many different animal species, has caused an increased interest in collective behaviour research in recent years because of its applicability to a broad spectrum of technical systems requiring robust multi-agent control. One of the main problems is understanding the mechanisms driving the emergence of collective behaviour of social animals. With the aim of deepening the knowledge in this field, we have designed a multi-robot system capable of interacting with honeybees within an experimental arena. The final product, stationary autonomous robot units, designed by specificaly considering the physical, sensorimotor and behavioral characteristics of the honeybees (lat. Apis mallifera), are equipped with sensing, actuating, computation, and communication capabilities that enable the measurement of relevant environmental states, such as honeybee presence, and adequate response to the measurements by generating heat, vibration and airflow. The coordination among robots in the developed system is established using distributed controllers. The cooperation between the two different types of collective systems is realized by means of a consensus algorithm, enabling the honeybees and the robots to achieve a common objective. Presented results, obtained within ASSISIbf project, show successful cooperation indicating its potential for future applications.

A coordinated consistency voltage stability control method of active distribution grid

The presence of distributed generators (DGs) with high penetration poses new challenges in the management and operation of electrical grids. Due to the local character of DGs, they could in principle be used in emergency situations to prevent a voltage instability event of the grid. In this paper, a certain method is proposed to coordinate the operation of virtual power plant (VPP) and conventional voltage regulation device to improve the static voltage stability of distribution network with the multi-agent framework. The concept and the general framework of this coordinated control system is introduced, and the voltage instable nodes are determined based on the voltage instability indicator. The voltage coordinated control model of the distribution system is established according to the multi-agent consistency control theory and the coordinated controllers for agents are designed by solving a problem with bilinear matrix inequality constraints. The suggested method is implemented on an IEEE 33 nodes test system and the simulation results show its efficiency and validity.

Simple synchronization protocols for heterogeneous networks: beyond passivity

Synchronization among autonomous agents via local interactions is one of the benchmark problems in multi-agent control. Whereas synchronization algorithms for identical agents have been thoroughly studied, synchronization of heterogeneous networks still remains a challenging problem. The existing algorithms primarily use the internal model principle, assigning to each agent a local copy of some dynamical system (internal model). Synchronization of heterogeneous agents thus reduces to global synchronization of identical generators and local synchronization between the agents and their internal models. The internal model approach imposes a number of restrictions and leads to sophisticated dynamical (and, in general, nonlinear) controllers. At the same time, passive heterogeneous agents can be synchronized by a very simple linear protocol, which is used for consensus of first-order integrators. A natural question arises whether analogous algorithms are applicable to synchronization of agents that do not satisfy the passivity condition. In this paper, we study the synchronization problem for heterogeneous agents that are not passive but satisfy a weaker input feedforward passivity (IFP) condition. We show that such agents can also be synchronized by a simple linear protocol, provided that the interaction graph is strongly connected and the couplings are sufficiently weak. We demonstrate how stability of cooperative adaptive cruise control algorithms and some microscopic traffic flow models reduce to synchronization of heterogeneous IFP agents.

Differential Privacy and Minimum-Variance Unbiased Estimation in Multi-agent Control Systems

In a discrete-time linear multi-agent control system, where the agents are coupled via an environmental state, knowledge of the environmental state is desirable to control the agents locally. However, since the environmental state depends on the behavior of the agents, sharing it directly among these agents jeopardizes the privacy of the agents’ profiles, defined as the combination of the agents’ initial states and the sequence of local control inputs over time. A commonly used solution is to randomize the environmental state before sharing - this leads to a natural trade-off between the privacy of the agents’ profiles and the variance of estimating the environmental state. By treating the multi-agent system as a probabilistic model of the environmental state parametrized by the agents’ profiles, we show that when the agents’ profiles is e-differentially private, there is a lower bound on the l1 induced norm of the covariance matrix of the minimum-variance unbiased estimator of the environmental state. This lower bound is achieved by a randomized mechanism that uses Laplace noise.

Online Abstractions for Interconnected Multi-Agent Control Systems

In this paper, we aim at the development of an online abstraction framework for multi-agent systems under coupled constraints. The motion capabilities of each agent are abstracted through a finite state transition system in order to capture reachability properties of the coupled multi-agent system over a finite time horizon in a decentralized manner. In the first part of this work, we define online abstractions by discretizing an over approximation of the agents’ reachable sets over the horizon. Then, sufficient conditions relating the discretization and the agents’ dynamics are provided, in order to quantify the agents’ transition possibilities.

Controllability of multiagent systems based on path and cycle graphs

SummaryThis paper studies the controllability of multiagent systems based on path and cycle graphs. For paths or cycles, sufficient and necessary conditions are presented for determining the locations of leaders under which the controllability can be realized. Specifically, the controllability of a path is shown to be determined by a set generated only from its odd factors, and the controllability of a cycle is determined by whether the distance between 2 leaders belongs to a set generated from its even (odd) factors when the number of its nodes is even (odd). For both graphs, the dimension of the controllable subspace is also derived. Moreover, the technique used in the derivation of the above results is further used to get sufficient and necessary conditions for several different types of graphs generated from path and cycle topologies. These types of graphs can be regarded as typical topologies in the study of multiagent controllability, and accordingly the obtained results have meaningful enlightenment for the research in this field.

Системы управления знаниями и многоагентные системы

Работа посвящена анализу подходов к построению систем управления знаниями научно-образовательных организаций. Рассмотрены особенности СУЗ вузов и проблемы их реализации. Подчеркивается важность мотивации персонала в отчуждении знаний. Показана перспективность многоагентного подхода к созданию СУЗ вузов, в том числе с точки зрения мотивации персонала. Сформулированы основные принципы построения многоагентной СУЗ. Поставлена задача моделирования такой СУЗ в реальных условиях. Рассмотрена возможная функциональность агентов многоагентной СУЗ.

Student research highlights: Quadrotor electronics and intelligent control

Unmanned aerial vehicles (UAVs) have seen diverse applications, including radiation or border surveillance, with quadrotors offering small size, relatively low cost, high mobility, mass production, and easy access to challenging environments. In addition, quadrotors can carry sensors, cameras, or other gadgets for information collection, data fusion, and subsequent actions of targets, as was exemplified during the Fukushima nuclear plant disaster, by visual and sensory surveillance or for an air delivery system. UAVs have also spurred extensive research on physical design, operation principles, components, and control algorithms, partially due to complex stabilization problems, including the Stanford testbed of autonomous rotorcraft for multiagent control [1] quadrotor with linear controllers for position control systems, the X-4 flyer [2] with discrete model controllers, and the MikroKopter [3] with an open-source code development, as well as implementation and control with state estimation. Inherently nonlinear UAV dynamics can also be governed by intelligent controllers that can handle complex, nonlinear systems for superior outcomes. As an intelligent framework, fuzzy logic (FL) reflects real-life vagueness and imprecise data to provide decisions via nonlinear mapping via fuzzy sets, rules, and membership functions [4]. This project describes the development and implementation of a practical, low cost, lightweight quadrotor electronic test platform design to evaluate the FL closed-loop controller for roll and pitch stabilization.

A New Perspective to Graphical Characterization of Multiagent Controllability.

Recently, graphical characterization of multiagent controllability has been studied extensively. A major effort in the study is to determine controllability directly from topology structures of communication graphs. In this paper, we proposed the concept of controllability destructive nodes, which indicates that the difficulty in graphical characterization turns out to be the identification of topology structures of controllability destructive nodes. It is shown that each kind of double and triple controllability destructive nodes happens to have a uniform topology structure which can be defined similarly. The definition, however, is verified not to be applicable to the topology structures of quadruple controllability destructive (QCD) nodes. Even so, a design method is proposed to uncover topology structures of QCD nodes for graphs with any size, and a complete graphical characterization is presented for the graphs consisting of five vertices. One advantage of the established complete graphical characterization is that the controllability of any graph with any selection of leaders can be determined directly from the identified/defined destructive topology structures. The results generate several necessary and sufficient graphical conditions for controllability. A key step of arriving at these results is the discovery of a relationship between the topology structure of the controllability destructive nodes and a corresponding eigenvector of the Laplacian matrix.

Improving multi-agent manufacturing control system by indirect communication based on ant agents

Manufacturing control systems are facing difficulty on adopting agility in today’s dynamic environment. Much research has been conducted on the use of multi-agent systems to facilitate this transition. In our previous work, a radio frequency identification–enabled multi-agent-based manufacturing control system was introduced for flexible manufacturing systems. In that control system, a direct coordination mechanism coordinates the agents. However, the decision-making response of the control system to disturbances was low and caused the agents to spend more time on processing messages than doing the actual job. Thus, in this article, we propose an indirect coordination mechanism based on ant colony intelligence aiming to improve the self-organization and processing time of the system. The new system utilizes the stigmergy mechanism for cooperation among the agents. Furthermore, ant agents are created by the existing agents to enable communication among them. The proposed new mechanism was implemented in a flexible assembly line. The simulation results of a case study considering shop floor problems indicates, for most performance measures, that the radio frequency identification–enabled multi-agent-based manufacturing control system integrated with indirect coordination mechanism performs better in comparison with the radio frequency identification–enabled multi-agent-based manufacturing control system integrated with direct coordination mechanism.

Multi-agent based operational cost and inconvenience optimization of PV-based microgrid

The integration of solar power generation into microgrid systems has become very popular due to its positive environmental aspects and cost effectiveness. Nevertheless the existence of natural intermittency and fluctuations in PV generation incurs extra cost or service interruption in PV-based microgrids. The power generation of PV systems follows a natural schedule based on a sunny day. Similarly, the usage profiles in a microgrid are known from experience. When there is a mismatch in load or generation schedule, the system has to react to maintain a balance. In this work, both a centralized and a decentralized demand-responsive multi-agent control and management system are devised which include backup diesel generation and load curtailment. The latter affects user satisfaction. Wpose new realistic models to measure user satisfaction depending on the type of appliance curtailed. Our simulation shows that the inclusion of demand-side management lowers the cost of a mismatch even when user satisfaction is considered. Expectedly, the centralized implementation achieves a lower cost in more difficult conditions - when the peak consumption happens earlier than anticipated - but the decentralised approach provides acceptable cost levels when a centralized model cannot be implemented.

マルチエージェントシステムの制御 -VI 分散最適化

マルチエージェントシステムの制御 -VI 分散最適化

Mind as multiresolution system based on multiagents architecture

Abstract The paper presents four most advanced computer science perspective initiatives aiming for the mind understanding and replication of its functions. The second goal of the paper is to introduce a hybrid architecture based on multiagent system and real-time control units organized in multiresolution layers. Collaborative functional specialized modules implemented as agents solve complex tasks. The agent society deployed as cloud provides an on-demand runtime agent instantiation in order to compete or cooperate with others during the problem solving process. Complex tasks decomposition leads to hierarchical specialized agent structures which fulfill the goal by collaboration. The purpose is to offer a generic framework for research of neurocognitive domains like perception, learning, memory.

A flocking algorithm with individual agent destinations and without a centralized leader

We present a multi-agent control method that addresses the combined problem of flocking and destination seeking. The method is completely decentralized, that is, each agent’s controller relies on local sensing to determine the relative positions and velocities of nearby agents but does not rely on a centralized flock leader. Each agent has double-integrator dynamics and a potentially unique destination (i.e., position) that the agent must reach. We demonstrate that the flocking-and-destination-seeking control method accomplishes 2 objectives: (i) if an agent is far from its destination, then that agent flocks with nearby agents, and (ii) if an agent is close to its destination, then that agent approaches its destination. The flocking-and-destination-seeking algorithm is demonstrated with several numerical examples.

Multi-Agent-Based Controller for Voltage Enhancement in AC/DC Hybrid Microgrid Using Energy Storages

Development of renewable energies and DC loads have led microgrids toward the creation of DC networks. The predictions show that the hybrid microgrids will be used widely in the future. This article has studied the voltage stability in the presence of sources of energy storage in AC/DC hybrid networks. However, because the different dynamics of hybrid networks applying centralized and distributed controllers will be faced with different problems, in this study, a multi-agent control for the microgrid has been used. A new structure referred to here as an event-driven microgrid control management (EDMCM) has been developed to control the microgrid. This method increases response speed and accuracy of decision making. Hybrid Network Simulation results confirm the validity of the developed model.

Game-Theoretic Multi-Agent Control and Network Cost Allocation Under Communication Constraints

Multi-agent networked linear dynamic systems have attracted the attention of researchers in power systems, intelligent transportation, and industrial automation. The agents might cooperatively optimize a global performance objective, resulting in social optimization, or try to satisfy their own selfish objectives using a noncooperative differential game. However, in these solutions, large volumes of data must be sent from system states to possibly distant control inputs, thus resulting in high cost of the underlying communication network. To enable economically viable communication, a game-theoretic framework is proposed under the communication cost , or sparsity , constraint, given by the number of communicating state/control input pairs. As this constraint tightens, the system transitions from dense to sparse communication, providing the tradeoff between dynamic system performance and information exchange. Moreover, using the proposed sparsity-constrained distributed social optimization and noncooperative game algorithms, we develop a method to allocate the costs of the communication infrastructure fairly and according to the agents’ diverse needs for feedback and cooperation. Numerical results illustrate utilization of the proposed algorithms to enable and ensure economic fairness of wide-area control among power companies.

Self Organizing Networks: A Reinforcement Learning approach for self-optimization of LTE Mobility parameters

With the evolution of broadband mobile networks towards LTE and beyond, the support for the Internet and Internet based services is growing. Self Organizing Network (SON) functionalities intend to optimize the network performance for the improved user experience while at the same time reducing the network operational cost. This paper proposes a Reinforcement Learning (RL) based framework to improve throughput of the mobile users. The problem of spectral efficiency maximization is modeled as co-operative Multi-Agent control problem between the neighbouring eNodeBs (eNBs). Each eNB has an associated agent that dynamically changes the outgoing Handover Margin (HM) to its neighbouring cells. The agent uses the RL technique of Fuzzy Q-Learning (FQL) to learn the optimal mobility parameter i.e., HM value. The learning framework is designed to operate in an environment with the variations in traffic, user positions and propagation conditions. Simulation results have shown the proposed approach improves the network capacity and user experiences in terms of throughput.