As more industrial processes, transportation and appliances have been automated or equipped with some level of artificial intelligence, the number and scale of interconnected systems has grown in the recent past. This is a development which can be expected to continue and therefore the research in performance of interconnected systems and networks is growing. Due to increased automation and sheer scale of networks, dynamically scaling networks is an increasing field and research into scalable performance measures is advancing. Recently, the notion gamma-robustness, a scalable network performance measure, was introduced as a measurement of interconnected systems robustness with respect to external disturbances. This thesis aims to investigate how the distribution of control effort and cost, within interconnected system, affects network performance, measured with gamma-robustness. Further, we introduce a notion of fairness and a measurement of unfairness in order to quantify the distribution of network properties and performance. With these in place, we also present distributed algorithms with which the distribution of control effort can be controlled in order to achieve a desired network performance. We close with some examples to show the strengths and weaknesses of the presented algorithms. / I och med att fler och fler system och enheter blir utrustade med olika grader av intelligens så växer både förekomsten och omfattningen av sammankopplade system, även kallat Multi-Agent Systems. Sådana system kan vi se exempel på i traffikledningssystem, styrning av elektriska nätverk och fordonståg, vi kan också hitta fler och fler exempel på så kallade sensornätverk i och med att Internet of Things och Industry 4.0 används och utvecklas mer och mer. Det som särskiljer sammankopplade system från mer traditionella system med flera olika styrsignaler och utsignaler är att dem sammankopplade systemen inte styrs från en central styrenhet. Istället styrs dem sammankopplade systemen på ett distribuerat sätt i och med att varje agent styr sig själv och kan även ha individuella mål som den försöker uppfylla. Det här gör att analysen av sammankopplade system försvåras, men tidigare forskning har hittat olika regler och förhållninssätt för agenterna och deras sammankoppling för att uppfylla olika krav, såsom stabilitet och robusthet. Men även om dem sammankopplade systemen är både robusta och stabila så kan dem ha egenskaper som vi vill kunna kontrollera ytterligare. Specifikt kan ett sådant prestandamått vara systemens motståndskraft mot påverkan av yttre störningar och i vanliga olänkade system finns det en inneboende avvägning mellan kostnad på styrsignaler och resiliens mot yttre störningar. Samma avvägning hittar vi i sammankopplade system, men i dessa system hittar vi också ytterligare en dimension på detta problem. I och med att ett visst mått av en nätverksprestanda inte nödvändigtvis betyder att varje agent i nätverket delar samma mått kan agenterna i ett nätverk ha olika utväxling mellan styrsignalskostnad och resiliens mot yttre störningar. Detta gör att vissa agenter kan ha onödigt höga styrsignalskonstander, i den mening att systemen skulle uppnå samma nätverksprestanda men med lägre styrsignalskostnad om flera av agenterna skulle vikta om sina kontrollinsatser. I det här examensarbetet har vi studerat hur olika val av kontrollinsats påverkar ett sammankopplat systems prestanda. Vi har gjort detta för att undersöka hur autonoma, men sammankopplade, agenter kan ändra sin kontrollinsats, men med bibehållen nätverksprestanda, och på det sättet minska sina kontrollkostnader. Detta har bland annat resulterat i en distruberad algoritm för att manipulera agenternas kontrollinsats så att skillnaderna mellan agenternas resiliens mot yttre störningar minskar och nätverksprestandan ökar. Vi avslutar rapporten med att visa ett par exempel på hur system anpassade med hjälp av den framtagna algoritmen får ökad prestanda. Avslutningsvis följer en diskussion kring hur vissa antaganden kring systemstruktur kan släppas upp, samt kring vilka områden framtida forskning skulle kunna fortsätta med.

A cooperative multi-agent system is a collection of interacting agents deployed in a mission space where each agent is allowed to control its local state so that the fleet of agents collectively optimizes a common global objective. While optimization problems associated with multi-agent systems intend to determine the fixed set of globally optimal agent states, control problems aim to obtain the set of globally optimal agent controls. Associated non-convexities in these problems result in multiple local optima. This dissertation explores systematic techniques that can be deployed to either escape or avoid poor local optima while in search of provably better (still local) optima. First, for multi-agent optimization problems with iterative gradient-based solutions, a distributed approach to escape local optima is proposed based on the concept of boosting functions. These functions temporarily transform gradient components at a local optimum into a set of boosted non-zero gradient components in a systematic manner so that it is more effective compared to the methods where gradient components are randomly perturbed. A novel variable step size adjustment scheme is also proposed to establish the convergence of this distributed boosting process. Developed boosting concepts are successfully applied to the class of coverage problems. Second, as a means of avoiding convergence to poor local optima in multi-agent optimization, the use of greedy algorithms in generating effective initial conditions is explored. Such greedy methods are computationally cheap and can often exploit submodularity properties of the problem to provide performance bound guarantees to the obtained solutions. For the class of submodular maximization problems, two new performance bounds are proposed and their effectiveness is illustrated using the class of coverage problems. Third, a class of multi-agent control problems termed Persistent Monitoring on Networks (PMN) is considered where a team of agents is traversing a set of nodes (targets) interconnected according to a network topology aiming to minimize a measure of overall node state. For this class of problems, a gradient-based parametric control solution developed in a prior work relies heavily on the initial selection of its `parameters' which often leads to poor local optima. To overcome this initialization challenge, the PMN system's asymptotic behavior is analyzed, and an off-line greedy algorithm is proposed to systematically generate an effective set of initial parameters. Finally, for the same class of PMN problems, a computationally efficient distributed on-line Event-Driven Receding Horizon Control (RHC) solution is proposed as an alternative. This RHC solution is parameter-free as it automatically optimizes its planning horizon length and gradient-free as it uses explicitly derived solutions for each RHC problem invoked at each agent upon each event of interest. Hence, unlike the gradient-based parametric control solutions, the proposed RHC solution does not force the agents to converge to one particular behavior that is likely to be a poor local optimum. Instead, it keeps the agents actively searching for the optimum behavior. In each of these four parts of the thesis, an interactive simulation platform is developed (and made available online) to generate extensive numerical examples that highlight the respective contributions made compared to the state of the art.

This dissertation addresses the flow control problem in hyperbolic heat transfer mechanisms. It raises in concentrated distributed solar collectors to enhance their production efficiency under the unpredictable variations of the solar energy and the external disturbances. These factors which are either locally measured (the solar irradiance) or inaccessible for measurement (the collectors’ cleanliness) affect the source term of the distributed model and represent a major difficulty for the control design. Moreover, the temperature in the collector can only be measured at the boundaries. In this dissertation, we propose new adaptive control approaches to provide the adequate level of heat while coping with the unpredictable varying disturbances. First, we design model based control strategies for a better efficiency, in terms of accuracy and response time, with a relatively reduced complexity. Second, we enhance the controllers with on-line adaptation laws to continuously update the efficient value of the external conditions. In this study, we approach the control problem using both, the infinite dimensional model (late lumping) and a finite dimensional approximate representation (early lumping). For the early lumping approach, we introduce a new reduced order bilinear approximate model for system analysis and control design. This approximate state representation is then used to derive a nonlinear state feedback resorting to Lyapunov stability theory. To compensate for the external disturbances and the approximation uncertainties, an adaptive controller is developed based on a phenomenological representation of the system dynamics. For the late lumping approach, we propose two PDE based controllers by stabilization of the reference tracking error distributed profile. The control laws are explicitly defined as functions of the available measurement. The first one is obtained using a direct approach for error stabilization while the second one is derived through a nonlinear mapping. Furthermore, we endow the nonlinear controllers with an adaptation law to cope with the unpredictable unmeasured disturbances. The proposed adaptation law is based on a Proportional plus Integral correction feedback. We show that the control objectives with the required performance can be achieved following both approaches, but yet are conditioned with the physical limitations of the system.

This dissertation focuses on the study of decentralized coordination algorithms of multiple autonomous vehicles. Here, the term decentralized coordination is used to refer to the behavior that a group of vehicles reaches the desired group behavior via local interaction. Research is conducted towards designing and analyzing distributed coordination algorithms to achieve desired group behavior in the presence of none, one, and multiple group reference states. Decentralized coordination in the absence of any group reference state is a very active research topic in the systems and controls society. We first focus on studying decentralized coordination problems for both single-integrator kinematics and double-integrator dynamics in a sampled-data setting because real systems are more appropriate to be modeled in a sampled-data setting rather than a continuous setting. Two sampled-data consensus algorithms are proposed and the conditions to guarantee consensus are presented for both fixed and switching network topologies. Because a number of coordination algorithms can be employed to guarantee coordination, it is important to study the optimal coordination problems. We further study the optimal consensus problems in both continuous-time and discrete-time settings via an linear-quadratic regulator (LQR)-based approach. Noting that fractional-order dynamics can better represent the dynamics of certain systems, especially when the systems evolve under complicated environment, the existing integer-order coordination algorithms are extended to the fractional-order case. Decentralized coordination in the presence of one group reference state is also called coordinated tracking, including both consensus tracking and swarm tracking. Consensus tracking refers to the behavior that the followers track the group reference state. Swarm tracking refers to the behavior that the followers move cohesively with the external leader while avoiding inter-vehicle collisions. In this part, consensus tracking is studied in both discrete-time setting and continuous-time settings while swarm tracking is studied in a continuous-time setting. Decentralized coordination in the presence of multiple group reference states is also called containment control, where the followers will converge to the convex hull, i.e., the minimal geometric space, formed by the group references states via local interaction. In this part, the containment control problem is studied for both single-integrator kinematics and double-integrator dynamics. In addition, experimental results are provided to validate some theoretical results.

京都大学 / 0048 / 新制・課程博士 / 博士(情報学) / 甲第19124号 / 情博第570号 / 新制||情||100(附属図書館) / 32075 / 京都大学大学院情報学研究科システム科学専攻 / (主査)教授 杉江 俊治, 教授 太田 快人, 教授 大塚 敏之, 准教授 東 俊一 / 学位規則第4条第1項該当 / Doctor of Informatics / Kyoto University / DFAM

Shipboard Power System (SPS) is known as an independent controlled small electric network powered by the distributed onboard generation system. Since many electric components are tightly coupled in a small space and the system is not supported with a relatively stronger grid, SPS is more susceptible to unexpected disturbances and physical damages compared to conventional terrestrial power systems. Among different distribution configurations, power-electronic based DC distribution is considered the trending technology for the next-generation U.S. Navy fleet design to replace the conventional AC-based distribution. This research presents appropriate control management frameworks to improve the Medium-Voltage DC (MVDC) shipboard power system performance. Model Predictive Control (MPC) is an advanced model-based approach which uses the system model to predict the future output states and generates an optimal control sequence over the prediction horizon. In this research, at first, a centralized MPC is developed for a nonlinear MVDC SPS when a high-power pulsed load exists in the system. The closed-loop stability analysis is considered in the MPC optimization problem. A comparison is presented for different cases of load prediction for MPC, namely, no prediction, perfect prediction, and Autoregressive Integrated Moving Average (ARIMA) prediction. Another centralized MPC controller is also designed to address the reconfiguration problem of the MVDC system in abnormal conditions. The reconfiguration goal is to maximize the power delivered to the loads with respect to power balance, generation limits and load priorities. Moreover, a distributed control structure is proposed for a nonlinear MVDC SPS to develop a scalable power management architecture. In this framework, each subsystem is controlled by a local MPC using its state variables, parameters and interaction variables from other subsystems communicated through a coordinator. The Goal Coordination principle is used to manage interactions between subsystems. The developed distributed control structure brings out several significant advantages including less computational overhead, higher flexibility and a good error tolerance behavior as well as a good overall system performance. To demonstrate the efficiency of the proposed approach, a performance analysis is accomplished by comparing centralized and distributed control of global and partitioned MVDC models for two cases of continuous and discretized control inputs.

Distributed computing systems (DCS) host a wide variety of enterprise applications in dynamic and uncertain operating environments. These applications require stringent reliability, availability, and quality of service (QoS) guarantee to maintain their service level agreements (SLAs). Due to the growing size and complexity of DCS, an autonomic performance management system is required to maintain SLAs of these applications. A model-based autonomic performance management structure is developed in this dissertation for applications hosted in DCS. A systematic application performance modeling approach is introduced in this dissertation to define the dependency relationships among the system parameters, which impact the application performance. The developed application performance model is used by a model-based predictive controller for managing multi-dimensional QoS objectives of the application. A distributed control structure is also developed to provide scalability for performance management and to eliminate the requirement of approximate behavior modeling in the hierarchical arrangement of DCS. A distributed monitoring system is also introduced in this dissertation to keep track of computational resources utilization, application performance statistics, and scientific application execution in a DCS, with minimum latency and controllable resource overhead. The developed monitoring system is self-configuring, self-aware, and fault-tolerant. It can also be deployed for monitoring of DCS with heterogeneous computing systems. A configurable autonomic performance management system is developed using modelintegrated computing methodologies, which allow administrators to define the initial settings of the application, QoS objectives, system components’ placement, and interaction among these components in a graphical domain specific modeling environment. This configurable performance management system facilitates reusability of the same components, algorithms, and application performance models in different deployment settings.

No description available.

As we begin to analyze and construct extremely complex systems, a theory for understanding and designing the underlying architecture becomes very important. To move in the direction of a precise theory of architecture, this thesis will provide some concrete tools to analyze and design complex systems with a given network structure. The first main result of this thesis analyzes the vulnerability of a system and shows that a system's vulnerability depends on its network structure. We will consider destabilization attacks acting on a single link in a system's logical interconnection structure. The concept of a vulnerable link is characterized and necessary and sufficient conditions for identifying these links are provided. The vulnerability of various system architectures are then characterized by the vulnerability of their weakest link, and it is shown that every transfer function has a completely secure architecture with no vulnerable links. The second part of this thesis focuses on synthesizing controllers with a specified network structure. It presents a new approach to distributed controller design that exploits the dynamical structure function representation of linear time invariant systems to characterize the structure of a system. The design technique sequentially constructs each link in an arbitrary controller signal structure, and the main theorem proves that either the resulting controller is stabilizing or that no controller with the desired structure can stabilize the system.

This dissertation concerns the formation problems for multiple reduced attitudes, which are extensively utilized in many pointing applications and under-actuated scenarios for attitude maneuvers. In contrast to most existing methodologies on formation control, the proposed method does not need to contain any formation errors in the protocol. Instead, the constructed formation is attributed to geometric properties of the configuration space and the designed connection topology. We refer to this type of formation control as intrinsic formation control. Besides, the control protocols proposed in this work are designed directly in space S2, avoiding to use any attitude parameterisations. At last but not least, along the studies, some elementary tools for reduced attitudes control are developed.In paper A, a continuous control law is provided for a reduced attitude systems, by which a regular tetrahedron formation can achieve asymptotic stability under a quite large family of gain functions in the control. Then, with a further restriction on the control gain, almost global stability of the tetrahedron formation is also obtained. In this work, we introduce a novel coordinates transformation that represents the relative reduced attitudes be-tween the agents. The proposed method is an intrinsic formation control that does not need to involve any information of the desired formation before-hand. Another virtue of the method proposed is that only relative attitude measurement is required.Paper B further concerns the formation control of all regular polyhedral configurations (also called Platonic solids) for reduced attitudes. According to the symmetries possessed by regular polyhedra, a unified framework is proposed for their formations. Via using the coordinates transformation previously proposed, it is shown that the stability of the desired formations can be provided by stabilizing a constrained nonlinear system. Then, a methodology to investigate the stability of this type of constrained systems is also presented. Paper C considers the problem of tracking and encircling a moving target by agents in 3-dimensional space. By this work, we show that similar design techniques proposed for reduced attitudes formations can also be applied to the formation control for point mass systems. Therein, a group of agents are driven to some desired formation on a spherical surface and simultaneously keep the center of this spherical formation coinciding with the target to be tracked. By properly designing communication topology, the agents constitute a cyclic formation along the equator of an encircling sphere. / <p>QC 20170302</p>