دانلود رایگان ترجمه مقاله معماری کنترل توزیع شده بر اساس چند عامل برای مدیریت انرژی و بهینه سازی میکروگرید – الزویر 2016

معماری کنترل توزیع شده مبتنی بر چند عامل برای مدیریت و بهینه سازی انرژی میکروگرید

Multi-agent based distributed control architecture for microgrid energy management and optimization

Abstract

Most energy management systems are based on a centralized controller that is difficult to satisfy criteria such as fault tolerance and adaptability. Therefore, a new multi-agent based distributed energy management system architecture is proposed in this paper. The distributed generation system is composed of several distributed energy resources and a group of loads. A multi-agent system based decentralized control architecture was developed in order to provide control for the complex energy management of the distributed generation system. Then, non-cooperative game theory was used for the multi-agent coordination in the system. The distributed generation system was assessed by simulation under renewable resource fluctuations, seasonal load demand and grid disturbances. The simulation results show that the implementation of the new energy management system proved to provide more robust and high performance controls than conventional centralized energy management systems.

1 Introduction

Widespread integration of distributed multi-source generators pose a challenge to the present electrical grid system. More integration of renewable energy (RE) generation systems into the present distribution networks adds new dynamic elements due to the intermittencies and inherent unpredictability of the renewable energy system (RES) operations. Therefore, in order to improve the present performance of such systems, it is crucial for the energy management system (EMS) to have an effective and optimal control strategy. The EMS should primarily be able to provide balance between the electricity supply and load demand. Additionally, the system must also comply with other requirements such as reliability, flexibility, fault tolerance and operating costs reduction. Typically, most energy management systems are based on centralized controllers. For instance, a generic centralized EMS utilized for managing power converters in a microgrid that consists of wind and Photovoltaic (PV) systems as described in [1]. In [2], the centralized EMS was used to coordinate the micro-generators with the main grid for minimizing the greenhouse gases (GHG) emissions, energy costs, and maximizing the power output of renewable energy systems. Furthermore, in [3], a microgrid central controller was used to optimize supply and demand profiles for mitigating fuel consumption costs. Conventional EMS architecture is summarized in Fig. 1. The central supervisory controller is used to optimize the usage of fossil fuel based distributed energy resources (DERs), renewable DERs, and energy storage in the microgrid. It commonly consists of a communication network that monitors the DERs and also sends commands to local controllers in order for the dispatchable resources to deliver power to the load in the most promising economical method. Despite its universal successes, this system’s topdown approach has several drawbacks. It represents a ‘single point of failure’, which means that it has to be securely planned with proper redundancy built in [1,4]. In addition, the complexity of the centralized energy management system grows exponentially with the growing number of generators and loads causing the higher cost of communication for scheduling and online monitoring [5,6]. Moreover, the centralized controller needs to be updated and reconfigured for any changes in the microgrid structure or when new generators or loads are installed [6]. Although centralized control methods may be used to find the best control solutions, it require powerful computing ability in order to deal with a huge amount of data as the systems become bigger and more complex [6]. It also needs a network with a highly distributed control strategy and communication capabilities [7]. On the other hand, decentralized control based on a bottom-up approach for energy management systems is more robust and less complex than centralized management [8,9]. The distributed management components are viewed as a unit with some intelligence that enables them to provide basic computation, planning action and decision-making. It not only minimizes the communication and computation capability; but also completely respects other various parts requirements and operational performance [7]. Therefore, it is expected that an intelligent, dynamic and open system that is self-adaptive is essential in order for the hybrid RE system to function effectively under various conditions to meet dynamic load variations and renewable sources intermittency. This paper focuses on multi-agent system (MAS) based EMS architecture for optimizing hybrid RE system performance. Extensive research has been conducted on energy management for micro-grids and distributed generation. Many researchers have implemented a hierarchical control structure for controlling microgrid and distributed generations. In [11], a hierarchical EMS was used to control a microgrid that includes three control layers such as supervisory, optimizing and execution. In addition, a hierarchical EMS comprised of master and slave control strategy was used in [12] to control a microgrid composed of PV, wind, hydrogen and battery system. There were also several energy management methods utilizing soft-computing approaches to control the microgrid such as genetic algorithm, fuzzy logic, particle swarm optimization, and neural networks. For instance, a genetic algorithm based EMS employed to manage and optimize the generation dispatch of a microgrid with multiple generators was presented in [13]. Meanwhile in [14], fuzzy logic EMS was used to optimize the operation of microgrid components and sizes. In [15], particle swarm optimization was used in the EMS to optimize the power output between the distributed generators aiming to improve the power quality in the microgrid. Moreover, in [16], a neural network based EMS applied in the PV microgrid to optimally manages the system’s operation by adapting to the input variable such as PV output power and load demand. MASs have been widely studied in the field of computer science [17]. However, in recent years, the development of the MAS has gained attention from power system researchers for application in the field of hybrid energy systems and microgrids for distributed control and energy management [7,18]. A multi-agent system for optimizing the hybrid RE system was presented in [7]. Meanwhile in [18], a distributed management solution based on MAS was proposed to provide better system reliability than conventional centralized EMS. In [9], a MAS based hierarchical decentralized coordinated control was presented to solve the energy management issue of a distributed generation system (DGS) by ensuring energy supply with high security. A MAS and fuzzy cognitive map were used in [8] for a decentralized energy management system of an autonomous poly-generation microgrid. In [19], a decentralized MAS was used for demand side integration that was able to reduce the energy cost, improve energy efficiency and increase security and quality of supply. Furthermore, MAS has also been used for reactive power management in distribution networks with renewable energy sources to enhance the dynamic voltage stability [20]. All the mentioned researchers concluded that the MAS-based decentralized energy management control structure is capable of handling complex DGS or hybrid energy systems more effectively, since it can deliver several key advantages to the system [8,9,18,21]. Firstly, one of the major advantages is the high reliability and robustness of the system. For instance, if one of the controllers fails, the rest of the system can still operate in part, and it does not affect the entire system’s performance. As such, the managed microgrid has a higher likelihood of partial operation in cases when malfunctions take place in different parts of the system. Moreover, the MAS DGS also offers flexibility since the different levels of agents not only can identify and respond quickly to the environmental variations, but also depend on each other to regulate the operational status in reaction to the changes. The MAS based distributed scheme is also more feasible to handle than a centralized scheme. The agents are not only able to strategize their own asynchronous decision-making simultaneously, but can also attain the goal of the whole system in a cooperative way. The MAS can also minimize the communication and computation burden since the dynamic information, basic computation, action-planning and decision making can be processed by individual agents locally. Finally, the openness and adaptability of the MAS-based energy management system allows the integration of new DER units or loads without reconfiguring the entire system. Therefore, in this paper, a new MAS based decentralized control architecture was implemented for a microgrid in order to handle the complex energy management issue of the DGS. The remaining sections of this paper are organized as follows: In Section 2, the MAS concept such as the definition, agent’s characteristics and the MAS contributions for the EMS are presented. Moreover, details of the distributed generation model that comprises of components such as diesel generator, PV system, wind system, micro-hydropower system, battery storage system, and loads are included in Section 3. The utilization of MAS in the microgrid with agent model for each component, MAS architecture and global objective function are shown in Section 4. The game theory implementation for multi-agent coordination in the microgrid is described in Section 5. Meanwhile, the simulation and results of this paper that includes the case study, performance evaluations through different scenarios, and comparison with a centralized system are discussed in Section 6. Finally, Section 7 presents the conclusions of this paper.