A multi-agent system design and implementation for flexible network management

  • Minjiang Chen

Student thesis: Doctoral Thesis

Abstract

With the introduction of renewable energy technologies to reduce greenhouse gas emissions, a significant amount of Distributed Generation (DG) has connected to distribution networks. Hence, the operation of electrical power distribution systems has become complicated and increasingly challenging due to the uncertainty of bi-directional power flow, voltage fluctuations, and frequency deviations. To help manage these issues, the system requires more intelligent functions and flexibility in order to support solutions for network management and operation. Moreover, distribution network operators are looking for an active approach to maximise the utilisation of network capacity while solving network issues for more DG connections. As a result, Active Network Management (ANM) has been proposed to facilitate DG connections without breaching network operation limits. This is achieved by managing network control functions in line with operational objectives and it is often seen as a way to avoiding the high costs of reinforcing the existing infrastructure. However, as the network continues to change, such as increasing DG connections, and control functions keep evolving overtime. ANM is considered to be part of the solution for at least the medium term.ANM is considered to be part of the solution for at least the medium term. Therefore, ANM requires sufficient flexibility to adapt changes to its environment and extensibility to upgrade control functions overtime for future needs. A key aspect of this is interoperability to allow ANM to interact with different control devices, such as intelligent electronic devices, to collect data and realise control purposes as they also evolve.Multi-agent Systems (MAS) is one of the most relevant technologies to address the above challenges as it provides autonomous and proactive behaviour in open and dynamic environments, which is analogous to new DG connections. In addition, MAS offers a flexible and extensible platform that is the advantage of other relevant technologies, such as a service-oriented architecture. Therefore, it is proposed as part of the work of this thesis that the requirement for flexibility and extensibility can be achieved through the development of control functions as intelligent agents with scalable capabilities brought about through the use of MAS technology. The novel solver agent developed as part of the work of this thesis is an essentialcomponent of the MAS architecture considered in this thesis incorporates an integrated control algorithm and negotiation capability to solve conflicts between various control solutions.This thesis presents a fully integrated MAS architecture for ANM. It is developed by following a comprehensive design methodology and each stage of the proposed MAS architecture is detailed through specification to implementation. Selected control algorithms are developed as intelligent agents to achieve multiple ANM solutions. In order to provide a common understanding of terminology for agent communications, ontologies for power system control applications, including thermal overload and voltage violation, have been created. A novel IEC 61850 interface has been developed and embedded inside the agent to address interoperability issue between devices for data collection and control.To evaluate the performance of the developed MAS architecture, along with the novel elements of this thesis a range of simulation case studies are explored. Studies are based on a closed-loop simulation environment with a power system simulator: one case study is based on an operational UK 11 kV radial distribution network to demonstrate the application of MAS for various power system controls; another examines the self-organising ability of the developed MAS architecture by using the ε decomposition algorithm for distributed voltage regulation. The results demonstrate the flexib
Date of Award1 Sep 2015
LanguageEnglish
Awarding Institution
  • University Of Strathclyde
SponsorsEPSRC (Engineering and Physical Sciences Research Council) & University of Strathclyde
SupervisorStephen McArthur (Supervisor) & Ivana Kockar (Supervisor)

Cite this