A Research On The Utilization Of Multi-Agent System In Micro-Grid

Micro-Grid integrates renewable energy,storage devices and controllable generations,it provides efficent utilization of clean energy while keeps stable external characteristics.Capability of continuous power supply,high scalability and flexible operation modes can satifiy the current demand of joint operation between renewable energy and large grid,and will provide a solid foundation for smart grid technology in the future.Thus,MicroGrid is an excellent integration of renewable energy with a bright future;Multi-Agent System is a new collection of intelligent agents.Multi-Agent System can completely cover all the devices within a Micro-Grid in topology,its flexible operating modes meet the various needs of Micro-Grid,and the capability of distributed computing supports intelligent functions for Micro-Grid in the future.Therefore,developing premium functions for Multi-Agent System utilized as functional platform in Micro-Grid will promote the development of both Micro-Grid and Smart Grid technologies.Basing on actual operations of Micro-Grid,this thesis focuses on development and performance improvement of Multi-Agent System functions for basic and advanced demands of Micro-Grid.Main contributions of this thesis include:(1)A complete function developing and emulating system for Multi-Agent platform is designed with three parts: modules of basic functions for agents,debug tool collection and automatic data processing fuctions.This system provides convenient ways for maintaining and expanding basic functions for agents,facilitates development and debugging of compound fuctions for Multi-Agent System and enables automatic data processing along with repeated experimenting capability.The system laies a foundation for implementation of the methods proposed in following chapters.(2)A real-time physical quantitiy tracking method within a Micro-Grid for MultiAgent System is proposed.This method is designed basing on Taylor Series using current sampling techonology.Device Agents upload changes of variation rate of physical quantity,Monitor Agent emulates real-time variation of physical quantity basing on the uploaded information.This process reduces the number of messages transmitted;Meanwhile,Device Agents and Monitor Agent correct tracking errors synchronously to achieve an accurate physical quantity tracking.Emulation results show that the real-time performance of the proposed method is significantly higher than the method for comparison which uploads the variation of physical quantity directly.This tracking method is a general tool which provides real-time information about Micro-Grid status and helps improve performance for all kinds of control using Multi-Agent System.(3)A new method of active power coordination within a Micro-Grid using MultiAgent System is proposed to minimize the damage Micro-Grid suffers.When there is an imbalance of active power,Multi-Agent System orders storage devices to get involved immediately to reduce the imbalance;Meanwhile,all the controllable generations within the Micro-Grid increase or decrease outputs at their maximal speeds as soon as they receive the orders.After an active power balance appears,storage agent tracks the variation of total output of all controllable generations and makes output power of storage device be replaced by controllabe generations while keeps the balance all the time.Finally,storage devices quit their operations seamlessly and the Micro-Grid reaches a stable status.Emulation results indicate that this coordinating method outperforms the traditional method and siginificantly reduces the damage a Micro-Grid suffers during the coordination.(4)A cost rate(cost generated during unit time)optimization method for a Micro-Grid basing on hierarchical processing capability of Multi-Agent System is introduced with fast calculating speed,high result quality and compatibility to all kinds of cost rate functions.This method contains multiple rounds of calculations.In each round,a controllable generation agent searches different intervals to find out output variation with the optimal average cost rate locally;Then Information Collecting Agent chooses output variation with globally optimal cost rate among those local optimals sent by controllable generation agents.The aforementiond process stops its repetitions when the Micro-Grid reaches a balanced status.Then Information Collecting Agent performs a fine-tuning to adjust output of each controllable generation while keeps the overall output balance and finally leads the Micro-Grid to the minimal cost rate status.Emulation results show that this method outperforms Genetic Algorithms and Interior Point Method in calculation speed,result quality and compatibility.It provides a practical solution to frequent economic dispatch request within a Micro-Grid.(5)A fast optimal Nash Equilibrium selecting method within an open-market MicroGrid using Multi-Agent System is introduced.It helps Micro-Grid provide stable service with high quality shortly after a fluctuation of the market environment.In this method,a distributed Newton Iteration method is developed to solve the equation set of partial derivatives of profit functions of all controllable generations;Then a dimensionality reduction directional search of Nash Equilibrium focusing on singular points with random process is applied to achieve the optimal selection.Emulation results indicate that the calculation speed of this mehod is fast.The tendency of retail price and power quantity in different emulation scenes is in accordance to the feature of an open market.This method provides a practical solution to optimal Nash Equilibrium calculation in open electricity market within a Micro-Grid.