شناسايي توزيع‌شده‌ پارامتري سيستم‌هاي چندعاملي مبتني بر اجماع حداكثري

چكيده انگليسي :

Parametric Distributed Identification of Multi Agent Systems Based on Max Consensus Amirhosein Golfar a golfar@ec iut ac ir September 2019 Department of Electrical and Computer Engineering Isfahan University of Technology Isfahan 84156 83111 IranSupervisor Dr Jafar Ghaisari ghaisari@cc iut ac irAdvisor Dr Iman Izadi iman izadi@cc iut ac ir Advisor Dr Mehdi Mahdavi m mahdavi@cc iut ac ir Department Graduate Program Coordinator Dr Gholamreza Yousefi Department of Electrical and Computer Engineering Isfahan University of Technology Isfahan 84156 83111 IranAbstractWith the advent of model based control in vehicle systems power networks systems biology and transportation systems employing the modelling of multi agent systems MASs has been increased In MASs with wireless communication networksand dynamical dependency among agents employing the classical identification methods are not applicable due to distributeddeployment of agents inaccessibly of measurements and communication topology In this thesis a parametric distributedidentification of multi agent systems based on max consensus algorithm MCA is proposed The provided method is consistentwith the constraints of MASs in which agent may find the dynamic of other agents in online manner by employing localinformation of its neighbors through non ideal communications networks It is assumed that the wireless communicationnetworks among agents have Bernoulli dropouts It is proved that the sufficient condition for the convergence of suggestedmethod corresponds to the dynamical characteristic of identified agent and the convergence time of MCA It is stated that inthe presence of communication networks dropouts with Bernoulli distribution the MCA converges with probability one in thefinite time Furthermore the upper bound for the convergence time of MCA is given by means of probabilistic expressions By employing the suggested method the solution is provided for the power tracing problem in microgrids as a case study Key WordsParametric Distributed Identification Multi Agent Systems Online Identification Non Ideal Communication Networks Max Consensus Algorithm Microgrid IntroductionThe Multi Agent Systems MASs play an increasingly important role in many fields of en gineering and science such as power systems Catterson et al 2012 advanced transportation Bosankic et al 2015 flying formation Zhao et al 2015 systems biology Hasenauer et al 2010 and economics Deng et al 2015 Since agents of MASs are distributed spatially andcommunicate with each other through wireless communication networks developing effectivemodel based control methods in MASs e g adaptive or robust methods requires identifica tion algorithms that deal with uncertainties induced by communication networks distributeddeployment of agents and inaccessibly of measurements Furthermore in MASs that modelsof agents are time variable the identification algorithms should be executed in online mannerto update the models of agents in time steps Although the control synthesis of MASs has attracted attention in recent literature Huanget al 2016 Li and Li 2015 Liu et al 2015 Ma et al 2015 Mahmoud and Khan 2015 Wenet al 2016 the identification and model verification of these systems on the basis of mea surements have not been explored in much detail yet Some papers in the field of identificationof MASs e g Massioni and Verhaegen 2009 2008 study the offline identification meth ods of MASs In the offline identification the measurements of agents are gathered for sometime steps and then the identification method applies to the batch of measurements Therefore the mentioned papers are not applicable in the case of time variable models of agents and do

گلفر، اميرحسين

شناسايي توزيع‌شده‌ پارامتري سيستم‌هاي چندعاملي مبتني بر اجماع حداكثري

شناسايي توزيع‌شده و برخط , سيستم چندعاملي , شبكه‌هاي ارتباطي غيرايده‌آل , الگوريتم اجماع حداكثري , ريزشبكه قدرت