Convergence Analysis And Optimization For Distributed Control Of Virtual Power Plant

The combination of virtual power plant (VPP) technology and distributed control brings about an effective approach to manage a large amount of small scale distributed generators (DGs). It takes several iterations to attain the control goal, so convergence speed is of concern. Consensus algorithm is a distributed algorithm with extensive application in the fields of control and computer science, and can be applied to active/reactive power control.Firstly, VPP technology is combined with discrete first-order consensus algorithm to control DGs in a distributed fashion, which can dispatch active power among DGs economically.Secondly, In order to reduce the iterations, spectral radius measurement is used to analysis the convergence of consensus algorithms with and without time-delay, which reveals that the convergence speed is affected by topology and edge weights of communication network (CN), feedback coefficient, position of leader DG, and cost functions of DGs. Convergence speed measurement for discrete first-order consensus algorithms with and without time-delay is proposed.thirdly, convergence speed optimization is transformed into eigenvalue optimization. Communication network optimization which aims at accelerate the distributed control of VPP is proposed, including communication network topology designing, edge weights optimization, and leader selecting. The optimization method is proved to be effective through simulation cases.