Distributed Model Predictive Emergency Control For Long-term Voltage Stability Based On Multi-agent Technology

Due to the use of centralized modeling and centralized solving method, thecomputational cost of the centralized model predictive control receding optimization problemis expensive, whereas the control effect of the decentralized model predictive control isunsatisfactory because of neglecting mutual influence among each subsystems. Theapplication of these two model predictive control method in large-scale interconnectedsystems are restricted, therefore, this paper focuses on the distributed model predictive controlmethod and applies this method to design emergency controller for long-term voltagestability.Firstly, a brief introduction to the basic principles and processes of the power systemquasi-steady-state simulation is presented. The components and their quasi-steady-state modelthat describe the dynamic process of long-term voltage stability are also given in this part.Secondly, we discuss how to change the centralized model predictive control model intothe distributed model predictive control model. The emphasis is how to convert onlinereceding optimization problem in large-scale systems into an online collaborativeoptimization problem in several interconnected subsystems. During optimization process, theconvex combination of objective functions from each subsystem is constructed into a globalobjective function, and this optimization problem is solved by iteration and networkcommunication. Thus the convergence to obtain the Nash equilibrium point is enhanced.Furthermore, a multi-agent system is established to implement parallel computing of thereceding optimization problem and to cope with the communication problem among thesesubsystems by MATLAB and JADE according to the commonality between multi-agenttechnology and distributed model predictive control algorithm.Finally, the effectiveness of the proposed distributed model predictive control algorithmand multi-agent system architecture are verified through simulations on New England10machine39-bus system and IEEE145bus system.