Multi-agent system based solution of the optimal reactive power dispatch problem in autonomously operated multi-area systems

Abstract

Cooperation of regional operators in solving optimal reactive power dispatch (ORPD) problems in large systems is beneficial despite of many challenges. One major challenge is the fact that each regional operator is typically not willing to reveal the local system data. To deal with such challenge and to try to attain a dispatch resulting in overall optimum, this paper proposes a control scheme in which a large area is partitioned into multi-agent based system where each operator is considered as a control agent and uses a model of its local system and communication links with its neighboring control agents to come to agreement on the evolution of interconnections and to determine optimal local control actions and states. At each agent, a linearized objective function enforced with constraints has to be solved to determine control variables, which are transformer tap positions and reactive power injections. Calculation procedure at each agent based on sensitivity coefficients, thus impact of change magnitudes of control variables on control performance would be significant and hence is fully analyzed. The proposed algorithm is applied to the modified IEEE 30-bus system and numerical results are presented.Index Terms-Mean-variance mapping optimization, online adaptive control, reactive power management, wind power plant.

Multi-Area and Multi-Period Optimal Reactive Power Dispatch in Electric Power Systems

“…For each of the areas, 24 Ward equivalents related to the 24 periods included in the multi-period analysis were calculated [23]. Communication between areas is performed through a multi-agent approach, which allows communication between each optimizer in each of the areas [42]. Multi-agent master-slave communication was implemented in this work as a system that allows coordination and communication between the different optimization processes involved.…”

“…are the Ward equivalents in two different areas at time t. If these nodes have equal voltage, there is equivalence between the complete system and each of the areas. This measurement is performed after carrying out the optimization of each of the areas [42].…”