Multi-agent-based optimal power scheduling of shipboard power systems

Abstract

This paper presents a multi-agent-based algorithm for optimal power scheduling of multi-agent-based medium voltage direct current shipboard power systems of all-electric ships. shipboard power system is the crucial part of every all-electric ship due to its vital duty in generating electrical energy for various parts of the ship, particularly in long-distance travels. Moreover, the shipboard power system should be efficiently operated to reduce operating costs and energy to serve heavy loads for long periods of time. Thus, a system with resilient, secure, and efficient performance is needed to operate this essential power system. Traditionally, shipboard power systems were operated using centralized systems that were threatened by single-point failures, security issues, and central communication and computation burden. Considering the existing concerns about operating centralized systems, a multi-agent-based operation system could be best fitted for this goal. To this aim, this paper introduces a new multi-agent-based operation system based on the primal-dual method of multipliers, in which the generator stations are operated by the communication of agents with each other and without any central commander. The presence of energy storage systems is considered to enhance the reliability of the proposed power system. Furthermore, single- and twin-shaft turbogenerators are assumed to construct the main power generators of the shipboard power system as reliable, efficient, and maneuverable types of generators. The local speed and efficiency of the individual turbogenerators are also calculated using the proposed multi-agent-based method. The proposed method is simulated on a five-zone shipboard power test system in the presence of an energy storage systems and five turbogenerators, and is compared with a precise centralized scheme and a well-known multi-agent-based method during a 24-h time period. The simulation results demonstrate the accuracy of the proposed method with the total 24-h scheduling cost error of 0.2% in comparison with the optimal results gained by the centralized method. Besides, the proposed method decreases the error in calculating the total 24-h scheduling cost by almost 71% in fewer required iterations compared to the other well-known multi-agent-based scheme.