Memetic Algorithm Optimization of Electric Ship Power System

The use of metaheuristic algorithms allows for the optimization of systems that are complex to solve traditionally. With the addition of an accurate digital twin (DT) of the system under study, the metaheuristic algorithm can be used in a simulation-based optimization (SBO) to improve performance. This work presents an optimization engine powered by a memetic algorithm (MA) that aligns the control setpoints and configuration of a dc microgrid subject to a multiobjective cost function. The algorithm generates the fitness of candidate solutions through a co-simulated pair of models of the power system, a discrete events simulation, and an abstracted dynamics model. The developed algorithm, the posture-based prealignment Version 1 (PBPAv1), performs well in simulation studies, outperforming a static or lookup-table-based alignment approach. The algorithm, which is intended for real-time deployment, executes quickly simulating hours of simulated system time in under a minute. The DT models and the algorithm are demonstrated in a microgrid testbed designed to emulate a zonal dc ship electric distribution system. The algorithm performs well at the task of aligning control setpoints and system configuration, and the DT models are validated against the physical hardware.