Genetic Algorithm Design of an Adaptive, Multirate LQR Controller for a Multi-Machine MVDC Shipboard Electric Distribution System with Constant Power Loads

Abstract

Over the last two decades, the US Navy has invested significantly in developing Integrated Power Systems (IPS) on warships. Future warship classes equipped with IPS are expected to have Medium Voltage DC distribution. DC distribution systems are host to scores of power converters which provide tailored voltage, frequency, and power quality to connected loads. When coupled to high-bandwidth controllers, power converters appear as constant power loads to the distribution system. The negative non-linear impedance associated with constant power loads reduces stability margins and limits the efficacy of linear control methods. The inclusion of megawatt level pulsed loads, such as laser weapons or railguns, further exacerbates the challenge to designers. A recently introduced control scheme, Adaptive Select-Matrix LQR (LQR-SM), is a flexible and adaptable centralized control approach to multi-input, multi-rate, high order systems. This paper presents the design and implementation of the LQR-SM control scheme in a complex hypothetical naval zonal MVDC distribution system that includes: multiple power generating units, multiple active energy storage devices, multiple load zones, and pulsed loads.