The invention discloses a large-scale
cruise ship water chilling unit
pipe network
cold storage intelligent energy-saving control method based on double-layer deep
reinforcement learning, which comprises the following steps: 1, carrying out inherent
cold storage engineering modeling on a
chilled water pipe network of a water chilling unit to obtain an
energy balance relationship between equivalent
cold storage capacity and cold release sustainable time of the
pipe network and bearable cold capacity; the relationship is used as a physical prior to introduce strategy learning and a security boundary; 2, constructing a double-layer deep
reinforcement learning architecture; 3, a safety layer is additionally arranged in the double-layer deep
reinforcement learning architecture, the safety layer cuts and projects output feasible regions of a target layer and a sublayer through a hard constraint and barrier mechanism, the
water supply temperature, the
temperature difference, the start-stop interval, the
condensation temperature and the power change rate are ensured, and the maritime affair safety requirement is met; and 4, on the basis of a pipe network inherent cold
storage model, a double-layer deep reinforcement learning framework and a safety layer,
peak load shifting control of low-load cold charging-high-load cold discharging is achieved, and unit cold
energy consumption of the water chilling unit is reduced.