The invention provides an
elevator comprehensive
energy storage device
capacity optimization configuration method based on element
reinforcement learning, and belongs to the field of
elevator energy-saving control. The method comprises the following steps that
elevator operation data, environment data and
energy storage device states are collected in real time, an
elevator control element
reinforcement learning model is built, and in the element
reinforcement learning model training stage, the
elevator control element reinforcement learning model is built; a strategy is pre-trained in a historical task through an
algorithm, a new scene is rapidly adapted, the strategy is updated based on real-
time data, future load is predicted in combination with LSTM, the
energy storage capacity is dynamically adjusted, the charging and discharging threshold value and the capacity distribution proportion of the energy storage device are adjusted according to an optimization result, and a control instruction is output to an
energy management system. Cross-scene strategy migration is achieved through an MAML
algorithm, the defect that a traditional method needs to be redesigned in a new scene is overcome, element strategy learning and online fine adjustment are separated, global generalization and local optimization are considered,
energy consumption, cost and equipment service life are balanced, a
variable capacity distribution proportion is introduced instead of a fixed
capacity value, and the
utilization rate of an energy storage device is increased.