Technical principle: First, multi-source data fusion, collecting data such as gate passage, elevator call, floor reservation and passenger density; Second, AI algorithm prediction, training models based on historical passenger flow to predict peak hours and floor demand; Third, dynamic grouping dispatch, merging passengers in the same direction to reduce stops; Fourth, multi-elevator collaboration, linkage optimization of multiple elevators to avoid empty load and repeated stops; Fifth, scene adaptation, supporting different scenarios such as office buildings, shopping malls and hospitals, automatically switching scheduling strategies.