CN-121974208-A - Group control elevator intelligent dispatching energy-saving method based on AI learning

CN121974208ACN 121974208 ACN121974208 ACN 121974208ACN-121974208-A

Abstract

The application discloses an intelligent dispatching energy-saving method of group control elevators based on AI learning, which relates to the technical field of intelligent dispatching of elevators, and the method comprises the steps of sequentially executing the steps of obtaining running state data of each elevator to form an interval passing track string, segmenting continuous intervals, generating interval relay records, constructing a concessional section table, displaying a floor section still provided with transfer conditions at the current moment, and avoiding blind probing; A freezing window is generated for each elevator, a near-end stopping node of the elevator is locked to be a reserved node, the elevator is ensured to respond to registered and upcoming service preferentially in the current decision period, and the problems that the near-end registered stopping task is easily repeatedly changed by other elevators in the group in the traditional group control scheduling, so that the running track of the elevator is frequently adjusted, the elevator taking experience is reduced, and the ineffective running times of the elevator are increased are fundamentally solved.

Inventors

YANG ZHITAO
BAI HONGZHE
WANG WEI

Assignees

奥远梯米数字科技(大连)有限公司

Dates

Publication Date: 20260505
Application Date: 20260330

Claims (10)

1. An intelligent dispatching energy-saving method for group control elevators based on AI learning is characterized by comprising the following steps: s1, acquiring running state data of each elevator to form an interval passing track string; S2, dividing the passage track string of the layer section into a plurality of continuous layer sections, generating a served layer section corresponding to the elevator according to the continuous layer sections, matching the served layer sections of the elevators in the elevator group, identifying a release layer section, generating a receiving layer section according to the release layer section, and generating an layer section relay record according to the release layer section and the receiving layer section; S3, acquiring relay records of all layers to generate a concessible section table; s4, acquiring a decision period and a current decision period in the elevator group, and generating a corresponding freezing window for each elevator in the current decision period; s5, generating relay candidate sequences of the elevator according to the concessional section table and the freezing window; S6, executing a first-stage confirmation process on the relay candidate sequences of the elevators, wherein the first-stage confirmation process is used for locking reserved nodes of the elevators in a freezing window and forming a freezing confirmation result shared in a group; S7, executing second-stage relay processing on the relay candidate sequences of the elevators according to the freezing confirmation result, wherein the second-stage relay processing is used for reassigning the stop floors marked as transferable nodes in the elevator group to form relay submitting results corresponding to the elevators; s8, extracting the next stopping floor of each elevator from the relay submitting results as a guiding result according to the relay submitting results corresponding to each elevator.
2. The intelligent scheduling energy-saving method for group control elevators based on AI learning as claimed in claim 1, wherein in S1, the running state data of each elevator is obtained to form a layer section passing track string as follows: acquiring running state data of each elevator in the same elevator group, wherein the running state data comprise historical floor arrival records, historical door zone state records, historical load change records, current floor position data, current running direction data and current registered stop records; The currently registered stop records include stop floors; And reorganizing the historical floor arrival record, the historical door zone state record, the historical load change record, the current floor position data, the current running direction data and the current registered stop record according to a unified time sequence to generate a layer section passing track string respectively corresponding to each elevator.
3. The intelligent scheduling energy-saving method for group control elevators based on AI learning as claimed in claim 2, wherein in S2, the passage track string of the layer section is divided into a plurality of continuous layer sections as follows: traversing the layer section passing track strings of each elevator; The interval passing track string consists of a plurality of records arranged according to a unified time sequence; The records comprise a history floor arrival record, a history door zone state record, a history load change record, current floor position data, current running direction data and a current registered stop record; taking the historical floor arrival record and the historical gate area state record as segmentation nodes, starting from the record with the earliest time sequence in the passage track string of the layer section, and traversing each record in sequence along the time increment direction; When traversing to a history floor arrival record or history door zone state record, forming a continuous record subsequence by all records which are not included in any subsequence from the last segmentation node to the current history floor arrival record or history door zone state record; the continuous recording sub-sequence does not comprise the last segmentation node, comprises the current historical floor arrival record or the historical door zone state record, and is used as a continuous layer section.
4. The intelligent scheduling energy-saving method for group control elevators based on AI learning according to claim 3, wherein in S2, the served layer segments of the corresponding elevators are generated according to the continuous layer segments as follows: Acquiring an elevator identifier corresponding to the passage track string of the layer section as an elevator identifier; Acquiring a floor corresponding to a first record in a continuous layer as a layer starting floor; Acquiring a floor corresponding to the last record in the continuous layer as a layer ending floor; acquiring current running direction data in a continuous layer section as an layer section running direction, and if the continuous layer section comprises a plurality of pieces of current running direction data, taking a direction indicated by the piece of current running direction data with the earliest time sequence as the layer section running direction; acquiring the time of a first record in a continuous interval as the time of entering the interval; Acquiring the last recorded time in the continuous interval as the interval leaving time; elevator marks, continuous floors, floor starting floors, floor ending floors the interval travel direction, the entering interval time, the exiting interval time form a served interval.
5. The intelligent scheduling energy-saving method for group control elevators based on AI learning as set forth in claim 4, wherein in S2, matching is performed between served floors of each elevator in the elevator group, a release floor is identified, a receiving floor is generated according to the release floor, and a floor relay record is generated according to the release floor and the receiving floor, as follows: matching between served floors of elevators in an elevator group, identifying a release floor as follows: taking any elevator in the elevator group and marking the elevator as an elevator member in the elevator group; Taking any one served layer section of the elevator member and marking the served layer section as a layer section to be detected; Taking any one elevator except the elevator members in the elevator group and recording the elevator as another elevator member; taking all served floors of another elevator member to form a matched object set; comparing the to-be-detected interval with each served interval in the matched object set one by one, and marking the to-be-detected interval as a release interval when the following conditions are simultaneously satisfied: the condition A is that no history floor arrival record exists in a continuous layer section corresponding to the layer section to be detected; The condition B is that at least one served layer section exists in the matched object set, a layer section formed by the layer section starting floor to the layer section ending floor of the served layer section and a layer section formed by the layer section starting floor to the layer section ending floor of the layer section to be detected are intersected, the intersection is marked as an overlapped layer section, and the served layer section is used as a candidate receiving layer section; The condition C is that the time of leaving the interval to be detected is earlier than the time of entering the candidate bearing interval; The layer section running direction of the layer section to be detected is the same as the layer section running direction of the candidate bearing layer section in the intersection of the floor sections; When the to-be-detected interval is marked as a release interval, recording the candidate bearing interval matched with the release interval as a to-be-confirmed bearing interval; and marking the to-be-confirmed accepting layer section as an accepting layer section when the following conditions are simultaneously met: Acquiring a decision period in an elevator group, wherein the time difference between the time of entering an accepting layer section to be confirmed and the time of leaving the associated releasing layer section is smaller than or equal to one decision period; the condition F is that a history floor arrival record exists in a continuous layer section corresponding to the receiving layer section to be confirmed; In the continuous layer section corresponding to the bearing layer section to be confirmed, historical load change records exist before and after the historical floor reaches the record, and the load capacity value recorded by the historical load change record after the historical floor reaches the record is larger than the load capacity value recorded by the historical load change record before the historical floor reaches the record; associating a release interval and a receiving interval matched with each other into an associated interval relay; acquiring a floor interval intersection of the release layer section and the corresponding bearing layer section as a relay floor interval; Taking an elevator identifier of the release layer section as a release elevator identifier; Taking an elevator identifier of the bearing layer section as a bearing elevator identifier; taking the leaving interval time of the release interval as the release time; taking the time of entering the bearing layer section as the bearing time; And forming a layer section relay record by the associated layer section relay, the relay floor section, the elevator identifier release, the elevator identifier bearing, the elevator identifier release time and the elevator bearing time.
6. The intelligent scheduling energy-saving method for group control elevators based on AI learning as claimed in claim 1, wherein in S3, all layer segment relay records are acquired to generate a concessible section table as follows: acquiring relay records of all the layers, and extracting a relay floor interval from each layer relay record; removing the weight of all relay floor intervals to form a historical relay floor interval set; according to the current registered stop records of each elevator, obtaining all floors which are currently registered by the elevators to stop, and forming an occupied floor set; And for each relay floor interval in the history relay floor interval set, eliminating floors belonging to the occupied floor set in the relay floor interval, if the floor segments which are continuously arranged still exist in the rest floors in the relay floor interval after eliminating, taking the continuously arranged floor segments as the resolvable sections, and summarizing all the resolvable sections to form a resolvable section table.
7. The intelligent scheduling energy-saving method for group control elevators based on AI learning according to claim 2, wherein in S4, a corresponding freezing window is generated for each elevator in the current decision period as follows: In the current decision period, for each elevator, according to the current floor position data, the current running direction data and the current registered stop records; Taking the floor indicated by the current floor position data as a starting point, selecting a plurality of stop floors which are arranged continuously first from the current registered stop records along the direction indicated by the current running direction data to form a stop floor interval, and taking the stop floor interval as a freezing window of the elevator in the current decision period; The stopping floors in the freezing window must not be accepted or overwritten by other elevators in the elevator group.
8. The intelligent scheduling energy-saving method of group control elevator based on AI learning of claim 2, wherein in S5, a relay candidate sequence of the elevator is generated according to a concessional section table and a freezing window, as follows: Generating a mark as a reserved node for the stop floor in the freezing window according to the stop floor in the current registered stop record, generating a mark as a transferable node for the stop floor which is positioned outside the freezing window and falls into the concessible section table, and enabling the stop floor marked as the transferable node to be accepted or rewritten by other elevators in the elevator group; combining and arranging the stop floors marked as reserved nodes and the stop floors marked as transferable nodes according to the original sequence in the current registered stop records to form a relay candidate sequence of the elevator.
9. The intelligent scheduling energy-saving method for group control elevators based on AI learning according to claim 1, wherein in S6, a first-stage confirmation process is performed on the relay candidate sequences of each elevator, the first-stage confirmation process is used for locking the reserved nodes of each elevator in the freezing window, and forming a freezing confirmation result shared in the group, as follows: Taking the floor indicated by the current floor position data of each elevator as a starting point, and taking the front stop floor in the current registered stop records as a target floor along the direction indicated by the current running direction data of each elevator; Acquiring elevator operation parameters; According to the elevator operation parameters, the estimated time consumption of the elevator from the starting point to the target floor door zone is obtained; sequencing all elevators in an elevator group from small to large according to predicted time consumption to obtain an elevator door zone sequence, and taking the elevator door zone sequence as a door zone arrival sequence; The first-stage confirmation process is performed on the relay candidate sequences of the elevators as follows: sequentially confirming reserved nodes in each elevator freezing window according to the ordering sequence in the arrival sequence of the door zone to obtain confirmed reserved nodes; The confirmed reservation nodes are formed into a freeze confirmation result shared in the group, and the freeze confirmation result makes the follow-up elevator unable to make a carrying or rewriting request to the confirmed reservation nodes again.
10. The intelligent scheduling energy-saving method for group control elevators based on AI learning according to claim 9, wherein in S7, according to the freeze confirmation result, a second-stage relay process is performed on the relay candidate sequences of the elevators, and the second-stage relay process is used for reassigning the stop floors marked as transferable nodes in the elevator group to form the relay submission results corresponding to the elevators as follows: executing second-stage relay treatment on the relay candidate sequences of the elevators according to the freezing confirmation result, wherein the second-stage relay treatment is used for redistributing stop floors marked as transferable nodes in the elevator group to form relay submitting results corresponding to the elevators; The allocation condition is that each stop floor marked as a transferable node is sequentially processed according to the ordering sequence in the arrival sequence of the gate area, and the stop floor marked as the transferable node is used as a target floor; an elevator satisfying all of the following conditions simultaneously is selected as a target elevator from an elevator group: the running direction in the current running direction data of the elevator is the same as the running direction in the current running direction data corresponding to the target floor under the condition ①; Conditional ②, after adding the destination floor to the relay candidate sequence of the elevator, the relay candidate sequence keeps monotonous and no turn-back along the running direction in the current running direction data of the elevator; The condition ③ the target floor is not occupied by any confirmed reserved nodes in the freeze confirmation result; if a plurality of elevators meeting all the conditions exist, taking the elevator with the first sequence as a target elevator according to the sequence of the door zone in the arrival sequence; removing the target floor from the relay candidate sequence of the original elevator and writing the target floor into the relay candidate sequence of the target elevator; After all the target floors are processed, a final relay candidate sequence of each elevator is formed and is used as a relay submitting result of each elevator.

Description

Group control elevator intelligent dispatching energy-saving method based on AI learning Technical Field The invention relates to the technical field of intelligent elevator dispatching, in particular to an intelligent group control elevator dispatching energy-saving method based on AI learning. Background Along with popularization of high-rise buildings and improvement of intelligent level of elevators, dispatching optimization of group control elevators becomes a key link for improving elevator operation efficiency and reducing energy consumption, and currently, the mainstream intelligent dispatching technology of group control elevators in industry mainly comprises the steps of directly outputting dispatching results around a learning model, pre-stopping at hot floors, expanding fuzzy logic/particle swarm/reinforcement learning optimization and other technical routes, wherein the technical routes are all realized by adopting electronic data processing modes of elevator operation data acquisition, model calculation and decision output. The elevator stopping floor is determined by means of learning model prediction or various algorithm optimization, model training and parameter tuning are needed to be conducted based on a large amount of data, scheduling deviation is easy to occur due to insufficient scene suitability and the scheduling result is disjointed with the actual running requirement of the elevator. When the existing group control elevator dispatching technology processes the near-end stopping tasks of the elevators, as real-time dynamic global optimal solution calculation is carried out in a continuous decision period through modes of learning model prediction, optimization of various algorithms and the like, all elevators in an elevator group can participate in the fight and reassignment of all stopping tasks including the near end, and a dispatching system can continuously update global calculation results along with real-time changes of operation parameters and registration tasks, so that the registered near-end stopping tasks of the elevators are always in a state capable of being changed at any time, further, the registered near-end stopping tasks are easily repeatedly changed by other elevators in the group, the operation track of the elevators is frequently adjusted, the riding experience is reduced, and the ineffective operation times of the elevators are increased. Disclosure of Invention In order to solve the technical problems in the background technology, the invention provides an intelligent dispatching energy-saving method for a group control elevator based on AI learning. The invention provides an intelligent dispatching energy-saving method for a group control elevator based on AI learning, which comprises the following steps: s1, acquiring running state data of each elevator to form an interval passing track string; s2, cutting the interval passing track string into a plurality of continuous intervals; generating a served layer section corresponding to the elevator according to the continuous layer section; matching the served floors of each elevator in the elevator group, and identifying a release floor; Generating a receiving layer section according to the release layer section; Generating an interval relay record according to the release interval and the bearing interval; S3, acquiring relay records of all layers to generate a concessible section table; S4, acquiring a decision period and a current decision period in the elevator group; Generating a corresponding freezing window for each elevator in the current decision period; s5, generating relay candidate sequences of the elevator according to the concessional section table and the freezing window; S6, executing a first-stage confirmation process on the relay candidate sequences of the elevators, wherein the first-stage confirmation process is used for locking reserved nodes of the elevators in a freezing window and forming a freezing confirmation result shared in a group; S7, executing second-stage relay processing on the relay candidate sequences of the elevators according to the freezing confirmation result, wherein the second-stage relay processing is used for reassigning the stop floors marked as transferable nodes in the elevator group to form relay submitting results corresponding to the elevators; s8, extracting the next stopping floor of each elevator from the relay submitting results as a guiding result according to the relay submitting results corresponding to each elevator. Preferably, in S1, running state data of each elevator is acquired to form an interval traffic track string, as follows: acquiring running state data of each elevator in the same elevator group, wherein the running state data comprise historical floor arrival records, historical door zone state records, historical load change records, current floor position data, current running direction data and current registered stop records; By way of i