A smart pre-scheduling system and method for elevator groups

Abstract

This invention relates to the field of elevator group control technology, and more particularly to an intelligent pre-scheduling system and method for elevator groups. The system comprises: a VLC spatiotemporal sensing network deployed across multiple floors and elevator cars within a building; an edge computing gateway generating intent events representing passengers' waiting intentions; a pre-scheduling decision module generating a pre-scheduling strategy for at least one elevator; a dynamic optical guidance module generating dynamic optical guidance information corresponding to the pre-scheduling strategy; and a group control interface module receiving elevator status information from the elevator group control system in response to the pre-scheduling strategy. This system and method improve elevator operating efficiency and user experience, achieving high-precision and robust spatiotemporal sensing. It also constructs a closed-loop intelligent decision-making and adaptive guidance mechanism, balancing privacy protection and infrastructure reuse, reducing implementation costs, and ensuring high-security collaboration with existing elevator control systems, achieving a perfect balance between intelligent upgrades and operational safety.

Description

Technical Field

[0001] This invention relates to the field of elevator group control technology, and in particular to an intelligent pre-scheduling system and method for elevator groups. Background Technology

[0002] With the acceleration of urbanization, high-rise and super high-rise buildings are constantly emerging. As a core tool for vertical transportation, the operating efficiency of elevators directly affects the overall service quality and user experience of buildings. Traditional elevator group control systems mainly adopt a passive "call-response" working mode, meaning that the system only begins to make elevator dispatch decisions after a passenger presses the physical call button at a floor. This mode has an inherent information lag, resulting in the dispatch system lacking the ability to predict key information such as the number, location, movement trends, and target floors of passengers in the waiting area. Therefore, in high-traffic scenarios such as morning and evening rush hours, problems such as long waiting times for passengers, high elevator empty-run rates, and uneven crowding in elevator cars are likely to occur, making it difficult to achieve proactive allocation and optimization of transport capacity.

[0003] To improve the intelligence level of dispatching, existing technologies attempt to introduce various sensing and communication methods. For example, elevator systems based on radio frequency communication (such as Wi-Fi) can achieve some data transmission, but the metal shielding environment of the elevator shaft causes severe attenuation of wireless signals, resulting in poor communication reliability and low positioning accuracy (usually at the meter level), which cannot support refined passenger intention inference. Intelligent management systems based on machine vision acquire images through cameras and use deep learning algorithms for personnel tracking. Although they can obtain relatively rich dynamic information, their performance is unstable under complex lighting conditions such as strong light, backlight, and low illumination, and the continuous acquisition of passenger images raises serious privacy protection controversies, limiting their widespread adoption in practical applications.

[0004] Furthermore, the application of visible light communication (VLC) technology in the elevator field is still in its early stages. Existing solutions are mostly limited to replacing wired communication or realizing single functions such as single-point light control buttons. They have failed to integrate the multiple capabilities of VLC, such as communication, high-precision positioning, lighting and dynamic display, and have not formed a complete closed-loop system from spatiotemporal perception, intention prediction, pre-scheduling decision-making to passenger diversion guidance.

[0005] Therefore, how to achieve proactive perception of passenger intentions in elevator scenarios with high electromagnetic interference and strong privacy concerns, and generate dynamic and adaptive pre-scheduling strategies accordingly, while guiding passengers to efficiently divert in an intuitive way, has become a technical problem that urgently needs to be solved in this field. Summary of the Invention

[0006] The main objective of this invention is to overcome the shortcomings of the prior art and provide an intelligent pre-scheduling system and method for elevator groups.

[0007] The technical solution adopted by this invention to achieve its technical objective is: an intelligent pre-scheduling system for elevator groups, comprising:

[0008] VLC spatiotemporal sensing network is deployed on multiple floors and elevator cars within a building to acquire spatiotemporal data of at least one target in the waiting area in real time using visible light signals.

[0009] An edge computing gateway, which is communicatively connected to the VLC spatiotemporal awareness network, is used to receive the spatiotemporal data and generate intent events representing passengers' waiting intentions for elevators based on preset intent recognition rules.

[0010] The pre-scheduling decision module is communicatively connected to the edge computing gateway and is used to receive the intent event and, based on the intent event and the elevator group status, generate a pre-scheduling strategy for at least one elevator before the passenger triggers a physical call signal.

[0011] A dynamic optical guidance module is deployed in the waiting area and is communicatively connected to the pre-scheduling decision module. It is used to receive the pre-scheduling strategy and generate dynamic optical guidance information corresponding to the pre-scheduling strategy to guide passengers to the target elevator.

[0012] The group control interface module is communicatively connected to the pre-scheduling decision module and the elevator group control system, respectively, and is used to send the pre-scheduling strategy to the elevator group control system in the form of suggested instructions, and to receive elevator status information fed back by the elevator group control system.

[0013] Preferably, the VLC spatiotemporal sensing network includes:

[0014] Multiple floor-station VLC beacon nodes are installed on the ceiling of the waiting area on each floor to broadcast modulated visible light beacon signals containing their identification and three-dimensional location coordinates at a preset frequency;

[0015] At least one car VLC node is installed on the top outside the elevator car to receive the signal from the floor VLC beacon node and measure the received signal strength indication;

[0016] The spatiotemporal data includes the target's real-time three-dimensional position, continuous motion trajectory, dwell time in a specific area, and motion direction vector relative to the elevator door.

[0017] Preferably, the deployment density of the layer station VLC beacon nodes is one every 1.5-2.5 meters, the transmission power is 1-5 watts, and a Lambertian radiation model of order m=1-3 is adopted;

[0018] The positioning algorithm combines Received Signal Strength Indicator (RSSI) and Angle of Arrival (AOA) measurement to achieve decimeter-level positioning accuracy, wherein the Angle of Arrival is obtained by measuring the incident angle of the light signal using an image sensor or photodiode array.

[0019] Preferably, the spatiotemporal data is obtained in the following manner:

[0020] The distance between the target and the beacon node is calculated based on the Lambertian radiation model, and the real-time position of the target is determined by combining the angle of arrival (AOA) measurement or the polygonal positioning algorithm.

[0021] The motion trajectory is fitted by the position sequence within a continuous time window, and the tangent direction of the trajectory is calculated as the motion direction vector.

[0022] When the target stays in the waiting area for more than a preset threshold T, the stay time is recorded.

[0023] Preferably, the intent event is a structured data packet, including at least:

[0024] Source floor information, indicating the current building floor of the passenger;

[0025] Waiting area signage indicates the specific elevator door area where passengers are located;

[0026] The estimated time to arrive at the door (ETA) is based on the current motion trajectory and the predicted timestamp of arrival at the elevator door.

[0027] The probability distribution of the destination layer is a vector of probability values ​​for each possible destination layer, calculated based on historical elevator data and real-time context.

[0028] Preferably, the generation of the target layer probability distribution includes:

[0029] Obtain prior probabilities based on historical OD matrices;

[0030] Likelihood probabilities are obtained based on real-time contextual information, including the current time period, floor functional area, and number of people traveling together.

[0031] Bayesian inference is used to calculate the posterior probability distribution;

[0032] When the highest probability value exceeds the preset threshold P, the target layer is used as a deterministic input to generate a pre-scheduling strategy.

[0033] Preferably, the dynamic light guiding module includes:

[0034] LED arrays or light strips installed on the floor of the waiting area, or projectors installed on the walls;

[0035] The control unit is used to generate dynamic light guidance information according to the pre-scheduling strategy. The dynamic light guidance information is presented in the form of dynamically changing color codes, brightness gradients, flowing arrow patterns or flashing frequencies, wherein different colors or patterns correspond to different target elevators or elevator directions.

[0036] Preferably, the system further includes a closed-loop feedback mechanism:

[0037] The VLC spatiotemporal sensing network monitors the target's response behavior to the dynamic optical guidance information in real time. The response behavior includes the angle between the target's movement direction and the optical guidance indication direction, the rate of change of movement speed, and the dwell time on the guidance path.

[0038] The monitored response data is fed back to the edge computing gateway or the pre-scheduling decision module;

[0039] The pre-scheduling decision module determines whether the target deviates from the preset guidance path based on the response data. If so, it triggers an update to the pre-scheduling strategy.

[0040] Preferably, determining whether the target deviates from the preset guidance path includes:

[0041] Calculate the angle between the target's actual movement direction and the optical guidance direction. ;

[0042] like Exceeding the preset deviation angle threshold Continuing beyond the time window If the similarity between the actual trajectory of the target and the preset guidance path is lower than the threshold S, it is judged as a deviation;

[0043] Triggering pre-scheduling strategy updates may involve recalculating the ETA, changing the target elevator, or adjusting the optical guidance path.

[0044] Preferably, the edge computing gateway is further used for:

[0045] Receive data from the VLC spatiotemporal sensing network;

[0046] Receive data from other sensors, including at least one of millimeter-wave radar, infrared sensors, or depth cameras;

[0047] Kalman filtering or DS evidence theory fusion algorithms are used to perform spatiotemporal alignment and weight allocation on multi-source data to generate fused spatiotemporal data.

[0048] Preferably, the pre-scheduling decision module generates the pre-scheduling strategy using a multi-objective optimization algorithm or a reinforcement learning model, wherein:

[0049] The multi-objective optimization algorithm establishes an objective function with the optimization objectives of minimizing the average waiting time, reducing the elevator empty run rate, and balancing the elevator utilization rate.

[0050] The reinforcement learning model adopts a deep Q-network. Its state space includes the current position, direction, load of the elevator group and the queue of intention events in the waiting area. Its action space includes elevator assignment instructions. The reward function comprehensively considers the waiting time penalty and energy consumption penalty.

[0051] Preferably, the system further includes a passenger terminal interaction module, used for:

[0052] Establish near-field communication connection with passengers' mobile terminals;

[0053] Receive destination information or confirmation information actively entered by passengers;

[0054] The information is transmitted to the edge computing gateway or the pre-scheduling decision module to correct the destination layer probability distribution in the intent event or to confirm the pre-scheduling strategy.

[0055] Preferably, the group control interface module includes a safety interlock unit, used for:

[0056] Real-time monitoring of elevator system operation mode signals;

[0057] When the elevator system is detected to be in fire protection mode, maintenance mode, or fault mode, the suggested instructions output by the pre-scheduling decision module are automatically bypassed, and the control connection between this system and the elevator group control system is cut off to ensure the absolute priority of the original safety control circuit of the elevator.

[0058] This invention also provides an intelligent pre-scheduling method for elevator groups, comprising the following steps:

[0059] S1. Using the VLC spatiotemporal sensing network, based on the Lambertian radiation model and angle of arrival measurement, real-time spatiotemporal data of targets within the waiting area are acquired.

[0060] S2. Based on the spatiotemporal data, by analyzing whether the target dwell time exceeds the threshold T and whether the movement direction is towards the elevator door, an intention event representing the passenger's intention to wait for the elevator is generated.

[0061] S3. Based on the intent event and the elevator group status, a multi-objective optimization algorithm or reinforcement learning model is used to generate a pre-scheduling strategy for at least one elevator before the passenger triggers the physical elevator call signal.

[0062] S4. Based on the pre-scheduling strategy, generate dynamic light guidance information in the waiting area to guide passengers to the target elevator in the form of dynamic color, brightness or arrow.

[0063] S5. Send the pre-scheduling strategy to the elevator group control system in the form of a suggested instruction, and receive the elevator status information fed back by the system.

[0064] S6. Monitor the target's response to the dynamic optical guidance information in real time, calculate the angle between the actual movement direction and the guidance direction, and adjust the pre-scheduling strategy in a closed loop based on the response behavior.

[0065] The working principle of the intelligent pre-scheduling system for elevator groups based on VLC spatiotemporal perception and dynamic light guidance is as follows: A VLC spatiotemporal perception network deployed on building floors and elevator cars captures real-time spatiotemporal data such as the precise location, movement trajectory, and dwell status of passengers in the waiting area. The edge computing gateway generates an "intent event" based on this data, including the source floor, estimated time to arrival (ETA), and probability of the destination floor. Before a passenger triggers the physical call button, the pre-scheduling decision module, combined with the elevator group status, generates a pre-scheduling strategy through a multi-objective optimization algorithm. On one hand, the system sends scheduling suggestions to the elevator group control system via the group control interface module to enable elevators to be ready for standby or dispatched in advance. On the other hand, it generates dynamic light guidance information through LED light strips or projection devices on the waiting area floor to guide passengers to designated elevators. Simultaneously, the VLC network continuously monitors passenger response to the guidance and feeds the data back to the decision module. If a passenger deviates from the path, a strategy update is triggered, forming a complete closed-loop workflow of "perception-intent-pre-scheduling-light guidance-feedback correction."

[0066] Compared with the prior art, the beneficial effects of the present invention are:

[0067] This intelligent pre-scheduling system and method for elevator groups significantly improves elevator operating efficiency and user experience. By generating intent events and performing pre-scheduling based on passengers' spatiotemporal data before they trigger physical call signals, it transforms the model from "people looking for elevators" to "elevators waiting for people." It effectively shortens the average waiting time for passengers, significantly reduces long waiting rates, and uses dynamic optical guidance to divert passenger flow, balancing the load on each elevator, thereby comprehensively improving the transportation efficiency and riding experience of vertical transportation.

[0068] This intelligent pre-scheduling system and method for elevator groups achieves high-precision and robust spatiotemporal perception. Leveraging the inherent advantages of VLC sensing network technology in the strong electromagnetic shielding environment of elevator shafts, and combining it with Lambertian model-based, angle-of-arrival measurement, or polygonal positioning algorithms, it can achieve indoor positioning accuracy at the decimeter or even centimeter level, far exceeding traditional radio frequency or vision solutions. Simultaneously, through a multi-source sensor fusion mechanism using optional millimeter-wave radar, depth cameras, and other sensors, it effectively overcomes perception blind spots in complex environments such as occlusion and changes in lighting, ensuring the system's stable, reliable, and all-weather operation.

[0069] This intelligent pre-scheduling system and method for elevator groups constructs a closed-loop intelligent decision-making and adaptive guidance mechanism: the system creatively combines a VLC sensing network with a dynamic optical guidance module to form a complete closed loop of "perception-decision-guidance-feedback". The pre-scheduling strategy is not only used to control the elevators, but also transformed into visual optical guidance information to guide passenger behavior; while the VLC sensing network monitors passenger responses to guidance in real time and uses the data feedback for immediate correction and re-optimization of the strategy, thus enabling it to adapt to changes in passenger flow and the randomness of individual behavior, achieving efficient and flexible management of passenger flow.

[0070] This intelligent pre-scheduling system and method for elevator groups balances privacy protection and infrastructure reuse, reducing implementation costs. Based on VLC optical signals for positioning and communication, this invention differs from visual solutions that rely on camera image acquisition. It only needs to analyze the physical characteristics of the optical signals (such as intensity and angle of incidence), without involving passengers' biometric information, thus fundamentally avoiding the risk of privacy leaks and increasing public acceptance. Furthermore, the system can reuse existing LED lighting infrastructure within buildings as VLC beacon nodes, significantly reducing the complexity and cost of hardware deployment and modification, demonstrating good economic efficiency and scalability.

[0071] This intelligent pre-scheduling system and method for elevator groups ensures high-safety coordination with existing elevator control systems. By setting up a group control interface module containing a safety interlock unit, this invention interacts with the original elevator group control system in the form of "suggested instructions" rather than directly taking over the elevator's safety circuit. When special modes such as fire protection, maintenance, or malfunction are detected, the output of this system can be automatically bypassed, ensuring that the safety priority of the original elevator control system is not affected, achieving a perfect balance between intelligent upgrades and operational safety. Attached Figure Description

[0072] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0073] Figure 1 This is a system architecture diagram of an intelligent pre-scheduling system for elevator groups.

[0074] Figure 2 This is a system architecture diagram of the VLC spatiotemporal sensing network.

[0075] Figure 3 A flowchart illustrating the steps of an intelligent pre-scheduling method for elevator groups. Detailed Implementation

[0076] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. However, it should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of the invention. Furthermore, descriptions of well-known structures and technologies are omitted in the following description to avoid unnecessarily obscuring the concept of the invention.

[0077] In the description of this invention, it should be noted that when an element is referred to as being "fixed to" or "set on" another element, it can be directly on or indirectly on the other element. When an element is referred to as being "connected to" another element, it can be directly connected to or indirectly connected to the other element.

[0078] In the description of this invention, it should be noted that the terms "center," "length," "width," "thickness," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," and "outer," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings, or the orientation or positional relationship commonly used when the product of this invention is in use. They are used only for the convenience of describing the invention and for simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on the invention. Furthermore, the terms "first," "second," and "third," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified. "Several" means one or more, unless otherwise explicitly specified.

[0079] In the description of this invention, it should also be noted that, unless otherwise explicitly specified and limited, the terms "set," "install," "connect," and "link" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in this invention based on the specific circumstances.

[0080] Example 1:

[0081] Please see Figures 1-2 An intelligent pre-scheduling system for elevator groups includes a VLC spatiotemporal sensing network, an edge computing gateway, a pre-scheduling decision module, a dynamic optical guidance module, and a group control interface module.

[0082] In this embodiment, the VLC spatiotemporal sensing network is deployed on multiple floors and elevator cars within the building to acquire spatiotemporal data of at least one target in the waiting area in real time using visible light signals.

[0083] The VLC spatiotemporal sensing network includes:

[0084] Multiple floor-station VLC beacon nodes are installed on the ceiling of the waiting area on each floor to broadcast modulated visible light beacon signals containing their identification and three-dimensional location coordinates at a preset frequency;

[0085] At least one car VLC node is installed on the top outside the elevator car to receive the signal from the floor VLC beacon node and measure the received signal strength indication;

[0086] The spatiotemporal data includes the target's real-time three-dimensional position, continuous motion trajectory, dwell time in a specific area, and motion direction vector relative to the elevator door.

[0087] The VLC beacon nodes of the stratum station are deployed at a density of one every 1.5-2.5 meters, with a transmission power of 1-5 watts, and adopt a Lambertian radiation model of order m=1-3;

[0088] The positioning algorithm combines Received Signal Strength Indicator (RSSI) and Angle of Arrival (AOA) measurement to achieve decimeter-level positioning accuracy, wherein the Angle of Arrival is obtained by measuring the incident angle of the light signal using an image sensor or photodiode array.

[0089] The spatiotemporal data is obtained through the following methods:

[0090] The distance between the target and the beacon node is calculated based on the Lambertian radiation model, and the real-time position of the target is determined by combining the angle of arrival (AOA) measurement or the polygonal positioning algorithm.

[0091] The motion trajectory is fitted by the position sequence within a continuous time window, and the tangent direction of the trajectory is calculated as the motion direction vector.

[0092] When the target stays in the waiting area for more than a preset threshold T, the stay time is recorded.

[0093] In this embodiment, the edge computing gateway is communicatively connected to the VLC spatiotemporal awareness network to receive the spatiotemporal data and generate intent events representing passengers' waiting intentions for elevators based on preset intent recognition rules.

[0094] The edge computing gateway is also used for:

[0095] Receive data from the VLC spatiotemporal sensing network;

[0096] Receive data from other sensors, including at least one of millimeter-wave radar, infrared sensors, or depth cameras;

[0097] Kalman filtering or DS evidence theory fusion algorithms are used to perform spatiotemporal alignment and weight allocation on multi-source data to generate fused spatiotemporal data.

[0098] The intent event is a structured data packet, which includes at least:

[0099] Source floor information, indicating the current building floor of the passenger;

[0100] Waiting area signage indicates the specific elevator door area where passengers are located;

[0101] The estimated time to arrive at the door (ETA) is based on the current motion trajectory and the predicted timestamp of arrival at the elevator door.

[0102] The probability distribution of the destination layer is a vector of probability values ​​for each possible destination layer, calculated based on historical elevator data and real-time context.

[0103] The generation of the target layer probability distribution includes:

[0104] Obtain prior probabilities based on historical OD matrices;

[0105] Likelihood probabilities are obtained based on real-time contextual information, including the current time period, floor functional area, and number of people traveling together.

[0106] Bayesian inference is used to calculate the posterior probability distribution;

[0107] When the highest probability value exceeds the preset threshold P, the target layer is used as a deterministic input to generate a pre-scheduling strategy.

[0108] In this embodiment, the pre-scheduling decision module is communicatively connected to the edge computing gateway and is used to receive the intent event and, based on the intent event and the elevator group status, generate a pre-scheduling strategy for at least one elevator before the passenger triggers a physical call signal.

[0109] The pre-scheduling decision module generates the pre-scheduling strategy using a multi-objective optimization algorithm or a reinforcement learning model, wherein:

[0110] The multi-objective optimization algorithm establishes an objective function with the optimization objectives of minimizing the average waiting time, reducing the elevator empty run rate, and balancing the elevator utilization rate.

[0111] The reinforcement learning model adopts a deep Q-network. Its state space includes the current position, direction, load of the elevator group and the queue of intention events in the waiting area. Its action space includes elevator assignment instructions. The reward function comprehensively considers the waiting time penalty and energy consumption penalty.

[0112] In this embodiment, a dynamic light guidance module is deployed in the waiting area and is communicatively connected to the pre-scheduling decision module. It is used to receive the pre-scheduling strategy and generate dynamic light guidance information corresponding to the pre-scheduling strategy to guide passengers to the target elevator.

[0113] The dynamic light guiding module includes:

[0114] LED arrays or light strips installed on the floor of the waiting area, or projectors installed on the walls;

[0115] The control unit is used to generate dynamic light guidance information according to the pre-scheduling strategy. The dynamic light guidance information is presented in the form of dynamically changing color codes, brightness gradients, flowing arrow patterns or flashing frequencies, wherein different colors or patterns correspond to different target elevators or elevator directions.

[0116] In this embodiment, the group control interface module is communicatively connected to the pre-scheduling decision module and the elevator group control system, respectively, and is used to send the pre-scheduling strategy to the elevator group control system in the form of a suggestion instruction, and receive the elevator status information fed back by the elevator group control system.

[0117] The group control interface module includes a safety interlock unit for:

[0118] Real-time monitoring of elevator system operation mode signals;

[0119] When the elevator system is detected to be in fire protection mode, maintenance mode, or fault mode, the suggested instructions output by the pre-scheduling decision module are automatically bypassed, and the control connection between this system and the elevator group control system is cut off to ensure the absolute priority of the original safety control circuit of the elevator.

[0120] In this embodiment, the system further includes a closed-loop feedback mechanism:

[0121] The VLC spatiotemporal sensing network monitors the target's response behavior to the dynamic optical guidance information in real time. The response behavior includes the angle between the target's movement direction and the optical guidance indication direction, the rate of change of movement speed, and the dwell time on the guidance path.

[0122] The monitored response data is fed back to the edge computing gateway or the pre-scheduling decision module;

[0123] The pre-scheduling decision module determines whether the target deviates from the preset guidance path based on the response data. If so, it triggers an update to the pre-scheduling strategy.

[0124] The determination of whether the target deviates from the preset guidance path includes:

[0125] Calculate the angle between the target's actual movement direction and the optical guidance direction. ;

[0126] like Exceeding the preset deviation angle threshold Continuing beyond the time window If the similarity between the actual trajectory of the target and the preset guidance path is lower than the threshold S, it is judged as a deviation;

[0127] Triggering pre-scheduling strategy updates may involve recalculating the ETA, changing the target elevator, or adjusting the optical guidance path.

[0128] In this embodiment, the system further includes a passenger terminal interaction module, used for:

[0129] Establish near-field communication connection with passengers' mobile terminals;

[0130] Receive destination information or confirmation information actively entered by passengers;

[0131] The information is transmitted to the edge computing gateway or the pre-scheduling decision module to correct the destination layer probability distribution in the intent event or to confirm the pre-scheduling strategy.

[0132] Specifically, in use, when a passenger enters the elevator waiting area, the system automatically senses their presence and location via VLC light signals. If a passenger stops in front of an elevator door or continues to move towards the door area, the system determines that they intend to take the elevator and generates an intent event.

[0133] Based on this, the system backend calculates the optimal elevator dispatch plan in advance (e.g., "Please go to elevator A") even if the passenger does not press any buttons, and issues diversion guidance to the passenger through dynamic light strips or flashing arrows on the ground. As the passenger follows the guidance to the designated elevator door, the system continuously tracks their path to ensure that they "follow the map".

[0134] If a passenger deviates from the path or the elevator status changes abruptly, the system will adjust the guidance and elevator dispatch plan in real time. When the passenger finally arrives at the door, the corresponding elevator is already waiting, and the passenger only needs to press the regular button to enter, and can experience elevator service with almost no waiting.

[0135] Throughout the entire process, passengers do not need to perform any active operations. The system recognizes intentions through light perception and achieves human-computer interaction through light guidance.

[0136] Example 2:

[0137] Please see Figure 3 Based on the above embodiments, this invention also provides an intelligent pre-scheduling method for elevator groups, comprising the following steps:

[0138] S1. Using the VLC spatiotemporal sensing network, based on the Lambertian radiation model and angle of arrival measurement, real-time spatiotemporal data of targets within the waiting area are acquired.

[0139] S2. Based on the spatiotemporal data, by analyzing whether the target dwell time exceeds the threshold T and whether the movement direction is towards the elevator door, an intention event representing the passenger's intention to wait for the elevator is generated.

[0140] S3. Based on the intent event and the elevator group status, a multi-objective optimization algorithm or reinforcement learning model is used to generate a pre-scheduling strategy for at least one elevator before the passenger triggers the physical elevator call signal.

[0141] S4. Based on the pre-scheduling strategy, generate dynamic light guidance information in the waiting area to guide passengers to the target elevator in the form of dynamic color, brightness or arrow.

[0142] S5. Send the pre-scheduling strategy to the elevator group control system in the form of a suggested instruction, and receive the elevator status information fed back by the system.

[0143] S6. Monitor the target's response to the dynamic optical guidance information in real time, calculate the angle between the actual movement direction and the guidance direction, and adjust the pre-scheduling strategy in a closed loop based on the response behavior.

[0144] The solution in this embodiment can be selectively combined with solutions in other embodiments.

[0145] The specific workflow of this intelligent pre-scheduling system for elevator groups based on VLC spatiotemporal perception and dynamic light guidance is as follows: By deeply integrating high-precision spatiotemporal perception, intent inference, multi-objective optimization decision-making, and human-computer interaction guidance, a closed-loop ecosystem of "perception-decision-guidance-feedback" is constructed. Specifically, the system first relies on the VLC spatiotemporal perception network deployed on the ceilings of the waiting areas on each floor of the building and in the elevator cars. These LED beacons not only provide lighting but also continuously broadcast modulated light signals containing identification and location coordinates. Passengers' mobile terminals (such as mobile phones) or VLC nodes on the elevator cars act as receivers. By analyzing these signals and combining them with distance estimation based on the Lambertian radiation model, angle of arrival (AOA) measurement, or polygonal positioning algorithms, the system can capture in real time fine-grained spatiotemporal data such as centimeter-level high-precision location, movement trajectory, dwell time, and direction of movement of passengers.

[0146] After the edge computing gateway aggregates this raw data, it fuses information from multiple sources such as millimeter-wave radar or depth cameras, uses Kalman filtering for trajectory smoothing, and performs behavior recognition based on deep learning models such as Bi-LSTM or Transformer. This intelligently generates a structured "intent event" before the passenger physically triggers the elevator call button. This event not only includes the source floor and waiting area identifiers but also accurately calculates the estimated time to arrival (ETA) of the passenger at the elevator door and the probability distribution of the destination floor based on the historical OD matrix, real-time context, and optional mobile phone input. Subsequently, the pre-scheduling decision module receives these intent events and the real-time status of the elevator group (such as location, direction, and load). Using advanced algorithms such as the NSGA-II multi-objective genetic algorithm, Deep Q-Network (DQN), or Model Predictive Control (MPC), it solves for the optimal pre-scheduling strategy—that is, before the passenger presses the button, it pre-assigns one or more elevators to the corresponding floor to wait or adjusts their direction of travel.

[0147] Once generated, the strategy is executed through two paths: First, it is sent to the original elevator group control system as a "suggested instruction" via the group control interface module (using industrial protocols such as CANopen or Modbus TCP). The original system adopts and executes the suggestion under the premise of ensuring safety. The interface has a built-in safety interlock unit that can automatically bypass in fire or maintenance mode to ensure absolute safety. Second, the pre-scheduling strategy is transmitted to the dynamic light guidance module (such as LED array, light strip or projection device) deployed on the floor or wall of the waiting area. The module then generates a clear and intuitive visual path for passengers in the form of dynamically changing colors, brightness, flowing arrows or digital codes, guiding them to the designated target elevator door area to wait, thereby achieving the efficient "elevator waits for passengers" mode.

[0148] Crucially, the entire system features a dynamic closed-loop feedback mechanism: the VLC spatiotemporal awareness network continuously monitors passenger responses to optical guidance information (such as whether they move in the indicated direction or stay in the designated area) and transmits this response data back in real time. Once the edge computing gateway or pre-scheduling decision module detects that a passenger has deviated from the preset path or experienced abnormal stagnation, it triggers immediate correction—either dynamically updating the optical guidance path or readjusting the elevator dispatch candidate strategy, and then conveying the updated information back to the passenger through the optical guidance module. Thus, through a cycle of "continuous perception - intent update - strategy optimization - optical guidance adjustment - re-perception," the system achieves adaptive adjustment to dynamic passenger flow and uncertain behaviors, ensuring that the entire elevator group system maintains efficient, smooth, and intelligent operation even in complex scenarios such as peak hours, multi-floor cross-requests, and emergency rescues.

[0149] It should be noted that although the above embodiments have been described herein, this does not limit the scope of patent protection of this invention. Therefore, any changes and modifications made to the embodiments described herein based on the innovative concept of this invention, or equivalent structural, procedural, or functional transformations made using the description and drawings of this invention, directly or indirectly applying the above technical solutions to other related technical fields, are all included within the scope of protection of this invention.

Claims

1. An intelligent pre-dispatching system for an elevator group, characterized in that, include: VLC spatiotemporal sensing network is deployed on multiple floors and elevator cars within a building to acquire spatiotemporal data of at least one target in the waiting area in real time using visible light signals. An edge computing gateway, which is communicatively connected to the VLC spatiotemporal awareness network, is used to receive the spatiotemporal data and generate intent events representing passengers' waiting intentions for elevators based on preset intent recognition rules. The pre-scheduling decision module is communicatively connected to the edge computing gateway and is used to receive the intent event and, based on the intent event and the elevator group status, generate a pre-scheduling strategy for at least one elevator before the passenger triggers a physical call signal. A dynamic optical guidance module is deployed in the waiting area and is communicatively connected to the pre-scheduling decision module. It is used to receive the pre-scheduling strategy and generate dynamic optical guidance information corresponding to the pre-scheduling strategy to guide passengers to the target elevator. The group control interface module is communicatively connected to the pre-scheduling decision module and the elevator group control system, respectively, and is used to send the pre-scheduling strategy to the elevator group control system in the form of suggested instructions, and to receive elevator status information fed back by the elevator group control system.

2. The intelligent pre-dispatching system for elevator group according to claim 1, characterized in that, The VLC spatiotemporal sensing network includes: Multiple floor-station VLC beacon nodes are installed on the ceiling of the waiting area on each floor to broadcast modulated visible light beacon signals containing their identification and three-dimensional location coordinates at a preset frequency; At least one car VLC node is installed on the top outside the elevator car to receive the signal from the floor VLC beacon node and measure the received signal strength indication; The spatiotemporal data includes the target's real-time three-dimensional position, continuous motion trajectory, dwell time in a specific area, and motion direction vector relative to the elevator door.

3. The intelligent pre-scheduling system for elevator groups according to claim 1, characterized in that, The intent event is a structured data packet, which includes at least: Source floor information, indicating the current building floor of the passenger; Waiting area signage indicates the specific elevator door area where passengers are located; The estimated time to arrive at the door (ETA) is based on the current motion trajectory and the predicted timestamp of arrival at the elevator door. The probability distribution of the destination layer is a vector of probability values ​​for each possible destination layer, calculated based on historical elevator data and real-time context.

4. The intelligent pre-scheduling system for elevator groups according to claim 1, characterized in that, The dynamic light guiding module includes: LED arrays or light strips installed on the floor of the waiting area, or projectors installed on the walls; The control unit is used to generate dynamic light guidance information according to the pre-scheduling strategy. The dynamic light guidance information is presented in the form of dynamically changing color codes, brightness gradients, flowing arrow patterns or flashing frequencies, wherein different colors or patterns correspond to different target elevators or elevator directions.

5. The intelligent pre-scheduling system for elevator groups according to claim 1, characterized in that, The system also includes a closed-loop feedback mechanism: The VLC spatiotemporal sensing network monitors the target's response behavior to the dynamic optical guidance information in real time. The response behavior includes the angle between the target's movement direction and the optical guidance indication direction, the rate of change of movement speed, and the dwell time on the guidance path. The monitored response data is fed back to the edge computing gateway or the pre-scheduling decision module; The pre-scheduling decision module determines whether the target deviates from the preset guidance path based on the response data. If so, it triggers an update to the pre-scheduling strategy.

6. The intelligent pre-scheduling system for elevator groups according to claim 1, characterized in that, The edge computing gateway is also used for: Receive data from the VLC spatiotemporal sensing network; Receive data from other sensors, including at least one of millimeter-wave radar, infrared sensors, or depth cameras; Kalman filtering or DS evidence theory fusion algorithms are used to perform spatiotemporal alignment and weight allocation on multi-source data to generate fused spatiotemporal data.

7. The intelligent pre-scheduling system for elevator groups according to claim 1, characterized in that, The pre-scheduling decision module generates the pre-scheduling strategy using a multi-objective optimization algorithm or a reinforcement learning model, wherein: The multi-objective optimization algorithm establishes an objective function with the optimization objectives of minimizing the average waiting time, reducing the elevator empty run rate, and balancing the elevator utilization rate. The reinforcement learning model employs a deep learning network. Its state space includes the current position, direction, load of the elevator group, and the queue of intentional events in the waiting area. Its action space includes elevator assignment instructions. The reward function comprehensively considers waiting time penalties and energy consumption penalties.

8. The intelligent pre-scheduling system for elevator groups according to claim 1, characterized in that, The system also includes a passenger terminal interaction module, used for: Establish near-field communication connection with passengers' mobile terminals; Receive destination information or confirmation information actively entered by passengers; The information is transmitted to the edge computing gateway or the pre-scheduling decision module to correct the destination layer probability distribution in the intent event or to confirm the pre-scheduling strategy.

9. The intelligent pre-scheduling system for elevator groups according to claim 1, characterized in that, The group control interface module includes a safety interlock unit for: Real-time monitoring of elevator system operation mode signals; When the elevator system is detected to be in fire protection mode, maintenance mode, or fault mode, the suggested instructions output by the pre-scheduling decision module are automatically bypassed, and the control connection between this system and the elevator group control system is cut off to ensure the absolute priority of the original safety control circuit of the elevator.

10. An intelligent pre-scheduling method for elevator groups, applied to the system according to any one of claims 1 to 9, characterized in that, Includes the following steps: S1. Real-time acquisition of spatiotemporal data of targets within the waiting area through VLC spatiotemporal sensing network; S2. Based on the spatiotemporal data, generate an intent event representing the passenger's intention to wait for the elevator; S3. Based on the intent event and the elevator group status, generate a pre-scheduling strategy for at least one elevator before the passenger triggers a physical elevator call signal. S4. Based on the pre-scheduling strategy, generate dynamic optical guidance information in the waiting area to guide passengers to the target elevator; S5. Send the pre-scheduling strategy to the elevator group control system in the form of a suggested instruction, and receive the elevator status information fed back by the system. S6. Monitor the target's response behavior to the dynamic optical guidance information in real time, and adjust the pre-scheduling strategy in a closed loop based on the response behavior.

Images