Multi-elevator linkage dynamic response intelligent scheduling system and scheduling method thereof

Abstract

The invention discloses a multi-elevator linkage dynamic response intelligent scheduling system and a scheduling method thereof. The scheduling method comprises the following steps: acquiring an operation rule of an elevator, constructing a historical operation database and updating the historical operation database in real time; capturing a field state in real time through a monitoring system, and generating a real-time data stream according to a time sequence; a digital twinborn model is constructed through a historical database and real-time data flow, and the elevator calling probability and the congestion risk of each floor are predicted; the elevator calling probability of each floor in the next five minutes is predicted based on a digital twinborn model, the congestion risk is calculated, a real-time demand thermodynamic diagram fusing the space-time sequence dimension is generated, and intelligent scheduling is conducted on the elevator through the thermodynamic diagram; by reducing the waiting time, the overall operation efficiency is improved, the user satisfaction is enhanced, and smoother vertical traffic experience is provided for high-rise buildings.

Description

TECHNICAL FIELD

[0001] The present application relates to a multi-elevator linkage scheduling method, in particular to a multi-elevator linkage dynamic response intelligent scheduling system and a scheduling method thereof. BACKGROUND

[0002] Under the background of the increasing popularity of high-rise buildings, the operation efficiency of the elevator system has become a key factor affecting the building internal traffic experience and energy consumption. The traditional elevator scheduling method mostly adopts independent operation or simple group control strategy, such as fixed time sequence or single call response distribution mechanism, which is difficult to cope with the complex and variable passenger flow dynamics and the concentrated demand in peak period in modern buildings. Such methods often lead to problems such as delayed response, high empty load rate, prolonged passenger waiting time, and low system energy efficiency. With the rapid development of Internet of Things technology and artificial intelligence algorithms, a multi-elevator linkage dynamic response intelligent scheduling system has emerged, which collects real-time data such as elevator operation status, car load, floor call signal, and historical passenger flow patterns, and builds a scheduling model with prediction and adaptive capabilities. The system aims to optimize collaboration, uses machine learning, fuzzy logic, or reinforcement learning algorithms to achieve multi-elevator collaborative distribution, dynamic path planning, and load balancing, thereby significantly improving transportation efficiency, reducing energy consumption, and improving user experience, becoming an important part of the core infrastructure of modern intelligent buildings. SUMMARY

[0003] The present application overcomes the shortcomings of the prior art and provides a multi-elevator linkage dynamic response intelligent scheduling system and a scheduling method thereof.

[0004] To achieve the above purpose, the technical solution adopted by the present application is as follows: a multi-elevator linkage dynamic response intelligent scheduling system and a scheduling method thereof, comprising the following steps:

[0005] S1: Obtain the operation rules of the elevator, build a historical operation database and update it in real time;

[0006] S2: Capture the real-time state through the monitoring system, and generate real-time data stream according to the time sequence;

[0007] S3: Build a digital twin model through the historical database and real-time data stream to predict the call probability of each floor and the congestion risk;

[0008] S4: Based on the digital twin model, predict the call probability of each floor in the next five minutes and calculate the congestion risk, generate a real-time demand heat map integrating space-time dimensions, and intelligently schedule the elevator through the heat map.

[0009] In one preferred embodiment of the present application, in S1, the elevator operation rules include the call frequency, stopping times, operation conditions, and special events of each floor every day.

[0010] In a preferred embodiment of the present application, in S1, the historical operation database is specifically constructed to record the elevator operation rules in the last month, including the call frequency of each floor in different time periods, the actual number of elevator stops, the number of waiting passengers based on real-time monitoring, the door opening and closing time to the second, the car full load rate monitored by the pressure sensor, and the marking of special events; the historical database also includes the passenger demand profile.

[0011] In a preferred embodiment of the present application, the passenger demand profile records the spatial demand characteristics of different passengers, including the average standing area and luggage carrying mode.

[0012] In a preferred embodiment of the present application, in S2, the monitoring system adopts a multi-spectral depth camera array to form a heterogeneous sensor network, and calculates the effective standing area of each passenger and the real-time dynamic remaining area in the elevator car through real-time point cloud modeling and human contour recognition algorithm.

[0013] In a preferred embodiment of the present application, in S2, the field state includes the real-time crowd distribution of each floor, the real-time gathering situation of the elevator entrance, and the passenger carrying state inside the elevator, and a timestamp is generated for each detected passenger to record the time of entering the monitoring area. The captured field state is specifically real-time analysis of crowd density and spatial demand at each floor elevator entrance through depth camera and computer vision algorithm, dynamic calculation of passenger capacity and remaining effective area through infrared array sensor inside the car, and real-time transmission of call instruction and car selection instruction by the elevator control system.

[0014] In a preferred embodiment of the present application, in S3, the digital twin system construction includes mapping the computable model of building structure, elevator physical characteristics and traffic flow pattern, and analyzing the periodicity of historical data through LSTM neural network, while dynamically correcting according to real-time data flow.

[0015] In a preferred embodiment of the present application, in S3, the call probability prediction is specifically based on the periodicity trend of historical data and fused with real-time spatial demand characteristics; the congestion risk is specifically determined by the current floor real-time crowd gathering degree, the total amount of spatial demand, and the prediction based on historical data, and the waiting time anxiety factor is introduced.

[0016] In a preferred embodiment of the present invention, in S4, the real-time demand heat map calculates the heat score of each floor through a multi-dimensional comprehensive calculation, wherein the multiple dimensions include four dimensions: elevator call probability, real-time congestion coefficient, area adaptation index and anxiety factor, specifically the average number of elevator calls for the floor in the past month, the real-time congestion coefficient is the current number of waiting people, the area adaptation index is specifically the current idle or nearly idle elevator queue, the ratio of the theoretical remaining space after each elevator arrives at the floor to the space required by the waiting crowd, and the anxiety factor is adjusted based on the waiting time. The longer the waiting time, the greater the anxiety factor.

[0017] In a preferred embodiment of the present invention, in S4, intelligent scheduling is for the system to schedule elevators according to the thermal level. Specifically, when the elevator task queue is cleared, a 10-second countdown is started. If there is a new call during this period, the call waiting period is interrupted to respond to the new demand. When the countdown ends, the thermal map is queried to obtain the highest-scoring floor, and the elevator is instructed to move to the target floor to wait for the call. During the scheduling process, under the premise of ensuring waiting time priority, the passenger carrying efficiency is improved through space adaptation optimization. When the body size difference exceeds 40% and the space utilization rate can be improved by more than 30%, the order is allowed to be adjusted.

[0018] The present invention solves the defects existing in the background technology and has the following beneficial effects:

[0019] (1) The present invention provides a dynamic response intelligent dispatching system for multiple elevators and its dispatching method. By constructing a historical operation database and updating it in real time, and using a digital twin model to predict the elevator call probability and congestion risk of each floor in the next five minutes, the system can deeply integrate historical rules and real-time data. Through this predictive ability, the system instructs idle elevators to move to high-demand floors in advance and wait, changing passive response to active service, significantly shortening the waiting time for passengers, achieving the optimized state of "elevators waiting for people", and improving transportation efficiency. Compared with existing technologies, traditional elevator dispatching systems only respond based on current elevator call requests, lack foresight, and often lead to extended waiting time and dispatch delays. By reducing waiting time, not only the overall operating efficiency is improved, but also user satisfaction is enhanced, and a smoother vertical transportation experience is provided for high-rise buildings.

[0020] (2) The application provides a multi-elevator linkage dynamic response intelligent scheduling system and a scheduling method thereof, through the introduction of area adaptation index and space adaptation optimization logic, accurate calculation is carried out based on passenger space demand characteristics, it is ensured that the remaining space of the elevator and the demand of the waiting crowd are considered during scheduling, the system intelligently adjusts the service order on the premise of ensuring fairness, maximizes the elevator space utilization, improves the passenger carrying efficiency, reduces the empty running or inefficient operation, compared with the prior art, the existing system usually simply counts the number of passengers, ignores the space factor, is easy to cause space waste or uneven passenger carrying, thereby optimizing the loading order, the system maintains service fairness while improving the operation capacity, thereby reducing the operation cost and prolonging the service life of the equipment.

[0021] (3) The application provides a multi-elevator linkage dynamic response intelligent scheduling system and a scheduling method thereof, through the adoption of an anxiety factor to quantify the waiting time of passengers, and the passengers with a longer waiting time are preferentially processed in scheduling, which ensures service fairness, avoids any passenger being delayed for a long time, reduces passenger anxiety, improves user experience and satisfaction, compared with the prior art, the traditional scheduling algorithm such as first come first served may not effectively handle the problems of sudden congestion or individual long waiting, and by combining space optimization, the system maintains high efficiency while humanizing the care of passenger emotions, thereby constructing a more harmonious building environment and enhancing system reliability. BRIEF DESCRIPTION OF DRAWINGS

[0022] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiment or prior art description, obviously, the drawings in the following description are only some embodiments of the present application, and for those skilled in the art, other drawings can be obtained without creative labor on the basis of these drawings;

[0023] Figure 1 is a perspective structural view of the preferred embodiment of the present application. DETAILED DESCRIPTION

[0024] The technical solutions in the embodiments of the present application will be described clearly and completely in combination with the drawings in the embodiments of the present application, obviously, the described embodiments are only some of the embodiments of the present application, not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the protection scope of the present application.

[0025] In the following description, many specific details are set forth in order to provide a thorough understanding of the present application, however, the present application can also be implemented in other ways different from those described herein, therefore, the protection scope of the present application is not limited by the specific embodiments disclosed below.

[0026] In the description of the present application, it should be understood that the terms "center", "longitudinal", "transverse", "upper", "lower", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer" and the like indicate the orientation or positional relationship shown in the drawings, and are only for the convenience of describing the present application and simplifying the description, and do not indicate or imply that the device or element referred to must have a particular orientation, be constructed and operated in a particular orientation, and therefore cannot be understood as limiting the scope of protection of the present application. In addition, the terms "first", "second" and the like are only for descriptive purposes and cannot be understood as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Therefore, the features defined with "first", "second" and the like can explicitly or implicitly include one or more features. In the description of the present application, unless otherwise specified, the meaning of "a plurality of" is two or more.

[0027] In the description of the present application, it should be noted that unless otherwise specified and limited, the terms "mounting", "connection", "connection" should be understood broadly, for example, it can be fixedly connected, or it can be detachably connected, or integrally connected; it can be mechanically connected, or it can be electrically connected; it can be directly connected, or it can be indirectly connected through an intermediate medium, and it can be the communication between the two elements inside. For those skilled in the art, the specific meaning of the above terms in the present application can be understood through specific circumstances.

[0028] As shown in the figure, a multi-elevator linkage dynamic response intelligent scheduling system and its scheduling method, comprising the following steps:

[0029] S1: Obtain the operation law of the elevator, build a historical operation database and update it in real time;

[0030] S2: Capture the real-time state through the monitoring system, and generate real-time data stream according to time sequence;

[0031] S3: Build a digital twin model through the historical database and real-time data stream, predict the elevator call probability of each floor and congestion risk;

[0032] S4: Based on the digital twin model, predict the elevator call probability of each floor in the next five minutes and calculate the congestion risk, generate a real-time demand heat map integrating space-time dimension, and intelligently schedule the elevator through the heat map.

[0033] It should be noted that the dynamic response intelligent scheduling method of the multiple elevators is based on the deep fusion of historical operation data and real-time monitoring information in the scheduling process. The system uses digital twin technology to construct a virtual mapping model, analyzes long-term periodicity, and combines real-time captured field conditions to predict the potential demand of each floor in the next five minutes. Based on this prediction, the system generates a dynamic heat map that integrates spatial demand and time urgency, which serves as the core basis for scheduling decisions.

[0034] In specific implementation, the idle elevator is instructed to move to the predicted high demand floor in advance to standby, so that the transportation deployment is completed before the actual call request of the passenger, changing passive response to active service, realizing the predictive action of the elevator, and the whole process builds a closed loop of self-learning and continuous optimization. The actual results of each scheduling, such as passenger waiting time, actual passenger capacity of the car and the deviation of the predicted value, are fed back to the historical database for continuous training and optimization of the prediction model. By accurately calculating the overall spatial demand of the waiting passengers instead of simply counting, and combining individual waiting time to give different anxiety weights, the system maximizes transportation efficiency while ensuring service fairness, effectively avoiding any passenger being delayed for a long time.

[0035] Through multi-dimensional data fusion and intelligent prediction algorithm, the dynamic optimization scheduling of the elevator system is realized. The running state of the elevator, the distribution of the waiting crowd and the individual spatial demand characteristics are collected in real time through a heterogeneous sensor network (multi-spectrum depth camera, infrared array, pressure sensor, etc.), forming high-precision space-time data flow. At the same time, a knowledge base including historical call patterns, passenger flow periodicity and passenger behavior characteristics is established. Then, the physical elevator system is mapped to a computable model using digital twin technology, the LSTM neural network is used to analyze the time sequence rules in the historical data, and the real-time data flow is dynamically corrected to realize accurate prediction of the call probability and congestion risk of each floor in the next five minutes. On this basis, the four-dimensional heat map algorithm (integrating historical call probability, real-time congestion coefficient, spatial adaptation index and waiting anxiety factor) is used to quantify the scheduling priority, the "predictive standby" mechanism is used to instruct the idle elevator to be deployed in advance to the high demand area, and the spatial optimization strategy is introduced to ensure basic fairness, realizing the coordinated optimization of transportation efficiency and user experience.

[0036] S1: Obtain the running rules of the elevator, construct a historical operation database and update it in real time;

[0037] In the present application, in S1, the elevator running rules include the call frequency, stopping times, running conditions and special events of each floor every day.

[0038] In the present application, in S1, the historical operation database is specifically constructed to record the elevator operation rules in the last month, including the elevator call frequency of each floor in different time periods, the actual elevator stopping times, the number of waiting passengers based on real-time monitoring, the door opening and closing time to the second level, the car full load rate monitored by the pressure sensor, and the marking of special events; the passenger demand profile is also included in the historical database.

[0039] In the present application, the passenger demand profile records the space demand characteristics of different passengers, including the average standing area and luggage carrying mode.

[0040] It should be noted that the core of the present scheme is to comprehensively obtain and continuously update the operation rules of the elevator, forming a complete historical operation database, which records detailed operation data of each floor in different time periods in the last month, including the elevator call frequency of each floor in different time periods, the actual elevator stopping times, the number of waiting passengers based on real-time monitoring, the door opening and closing time to the second level, the car full load rate monitored by the pressure sensor, and the marking of special events such as large meeting dispersal and lunch peak, and the passenger demand profile is also introduced in the database. This profile records and summarizes the space demand characteristics of different passengers, such as their average standing area and common luggage carrying mode, through long-term observation and machine learning, thereby laying a foundation for accurate space calculation.

[0041] This step is based on big data analysis and machine learning, through time series analysis, clustering and pattern recognition of massive historical data, the rules of elevator operation are extracted, for example, the system can learn that the probability of calling an elevator from the conference room floor to the lobby is as high as 70% at 10:00-10:15 every Wednesday, and that passengers carrying suitcases usually need 1.5 times the standard standing space.

[0042] The database constructed based on S1 provides data support for the training and learning of the digital twin model in S3, so that the prediction of the call probability can be accurately calculated, and through the cooperation of the passenger space demand profile and the subsequent real-time data stream, the system can predict the number of people taking the elevator on each floor in the next five minutes, and can further predict the demand for the card based on this, so as to perform optimal scheduling. At the same time, the feedback data of each scheduling result will flow back to this historical database for continuous correction and enrichment of the model, forming a positive cycle that becomes smarter with use.

[0043] S2: Real-time capture of the on-site state through the monitoring system, and generation of real-time data stream according to time sequence;

[0044] In a preferred embodiment of the present application, in S2, the monitoring system adopts a multi-spectral depth camera array to form a heterogeneous sensor network, calculates the effective standing area of each passenger and the real-time dynamic remaining area in the elevator car through real-time point cloud modeling and human body contour recognition algorithm.

[0045] In a preferred embodiment of the present application, in S2, the field state includes the real-time crowd distribution of each floor, the real-time gathering situation of the elevator entrance, and the passenger carrying state inside the elevator, and a timestamp is generated for each detected passenger to record the time of entering the monitoring area. The specific capture of the field state is to analyze the crowd density and space demand of each floor elevator entrance in real time through a depth camera and computer vision algorithm, to dynamically calculate the passenger capacity and the remaining effective area through the infrared array sensor inside the car, and the elevator control system to transmit the call instruction and the car selection instruction in real time.

[0046] It should be noted that in step S2, the heterogeneous sensor network formed by the multi-spectral depth camera array uses spectral sensing technology to realize real-time sensing and quantization of the dynamic changes of the physical world, and then accurately measures the multi-spectral depth camera array. The heterogeneous sensor network formed by the array is specifically to deploy sensors in the waiting area of each floor and inside the elevator car, generate high-precision environment point cloud data by emitting and receiving infrared light or laser. The system processes the point cloud data in real time, uses computer vision detection and deep learning algorithm for human body contour recognition and tracking, so as to accurately calculate the three-dimensional space occupation of each passenger, that is, the effective standing area.

[0047] At the same time, the infrared array sensor inside the car continuously scans the internal space, calculates the total volume of the carried passengers and compares it with the fixed volume model of the car to dynamically calculate the current remaining effective area. The whole monitoring process not only captures static space information, but also records dynamic changes in time sequence. Each passenger entering the monitoring area is assigned a precise timestamp, so as to accurately record the waiting time; the elevator control system transmits all call and internal selection instructions in milliseconds. All these data streams, including the real-time distribution and density of the crowd on each floor, the instantaneous gathering state of the elevator entrance, the passenger load in the car, and each user request are synchronized and integrated to form a continuous, time-labeled real-time data stream.

[0048] Through the real-time sensing capability provided by the heterogeneous sensor network formed by the multi-spectral depth camera array in step S2, the data basis for the prediction of the subsequent digital twin model is provided, so that the digital twin model in step S3 can make accurate prediction for the near future according to the real-time data. The real-time data stream generated in S2 and the final execution result are fed back to the historical database again, providing a large amount of data for the self-learning of the system.

[0049] S3: Building a digital twin model through historical database and real-time data stream to predict elevator call probability and congestion risk of each floor;

[0050] In a preferred embodiment of the present application, in S3, the digital twin system builds a computable model including mapping of building structure, elevator physical characteristics and traffic flow patterns, and analyzes periodicity rules in historical data through LSTM neural network, and dynamically corrects according to real-time data stream.

[0051] In a preferred embodiment of the present application, in S3, the elevator call probability prediction is based on periodicity trend of historical data and fused with real-time space demand characteristics; the congestion risk is determined by real-time crowd gathering degree of current floor, total space demand and prediction based on historical data, and an waiting time anxiety factor is introduced.

[0052] It should be noted that in step S3, based on the data basis of the previous two steps, a digital twin is built to simulate, deduce and predict future elevator traffic conditions, a virtual environment is mapped through the digital twin system, which replicates the physical world, including the floor structure of the building, the physical characteristics of the elevator, i.e. the running speed, acceleration and car capacity, and the behavior patterns of people flow traffic. Then, through the system, the historical long-period data accumulated in step one is analyzed by LSTM (Long Short Term Memory) neural network for time series analysis, so as to deeply analyze and learn the periodicity and trend rules, such as fixed people flow tide in the morning peak every day, off-site mode in the evening every Friday, etc.

[0053] And the digital twin model built receives and fuses real-time data stream generated in step two in real time, and dynamically corrects and calibrates, which makes the prediction no longer purely rely on historical experience, but combines the current actual situation to correct and calibrate, for example, in the prediction process of elevator call probability, it is the fusion product of historical periodicity rules and real-time space demand characteristics (such as a large number of passengers carrying luggage have gathered in a certain floor), then after detecting the real-time space demand characteristics, scheduling is performed to avoid congestion in sudden situations (which cannot be predicted by periodicity rules). Through the evaluation of congestion risk, a more comprehensive judgment is made, considering the real-time crowd gathering degree of the current floor, the total space demand of this crowd, and whether more people will rush in based on historical data prediction in this period, and on this basis, an waiting time anxiety factor is introduced to convert the waiting time of passengers into a quantitative risk value. The longer the waiting time, the higher the risk coefficient, the stronger the scheduling urgency.

[0054] The prediction capability provided by step S3 enables the elevator to take proactive action, after predicting where high demand will occur, the system instructs the elevator to go to standby in advance, and the output result fuses the risk index in the space and time dimensions, thereby ensuring the accuracy and humanization of the scheduling strategy, achieving the effect of efficiently evacuating people and prioritizing the problem of passengers waiting for a long time.

[0055] S4: predicting the call probability of each floor in the next five minutes based on the digital twin model, calculating the congestion risk, generating a real-time demand heat map that fuses the space-time dimension, and intelligently scheduling the elevator through the heat map.

[0056] In a preferred embodiment of the present application, in S4, the real-time demand heat map calculates the heat score of each floor through multi-dimensional comprehensive calculation, wherein the multiple dimensions include call probability, real-time congestion coefficient, area adaptation index, and anxiety factor, specifically, the average number of calls in the past month, the real-time congestion coefficient is the current waiting number, the area adaptation index is specifically the idle or nearly idle elevator queue, the ratio of the theoretical remaining space of each elevator after arriving at the floor to the space required by the waiting crowd, and the anxiety factor is adjusted based on the waiting time, and the longer the waiting time, the greater the anxiety factor.

[0057] In a preferred embodiment of the present application, in S4, intelligent scheduling is scheduling the elevator according to the heat level, specifically, starting a 10-second countdown after the elevator task queue is empty, if there is a new call during the countdown, interrupting the standby response to new demand, and when the countdown is over, querying the heat map to get the highest score floor, and instructing the elevator to move to the target floor standby, in the scheduling process, under the premise of ensuring waiting time priority, optimizing the space adaptation to improve the passenger carrying efficiency, when the body size difference exceeds 40% and the space utilization rate can be improved by more than 30%, the order is allowed to be adjusted.

[0058] It should be noted that through the prediction result output based on the digital twin model, a fusion calculation model based on rules and priority sorting is used to synthesize the final decision basis, i.e., the real-time demand heat map, which fuses space, time, history, and real-time state, and comprehensively judges the demand of each floor through four dimensions of call probability based on historical data, real-time congestion coefficient reflecting current crowd pressure, area adaptation index calculating the matching degree of capacity and demand, and anxiety factor quantifying passenger waiting anxiety, wherein the area adaptation index dynamically evaluates the idle or soon-to-be-idle elevator, the ratio of its theoretical remaining space after arriving at a floor to the total space demand of the waiting crowd at the floor, thereby introducing accurate space calculation capability in decision-making to ensure that the dispatched elevator can be accommodated, and the anxiety factor ensures that passengers waiting for a long time can be responded to in priority.

[0059] In the elevator dispatching process, the heat map is updated in real time. When an elevator finishes a task, it does not stay in place, but starts a short 10-second countdown window, during which it is ready to respond to sudden calls. Once the countdown is over, the elevator automatically enters the "waiting for call state" and immediately queries the heat map, actively driving to the most urgent (i.e. highest heat score) floor. In the path planning for responding to multiple requests, the system prioritizes "waiting time" to ensure fairness, but also starts the "space adaptation optimization" logic: under the premise of not skipping long waiting passengers, if adjusting the service order (i.e. giving priority to two smaller passengers over one 40% larger passenger) can improve elevator space utilization by more than 30%, the system will intelligently fine-tune the loading order to maximize efficiency.

[0060] This allows the elevator dispatching to take proactive action, instructing the elevator to go to predicted high-demand points in advance, and adding space and anxiety factors to the heat score, allowing the use of every inch of car space and the consideration of passenger waiting emotions in the dispatching process. The actual effect after each dispatch is fed back to the historical database and digital twin model (S3) as new data, driving the model to iterate and continuously learn and optimize the entire system. Therefore, S4 is the final and most important step in translating intelligent algorithms into tangible user value.

[0061] The real-time demand heat map calculates the heat score of each floor through a multi-dimensional fusion calculation model based on rules and priority sorting. Specifically, it assigns values to four dimensions: call probability, real-time congestion coefficient, area adaptation index, and anxiety factor, and sorts them according to the assigned values.

[0062] The emergency response layer, i.e. the anxiety factor, sets the maximum tolerance threshold for passenger waiting time. As the waiting time increases, the heat value increases. When it reaches the maximum tolerance threshold, the heat value is marked as the highest level. The space emergency layer, i.e. the area adaptation index, calculates the remaining space in real time and predicts the remaining space of the next floor. As the space decreases, the heat value increases. When the remaining space cannot accommodate the crowd waiting at the floor, the heat value is maximum. The crush warning layer, i.e. the real-time congestion coefficient, calculates the current waiting number in real time and compares it with the historical data average. When the real-time waiting number exceeds the historical data average, the heat value increases. The prediction busy layer, i.e. the call probability, marks the heat value as the basic busy level if the call frequency in the past 5 minutes exceeds the daily average of the same period. The regular layer, i.e. no above situation, uses the historical call probability to go to the floor with the highest call probability in the next five minutes. In this way, when multiple rules are triggered at the same floor, only the score corresponding to the highest priority is taken. The heat of each floor is calculated through a fusion calculation model based on rules and priority sorting to dispatch the elevator.

[0063] A multi-elevator linkage dynamic response intelligent scheduling system according to the intelligent scheduling method of claims 1-8, comprising a data acquisition module, an analysis module, and a scheduling module, characterized in that the data acquisition module acquires and transmits data through a multi-spectral depth camera array, an infrared sensor network, a car pressure sensor, and a central server arranged in each floor waiting hall and inside the elevator car; the analysis module is used to map the digital twin model synchronized with the physical entity, real-time receives and fuses the real-time data stream from the data acquisition layer, dynamically calibrates and corrects the prediction; the scheduling module is used to generate a demand heat map; the demand heat map is analyzed through four dimensions of call probability, congestion coefficient, area adaptation index, and anxiety factor.

[0064] It should be noted that the multi-spectral depth camera array is deployed above the entrance of each elevator hall and inside the top of each elevator car. The array is not a single imaging device, but integrates visible light (RGB) image sensors, near-infrared (NIR) image sensors, and depth information capture modules. Its working wavelength range is carefully designed to cover 400-700 nm of the visible light spectrum for capturing color information and subtle features visible to the human eye. At the same time, it also covers 800-950 nm of the near-infrared spectrum to ensure stable image data acquisition under insufficient or complex lighting conditions and to detect certain non-visible light features. The depth information capture capability is a key component of the array, which is based on structured light or time-of-flight and can provide accurate three-dimensional coordinate data of objects in the scene relative to the sensor. For example, a sensor using the ToF principle can meet the fine identification needs of passenger position and posture in the elevator scene. Through the built-in high-performance image processing unit (such as a high-performance image processing unit with embedded GPU or NPU), the array can perform real-time, edge-side analysis on the simultaneously captured visible light images, near-infrared images, and three-dimensional depth point cloud data. Through the system, the number of passengers in the waiting area or inside the car, the accurate three-dimensional position of each passenger in space, their height, gait features, current moving direction, stay time in a certain area, and most importantly, the face orientation information can be obtained in real time and accurately. Among them, by capturing facial information, the passenger's attention focus is determined to evaluate the passenger's waiting emotion. For example, passengers who frequently look at the elevator direction or call button may have a relatively higher level of anxiety.

[0065] Further, the infrared array sensor is a non-contact, privacy-friendly sensing means integrated under the floor of the elevator hall waiting area and at the car entrance. The sensor array uses thermal imaging principles to sense the temperature distribution of objects by measuring the infrared radiation emitted by their surfaces. It is composed of several thermopile sensor elements, each with a field of view angle of about 60°, allowing the entire array to cover a wide area with a total measurement range of -20°C to 120°C and a temperature resolution of up to 0.1°C, ensuring accurate capture of small changes in human body temperature. Through real-time analysis of thermal radiation images, the infrared array sensor can accurately identify and count human targets in the waiting area and inside the car without contact, and evaluate their local heat density distribution. Local heat density distribution information is a key indicator of crowding: high heat density areas usually mean high personnel concentration. By comparing with the building space model, the crowding level of waiting or boarding can be judged, and it is an important input parameter for congestion risk assessment. For example, when the average heat radiation intensity per unit area exceeds the preset threshold, it indicates that there is a high risk of congestion in that area.

[0066] A pressure sensor array is embedded under the floor of each elevator car to provide fine perception of the load distribution and dynamic changes inside the car. The array is usually composed of several piezoelectric film sensor units with a response frequency of up to 100 Hz and a maximum bearing pressure of 200 kg / cm2. This high-density, high-sensitivity array can detect the weight distribution on the car floor and its dynamic changes in real time, such as the standing position, movement trajectory and even minor adjustments of the passenger's body posture. By accumulating and spatially analyzing the data of each sensor unit, the system can accurately calculate the total load inside the car, the passenger's center of gravity distribution, and the estimated weight of individual passengers. The estimated weight of individual passengers can be obtained by subtracting the car's self-weight from the total load, combined with the number of passengers and approximate volume information provided by the multi-spectral depth camera array. These information are used together to assess the actual passenger capacity, space utilization and passenger body posture changes inside the car, such as whether the passenger is moving or shifting the center of gravity in the car. Such information can indirectly reflect the comfort of passengers, for example, feeling uncomfortable in a small space and trying to adjust their posture.

[0067] Embodiment one

[0068] Through the deployment of sensor networks and elevator control systems on each floor, a historical operation database is continuously collected and constructed. This database not only records in detail the call frequency, actual stop times, precise door opening and closing times, full load rate and special events in different time periods and directions within the past month, but also innovatively establishes a passenger space demand profile. Through machine learning analysis of different passengers' regular standing area and luggage carrying habits, a solid foundation is laid for accurate space calculation.

[0069] Through the multi-spectrum depth camera array deployed in the waiting hall and the car, the system captures the real-time state and generates a time-stamped real-time data stream. Through point cloud modeling and computer vision algorithms, the system accurately calculates the effective standing area of each passenger and the real-time remaining effective area of the car, uploads the data, and the digital twin engine integrates historical rules and real-time data streams. Through LSTM neural network, the system predicts the call probability and congestion risk of each floor in the next five minutes, which takes into account real-time space demand characteristics and waiting time anxiety factors, and dynamically corrects sudden situations.

[0070] Based on the prediction results, the system generates a real-time demand heat map in space-time dimension. The heat map integrates historical call probability, real-time congestion coefficient, area adaptation index, and anxiety factor through a fusion calculation model based on rules and priority sorting to generate a dynamic heat score for each floor. When the elevator task queue is empty, the system starts a 10-second buffer countdown, during which it responds to sudden calls at any time. After the countdown ends, the elevator no longer idles and actively drives to the floor with the highest heat score. When performing multiple request services, the system strictly follows the "waiting time priority" fairness principle and intelligently starts the space adaptation optimization logic. Under the premise of ensuring not to skip passengers waiting for a long time, if adjusting the service order (such as allowing two smaller passengers to take the elevator first) can improve elevator space utilization by more than 30%, the system will fine-tune to maximize capacity.

[0071] Through the system, passive response is transformed into proactive predictive service, and the elevator is directed to predicted high-demand points in advance, achieving an optimized state of "elevator waiting people", significantly reducing waiting time. At the same time, the dispatching strategy that integrates space and anxiety factors not only scientifically utilizes car space but also humanizes passenger experience. Each dispatching result is fed back to the database, enabling the system to continuously learn and optimize the prediction model and decision algorithm, ultimately forming a virtuous cycle of increasing intelligence, redefining the vertical transportation efficiency and service quality of high-rise buildings.

[0072] Based on the ideal embodiments of the present application, the above description, relevant personnel can make various changes and modifications without deviating from the scope of the present application. The technical scope of the present application is not limited to the contents of the specification, and must be determined by the scope of the claims.

Claims

1. A dynamic response intelligent scheduling method for multi-elevator linkage, characterized in that: The following steps are involved: S1: Real-time collection of multi-dimensional operation data of elevators, forming a historical database with timestamps and spatial coordinates and updating it in real time; S2: Capture the on-site status in real time through the monitoring system and generate real-time data stream according to the time series; S3: Build a digital twin model using historical databases and real-time data streams to predict elevator call probability and congestion risk for each floor; S4: Based on the digital twin model, the elevator call probability for each floor in the next five minutes is predicted and the congestion risk is calculated. A real-time demand heat map that integrates spatial and temporal dimensions is generated, and elevators are dispatched based on the heat map.

2. The method for dynamic response intelligent scheduling of multiple elevator linkage according to claim 1, characterized in that: In S1, the multi-dimensional operation data includes the daily elevator call frequency, number of stops, operation status and special events on each floor; the construction of the historical operation database is specifically to digitize the elevator operation patterns in the past month, including the elevator call frequency in the up and down directions of each floor in different time periods, the actual number of elevator stops, the number of people waiting for the elevator based on real-time monitoring, the time taken to open and close the door accurate to the second, the car load rate monitored by the pressure sensor, and the marking of special events; the historical database also includes passenger demand files; the passenger demand files record the space demand characteristics of different passengers, including the average standing area and luggage carrying mode.

3. The method for dynamic response intelligent scheduling of multiple elevator linkage according to claim 1, characterized in that: In S2, the monitoring system uses a multispectral depth camera array to form a heterogeneous sensor network, and calculates the effective standing area of ​​each passenger and the real-time dynamic remaining area in the elevator car through real-time point cloud modeling and human contour recognition algorithm.

4. The method for dynamic response intelligent scheduling of multiple elevator linkage according to claim 3, characterized in that: In S2, the on-site status includes the real-time crowd distribution on each floor, the real-time gathering situation at the elevator entrance, and the passenger status inside the elevator. A timestamp is generated for each detected passenger to record the time when the passenger enters the monitoring area. The capture of the on-site status is specifically to analyze the crowd density and space requirements at the elevator entrance on each floor in real time through a depth camera and a computer vision algorithm, and dynamically calculate the passenger capacity and the remaining effective area through the infrared array sensor in the car. The elevator control system transmits the elevator call command and the floor selection command in the car in real time.

5. The method for dynamic response intelligent scheduling of multiple elevator linkage according to claim 1, characterized in that: In the S3, the digital twin system constructs a computable model that maps the building structure, elevator physical characteristics and traffic flow patterns, and uses an LSTM neural network to analyze the periodic patterns in historical data, while making dynamic corrections based on real-time data streams.

6. The method for dynamic response intelligent scheduling of multiple elevator linkage according to claim 1, characterized in that: In S3, the elevator call probability prediction is specifically based on the periodic regular trend of historical data and the integration of real-time space demand characteristics; The congestion risk is determined by the real-time crowd concentration on the current floor, the total space demand, and predictions based on historical data, and a waiting time anxiety factor is introduced.

7. The method for dynamic response intelligent scheduling of multiple elevator linkage according to claim 1, characterized in that: In said S4, the real-time demand heat map adopts a fusion calculation model based on rules and priority sorting to calculate the heat score of each floor by comprehensively considering four dimensions: elevator call probability, real-time congestion, area adaptation index and anxiety factor. The elevator call probability is specifically the average number of elevator calls for the floor in the past month, the real-time congestion coefficient is the current number of waiting people, the area adaptation index is specifically the current idle or nearly idle elevator queue, the ratio of the theoretical remaining space after each elevator arrives at the floor to the space required by the waiting crowd, and the anxiety factor is adjusted based on the waiting time. The longer the waiting time, the greater the anxiety factor.

8. The method for dynamic response intelligent scheduling of multiple elevator linkage according to claim 1, characterized in that: In S4, intelligent scheduling is the system scheduling elevators according to the thermal level. Specifically, when the elevator task queue is cleared, a 10-second countdown is started. If there is a new call during this period, the waiting call is interrupted to respond to the new demand. When the countdown ends, the thermal map is queried to obtain the highest-scoring floor, and the elevator is instructed to move to the target floor and wait for the call. During the scheduling process, under the premise of ensuring waiting time priority, the passenger carrying efficiency is improved through space adaptation optimization. When the body size difference exceeds 40% and the space utilization rate can be improved by more than 30%, the order is allowed to be adjusted.

9. A dynamic response intelligent dispatching system for multiple elevators, comprising the intelligent dispatching method according to claims 1-8, comprising: The data acquisition module, analysis module and intelligent scheduling module are characterized in that the data acquisition module collects and transmits data through a multi-spectral depth camera array, an infrared sensor network, a car pressure sensor and a central server respectively installed in the elevator lobby on each floor of the building and inside the elevator car; The analysis module is used to map out a digital twin model that is mapped and evolved synchronously with the physical entity, receive and integrate real-time data streams from the data acquisition layer in real time, and dynamically calibrate and correct the predictions; The scheduling module is used to generate a demand heat map.

10. The multi-elevator linkage dynamic response intelligent dispatching system according to claim 9, characterized in that: The demand heat map is analyzed through four dimensions: elevator call probability, congestion coefficient, area adaptation index and anxiety factor.

Images