Elevator dynamic dispatching control method and system based on passenger flow prediction

Abstract

The application relates to an elevator dynamic scheduling control method and system based on passenger flow prediction, which comprises the following steps: obtaining an intention pulse signal generated when a terminal accessing a preset physical area interacts with an elevator in the physical area, and constructing an intention pulse signal model based on the intention pulse signal; determining a space congestion coefficient of the physical area according to the effective access quantity of the terminal; performing dynamic reconstruction on a preset traffic delay distribution model based on the space congestion coefficient; generating a passenger flow arrival probability curve according to the intention pulse signal model and the dynamically reconstructed preset traffic delay distribution model; and controlling the corresponding elevator to go to a target landing position through the passenger flow arrival probability curve. The application can solve the problem that the existing elevator scheduling method uses a static traffic delay time, which cannot adapt to the dynamic change of the congestion degree of the physical space, and invalid elevator scheduling is prone to occur.

Description

Technical Field

[0001] This application relates to the field of elevator scheduling technology, and in particular to a method and system for dynamic elevator scheduling control based on passenger flow prediction. Background Technology

[0002] With the increasing size and density of modern high-rise buildings, the scheduling efficiency of elevator group control directly impacts building operational efficiency and energy consumption. Conventional elevators rely on passengers operating physical call buttons in the waiting hall, with the elevators responding passively. However, this scheduling model easily leads to delayed response, excessively long waiting times, and localized congestion during peak hours. Therefore, accurately anticipating passenger demand and proactively scheduling elevators is a key area for improvement in this field.

[0003] Existing technical solutions trigger elevator pre-scheduling through passenger interaction with a network terminal. This solution sets the travel delay time from the passenger's location at the time of triggering pre-scheduling to the waiting hall as a static prediction parameter, and then generates a pre-schedule dispatch instruction based on this parameter. However, this solution ignores the impact of physical space congestion on passenger walking speed. When physical space congestion increases, the actual travel delay time will deviate significantly from the aforementioned static prediction parameter, leading to inaccurate elevator predictions of passenger arrival times in the waiting hall and resulting in ineffective elevator scheduling.

[0004] The existing elevator scheduling method uses a static passage delay time, which makes it unable to adapt to the dynamic changes in the congestion of physical space and easily leads to ineffective elevator scheduling. There is currently no effective solution to this problem. Summary of the Invention

[0005] This application provides an elevator dynamic scheduling and control method and system based on passenger flow prediction to solve the technical problem that existing elevator scheduling methods use static passage delay time, which makes them unable to adapt to the dynamic changes in the degree of congestion in physical space and easily leads to ineffective elevator scheduling.

[0006] In the first aspect, this application provides an elevator dynamic scheduling and control method based on passenger flow prediction, including:

[0007] Acquire the intent pulse signal generated when a terminal within a preset physical area interacts with an elevator within the physical area, and construct an intent pulse signal model based on the intent pulse signal;

[0008] The intention pulse signal model represents the distribution of the elevator-riding intention of the passenger at discrete time points;

[0009] Based on the number of valid accesses of the terminals, the spatial congestion coefficient of the physical area is determined; based on the spatial congestion coefficient, a preset passage delay distribution model is dynamically reconstructed.

[0010] The preset passage delay distribution model represents the probability distribution of the time required for the passenger to reach the target elevator location;

[0011] Based on the intentional pulse signal model and the dynamically reconstructed preset passage delay distribution model, a passenger flow arrival probability curve is generated;

[0012] Based on the passenger flow arrival probability curve, the corresponding elevator is controlled to go to the target elevator location.

[0013] In some embodiments, acquiring the intent pulse signal generated when a terminal within a preset physical area interacts with an elevator within that physical area, and constructing an intent pulse signal model based on the intent pulse signal, includes:

[0014] Extract the trigger timestamps corresponding to the events representing the intention to ride the elevator from the log data of the interaction between each terminal and the elevator;

[0015] Based on the trigger timestamp, generate the corresponding intent pulse signal;

[0016] The intention pulse signals within a preset time period are deduplicated; the deduplicated intention pulse signals are then combined to construct the intention pulse signal model.

[0017] In some embodiments, determining the spatial congestion coefficient of the physical area based on the number of valid accesses of the terminals includes:

[0018] Obtain the interaction traffic between each terminal and the elevator;

[0019] Based on the interaction traffic, determine the active terminals among all the terminals;

[0020] The number of active terminals accessing the network is taken as the number of valid accesses.

[0021] The spatial congestion coefficient of the physical area is determined based on the number of valid accesses and the preset baseline number of terminals.

[0022] In some embodiments, the dynamic reconstruction of the preset passage delay distribution model based on the spatial congestion coefficient includes:

[0023] Obtain the baseline mathematical expectation parameter and baseline standard deviation parameter of the preset passage delay distribution model;

[0024] Wherein, the benchmark mathematical expectation parameter characterizes the average walking time of the passenger to the target elevator position under ideal empty-load conditions; the benchmark standard deviation parameter characterizes the degree of dispersion of time distribution caused by individual walking speed differences under ideal empty-load conditions.

[0025] Based on the spatial congestion coefficient, the baseline mathematical expectation parameter and the baseline standard deviation parameter are simultaneously amplified;

[0026] The preset passage delay distribution model is dynamically reconstructed based on the synchronously amplified benchmark mathematical expectation parameters and benchmark standard deviation parameters.

[0027] In some further embodiments, the method further includes:

[0028] Collect the actual walking delay time of the person taking the elevator;

[0029] The actual walking delay time is linked to the spatial congestion coefficient of the physical area when the intention to take the elevator is triggered, and training samples are constructed.

[0030] Based on the training samples, the baseline mathematical expectation parameter and the baseline standard deviation parameter of the preset traffic delay distribution model are subjected to closed-loop adaptive calibration.

[0031] In some embodiments, generating a passenger flow arrival probability curve based on the intention impulse signal model and the dynamically reconstructed preset passage delay distribution model includes:

[0032] In the intent pulse signal model, the intent pulse signal whose generation time is within the preset maximum retention time is retained to obtain the updated intent pulse signal model;

[0033] Based on a preset time sampling step, spatiotemporal convolution operations are performed on the updated intention pulse signal model and the dynamically reconstructed preset passage delay distribution model to generate the passenger flow arrival probability curve.

[0034] In further embodiments, the step of performing spatiotemporal convolution operations on the updated intent pulse signal model and the dynamically reconstructed preset passage delay distribution model based on a preset time sampling step size to generate the passenger flow arrival probability curve includes:

[0035] Based on the preset time sampling step size, the time axis corresponding to the updated intention pulse signal model is divided into multiple time slice nodes;

[0036] Determine the generation time corresponding to each retained intention pulse signal; determine the time difference between each generation time and each time slice node;

[0037] The preset passage delay distribution model is dynamically reconstructed by inputting the time difference, and the arrival probability of each intention pulse signal at the branch corresponding to the time slice node is determined accordingly.

[0038] The arrival probabilities of each branch are accumulated to generate the passenger flow arrival probability curve.

[0039] In some embodiments, controlling the corresponding elevator to proceed to the target elevator location based on the passenger flow arrival probability curve includes:

[0040] The passenger flow arrival probability curve is integrated over a preset look-ahead time window to obtain the area of ​​integration.

[0041] When the integral area exceeds a preset scheduling trigger threshold and no physical call signal is detected at the target elevator location, the target elevator is determined from all the elevators associated with the target elevator location.

[0042] Control the target elevator to proceed to the target elevator location.

[0043] In some further embodiments, determining the target elevator among all elevators associated with the target elevator location includes:

[0044] Identify all available elevators associated with the target elevator location;

[0045] The scheduling cost for all available elevators is calculated using the cost function of the integral area.

[0046] The available elevator with the lowest scheduling cost is identified as the target elevator.

[0047] In some further embodiments, the method further includes:

[0048] Monitoring is performed on the terminals accessing the physical area;

[0049] When the access status of the terminal changes and the signal strength of the terminal meets the preset time condition and the preset threshold condition, the integral area is updated according to the intention pulse signal corresponding to the terminal.

[0050] The target elevator is redefined based on the updated integral area;

[0051] If the re-determined target elevator is different from the previously determined target elevator, the scheduling of the previously determined target elevator shall be cancelled.

[0052] In further embodiments, when the access state of the terminal changes and the signal strength of the terminal meets a preset time condition and a preset threshold condition, updating the integral area according to the intent pulse signal corresponding to the terminal includes:

[0053] When the access location of the terminal is detected to change from within the physical area to outside the physical area, it is determined that the access status of the terminal has changed;

[0054] When it is detected that the signal strength of the terminal continuously increases within a preset time period, and the increased signal strength reaches the preset threshold, it is determined that the signal strength of the terminal satisfies the preset time condition and the preset threshold condition.

[0055] Determine the area of ​​contribution of the intention pulse signal corresponding to the terminal to the integral area;

[0056] The contributing area is removed from the integral area to obtain the updated integral area.

[0057] Secondly, this application provides an elevator dynamic scheduling and control system based on passenger flow prediction, including: network facilities, elevator system and control facilities;

[0058] The network infrastructure is used to provide intent pulse signals generated when interacting with elevators within a preset physical area;

[0059] The elevator system is communicatively connected to the control facility;

[0060] The control facility is communicatively connected to the network facility and is used to perform the method described in any one of the first aspects.

[0061] Compared with existing related technologies, the embodiments of this application have the following beneficial effects:

[0062] This application embodiment acquires intention pulse signals from elevator interactions within a preset physical area and constructs an intention pulse signal model. This model can detect elevator demand before physical call signals are generated, based on the distribution of passenger intentions at discrete time points. It determines the spatial congestion coefficient by the number of terminal connections within the physical area and dynamically reconstructs a preset travel delay distribution model using this coefficient. This allows for dynamic adjustment of the travel delay distribution model as spatial congestion levels change. By fusing the intention pulse signal model with the dynamically reconstructed preset travel delay distribution model, the generated passenger arrival probability curve reflects the probability distribution of passenger flow reaching the target elevator location at various future time points. Therefore, the passenger arrival probability curve allows for accurate allocation of target elevators and control for them to travel to the target elevator location at appropriate scheduling times, thereby reducing ineffective elevator scheduling.

[0063] Details of one or more embodiments of this application are set forth in the following drawings and description to make other features, objects and advantages of this application more readily apparent. Attached Figure Description

[0064] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings:

[0065] Figure 1 This is a flowchart of an elevator dynamic scheduling and control method based on passenger flow prediction provided in an embodiment of this application;

[0066] Figure 2 This is a flowchart illustrating the construction of an intent pulse signal model based on an intent pulse signal according to an embodiment of this application;

[0067] Figure 3 This is a flowchart of determining the spatial congestion coefficient of a physical area according to an embodiment of this application;

[0068] Figure 4 This is a flowchart of a dynamic reconstruction preset traffic delay distribution model provided in an embodiment of this application;

[0069] Figure 5 This is a flowchart of generating a passenger flow arrival probability curve provided in one embodiment of this application;

[0070] Figure 6 This is a flowchart illustrating the generation of a passenger flow arrival probability curve according to an embodiment of this application;

[0071] Figure 7 This is a flowchart of scheduling corresponding elevators based on passenger flow arrival probability curves provided in one embodiment of this application;

[0072] Figure 8 This is a flowchart of determining a target elevator provided in an embodiment of this application;

[0073] Figure 9 This is a flowchart of elevator dispatch secondary confirmation provided in one embodiment of this application;

[0074] Figure 10 This is a block diagram of an elevator dynamic scheduling and control system based on passenger flow prediction provided in one embodiment of this application;

[0075] Figure 11 This is a hardware structure block diagram of the terminal of the elevator dynamic scheduling and control method based on passenger flow prediction provided in an embodiment of this application.

[0076] In the diagram: 102, processor; 104, memory; 106, transmission device; 108, input / output device. Detailed Implementation

[0077] To better understand the purpose, technical solution, and advantages of this application, the application is described and illustrated below in conjunction with the accompanying drawings and embodiments.

[0078] Unless otherwise defined, the technical or scientific terms used in this application shall have the general meaning understood by one of ordinary skill in the art to which this application pertains. Words such as “a,” “an,” “an,” “the,” “the,” and “these” used in this application do not indicate quantitative limitation and may be singular or plural. The terms “comprising,” “including,” “having,” and any variations thereof used in this application are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or device that comprises a series of steps or modules (units) is not limited to the listed steps or modules (units) but may include steps or modules (units) not listed, or may include other steps or modules (units) inherent to these processes, methods, products, or devices. Words such as “connected,” “linked,” and “coupled” used in this application are not limited to physical or mechanical connections but may include electrical connections, whether direct or indirect. “Multiple” used in this application refers to two or more. “And / or” describes the relationship between related objects, indicating that three relationships may exist; for example, “A and / or B” can represent: A alone, A and B simultaneously, and B alone. Normally, the character " / " indicates that the objects before and after it are in an "or" relationship. The terms "first," "second," "third," etc., used in this application are merely to distinguish similar objects and do not represent a specific order.

[0079] This application can be applied to scenarios such as high-rise buildings and commercial complexes. In these scenarios, existing elevator dispatching systems generally cannot dynamically adjust the travel delay time for people to reach their target elevator position based on changes in the level of crowding within the physical area. When the population density in the physical area increases, the actual walking speed of people decreases, causing existing solutions to significantly lose accuracy when estimating the arrival time of people at the target elevator position based on a fixed travel delay time. Therefore, when dispatching elevators based on the above solutions, elevators may arrive at the target elevator position too early or leave prematurely, resulting in ineffective dispatching.

[0080] This embodiment provides a dynamic elevator scheduling and control method based on passenger flow prediction. Figure 1 This is a flowchart of the elevator dynamic scheduling and control method based on passenger flow prediction in this embodiment, as shown below. Figure 1 As shown, the process includes the following steps:

[0081] Step S110: Obtain the intent pulse signal generated when the terminal in the preset physical area interacts with the elevator in the physical area, and construct an intent pulse signal model based on the intent pulse signal.

[0082] Specifically, the intention pulse signal model characterizes the distribution of the elevator-riding intention of an object at discrete time points.

[0083] For example: A physical area can refer to at least a portion of the spatial range within the target floors related to elevator riding behavior, used to determine the actual detection range; for example, based on actual scheduling needs, the entire area of ​​a building can be set as the aforementioned physical area, or several floors of a building can be set as the aforementioned physical area. A terminal within the physical area can refer to a device that interacts directly or indirectly with the elevator within the physical area, such as a mobile device like a phone or tablet. An intent pulse signal can refer to a time stamp formed when an elevator-riding object interacts directly or indirectly with the elevator using a terminal; each time stamp corresponds to a potential elevator riding event. An elevator-riding object refers to the aforementioned person; correspondingly, the elevator-riding object's intent can refer to the elevator-riding object's expectation or tendency to use the elevator. An intent pulse signal model can refer to the set of the aforementioned intent pulse signals; in a real-world scenario, the intent pulse signal model can reflect the time sequence corresponding to the elevator riding intent within a certain period.

[0084] Step S120: Determine the spatial congestion coefficient of the physical area based on the number of valid terminal accesses; and perform dynamic reconstruction on the preset passage delay distribution model based on the spatial congestion coefficient.

[0085] Specifically, a pre-defined passage delay distribution model is used to characterize the probability distribution of the time required for an elevator passenger to reach the target elevator location.

[0086] For example: The effective access quantity can refer to the number of terminals in the target state within a physical area. In specific scenarios, the target state can be set according to actual screening needs, thereby filtering all accessing terminals to obtain the effective access quantity. The spatial congestion coefficient reflects the density of people within a physical area. The preset passage delay distribution model can be a probability distribution model, used to describe the probabilistic pattern of the time taken for a single passenger to reach the target passage location from the location triggering the intention to ride the elevator. Dynamic reconstruction can refer to the real-time adjustment of the model parameters of the preset passage delay distribution model using the spatial congestion coefficient, so that the dynamically reconstructed distribution characteristics reflect the changes in passage delay time under congestion conditions in real time.

[0087] Step S130: Generate a passenger flow arrival probability curve based on the intention pulse signal model and the preset passage delay distribution model after dynamic reconstruction.

[0088] For example: a passenger flow arrival probability curve can refer to a data sequence that changes continuously over time, and its value represents the predicted passenger flow density arriving at the target escalator location.

[0089] Step S140: Based on the passenger flow arrival probability curve, control the corresponding elevator to go to the target elevator location.

[0090] Specifically, this step determines whether the preset triggering conditions are met based on the numerical characteristics of the passenger flow arrival probability curve within a specific time interval. When the aforementioned conditions are met, the corresponding elevator is selected and a dispatching instruction is generated to drive the elevator to move closer to the target elevator location.

[0091] Unlike existing technologies, this embodiment constructs an intention pulse signal model from the elevator-riding intention reflected during terminal-elevator interaction through the above steps. Then, based on the number of valid terminal connections, the congestion level of the physical area is quantified in real time into a spatial congestion coefficient. A preset travel delay distribution model is dynamically reconstructed based on this spatial congestion coefficient, ensuring that the preset travel delay distribution model reflects changes in the congestion level of the physical area. Subsequently, based on the intention pulse signal model and the reconstructed preset travel delay distribution model, a passenger arrival probability curve is generated. This curve reflects the elevator-riding intention and changes in real time with fluctuations in passenger density, thereby accurately predicting the peak arrival times of passenger flow. Based on this, pre-scheduling of the corresponding elevators effectively solves the problem of existing elevator scheduling methods using static travel delay times, which cannot adapt to dynamic changes in physical space congestion and are prone to ineffective elevator scheduling.

[0092] In some of these embodiments, Figure 2 This is a flowchart of constructing an intent pulse signal model based on the intent pulse signal provided in this embodiment. Please refer to it. Figure 2 Step S110, which involves acquiring the intent pulse signal generated when a terminal within the preset physical area interacts with an elevator within the physical area, and constructing an intent pulse signal model based on the intent pulse signal, specifically includes the following steps:

[0093] Step S111: Extract the trigger timestamps corresponding to the events representing the intention to ride the elevator from the log data of the interaction between each terminal and the elevator.

[0094] Step S112: Generate the corresponding intent pulse signal based on the trigger timestamp;

[0095] Step S113: Perform deduplication processing on each intention pulse signal within a preset time period; combine the deduplicated intention pulse signals to construct an intention pulse signal model.

[0096] For example: Log data refers to system records of various interactive behaviors between the elevator user and the elevator system; in real-world scenarios, log data includes network authentication events and access control events. An elevator intention event indicates that the user intends to go to the target elevator location and has directly or indirectly interacted with the elevator using a terminal. A trigger timestamp refers to the timestamp when the elevator intention event occurred. Deduplication refers to retaining only the signals that meet certain conditions from multiple intention pulse signals generated for the same user and the same elevator characteristics within a preset time period.

[0097] This embodiment extracts the trigger timestamps corresponding to elevator intention events from scattered log data through the above steps, and generates discrete intention pulse signals accordingly. After deduplication, the signals are combined to construct an intention pulse signal model, thereby mapping the asynchronous interactive behaviors generated by various terminals in the building onto the same time axis, forming structured data that is convenient for subsequent calculations of passenger flow prediction.

[0098] In some feasible embodiments, the log data includes access control log data and network session log data. The access control log data is used to generate uplink intent pulse signals, and the network session log data is used to generate downlink intent pulse signals.

[0099] For example: When a passage record of an elevator user passing through the ground floor gate is extracted from the access control log data, and this passage record indicates that the user needs to enter other floors from the ground floor, the passage timestamp corresponding to this passage record is used as the trigger timestamp for an upward elevator intention, generating an upward intention pulse signal. When a terminal is detected disconnecting from the network access point corresponding to the office area from the network session log data, or when the signal strength at the network access point continuously weakens and falls below a preset departure threshold, the network action timestamp corresponding to this event is used as the trigger timestamp for a downward elevator intention, generating a downward intention pulse signal. By distinguishing the source of the log data, different elevator intentions can be captured.

[0100] In some feasible embodiments, before extracting the trigger timestamp, the method further includes: performing time base verification on the timestamps of each data source in the log data; when the time deviation of any data source is detected to exceed a preset time deviation threshold, performing offset correction on the timestamp of that data source; if correction cannot be completed, temporarily removing that data source from the generation of intent pulse signals. Here, "temporary removal" as mentioned herein does not mean deleting the data source, shutting down the corresponding device, or permanently abandoning the data from that data source. Rather, it means that when the clock deviation of a data source is detected to exceed a threshold, and its timestamp cannot be corrected to a unified time base through time offset compensation or other methods, the edge gateway suspends the use of newly generated data from that data source to construct intent pulses during the duration of this anomaly, and also prevents this data from participating in subsequent passenger flow prediction, convolution calculation, and scheduling cancellation judgment. Once the data source restores time synchronization or its timestamp can be reliably corrected, the reception and use of data from that data source are resumed.

[0101] For example, all data sources can be synchronized with a time synchronization server within the building. Time benchmark verification ensures that all intent pulse signals are on the same time axis during subsequent spatiotemporal convolution operations and scheduling cancellation decisions, preventing inaccurate passenger flow predictions due to time discrepancies.

[0102] In some feasible embodiments, step S113 further includes the following implementation details:

[0103] During deduplication, the preset time period can range from 30 seconds to 120 seconds, preferably 75 seconds. The qualifying signals are the earliest triggered intent pulse signal generated within the preset time period for the same passenger and in the same direction, or the intent pulse signal with the highest confidence weight. By setting reasonable deduplication periods and retention rules, duplicate counts caused by network roaming jitter or short-term terminal reconnection can be effectively eliminated.

[0104] In some feasible embodiments, the intent pulse signal model is constructed separately according to the escalator direction, including an upward intent pulse signal model and a downward intent pulse signal model; correspondingly, the passenger flow arrival probability curve is also generated separately according to the escalator direction. Integration and scheduling judgment are performed independently for the passenger flow arrival probability curves of different directions.

[0105] For example, the upward intention pulse signal mainly originates from access control events, while the downward intention pulse signal mainly originates from network status change events of office area terminals. By constructing models and generating curves separately for each direction, elevator scheduling decisions can be matched with the actual flow direction of passengers, avoiding interference between passenger flows from different directions.

[0106] In some feasible embodiments, the intent pulse signal model is constructed separately according to the elevator direction, including an uplink intent pulse signal model and a downlink intent pulse signal model, which are mathematically expressed on the system time axis as follows:

[0107] ;

[0108] in, The overall intention is a pulse signal model; This is a model of an uplink intention pulse signal; This is a downlink intention pulse signal model; To set the total number of times the upward elevator intention event is triggered within the observation and calculation period; This is the Dirac function, which takes the value of a unit impulse at a specific timestamp and zero at the rest of the timeline. For the current system, it is a continuous-time variable; For the access control system to trigger the first The trigger timestamp of the upward elevator ride intention event; To set the total number of downhill elevator intention events triggered within the observation and calculation period; Triggered by network session log The trigger timestamps of each downward elevator boarding intention event are used. In this way, the scattered elevator boarding intention events are uniformly abstracted into transient impact quantities discretely distributed on the time axis, providing a standardized input for subsequent spatiotemporal convolution operations.

[0109] In some of these embodiments, Figure 3 This is a flowchart for determining the spatial congestion coefficient of a physical area, provided in this embodiment. Please refer to it. Figure 3 In step S120, the sub-step of determining the spatial congestion coefficient of the physical area based on the number of valid terminal accesses specifically includes the following steps:

[0110] Step S121: Obtain the interaction traffic between each terminal and the elevator;

[0111] Step S122: Based on the interaction traffic, determine the active terminals among all terminals;

[0112] Step S123: The number of active terminals accessing the network is taken as the number of valid accesses.

[0113] Step S124: Determine the spatial congestion coefficient of the physical area based on the number of valid accesses and the preset baseline number of terminals.

[0114] For example: Interaction traffic can refer to the amount of data transmitted between the terminal and the elevator during direct or indirect interaction. In practical scenarios, interaction traffic can be used to measure the activity level of the interaction behavior of the terminal's corresponding elevator user, distinguishing between terminals in active use and terminals silently connected in the background. An active terminal indicates that its corresponding interaction activity level meets the requirements, which can be set according to actual needs.

[0115] This embodiment, through the above steps, filters active terminals from all terminals based on interaction traffic, and uses the number of active terminals accessing the system as the effective access count. This effective access count is then compared with a preset baseline number of terminals to obtain the spatial congestion coefficient. Therefore, the effective access count accurately reflects the number of active users within the physical area.

[0116] In some feasible embodiments, in step S124, the space congestion factor can be determined as follows:

[0117] The space congestion coefficient is obtained by quoting the number of valid accesses with the preset baseline number of terminals.

[0118] Specifically, when the spatial congestion coefficient is less than or equal to 1, it is determined that there is no obvious congestion in the physical area, and the unreconstructed preset passage delay distribution model can be directly used to participate in subsequent calculations; when the spatial congestion coefficient is greater than 1, it is determined that there is physical spatial congestion and dynamic reconstruction is triggered.

[0119] In some feasible embodiments, the space congestion factor is calculated according to the following formula:

[0120] ;in, This refers to the space congestion coefficient. This refers to the number of valid accesses obtained within the current statistical period. The preset baseline number of terminals. If it is determined that there is no obvious congestion in the physical area, the unreconstructed preset passage delay distribution model is directly adopted; When this occurs, it is determined that there is physical space congestion and dynamic reconstruction is triggered.

[0121] In some of these embodiments, Figure 4 This is a flowchart of the dynamic reconstruction of the preset passage delay distribution model provided in this embodiment. Please refer to it. Figure 4 In step S120, the sub-step of dynamically reconstructing the preset traffic delay distribution model based on the spatial congestion coefficient specifically includes the following steps:

[0122] Step S125: Obtain the baseline mathematical expectation parameters and baseline standard deviation parameters of the preset passage delay distribution model;

[0123] Specifically, the baseline mathematical expectation parameter characterizes the average walking time for an individual to reach the target elevator position under ideal empty-load conditions; the baseline standard deviation parameter characterizes the degree of dispersion of time distribution caused by individual walking speed differences under ideal empty-load conditions.

[0124] Step S126: Based on the spatial crowding coefficient, perform synchronous amplification on the baseline mathematical expectation parameter and the baseline standard deviation parameter;

[0125] Step S127: Dynamically reconstruct the preset passage delay distribution model based on the synchronously amplified benchmark mathematical expectation parameters and benchmark standard deviation parameters.

[0126] For example: Synchronous amplification can refer to using the spatial congestion coefficient as an amplification factor, while multiplicatively adjusting the baseline mathematical expectation parameter and the baseline standard deviation parameter. In real-world scenarios, when the population density in a physical area increases, the spatial congestion coefficient increases, and the synchronously amplified baseline mathematical expectation parameter increases accordingly, representing a longer average time required for passengers to walk to their target elevator position in a congested environment. The synchronously amplified baseline standard deviation parameter also increases, representing a more dispersed time for passengers to reach their target elevator position. Dynamic reconstruction can refer to substituting the synchronously amplified baseline mathematical expectation parameter and baseline standard deviation parameter into the original structure of the preset travel delay distribution model to form a new preset travel delay distribution model, enabling this preset travel delay distribution model to reflect changes in travel delay time under the current congestion state in real time.

[0127] This embodiment, through the above steps, synchronously amplifies the baseline mathematical expectation parameter and baseline standard deviation parameter in the preset travel delay distribution model using the spatial congestion coefficient, and dynamically reconstructs the distribution model based on the amplified parameters. Therefore, the reconstructed distribution model can reflect both the overall increase in travel delay caused by congestion and the widening difference in individual arrival times on the time axis, making the generation basis of the passenger flow arrival probability curve more closely reflect the actual traffic conditions within the physical area.

[0128] In some feasible embodiments, during synchronous amplification, the baseline mathematical expectation parameter is linearly amplified according to the spatial crowding coefficient; since the standard deviation is the arithmetic square root of the variance, the baseline standard deviation parameter is amplified according to the square root of the spatial crowding coefficient.

[0129] Specifically, the reconstructed dynamic mathematical expectation parameter is equal to the product of the spatial crowding coefficient and the baseline mathematical expectation parameter, and the reconstructed dynamic standard deviation parameter is equal to the product of the square root of the spatial crowding coefficient and the baseline standard deviation parameter. Through the above amplification method, the reconstructed distribution model can maintain a reasonable quantitative correspondence with the physical crowding degree in both mathematical expectation and discreteness.

[0130] In some feasible embodiments, the preset passage delay distribution model uses a Gaussian distribution for mathematical modeling, and its baseline expression in the unreconstructed state is:

[0131] ;

[0132] in, The default travel delay distribution model is the one that has not been reconstructed. This is a time interval variable representing the distance from the trigger point of the elevator intention event; The baseline mathematical expectation parameter; This is the baseline standard deviation parameter. In actual engineering configurations, the baseline mathematical expectation parameter is set based on the ratio of the physical distance from the specific trigger area to the target elevator position to the average walking speed of a single person; the baseline standard deviation parameter is empirically calibrated based on the deviation value of the statistical distribution of walking speed of a typical population, and can be taken as 10% to 20% of the baseline mathematical expectation parameter as the initial setting value.

[0133] In some feasible embodiments, during synchronous amplification, the reconstructed dynamic mathematical expectation parameter and the reconstructed dynamic standard deviation parameter are calculated according to the following formulas:

[0134] ; ;

[0135] in, These are the parameters of the reconstructed dynamic mathematical expectation; This refers to the space congestion coefficient. The baseline mathematical expectation parameter; The reconstructed dynamic standard deviation parameter; Here, the baseline expected value parameter is linearly amplified according to the spatial congestion coefficient, and the baseline standard deviation parameter is amplified according to the square root of the spatial congestion coefficient. The reason is that the variance of walking delay increases approximately linearly with the congestion coefficient under congested conditions, and the standard deviation is the arithmetic square root of the variance.

[0136] In some feasible embodiments, the mathematical expression of the preset passage delay distribution model after dynamic reconstruction is:

[0137] ;

[0138] in, This is the preset passage delay distribution model after dynamic reconstruction; This is a time interval variable representing the distance from the trigger point of the elevator intention event; These are the parameters of the reconstructed dynamic mathematical expectation; The reconstructed dynamic standard deviation parameter is used. Through the above dynamic reconstruction, the probability density curve of the distribution model undergoes a lag shift on the system time axis and the distribution shape becomes wider, reflecting the actual physical characteristics of increased walking delay and dispersion in crowded conditions.

[0139] In some further embodiments, the method further includes the following steps:

[0140] Collect the actual walking delay time of passengers using the elevator;

[0141] Training samples are constructed by binding the actual walking delay time with the spatial congestion coefficient of the physical area when the intention to take the elevator is triggered;

[0142] Based on the training samples, the baseline mathematical expectation parameters and baseline standard deviation parameters of the preset traffic delay distribution model are subjected to closed-loop adaptive calibration.

[0143] For example: Actual walking delay time can refer to the real time elapsed from the moment a passenger triggers their intention to ride the elevator until they actually arrive at their target elevator location. In real-world scenarios, the actual walking delay time can be calculated by comparing the timestamp of the passenger triggering the network interaction event with the timestamp of their actual entry into the elevator car. Closed-loop adaptive calibration can refer to iteratively updating the baseline mathematical expectation parameters and baseline standard deviation parameters using training samples, and then writing the updated parameters back into the system to replace the original parameters in the scheduling calculation of the next cycle. In real-world scenarios, a loss function can be constructed based on accumulated training samples, and the parameters can be gradually adjusted using gradient descent to make the distribution characteristics of the preset passage delay distribution model gradually approach the actual passage patterns of passengers.

[0144] This embodiment, through the above steps, collects the actual walking delay time of elevator users and binds it to historical spatial congestion coefficients to construct training samples. Based on this, it performs closed-loop adaptive calibration of the baseline mathematical expectation parameters and baseline standard deviation parameters of the preset passage delay distribution model. Thus, the baseline parameters can be continuously optimized based on real elevator data, ensuring that the preset passage delay distribution model maintains a match with the actual passage characteristics of the physical area during long-term operation. This avoids inaccurate passenger flow predictions due to the gradual deviation between preset fixed parameters and actual passage conditions.

[0145] In some feasible embodiments, when constructing training samples, the method further includes: setting an effective duration truncation window, and when the actual walking delay time obtained exceeds the effective duration truncation window, marking the data as an invalid sample and removing it.

[0146] For example, the effective duration cutoff window ranges from 10 to 15 minutes, preferably 13 minutes. In real-world scenarios, passengers may change their travel plans midway after triggering their intention to ride the elevator, resulting in an abnormally large actual walking delay. Such data cannot reflect normal traffic patterns, and removing it can avoid interfering with parameter calibration.

[0147] In some feasible embodiments, closed-loop adaptive calibration specifically includes: constructing a log-likelihood loss function using training samples, calculating the partial derivatives of the loss function with respect to the baseline mathematical expectation parameter and the baseline standard deviation parameter based on the gradient descent algorithm, and performing iterative parameter updates according to the set learning rate; writing the updated baseline mathematical expectation parameter and the baseline standard deviation parameter back to the system configuration file and putting them into the scheduling calculation of the next cycle.

[0148] For example, the learning rate ranges from 0.01 to 0.05 to ensure smooth convergence of the parameter update process. The trigger condition for closed-loop adaptive calibration can be set to the number of accumulated effective training samples reaching a preset threshold, or a timed trigger mechanism can be used to perform batch calibration during idle periods of the equipment. Through the above methods, self-learning optimization of model parameters can be achieved with minimal human intervention, maintaining a long-term match between the passenger flow prediction model and physical reality.

[0149] In some feasible embodiments, the closed-loop adaptive calibration employs the following log-likelihood loss function:

[0150] ;

[0151] in, The value is the log-likelihood loss function value; This represents the total number of valid training samples accumulated during the current calibration period. For the first The actual walking delay time corresponding to each valid training sample; To utilize the first A preset passage delay distribution model after dynamically reconstructing the historical spatial congestion coefficients bound to a valid training sample; The baseline mathematical expectation parameters to be optimized; The baseline standard deviation parameter is to be optimized. The physical meaning of this loss function is that by maximizing the likelihood probability of the actual walking delay time in the system's estimated probability distribution, the center and shape of the estimated distribution gradually approach the actual physical clustering characteristics of elevator use.

[0152] In some feasible embodiments, during closed-loop adaptive calibration, the iterative updates of the baseline mathematical expectation parameter and the baseline standard deviation parameter are performed according to the following formulas:

[0153] ;

[0154] ;

[0155] in, For the first The baseline mathematical expectation parameters updated in the next iteration; For the first The baseline mathematical expectation parameters before the next iteration update; The learning rate is the baseline mathematical expectation parameter, ranging from 0.01 to 0.05; For the first The baseline standard deviation parameter before the next iteration update; The learning rate for the baseline standard deviation parameter ranges from 0.01 to 0.05. After parameter iteration is complete, the updated baseline mathematical expectation parameter and baseline standard deviation parameter are written back to the system's underlying configuration file. In the next observation calculation cycle, they replace the original parameters in the forward passenger flow arrival probability curve generation and scheduling calculation.

[0156] In some feasible embodiments, the triggering condition for closed-loop adaptive calibration is: the total number of valid training samples accumulated by the system reaches two thousand, or a timed triggering mechanism is used to perform a batch calibration calculation once a day during the device's idle period. This triggering mechanism achieves a balance between system computational load and parameter timeliness.

[0157] In some of these embodiments, Figure 5 This is a flowchart of the process for generating the passenger flow arrival probability curve provided in this embodiment. Please refer to it. Figure 5 Step S130, which is the step of generating the passenger flow arrival probability curve based on the intention pulse signal model and the preset passage delay distribution model after dynamic reconstruction, specifically includes the following steps:

[0158] Step S131: In the intention pulse signal model, the intention pulse signals whose generation time is within the preset maximum retention time are retained to obtain the updated intention pulse signal model.

[0159] Step S132: Based on the preset time sampling step size, perform spatiotemporal convolution operation on the updated intention pulse signal model and the dynamically reconstructed preset passage delay distribution model to generate a passenger flow arrival probability curve.

[0160] For example: a preset maximum retention time is used to truncate historical data; in real-world scenarios, intention pulse signals generated earlier than the current time but exceeding the preset maximum retention time likely indicate that the corresponding passengers have already completed their elevator rides, and retaining such signals has limited reference value for current passenger flow prediction. Therefore, these signals are truncated and deleted. Spatiotemporal convolution operation can refer to using the updated intention pulse signal model as input stimulus and the dynamically reconstructed preset passage delay distribution model as response function, performing integration or summation on the time axis; in real-world scenarios, this operation diffuses each discrete intention pulse signal into a probability contribution according to the passage delay distribution model, and then superimposes each probability contribution at the corresponding time point to form a passenger flow arrival probability curve that changes continuously over time.

[0161] This embodiment uses the above steps to filter the timeliness of the intent pulse signal model using a preset maximum retention time, eliminating expired historical signals. Then, it fuses the filtered discrete intent pulse signals with the dynamically reconstructed preset travel delay distribution model through spatiotemporal convolution operations to generate a passenger flow arrival probability curve. Therefore, the passenger flow arrival probability curve reflects both the real-time changes in elevator intentions and the dynamic impact of physical area congestion on travel delays.

[0162] In some further embodiments, Figure 6 This is a flowchart illustrating the generation of the passenger flow arrival probability curve provided in this embodiment. Please refer to it. Figure 6 Step S132, which involves performing spatiotemporal convolution operations on the updated intention pulse signal model and the dynamically reconstructed preset passage delay distribution model based on a preset time sampling step size to generate a passenger flow arrival probability curve, specifically includes the following steps:

[0163] Step S132.1: Based on the preset time sampling step size, the time axis corresponding to the updated intention pulse signal model is divided into multiple time slice nodes;

[0164] Step S132.2: Determine the generation time corresponding to each retained intention pulse signal; determine the time difference between each generation time and each time slice node;

[0165] Step S132.3: Input the time difference into the preset passage delay distribution model after dynamic reconstruction, and determine the probability of arrival of each intention pulse signal at the branch corresponding to the time slice node.

[0166] Step S132.4: Accumulate the arrival probabilities of each branch to generate the passenger flow arrival probability curve.

[0167] For example: A preset time sampling step size refers to the time interval for discretizing the time axis, used to determine the temporal resolution of the passenger arrival probability curve. A time slice node refers to a discrete sampling point divided on the time axis according to the preset time sampling step size; in a real-world scenario, each time slice node corresponds to a future estimated time. A branch arrival probability refers to the probability contribution value of a single intention pulse signal at a certain time slice node, corresponding to the passenger arriving at the target elevator position; this contribution value is determined from the dynamically reconstructed preset passage delay distribution model based on the time difference between the generation time of the intention pulse signal and the time slice node. Accumulation involves summing the branch arrival probabilities corresponding to all retained intention pulse signals at the same time slice node to obtain the passenger arrival probability value corresponding to that time slice node; connecting the passenger arrival probability values ​​of each time slice node in chronological order forms the passenger arrival probability curve.

[0168] This embodiment decomposes the spatiotemporal convolution operation into four stages through the above steps: time axis division, time difference calculation, branch probability determination, and probability accumulation. The discrete intention pulse signal is unfolded on the time axis according to the preset passage delay distribution model after dynamic reconstruction, forming a continuous probability distribution. This reflects the overall pattern of the time taken for multiple elevator users to gradually arrive at the physical area after triggering their elevator intention at different times, under the influence of the degree of congestion in the physical area.

[0169] In some feasible embodiments, the spatiotemporal convolution operation is performed according to the following formula:

[0170] ;

[0171] in, The passenger arrival probability curve at time The value of ; For the current system, it is a continuous-time variable; The preset maximum retention period is used to define the longest traceability period for historical data involved in the calculation, and its value ranges from 5 minutes to 10 minutes. For the integration time variable in the integration calculation process; This represents the value of the intended pulse signal at the corresponding integral time variable. A preset passage delay distribution model is used for dynamic reconstruction after time offset operation. By configuring the lower limit of integration to the current time minus the preset maximum retention time, the risk of memory overflow caused by infinite data accumulation can be prevented. In actual digital computer operations, the above continuous integration operation is equivalent to an accumulation process based on discrete time steps.

[0172] In some feasible embodiments, before step S140, the method further includes: performing a status check on the elevators, removing elevator units in abnormal states, and generating a set of available elevators; abnormal states include at least one of full-load / overload state, hardware failure state, and manual maintenance state. Subsequent operations to determine the target elevator are all performed within the set of available elevators. Status check ensures that dispatching instructions are only issued to elevators with safe response capabilities, guaranteeing dispatching safety.

[0173] In some of these embodiments, Figure 7 This is a flowchart of elevator scheduling based on passenger arrival probability curves provided in this embodiment. Please refer to it. Figure 7 Step S140, which is the step of controlling the corresponding elevator to go to the target elevator position based on the passenger flow arrival probability curve, specifically includes the following steps:

[0174] Step S141: Integrate the passenger flow arrival probability curve within a preset look-ahead time window to obtain the area of ​​integration;

[0175] Step S142: When the integral area exceeds the preset scheduling trigger threshold and no physical call signal is detected at the target elevator location, determine the target elevator among all elevators related to the target elevator location.

[0176] Step S143: Control the target elevator to move to the target elevator location.

[0177] For example: A preset look-ahead time window, extending forward along the time axis from the current point in time, is used to assess the passenger flow volume approaching the target elevator location. The integral area refers to the cumulative probability value covered by the passenger arrival probability curve within the preset look-ahead time window, reflecting the expected number of passengers arriving. The preset scheduling trigger threshold refers to the minimum passenger flow standard used to determine whether pre-scheduling needs to be initiated. The physical call signal refers to the elevator call request signal triggered by a passenger at the target elevator location via a physical button. The target elevator, selected according to preset evaluation rules, refers to the elevator most suitable for responding to the current passenger flow among all available elevators related to the target elevator location.

[0178] This embodiment uses the integral area of ​​the passenger arrival probability curve within a preset look-ahead time window as the quantitative basis for scheduling decisions through the above steps. When the integral area reaches a trigger threshold and there is no physical call signal at the target elevator location, the target elevator is identified from the relevant elevators and controlled to move to the target elevator location in advance. Therefore, the elevator can be positioned before the passenger actually arrives, shortening the waiting time and avoiding duplicate responses to physical call signals, thus improving scheduling efficiency.

[0179] In some feasible embodiments, a preset look-ahead time window length parameter is configured based on the average response time of elevators in the building, with a value ranging from 15 seconds to 45 seconds, to cover the physical maneuvering time required for idle elevators to operate across floors.

[0180] The specific calculation method for the integral area is as follows: perform a definite integral operation on the passenger arrival probability curve within the time interval from the current time to the current time plus a preset forward time window, and the resulting integral value is the integral area.

[0181] In some feasible embodiments, the integral area is calculated according to the following formula:

[0182] ;

[0183] in, The integral area represents the cumulative passenger flow expected to reach the target elevator position within the preset forward time window; This refers to the current system time. The preset lookahead time window length parameter has a value range of 15 seconds to 45 seconds; The passenger arrival probability curve at time The value of the preset scheduling trigger threshold is typically set between 0.6 and 0.9 to filter out occasional probability calculation noise. When the integral area exceeds the preset scheduling trigger threshold, it indicates that the expected passenger flow has reached a level that requires the initiation of pre-scheduling.

[0184] In some further embodiments, Figure 8 This is a flowchart of the process for determining the target elevator provided in this embodiment. Please refer to it. Figure 8 In step S142, the sub-step of determining the target elevator among all elevators related to the target elevator location specifically includes the following steps:

[0185] Identify all available elevators associated with the target elevator location;

[0186] The scheduling cost for all available elevators is calculated using the cost function of integral area.

[0187] The available elevator with the lowest scheduling cost is identified as the target elevator.

[0188] For example: Available elevators are those that can currently respond normally to dispatch instructions among all elevators associated with the target elevator location. In real-world scenarios, elevators that are fully loaded, malfunctioning, or under maintenance will be excluded from the list of available elevators. The cost function, used to evaluate elevator dispatching costs, incorporates the integral area as a negative penalty in its calculation. In real-world scenarios, a larger integral area indicates more expected passengers, and the negative penalty reduces the dispatching cost of the corresponding elevator, thus prioritizing its dispatch. Dispatch cost refers to the comprehensive cost incurred by an available elevator to reach the target elevator location, taking into account factors such as waiting time and operating energy consumption.

[0189] This embodiment incorporates the predicted passenger flow into the elevator decision-making process through the above steps, so that the target elevator positions with large passenger flow receive corresponding priority weight in the scheduling cost assessment, thereby achieving reasonable allocation among multiple elevators that matches the passenger flow distribution.

[0190] In some feasible embodiments, the target elevator is selected from the set of available elevators using the following cost function:

[0191] ;

[0192] in, for The comprehensive assessment cost of all available elevators; This is a weighting coefficient for waiting time costs. To estimate the first The mechanical travel time parameter required for the available elevator to travel to the target elevator position; This is the weighting coefficient for mechanical energy consumption; To estimate the first The motor starting energy consumption parameters required for the elevator to perform this displacement; This is the passenger flow incentive weighting coefficient; The integral area; where the passenger flow incentive weighting coefficient is... The value is set as the weighting coefficient for waiting time cost. The factor is 0.5 to 1.5 times, allowing the passenger flow incentive to effectively influence the optimization result while avoiding abnormal elevator scheduling across multiple physical zones due to excessive incentives. This is achieved by introducing [a specific incentive] into the cost function. This item applies a cost deduction to target elevator locations with high passenger flow, resulting in idle elevators receiving a smaller evaluation value during the global traversal and minimization operation, thus prioritizing their assignment.

[0193] In some feasible embodiments, during the approach period when the target elevator is controlled to move toward the target passenger position, the method further includes: continuously detecting whether a physical call signal consistent with the estimated passenger direction is generated in the underlying hardware control loop corresponding to the target passenger position; when a physical call signal is collected, the currently executed virtual scheduling task is directly converted into a physical dispatch task to ensure the response to the real passenger request.

[0194] For example, virtual scheduling tasks are automatically generated by the system based on passenger flow predictions, which is different from physical elevator call signals triggered by passengers pressing physical buttons. By switching tasks when a physical elevator call signal is detected, command conflicts between virtual scheduling and actual elevator demand can be avoided, achieving a smooth handover of scheduling tasks.

[0195] In some further embodiments, Figure 9 This is a flowchart of the elevator dispatch secondary confirmation provided in this embodiment. Please refer to it. Figure 9 After step S142, the method further includes the following steps:

[0196] Monitor terminals within the access physical area;

[0197] When the access status of the terminal changes and the signal strength of the terminal meets the preset time condition and preset threshold condition, the integral area is updated according to the intention pulse signal corresponding to the terminal.

[0198] The target elevator is redefined based on the updated integral area.

[0199] If the newly determined target elevator is different from the previously determined target elevator, cancel the scheduling of the previously determined target elevator.

[0200] For example: A change in access status can refer to a change in the physical location of a terminal between different areas. In a real-world scenario, when a change in the access status of a terminal is detected, it indicates that the passenger corresponding to that terminal may have left the area where the target passenger location is located. Preset time conditions and preset threshold conditions, i.e., two preset dimensions of judgment conditions, are used to confirm that the change in access status is not caused by signal fluctuations, but rather that the terminal has stabilized outside the physical area, thereby updating the current integrated area. Correspondingly, updating the integrated area means deducting the area contributed by the intent pulse signal corresponding to the terminal with the changed access status from the original integrated area. Revoking the scheduling of the previously determined target elevator can refer to issuing a release command to the elevator system, releasing the target elevator from the locked state due to this scheduling, allowing it to freely respond to other scheduling requests.

[0201] This embodiment continuously monitors the terminal's access status and signal strength changes through the above steps. When it identifies that the terminal has left the physical area, it updates the integrated area and re-evaluates the scheduling decision. When the re-determined target elevator is inconsistent with the original target elevator, the scheduling of the original target elevator is promptly cancelled. Therefore, it can effectively identify cancellation events of elevator travel intentions, avoid continuing to execute invalid scheduling instructions due to passengers turning back midway, and reduce unnecessary elevator operation and resource consumption.

[0202] In some further embodiments, when the access state of the terminal changes and the signal strength of the terminal meets the preset time condition and preset threshold condition, the step of updating the integral area according to the intention pulse signal corresponding to the terminal specifically includes the following steps:

[0203] When the access location of a terminal is detected to change from within the physical area to outside the physical area, the access status of the terminal is determined to have changed.

[0204] When the signal strength of the terminal is continuously increased within a preset time period and the increased signal strength reaches a preset threshold, it is determined that the signal strength of the terminal meets the preset time condition and the preset threshold condition.

[0205] Determine the area of ​​contribution of the intention pulse signal corresponding to the terminal to the integral area;

[0206] The contributed area is removed from the integral area to obtain the updated integral area.

[0207] Specifically: When the access location changes from within a physical area to outside a physical area, the terminal disconnects from the network infrastructure within the physical area; for example, the terminal switches from a network access point in a waiting area to a network access point in an office area. A signal strength that continuously increases and reaches a preset threshold within a preset time period indicates that the terminal's signal reception strength at the new access location shows a stable upward trend and is within normal limits, thus eliminating misjudgments caused by brief fluctuations in the wireless signal. The contribution area refers to the cumulative probability value of the intentional pulse signal corresponding to the disconnected terminal, after being diffused through a preset passage delay distribution model, contributing to the integral area.

[0208] This embodiment uses the above steps to make a dual determination based on the change of terminal access location and signal strength to confirm the cancellation of elevator ride intention; then it accurately deducts the contribution area corresponding to the terminal from the integrated area to achieve quantitative update of the integrated area.

[0209] In some feasible embodiments, the updated integral area is calculated according to the following formula:

[0210] ;

[0211] in, The updated integral area; The area before the update is the integral area; The contribution area corresponding to the terminal whose elevator travel intention has been cancelled is calculated as follows: extract the single dynamic reconstruction preset passage delay distribution model corresponding to the terminal, and calculate the integral area of ​​this independent model within a preset look-ahead time window as the contribution area. Substitute the updated integral area into the cost function to recalculate the scheduling cost of each available elevator to determine whether the identified target elevator needs to be adjusted.

[0212] In some feasible embodiments, after the target elevator arrives at the target elevator position and completes the door opening action, the method further includes: calculating the tail integral area within the lag observation time window based on the passenger flow arrival probability curve; when the tail integral area meets the preset long-tail passenger flow conditions, issuing an extended door opening command to the door operator controller of the target elevator to extend the door opening holding time of the target elevator.

[0213] For example, the long-tail passenger flow condition can be set when the tail area exceeds a preset long-tail threshold. In real-world scenarios, during peak passenger flow periods, some passengers may arrive late due to overcrowding. Extending the door opening time can help connect these delayed passengers and reduce the number of times the elevator starts and stops twice.

[0214] This application provides an elevator dynamic scheduling and control system based on passenger flow prediction. Figure 10 This is a block diagram of the elevator dynamic scheduling and control system based on passenger flow prediction provided in this embodiment. Please refer to it. Figure 10 This includes: network infrastructure, elevator systems, and control facilities;

[0215] Network infrastructure for providing intent pulse signals generated when interacting with elevators within a pre-defined physical area;

[0216] The elevator system is connected in communication with the control facilities;

[0217] A control facility, communicatively connected to a network facility, is used to execute the aforementioned method embodiments and their preferred embodiments.

[0218] In some of these embodiments, please continue to refer to Figure 10 The network infrastructure includes an access control system and an authentication system. The access control system, located on the building's ground floor, controls the physical entry and exit of users and generates access log data. The authentication system centrally manages the authentication process for terminals connecting to network access points and generates network status session logs with identifiers. The control facilities establish bidirectional data communication connections with both the access control system and the authentication system via the building's internal LAN to obtain access log data and network status session logs in real time.

[0219] For example: access control system-generated access record data can be used as log data source for generating uplink intention pulse signals; authentication system-generated network status session logs can be used as log data source for generating downlink intention pulse signals.

[0220] In some feasible embodiments, please continue to refer to Figure 10 The network infrastructure also includes multiple network access points distributed across various floors of the building. These access points are located in office areas and elevator waiting areas to provide network access services to terminals. The control system monitors connection changes and signal strength variations between different network access points to obtain trigger timestamps corresponding to events representing elevator usage intentions. These network access points can be wireless.

[0221] In some of these embodiments, please continue to refer to Figure 10 The control facilities include an edge gateway. The edge gateway establishes data communication with the access control system and authentication system through pre-defined application programming interfaces (APIs) to obtain access record data and network status session logs. The edge gateway also establishes a communication connection with the elevator system through a standard API to issue dispatch commands.

[0222] For example, a standard application programming interface (API) includes at least the following fields: waiting floor field, elevator direction field, virtual call task identifier field, task validity period field, task cancellation field, points-based passenger flow weight field, and door hold time field. Through these API fields, the edge gateway can transmit complete scheduling parameters to the elevator system.

[0223] In some feasible embodiments, the edge gateway submits virtual call requests, dispatch weight deregistration requests, and door hold requests to the elevator system via a standard application programming interface (API). The edge gateway does not directly bypass or block the elevator system's safety hardware circuits; the safety hardware circuits include at least one of the following: door lock circuit, speed limit circuit, fire control circuit, maintenance control circuit, and overload protection circuit.

[0224] In some of these embodiments, please continue to refer to Figure 10 The elevator system includes a group control mainboard. The group control mainboard communicates with the edge gateway to receive dispatch instructions issued by the edge gateway, and converts the dispatch instructions into underlying dispatch actions when it is confirmed that the elevator system is not in a fire-fighting state, maintenance state, fault state, full-load state, or safety circuit abnormal state.

[0225] For example, the group control motherboard ensures that the virtual elevator call request submitted by the edge gateway is executed only when the elevator system is in a safe and responsive state through the above-mentioned status verification mechanism, thereby ensuring scheduling safety.

[0226] In some feasible embodiments, please continue to refer to Figure 10The elevator system also includes a door operator controller. The group control motherboard is physically connected to the door operator controller via the elevator's internal hardware control bus, enabling low-level communication. The door operator controller is located in the elevator car and is used to execute the opening and closing actions of the physical car doors.

[0227] For example: when the edge gateway determines that there is long-tail passenger flow based on the passenger flow arrival probability curve, it can send an extended door opening command to the door controller through the group control motherboard to suppress the normal timeout automatic door closing logic inside the door controller and extend the opening and holding time of the car door.

[0228] In some embodiments, the network infrastructure, control facilities, and elevator system are synchronized with each other via a time synchronization server within the building. When the control facility detects that the time deviation of any data source exceeds a preset time deviation threshold, it performs offset correction on the timestamp of that data source; if correction cannot be completed, the data source is temporarily excluded from the generation of the intent pulse signal to ensure that the spatiotemporal convolution operation and the scheduling cancellation judgment are on the same time axis.

[0229] The method embodiments provided in this example can be executed on a terminal, computer, or similar computing device. For example, it can run on a terminal. Figure 11 This is a hardware structure block diagram of the terminal for the elevator dynamic scheduling and control method based on passenger flow prediction in this embodiment. (See diagram for example.) Figure 11 As shown, a terminal may include one or more ( Figure 11 Only one is shown in the diagram. A processor 102 and a memory 104 for storing data are also included. The processor 102 may be, but is not limited to, a microprocessor (MCU) or a programmable logic device (FPGA). The terminal may also include a transmission device 106 for communication functions and an input / output device 108. Those skilled in the art will understand that… Figure 11 The structure shown is for illustrative purposes only and does not limit the structure of the terminal described above. For example, the terminal may also include components that are larger than... Figure 11 The more or fewer components shown, or having the same Figure 11 The different configurations shown are illustrated.

[0230] The memory 104 can be used to store computer programs, such as application software programs and modules, like the computer program corresponding to the elevator dynamic scheduling control method based on passenger flow prediction in this embodiment. The processor 102 executes various functional applications and data processing by running the computer program stored in the memory 104, thereby implementing the above-described method. The memory 104 may include high-speed random access memory, and may also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory 104 may further include memory remotely located relative to the processor 102, and these remote memories can be connected to the terminal via a network. Examples of such networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof.

[0231] The transmission device 106 is used to receive or send data via a network. This network includes a wireless network provided by the terminal's communication provider. In one example, the transmission device 106 includes a Network Interface Controller (NIC), which can connect to other network devices via a base station to communicate with the Internet. In another example, the transmission device 106 can be a Radio Frequency (RF) module used for wireless communication with the Internet.

[0232] This embodiment also provides a computer device, including a memory and a processor, wherein the memory stores a computer program and the processor is configured to run the computer program to perform the steps in any of the above method embodiments.

[0233] Optionally, the computer device may further include a transmission device and an input / output device, wherein the transmission device is connected to the processor and the input / output device is connected to the processor.

[0234] It should be noted that the specific examples in this embodiment can refer to the examples described in the above embodiments and optional implementations, and will not be repeated in this embodiment.

[0235] Furthermore, in conjunction with the elevator dynamic scheduling and control method based on passenger flow prediction provided in the above embodiments, this embodiment can also provide a storage medium for implementation. The storage medium stores a computer program; when executed by a processor, the computer program implements any of the elevator dynamic scheduling and control methods based on passenger flow prediction in the above embodiments.

[0236] It should be noted that all information and data involved in this application are authorized by the user or fully authorized by all parties and will be used legally.

[0237] It should be understood that the specific embodiments described herein are merely illustrative of the application and not intended to limit it. All other embodiments derived by those skilled in the art based on the embodiments provided in this application without inventive effort are within the scope of protection of this application.

[0238] Obviously, the accompanying drawings are merely some examples or embodiments of this application. Those skilled in the art can apply this application to other similar situations based on these drawings without any creative effort. Furthermore, it is understood that although the work done in this development process may be complex and lengthy, for those skilled in the art, certain design, manufacturing, or production modifications made based on the technical content disclosed in this application are merely conventional technical means and should not be considered as insufficient disclosure of this application.

[0239] The term "embodiment" in this application refers to a specific feature, structure, or characteristic described in connection with an embodiment that may be included in at least one embodiment of this application. The appearance of this phrase in various places in the specification does not necessarily imply the same embodiment, nor does it imply that it is mutually exclusive with or independent of other embodiments. It will be clearly or implicitly understood by those skilled in the art that the embodiments described in this application may be combined with other embodiments without conflict.

[0240] The above embodiments merely illustrate several implementation methods of this application, and their descriptions are relatively specific and detailed, but they should not be construed as limiting the scope of patent protection. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the scope of protection of this application.

Claims

1. A dynamic elevator scheduling and control method based on passenger flow prediction, characterized in that, include: The intention pulse signal generated when a terminal in a preset physical area interacts with an elevator in the physical area is acquired, and an intention pulse signal model is constructed based on the intention pulse signal. The intention pulse signal model represents the distribution of the elevator-riding intention of the passenger at discrete time points; Based on the number of valid accesses of the terminals, the spatial congestion coefficient of the physical area is determined; based on the spatial congestion coefficient, a preset passage delay distribution model is dynamically reconstructed. The preset passage delay distribution model represents the probability distribution of the time required for the passenger to reach the target elevator position. Based on the intention pulse signal model and the dynamically reconstructed preset passage delay distribution model, a passenger flow arrival probability curve is generated; Based on the passenger flow arrival probability curve, the corresponding elevator is controlled to go to the target elevator location.

2. The elevator dynamic scheduling and control method based on passenger flow prediction according to claim 1, characterized in that, The step of acquiring the intent pulse signal generated when a terminal within a preset physical area interacts with an elevator within that physical area, and constructing an intent pulse signal model based on the intent pulse signal, includes: Extract the trigger timestamps corresponding to the events representing the intention to ride the elevator from the log data of the interaction between each terminal and the elevator; Based on the trigger timestamp, generate the corresponding intent pulse signal; The intention pulse signals within a preset time period are deduplicated; the deduplicated intention pulse signals are then combined to construct the intention pulse signal model.

3. The elevator dynamic scheduling and control method based on passenger flow prediction according to claim 1, characterized in that, Determining the spatial congestion coefficient of the physical area based on the number of valid accesses of the terminals includes: Obtain the interaction traffic between each terminal and the elevator; Based on the interaction traffic, determine the active terminals among all the terminals; The number of active terminals accessing the network is taken as the number of valid accesses. The spatial congestion coefficient of the physical area is determined based on the number of valid accesses and the preset baseline number of terminals.

4. The elevator dynamic scheduling and control method based on passenger flow prediction according to claim 1, characterized in that, The step of dynamically reconstructing the preset traffic delay distribution model based on the spatial congestion coefficient includes: Obtain the baseline mathematical expectation parameter and baseline standard deviation parameter of the preset passage delay distribution model; Wherein, the benchmark mathematical expectation parameter characterizes the average walking time of the passenger to the target elevator position under ideal empty-load conditions; the benchmark standard deviation parameter characterizes the degree of dispersion of time distribution caused by individual walking speed differences under ideal empty-load conditions. Based on the spatial congestion coefficient, the baseline mathematical expectation parameter and the baseline standard deviation parameter are simultaneously amplified; The preset passage delay distribution model is dynamically reconstructed based on the synchronously amplified benchmark mathematical expectation parameters and benchmark standard deviation parameters.

5. The elevator dynamic scheduling and control method based on passenger flow prediction according to claim 4, characterized in that, The method further includes: Collect the actual walking delay time of the person taking the elevator; The actual walking delay time is linked to the spatial congestion coefficient of the physical area when the intention to take the elevator is triggered, and training samples are constructed. Based on the training samples, the baseline mathematical expectation parameter and the baseline standard deviation parameter of the preset traffic delay distribution model are subjected to closed-loop adaptive calibration.

6. The elevator dynamic scheduling and control method based on passenger flow prediction according to claim 1, characterized in that, The step of generating a passenger flow arrival probability curve based on the intention pulse signal model and the dynamically reconstructed preset passage delay distribution model includes: In the intent pulse signal model, the intent pulse signal whose generation time is within the preset maximum retention time is retained to obtain the updated intent pulse signal model; Based on a preset time sampling step, spatiotemporal convolution operations are performed on the updated intention pulse signal model and the dynamically reconstructed preset passage delay distribution model to generate the passenger flow arrival probability curve.

7. The elevator dynamic scheduling and control method based on passenger flow prediction according to claim 6, characterized in that, The step of performing spatiotemporal convolution operations on the updated intent pulse signal model and the dynamically reconstructed preset passage delay distribution model based on a preset time sampling step size to generate the passenger flow arrival probability curve includes: Based on the preset time sampling step size, the time axis corresponding to the updated intention pulse signal model is divided into multiple time slice nodes; Determine the generation time corresponding to each retained intention pulse signal; determine the time difference between each generation time and each time slice node; The preset passage delay distribution model is dynamically reconstructed by inputting the time difference, and the arrival probability of each intention pulse signal at the branch corresponding to the time slice node is determined accordingly. The arrival probabilities of each branch are accumulated to generate the passenger flow arrival probability curve.

8. The elevator dynamic scheduling and control method based on passenger flow prediction according to claim 1, characterized in that, The step of controlling the corresponding elevator to the target elevator location based on the passenger flow arrival probability curve includes: The passenger flow arrival probability curve is integrated over a preset look-ahead time window to obtain the area of ​​integration. When the integral area exceeds a preset scheduling trigger threshold and no physical call signal is detected at the target elevator location, the target elevator is determined from all the elevators associated with the target elevator location. Control the target elevator to proceed to the target elevator location.

9. The elevator dynamic scheduling and control method based on passenger flow prediction according to claim 8, characterized in that, Determining the target elevator among all elevators associated with the target elevator location includes: Identify all available elevators associated with the target elevator location; The scheduling cost for all available elevators is calculated using the cost function of the integral area. The available elevator with the lowest scheduling cost is identified as the target elevator.

10. The elevator dynamic scheduling and control method based on passenger flow prediction according to claim 8, characterized in that, The method further includes: Monitoring is performed on the terminals accessing the physical area; When the access status of the terminal changes and the signal strength of the terminal meets the preset time condition and the preset threshold condition, the integral area is updated according to the intention pulse signal corresponding to the terminal. The target elevator is redefined based on the updated integral area; If the re-determined target elevator is different from the previously determined target elevator, the scheduling of the previously determined target elevator shall be cancelled.

11. The elevator dynamic scheduling and control method based on passenger flow prediction according to claim 10, characterized in that, When the access state of the terminal changes and the signal strength of the terminal meets a preset time condition and a preset threshold condition, updating the integral area according to the intent pulse signal corresponding to the terminal includes: When the access location of the terminal is detected to change from within the physical area to outside the physical area, it is determined that the access status of the terminal has changed; When it is detected that the signal strength of the terminal continuously increases within a preset time period, and the increased signal strength reaches the preset threshold, it is determined that the signal strength of the terminal satisfies the preset time condition and the preset threshold condition. Determine the area of ​​contribution of the intention pulse signal corresponding to the terminal to the integral area; The contributing area is removed from the integral area to obtain the updated integral area.

12. An elevator dynamic scheduling and control system based on passenger flow prediction, characterized in that, include: Network infrastructure, elevator systems, and control facilities; The network infrastructure is used to provide intent pulse signals generated when interacting with elevators within a preset physical area; The elevator system is communicatively connected to the control facility; The control facility is communicatively connected to the network facility and is used to perform the method of any one of claims 1 to 11.

Images