Passenger flow estimation method based on multimodality and mobile signaling

Abstract

The present invention relates to the field of smart tourism management technology, specifically a passenger flow estimation method based on multimodality and mobile signaling. The method comprises obtaining passenger flow data from multiple sources of a target scenic spot, constructing a first passenger flow prediction model based on the passenger flow data from the multiple sources; obtaining valid mobile signaling data from the target scenic spot based on a preset mobile signaling adoption strategy, fusing the valid mobile signaling data with the prediction results of the first passenger flow prediction model, and constructing a second passenger flow prediction model; estimating the passenger flow of the target scenic spot based on the second passenger flow prediction model to obtain a passenger flow estimation result. Within a continuous monitoring cycle, if the estimation results in any two adjacent time periods reach a warning threshold, a warning signal is triggered, and warning level information corresponding to the passenger flow estimation result is pushed to a management terminal. The method solves the problem that existing scenic spot passenger flow statistics have large data deviations and cannot reflect the actual passenger flow.

Description

Technical Field

[0001] The present invention relates to the technical field of smart tourism management, and in particular to a passenger flow estimation method based on multimodality and mobile signaling. Background Art

[0002] With the rapid development of the tourism industry, visitor traffic to scenic spots has increased annually. Scientifically and rationally managing the flow of people and visitors to these attractions has become a critical issue. Traditional methods for monitoring visitor flow rely primarily on single data sources (such as ticketing systems or manual statistics), resulting in issues such as data lag, low accuracy, and limited coverage. For example, ticketing data cannot reflect the flow of unticketed visitors, video surveillance is limited by device deployment density and blind spots, and checkpoint and gate data struggles to capture the dynamic distribution of visitors within a scenic area. Furthermore, existing technologies lack effective mechanisms for integrating multi-source data, resulting in insufficiently robust prediction models and difficulty in addressing fluctuations in visitor flow during holidays or emergencies.

[0003] Mobile signaling data, as an emerging data source, can provide real-time user location information. However, its use alone can lead to issues such as signal drift and base station coverage mismatches. Therefore, the challenge of integrating multimodal data (such as video, ticketing, and checkpoint data) with mobile signaling data to build a dynamic weight allocation model and achieve accurate passenger flow estimation and early warning has become a pressing technical challenge.

[0004] Therefore, the present invention provides a passenger flow estimation method based on multimodality and mobile signaling to solve the above problems. Summary of the Invention

[0005] In view of the above situation, in order to overcome the defects of the existing technology, the present invention provides a passenger flow estimation method based on multimodality and mobile signaling to solve the problem that the existing passenger flow estimation of scenic spots has large deviations and cannot reflect the actual passenger flow.

[0006] In order to achieve the above-mentioned purpose, the technical solution adopted by the present invention is: a passenger flow estimation method based on multi-modality and mobile signaling, the method comprising: obtaining passenger flow data from multiple sources of data of the target scenic spot, and constructing a first passenger flow prediction model based on the passenger flow data from the multiple sources of data, wherein the multiple sources of data include scenic spot parking lot video data, scenic spot checkpoint gate data, scenic spot video surveillance data and scenic spot ticketing data; based on a preset mobile signaling adoption strategy, obtaining valid mobile signaling data of the target scenic spot, integrating the valid mobile signaling data with the prediction results of the first passenger flow prediction model, and constructing a second passenger flow prediction model; estimating the passenger flow of the target scenic spot based on the second passenger flow prediction model to obtain a passenger flow estimation result, and within a continuous monitoring period, if the estimation results in any two adjacent time periods reach the warning threshold, a warning signal is triggered, and the warning level information corresponding to the passenger flow estimation result is pushed to the management terminal.

[0007] A further improvement of the present application is that, before constructing the first passenger flow prediction model based on the passenger flow data from the multi-source data sources, it also includes: performing spatiotemporal alignment and outlier cleaning on the multi-source data, unifying the time base of each data source through timestamps, and filling in missing data using linear interpolation; calculating the initial weight of each data source based on historical data, and the weight distribution rule is: the weight of checkpoint gate data is greater than the weight of video surveillance data, the weight of video surveillance data is greater than the weight of ticketing data, and the weight of ticketing data is greater than the weight of parking lot data.

[0008] A further improvement of the present application lies in the method of calculating the initial weight of each data source based on historical data, comprising: determining the accuracy of each data source based on the deviation between the predicted value of each data source in the historical data and the actual passenger flow, and determining the linear correlation between the historical data of each data source and the actual passenger flow through the Pearson correlation coefficient; allocating initial weights according to the weighted comprehensive score of the accuracy and correlation coefficient of each data source; dynamically adjusting the weights using the gradient descent method, and optimizing with minimizing the prediction error rate as the objective function, terminating the optimization when the error rate decreases by less than a preset threshold for several consecutive iterations, and outputting the final weight allocation result.

[0009] A further improvement of the present application is that the expression of the first passenger flow prediction model is:

[0010] (1),

[0011] In expression (1), represents the first predicted passenger flow, Indicates the The contribution weight of each data source to passenger flow prediction is obtained by calculating the accuracy and correlation coefficient of each data source through historical data. Indicates the Real-time data from multiple data sources, represents the impact factor of time on passenger flow, Indicates the impact factor of weather on passenger flow.

[0012] A further improvement of the present application is that the preset mobile signaling adoption strategy includes daily mode, holiday mode and emergency mode; if the mobile signaling adoption strategy is daily mode, the first radius of the coverage sector of the base station outside the target scenic area is used as the coverage range of the base station sector, and the second radius of the coverage sector of the base station within the target scenic area is used as the coverage range of the base station sector, and the second radius is greater than the first radius, and the number of active signaling IDs within the first radius and the second radius is collected as valid mobile signaling data; if the mobile signaling adoption strategy is holiday mode, the first radius of the coverage sector of the base station outside the target scenic area is used as the center of the circle, and the number of active signaling IDs within the first radius and the second radius is collected as valid mobile signaling data. The coverage angle is expanded at the coverage sector angle to form an extended coverage range, and the extended coverage range and the second radius of the coverage sector of the base station within the target scenic area are collected as the number of active signaling IDs within the coverage range of the base station sector as valid mobile signaling data; if the mobile signaling adoption strategy is emergency mode, the first radius of the coverage sector of the base station outside the target scenic area is used as the center of the circle, and the coverage angle is expanded to 360° at the original coverage sector angle to form a full coverage range, and the full coverage range and the second radius of the coverage sector of the base station within the target scenic area are collected as the number of active signaling IDs within the coverage range of the base station sector as valid mobile signaling data.

[0013] A further improvement of the present application is that the triggering condition for the conversion of the preset mobile signaling adoption strategy is: if the ratio between the number of active signaling IDs within the first radius and the second radius and the number of passengers collected by the scenic area checkpoint gate reaches a preset threshold, the daily mode is converted to the holiday mode; if the occupancy rate of the scenic area parking lot reaches the warning threshold or the day is a statutory holiday, the holiday mode is converted to the emergency mode.

[0014] A further improvement of the present application is that the expression of the second passenger flow prediction model is:

[0015] (2),

[0016] In expression (2), represents the final estimated value of passenger flow, represents the first predicted passenger flow, Indicates the number of active signaling IDs. Indicates the fusion coefficient, ranging from 0.6 to 0.8.

[0017] A further improvement of the present application is that the warning level information includes yellow warning, red warning and overflow warning; if the final passenger flow estimate reaches the first threshold of the scenic area's full load design, a yellow warning message is sent to the management terminal; if the final passenger flow estimate reaches the second threshold of the scenic area's full load design, a red warning message is sent to the management terminal; if the final passenger flow estimate reaches the third threshold of the scenic area's full load design, an overflow warning message is sent to the management terminal.

[0018] The beneficial effects of the present invention are as follows: by integrating multimodal data (such as video, ticketing, and checkpoint data) with mobile signaling data, data complementarity and enhancement are achieved, effectively solving the problems of data lag, low accuracy, and limited coverage that exist in traditional passenger flow monitoring methods. At the same time, by constructing a dynamic weight allocation model and dynamically adjusting weights based on the accuracy and linear correlation of the data sources, the robustness and adaptability of the prediction model are improved, enabling it to more accurately reflect the actual passenger flow of tourist attractions. Furthermore, through a preset mobile signaling collection strategy, the present invention flexibly adjusts the collection scope of signaling data according to the needs of different scenarios, further improving the accuracy and practicality of passenger flow estimation. Ultimately, by achieving accurate estimation and early warning of passenger flow, powerful data support is provided for the management and decision-making of tourist attractions, helping to improve the operational efficiency of tourist attractions and the visitor experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 is a schematic flow chart of a passenger flow estimation method based on multimodality and mobile signaling according to the present invention;

[0020] Figure 2 is a schematic flow chart of an embodiment of a passenger flow estimation method based on multimodality and mobile signaling according to the present invention;

[0021] Figure 3 This is a schematic structural diagram of the passenger flow estimation system based on multimodality and mobile signaling of the present invention. DETAILED DESCRIPTION

[0022] The following will describe various embodiments of the present invention in detail with reference to the accompanying drawings. It should be understood by those skilled in the art that these embodiments are only used to explain the technical principles of the present invention and are not intended to limit the scope of protection of the present invention.

[0023] Based on the above problems, the inventors provide the following solutions:

[0024] First, we obtain visitor flow data for the target scenic spot from multiple sources, including but not limited to parking lot video data, checkpoint gate data, in-scenic area video surveillance data, and ticketing data. This data provides a rich foundation for subsequent visitor flow prediction. We then preprocess this multi-source data, including spatiotemporal alignment and outlier removal, to ensure data consistency and accuracy. We use timestamps to unify the time base of each data source, and linear interpolation is used to fill in missing data, thereby reducing the impact of data errors on the prediction results.

[0025] Next, the initial weights for each data source are calculated based on historical data. The weighting rules take into account the accuracy and relevance of different data sources for passenger flow forecasts, ensuring that the prediction model fully leverages the strengths of each data source. Specifically, checkpoint and gate data is given the highest weight because it directly reflects the entry and exit of tourists. Video surveillance data, while limited by equipment deployment and viewing angles, still provides a general overview of tourist flows and therefore receives the next highest weight. Ticketing data reflects the number of tourists who purchased tickets, but does not cover all tourists (those who did not enter on the day of purchase or who purchased tickets online and then had them refunded), so it receives the next highest weight. Parking lot data, serving as supplementary information, receives the lowest weight.

[0026] To improve the accuracy of the first passenger flow prediction model, the present invention further incorporates mobile signaling data. Based on pre-defined mobile signaling collection strategies, valid mobile signaling data for the target scenic area is acquired. These strategies include daily, holiday, and emergency modes to accommodate passenger flow estimation needs in different scenarios. By adjusting the scope of signaling data collection, the present invention can more accurately reflect the actual flow of tourists.

[0027] Finally, the valid mobile signaling data is integrated with the predictions from the first passenger flow prediction model to construct a second passenger flow prediction model. This model leverages the strengths of both multimodal and mobile signaling data to accurately estimate passenger flow. During a continuous monitoring cycle, if the estimated results for any two adjacent time periods reach the warning threshold, a warning signal is triggered, and the corresponding warning level information for the passenger flow estimate is pushed to the management terminal. This feature provides timely and accurate data support for the management and decision-making of tourist attractions.

[0028] The technical solution will be described in detail below in conjunction with specific embodiments. Example

[0029] refer to Figure 1 A passenger flow estimation method based on multimodality and mobile signaling includes the following steps S100-S300:

[0030] S100: Obtain passenger flow data from multiple sources for a target scenic spot, and construct a first passenger flow prediction model based on the passenger flow data from the multiple sources, wherein the multiple sources include scenic spot parking lot video data, scenic spot checkpoint gate data, scenic spot video surveillance data, and scenic spot ticketing data;

[0031] S200: Based on a preset mobile signaling adoption strategy, obtain valid mobile signaling data of the target scenic spot, fuse the valid mobile signaling data with the prediction result of the first passenger flow prediction model, and construct a second passenger flow prediction model;

[0032] S300. Estimate the passenger flow of the target scenic spot based on the second passenger flow prediction model to obtain a passenger flow estimation result. During a continuous monitoring period, if the estimation results in any two adjacent time periods reach the warning threshold, a warning signal is triggered, and the warning level information corresponding to the passenger flow estimation result is pushed to the management terminal.

[0033] Specifically, in step S100, when counting passenger flow using video data from the scenic area parking lot, the estimation rule is as follows: small cars are calculated as 2 people per car, small tourist buses (commercial vehicles, Jinbei, etc.) are calculated as 7 people per car, medium-sized tourist buses (6-9 meters long) are calculated as 22 people per car, and large tourist buses (over 9 meters long) are calculated as 40 people. When counting passenger flow using data from the scenic area checkpoint gate, the final number of people is calculated as the number of people inspected at the entrance (the number of tickets scanned and verified) plus 5% of this base number (children, the elderly, or other people who meet the conditions for free admission). When counting passenger flow using video surveillance data within the scenic area, a combination of Mask R-CNN (segmentation) + density estimation + wandering detection algorithms is used to process data in dense channels / popular areas. For other areas, a combination of lightweight YOLO + infrared thermal imaging + direction recognition algorithms is used to process data to determine the estimated passenger flow data. When counting passenger flow using scenic area ticketing data, 98% of the ticket purchase volume (excluding some ticket purchases that did not enter or were refunded) is used as the basis for estimation. It should be noted that the above quantity estimates are all calculated according to the same time period to ensure consistency.

[0034] In step S200, the preset mobile signaling collection strategy includes daily mode, holiday mode and emergency mode, and triangulation positioning technology is used to collect mobile signaling;

[0035] If the mobile signaling collection strategy is daily mode, the first radius (2.5km from the base station) of the base station's coverage sector outside the target scenic area (e.g., the sector angle is ±30°) is used as the base station sector coverage range, and the second radius (5km from the base station) of the base station's coverage sector within the target scenic area is used as the base station sector coverage range. The number of active signaling IDs within the first radius and the second radius are collected as valid mobile signaling data;

[0036] If the mobile signaling collection strategy is the holiday mode, the first radius of the coverage sector of the base station outside the target scenic area is used as the center of the circle, and the coverage angle is expanded (for example, by ±15°) on the original coverage sector angle (for example, the sector angle is ±30°) to form an extended coverage range. The number of active signaling IDs within the extended coverage range and the second radius of the coverage sector of the base station within the target scenic area as the coverage range of the base station sector is collected as valid mobile signaling data;

[0037] If the mobile signaling adoption strategy is emergency mode, the first radius of the coverage sector of the base station outside the target scenic area is used as the center of the circle, and the coverage angle is expanded to 360° on the original coverage sector angle to form a full coverage range. The full coverage range and the second radius of the coverage sector of the base station within the target scenic area are collected as the number of active signaling IDs within the coverage range of the base station sector as valid mobile signaling data.

[0038] The triggering condition for switching the preset mobile signaling policy is:

[0039] If the ratio between the number of active signaling IDs within the first radius and the second radius and the number of passengers collected by the scenic area checkpoint gate reaches the preset threshold (for example, the ratio between the number of active signaling IDs and the number of passengers collected by the scenic area checkpoint gate reaches 30%), the daily mode will be switched to the holiday mode; if the occupancy rate of the scenic area parking lot reaches the warning threshold (for example, the occupancy rate of the scenic area parking lot reaches 90%) or the day is a statutory holiday, the holiday mode will be switched to the emergency mode.

[0040] In step S300, a collection cycle is usually 5 minutes. If the estimation results in two consecutive 5-minute collection cycles reach the warning threshold, a warning signal is triggered. The warning level information includes yellow warning, red warning and over-limit warning;

[0041] If the final estimated passenger flow reaches the first threshold (80%) of the scenic area's full load design, a yellow warning message will be sent to the management terminal;

[0042] If the final estimated passenger flow reaches the second threshold (90%) of the scenic area's full load design, a red warning message will be sent to the management terminal;

[0043] If the final estimated passenger flow reaches the third threshold (100%) of the scenic area's full load design, an overflow warning message will be sent to the management terminal.

[0044] In one embodiment of the present application, Figure 2 As shown, a passenger flow estimation method based on multimodality and mobile signaling includes steps T100-T400:

[0045] T100. Obtain passenger flow data from multiple sources for the target scenic area, perform spatiotemporal alignment and outlier cleaning on the multi-source data, unify the time base of each data source through timestamps, and fill in missing data using linear interpolation. Calculate the initial weight of each data source based on historical data, using the following weight distribution rule: the weight of checkpoint gate data is greater than the weight of video surveillance data, which is greater than the weight of ticketing data, which is greater than the weight of parking lot data.

[0046] T200, constructing a first passenger flow prediction model based on the passenger flow data from the multi-source data source, where the multi-source data source includes scenic area parking lot video data, scenic area checkpoint gate data, scenic area video surveillance data, and scenic area ticketing data;

[0047] T300, based on a preset mobile signaling adoption strategy, obtain valid mobile signaling data of the target scenic spot, fuse the valid mobile signaling data with the prediction result of the first passenger flow prediction model, and construct a second passenger flow prediction model;

[0048] T400. Estimate the passenger flow of the target scenic spot based on the second passenger flow prediction model to obtain the passenger flow estimation result. During the continuous monitoring period, if the estimation results in any two adjacent time periods reach the warning threshold, the warning signal is triggered, and the warning level information corresponding to the passenger flow estimation result is pushed to the management terminal.

[0049] In step T100, the time granularity of each data source is usually unified to 5 minutes through timestamps, and missing data is filled using linear interpolation. For example, if video surveillance data is missing from 12:00 to 12:05 on May 1, 2024 due to equipment failure, linear interpolation is used to fill the gap (taking the average of the 5 minutes before and after).

[0050] The method for calculating the initial weight of each data source based on historical data includes steps T101-T103:

[0051] T101. Determine the accuracy of each data source based on the deviation between the predicted value of each data source and the actual passenger flow in the historical data, and determine the linear correlation between the historical data of each data source and the actual passenger flow through the Pearson correlation coefficient. ;

[0052] Among them, based on the deviation between the predicted value of the data source in the historical data and the actual passenger flow, the accuracy is defined as , is the accuracy of the i-th data source, is the predicted value of the i-th data source at time t, that is, the collected values ​​of each data source, such as the number of people collected by the checkpoint gate, is the actual passenger flow at time t;

[0053] In the above steps, The data values ​​within 5 minutes counted by each data source are preferred. However, this data value may be affected by equipment limitations and human factors and may have data anomalies (for example, people evading fares). Therefore, the accuracy of the data source is introduced.

[0054] Furthermore, the actual passenger flow at time t in the historical data is first counted The specific steps are as follows:

[0055] 1: Select the checkpoint cameras at the main entrances and exits of the scenic area, obtain the data through the background cameras, and count the number of people entering and leaving (manually determine the area boundaries). Collect the data into the database every minute, including the name of the point camera, the name of the scenic area, the number of people entering, and the number of people leaving.

[0056] 2: Based on the number of parking spaces in parking lots near scenic spots, data is collected every minute, including the parking lot name, the name of the associated scenic spot, the total number of parking spaces, the number of empty parking spaces, and the number of parking spaces.

[0057] 3: Using mobile base stations, according to the coverage range of regional base stations, the selection rule of regional base stations is standard coverage mode:

[0058] A. For base stations outside the scenic area, the base station sector (with a radius of 3000 meters) is used as the base station sector as the coverage range;

[0059] B. For base stations outside the scenic area, the area 5000 meters in the direction of the base station is selected as the base station sector coverage area based on the base station coverage sector (with a radius of 3000 meters).

[0060] C. The collected data includes area number (mobile definition), scenic spot name, 5-minute real-time traffic volume and cumulative traffic volume.

[0061] The actual passenger flow at time t in historical data , iterative optimization is performed through the daily sum (5 minutes * 5 minutes real-time traffic) / the actual cumulative traffic of the day (the actual cumulative traffic of the day is obtained by comparing and adjusting the cumulative traffic of methods 1, 2, and 3, and manually analyzing and adjusting).

[0062] T102. Assign initial weights based on the weighted comprehensive scores of the accuracy and correlation coefficients of each data source;

[0063] Specifically, the accuracy weight is selected to account for 60%, the correlation coefficient weight is selected to account for 40%, and the initial weight is assigned according to the weighted comprehensive score of the accuracy and correlation coefficient. The calculation formula is:

[0064] , is the initial weight of the i-th data source, and n is the total number of data sources;

[0065] T103. Use the gradient descent method to dynamically adjust the weights, and optimize with minimizing the prediction error rate as the objective function. When the error rate decreases by less than a preset threshold (1%) after several consecutive iterations (three times in this embodiment), the optimization is terminated and the final weight distribution result is output.

[0066] Specifically, the objective function is:

[0067] , which is a loss function used to measure the difference between the model prediction results and the real data. Its goal is to minimize the sum of squared prediction errors for all time periods by adjusting the model parameters, where represents the predicted passenger flow at the t-th time point, represents the actual passenger flow at time t.

[0068] In step T200, the expression of the first passenger flow prediction model is:

[0069] (1),

[0070] In expression (1), represents the first predicted passenger flow, Indicates the The contribution weight of each data source to passenger flow prediction is obtained by calculating the accuracy and correlation coefficient of each data source through historical data. Indicates the Real-time data from multiple data sources, Indicates the influence factor of time on passenger flow (e.g. 1.2 on holidays and 0.8 on non-holidays). Indicates the impact factor of weather on tourist flow (e.g. 1.0 for sunny days and 0.5 for rainy and snowy days). The specific value is determined according to the actual situation of the scenic area.

[0071] In step T300, the expression of the second passenger flow prediction model is:

[0072] (2),

[0073] In expression (2), represents the final estimated value of passenger flow, represents the first predicted passenger flow, Indicates the number of active signaling IDs. Indicates the fusion coefficient, the value range is 0.6-0.8. When it is close to 0.8, it is more dependent on the prediction results of multi-source data. When it is close to 0.6, more emphasis is placed on the real-time dynamics of mobile signaling.

[0074] in, The number of active signaling IDs represented is for devices that have stayed within the base station range for more than 10 minutes.

[0075] Compared to existing technologies, this method integrates multimodal data (such as video, ticketing, and checkpoint data) with mobile signaling data, achieving data complementarity and enhancement, effectively addressing the problems of data lag, low accuracy, and limited coverage that exist in traditional passenger flow monitoring methods. Furthermore, by constructing a dynamic weight allocation model and dynamically adjusting weights based on the accuracy and linear correlation of the data sources, the robustness and adaptability of the prediction model are improved, enabling it to more accurately reflect the actual passenger flow of tourist attractions. Furthermore, through a preset mobile signaling collection strategy, the present invention flexibly adjusts the signaling data collection scope based on the needs of different scenarios, further improving the accuracy and practicality of passenger flow estimation. Ultimately, by achieving accurate passenger flow estimation and early warning, it provides powerful data support for the management and decision-making of tourist attractions, helping to improve the operational efficiency and visitor experience of tourist attractions.

[0076] refer to Figure 3 A schematic structural diagram that satisfies steps S100-S300 is provided. Various embodiments of the systems and techniques described above can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), system-on-chip systems (SOCs), programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include: being implemented in one or more computer programs, which can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special-purpose or general-purpose programmable processor, which can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0077] The program code for implementing the method of the present disclosure can be written in any combination of one or more programming languages. These program codes can be provided to a processor or controller of a general-purpose computer, a special-purpose computer, or other programmable data processing device so that when the program code is executed by the processor or controller, the functions / operations specified in the flow chart and / or block diagram are implemented. The program code can be executed entirely on the machine, partially on the machine, as a stand-alone software package, partially on the machine and partially on a remote machine, or entirely on a remote machine or server.

[0078] In the context of the present disclosure, a machine-readable medium may be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device. A machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any suitable combination of the foregoing. More specific examples of machine-readable storage media may include an electrical connection based on one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), optical fibers, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0079] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user can provide input to the computer. Other types of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0080] The systems and techniques described herein can be implemented in a computing system that includes back-end components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes front-end components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such back-end components, middleware components, or front-end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include a local area network (LAN), a wide area network (WAN), and the Internet.

[0081] A computer system may include a client and a server. The client and server are generally remote from each other and typically interact through a communication network. The client-server relationship arises through computer programs running on the respective computers and having a client-server relationship with each other. The server may be a cloud server, a server in a distributed system, or a server integrated with a blockchain.

[0082] It should be understood that the various forms of the processes shown above can be used to reorder, add, or delete steps. For example, the steps described in this disclosure can be performed in parallel, sequentially, or in a different order, as long as the desired results of the technical solutions disclosed in this disclosure can be achieved. This is not a limitation herein.

[0083] The above specific embodiments do not constitute a limitation on the scope of protection of this disclosure. Those skilled in the art will appreciate that various modifications, combinations, sub-combinations, and substitutions may be made based on design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this disclosure shall be included within the scope of protection of this disclosure.

Claims

1. A passenger flow estimation method based on multimodality and mobile signaling, characterized in that: The method comprises: The initial weight of each data source is calculated based on historical data. The weight distribution rule is as follows: the weight of checkpoint gate data is greater than the weight of video surveillance data, the weight of video surveillance data is greater than the weight of ticketing data, and the weight of ticketing data is greater than the weight of parking lot data. Obtain passenger flow data from multiple sources of a target scenic spot, and construct a first passenger flow prediction model based on the passenger flow data from the multiple sources, wherein the multiple sources include scenic spot parking lot video data, scenic spot checkpoint gate data, scenic spot video surveillance data, and scenic spot ticketing data; the expression of the first passenger flow prediction model is: （1）， In expression (1), represents the first predicted passenger flow, Indicates the The contribution weight of each data source to passenger flow prediction is obtained by calculating the accuracy and correlation coefficient of each data source through historical data. Indicates the Real-time data from multiple data sources, represents the impact factor of time on passenger flow, Indicates the impact factor of weather on passenger flow; Based on the preset mobile signaling adoption strategy, the effective mobile signaling data of the target scenic spot is obtained, and the effective mobile signaling data is integrated with the prediction results of the first passenger flow prediction model to construct a second passenger flow prediction model; the expression of the second passenger flow prediction model is: （2）， In expression (2), represents the final estimated value of passenger flow, represents the first predicted passenger flow, Indicates the number of active signaling IDs. Indicates the fusion coefficient, the value range is 0.6-0.8; The passenger flow of the target scenic spot is estimated based on the second passenger flow prediction model to obtain a passenger flow estimation result. During a continuous monitoring cycle, if the estimation results in any two adjacent time periods reach the warning threshold, a warning signal is triggered, and the warning level information corresponding to the passenger flow estimation result is pushed to the management terminal; The preset mobile signaling adoption strategy includes daily mode, holiday mode and emergency mode. The triggering conditions for the preset mobile signaling adoption strategy conversion are: if the ratio between the number of active signaling IDs within the first radius and the second radius and the number of passenger flows collected by the scenic area checkpoint gate reaches a preset threshold, the daily mode is converted to the holiday mode; if the occupancy rate of the scenic area parking lot reaches the warning threshold or the day is a statutory holiday, the holiday mode is converted to the emergency mode; If the mobile signaling collection strategy is the daily mode, the first radius of the coverage sector of the base station outside the target scenic area is used as the coverage range of the base station sector, and the second radius of the coverage sector of the base station within the target scenic area is used as the coverage range of the base station sector, and the second radius is greater than the first radius. The number of active signaling IDs within the first radius and the second radius is collected as valid mobile signaling data; If the mobile signaling acceptance strategy is holiday mode, the first radius of the coverage sector of the base station outside the target scenic area is used as the center of the circle, and the coverage angle is expanded on the original coverage sector angle to form an extended coverage range. The number of active signaling IDs within the extended coverage range and the second radius of the coverage sector of the base station within the target scenic area is collected as valid mobile signaling data; If the mobile signaling adoption strategy is emergency mode, the first radius of the coverage sector of the base station outside the target scenic area is used as the center of the circle, and the coverage angle is expanded to 360° on the original coverage sector angle to form a full coverage range. The full coverage range and the second radius of the coverage sector of the base station within the target scenic area are collected as the number of active signaling IDs within the coverage range of the base station sector as valid mobile signaling data.

2. The passenger flow estimation method based on multimodality and mobile signaling according to claim 1, characterized in that: Before constructing the first passenger flow prediction model based on the passenger flow data from the multi-source data sources, it also includes: performing spatiotemporal alignment and outlier cleaning on the multi-source data, unifying the time base of each data source through timestamps, and filling in missing data using linear interpolation.

3. The passenger flow estimation method based on multimodality and mobile signaling according to claim 2, characterized in that: The method for calculating the initial weight of each data source based on historical data includes: Based on the deviation between the predicted value of each data source and the actual passenger flow in the historical data, the accuracy of each data source is determined, and the linear correlation between the historical data of each data source and the actual passenger flow is determined using the Pearson correlation coefficient; Assign initial weights based on the weighted comprehensive score of the accuracy and correlation coefficient of each data source; The gradient descent method is used to dynamically adjust the weights, and the optimization is performed with minimizing the prediction error rate as the objective function. When the error rate decreases by less than the preset threshold after several consecutive iterations, the optimization is terminated and the final weight distribution result is output.

4. The passenger flow estimation method based on multimodality and mobile signaling according to claim 1, characterized in that: The warning level information includes yellow warning, red warning and explosion warning; If the final estimated passenger flow reaches the first threshold of the scenic area's full load design, a yellow warning message is sent to the management terminal; If the final estimated passenger flow reaches the second threshold of the scenic area's full load design, a red warning message is sent to the management terminal; If the final estimated passenger flow reaches the third threshold of the scenic area's full load design, an overflow warning message will be sent to the management terminal.

Images