Method and system for comprehensively managing personnel in region based on big data

Abstract

The invention relates to an intra-region personnel comprehensive management method and system based on big data. The system comprises a system management module, a statistical analysis module and a personnel data prediction module which communicate with one another and are interconnected. The method comprises the following steps: acquiring passenger flow data transmitted by each monitoring camera in a region in real time, monitoring working states of the cameras, and when the cameras have equipment faults and cause missing of the passenger flow data, performing data recovery by adopting a graph neural network algorithm and a time sequence interpolation model based on a self-attention mechanism to obtain complete passenger flow data. Therefore, under the condition that the camera breaks down, data estimation can be carried out through an algorithm, and data change of personnel in a tracking area is facilitated. Algorithms related to an existing image processing mode can be imported, and the implementation difficulty is reduced. And a modularized implementation layout is provided, so that a user can conveniently carry out layout implementation according to the actual condition of the current area.

Description

Technical Field

[0001] The present invention relates to a comprehensive management method and system, and in particular to a comprehensive management method and system for personnel in a region based on big data. Background Art

[0002] At present, the location of people or entry and exit management in a specific area mainly relies on mobile phones as an auxiliary participant. During the implementation, the number of people in the area is determined by determining the location data of the mobile phone. However, this method is often only suitable for situations where the accuracy requirement is not high, and the results often have large deviations for small or irregularly shaped areas.

[0003] At the same time, there are also corresponding personnel counting methods that use image recognition technology to cooperate with cameras. Specifically, it relies on image recognition to determine the entry and exit behavior of relevant personnel in the monitored area based on their actions, and realize real-time passenger flow statistics.

[0004] However, this method is easily affected by hardware facilities such as cameras. For example, network outages and power outages can cause equipment failures, which can lead to errors in the results. At the same time, the commonly used image recognition algorithms still have certain errors, and cumulative errors are prone to occur when counting people. In this way, the error impact on downstream tasks will be further increased.

[0005] In view of the above-mentioned defects, the designer actively conducts research and innovation in order to create a comprehensive regional personnel management method and system based on big data, so as to make it more valuable for industrial use. Summary of the invention

[0006] In order to solve the above technical problems, the purpose of the present invention is to provide a method and system for comprehensive regional personnel management based on big data.

[0007] The regional comprehensive personnel management system based on big data of the present invention includes cameras installed at each entrance and exit of the region, wherein: It includes a system management module, a statistical analysis module, and a personnel data prediction module that communicate with each other. The system management module is used to obtain passenger flow data in the area, pre-process and store the passenger flow data, monitor and maintain the hardware, and process data loss caused by hardware failure; The process of preprocessing and storing the passenger flow data is that the system management module receives and stores the passenger flow inflow and outflow transmitted by the camera within a fixed period, and marks if the data is missing; The monitoring and operation process is to detect whether the camera has a picture at the beginning and end of a fixed period. If there is no picture, it is considered that the camera has a fault and report it to the system management module, and the system management module supplements the lost data through algorithm simulation; The statistical analysis module is used to compare and optimize the passenger flow data collected by the surveillance camera, and to construct a filter based on the difference between the passenger flow inflow and outflow. The change amount presented by the filter is set as the error. Fitting the variation is the error function of passenger flow variation. Eliminating the error function can get the optimized passenger flow. The personnel data prediction module is used to predict the passenger flow in the area and the passenger flow in the surrounding areas in the future time period.

[0008] Furthermore, in the above-mentioned regional comprehensive personnel management system based on big data, the system management module includes a data acquisition submodule, a hardware monitoring submodule, a data missing processing submodule, and a data storage submodule that communicate with each other. The data acquisition submodule is used to acquire passenger flow change data transmitted at fixed periods by the surveillance camera ports at each entrance and exit in the area; The hardware monitoring submodule is used to detect abnormalities of each entrance and exit monitoring camera device at a fixed period using an image recognition algorithm and report abnormalities; The image recognition algorithm detects the camera image and determines that an abnormality occurs if all cells in the image are of a single color; The abnormal report is that after the system management module finds that the camera image is lost, it reports it to the operation and maintenance unit corresponding to the camera; The data missing processing submodule is used to utilize the graph neural network algorithm and the multivariate time series interpolation algorithm based on the self-attention mechanism to recover data when a device abnormality occurs in the surveillance camera; The data storage submodule is used to store passenger flow change data transmitted in a fixed period and equipment abnormality records within a fixed period, wherein the abnormality records include data loss and camera image loss.

[0009] Furthermore, in the above-mentioned regional comprehensive personnel management system based on big data, the statistical analysis module includes a data optimization submodule, a data statistical comparison submodule, a passenger flow situation evolution submodule, a spatiotemporal hot spot clustering analysis submodule, and a personnel flow analysis submodule that communicate with each other. The data optimization submodule calculates the passenger flow change in the area through a centralized fitting mean algorithm, fits the surveillance camera image algorithm, and eliminates it through an error accumulation function, thereby improving the accuracy of passenger flow data; The data statistics comparison submodule performs data statistics on the passenger flow data, where the passenger flow data is the inflow and outflow collected by the camera; The passenger flow situation evolution submodule simulates the passenger flow quantity situation in the area through a smoothing convolution algorithm; The spatiotemporal hotspot clustering analysis submodule uses a multivariate time series clustering algorithm to count the spatiotemporal distribution of peak and trough passenger flow in the region and obtain spatiotemporal hotspot clustering analysis results. The spatiotemporal hotspot clustering is to cluster the personnel data within the corresponding time and point; The personnel flow analysis submodule obtains the personnel flow situation in the area in different time periods through the results of spatiotemporal hot spot clustering analysis.

[0010] Furthermore, in the above-mentioned regional comprehensive personnel management system based on big data, the data statistics comparison submodule can perform year-on-year and month-on-month data statistics, and can realize daily or monthly reports.

[0011] Furthermore, in the above-mentioned regional comprehensive personnel management system based on big data, the personnel data prediction module includes a multivariate spatiotemporal sequence prediction algorithm submodule and a surrounding passenger flow prediction submodule that communicate with each other. The multivariate spatiotemporal sequence prediction submodule uses a deep convolutional residual network model to model the proximity, periodicity, and trend of passenger flow data, and introduces external information through a fully connected layer to obtain passenger flow data in the future time period. The proximity is that the passenger flow at a set time is highly different from the passenger flow at adjacent time; The periodicity includes the time period of the current moment, the time period of yesterday at this moment, the time period of last week at this moment, and the time period of last month at this moment; The trend modeling is a modeling obtained by a smoothing convolution algorithm; The surrounding passenger flow prediction submodule uses the passenger flow data of each entrance and exit of the historical area and the number of people in the surrounding area to establish a generalized linear model, and predicts the passenger flow situation in the surrounding area based on the passenger flow situation in the area. The generalized linear model establishes a generalized linear relationship based on the positive correlation between the passenger flow in and out of the current area and the surrounding area, obtains the headcount of the surrounding area and performs correlation fitting with the camera passenger flow, and uses the fitting function to make predictions.

[0012] Furthermore, in the above-mentioned regional comprehensive personnel management system based on big data, the external information includes external weather information and holiday information.

[0013] The comprehensive regional personnel management method based on big data of the present invention is characterized by comprising the following steps: Step 1: Obtain passenger flow data transmitted by each surveillance camera in the area in real time and monitor the working status of the camera; Step 2: When a camera fails, resulting in missing passenger flow data, the graph neural network algorithm and the time series interpolation model based on the self-attention mechanism are used to recover the data and obtain complete passenger flow data. Step 3: Calculate the passenger flow in and out difference of the passenger flow data, fit the error in the passenger flow data by the central fitting mean, and eliminate it to obtain accurate passenger flow in and out data; Step 4: Count the accurate passenger flow in and out data to obtain the passenger flow change; Step 5: Use the smoothing convolution algorithm to obtain the overall situation of the accurate passenger flow in and out data; Step 6: Perform multivariate time series clustering on the complete passenger flow data to obtain the temporal and spatial distribution of passenger flow peaks and troughs in the region; Step 7: Analyze the temporal and spatial distribution of peak and trough passenger flow in the region to obtain the flow of people in the region; Step 8: For the complete passenger flow data, use the deep convolutional residual network model to predict the passenger flow in and out of each entrance and exit in the future period; Step nine: Use historical accurate passenger flow in and out data and relevant passenger flow data outside the region to establish a generalized linear model to predict relevant passenger flow data outside the region.

[0014] By means of the above scheme, the present invention has at least the following advantages: 1. In the event of a camera failure, an algorithm can be used to make a data estimate to facilitate tracking of data changes of people in the area.

[0015] 2. It can import algorithms related to existing image processing methods to reduce the difficulty of implementation.

[0016] 3. Provides a modular implementation layout to facilitate users to implement layout according to the actual situation of the current area.

[0017] The above description is only an overview of the technical solution of the present invention. In order to more clearly understand the technical means of the present invention and implement it according to the contents of the specification, the following is a detailed description of the preferred embodiments of the present invention in conjunction with the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 It is a schematic diagram of the implementation process of the comprehensive regional personnel management method based on big data in the present invention. DETAILED DESCRIPTION

[0019] The specific implementation of the present invention is further described in detail below in conjunction with the accompanying drawings and examples. The following examples are used to illustrate the present invention, but are not intended to limit the scope of the present invention.

[0020] like Figure 1The regional comprehensive personnel management system based on big data includes cameras installed at each entrance and exit of the area. Its uniqueness lies in: it includes a system management module, a statistical analysis module, and a personnel data prediction module that communicate with each other.

[0021] The system management module is used to obtain passenger flow data in the area, pre-process and store the passenger flow data, monitor and maintain the hardware, and handle data loss caused by hardware failure.

[0022] The process of using passenger flow data for data preprocessing and storage is that the system management module receives and stores the passenger flow inflow and outflow transmitted by the camera within a fixed period, and marks if the data is missing. During the implementation period, the monitoring and operation process is to detect whether the camera has a picture at the beginning and end of the fixed period. If there is no picture, it is considered that the camera has a fault, report it to the system management module, and use the system management module to fill the missing data through algorithm simulation. At the same time, the statistical analysis module is used to perform statistical comparison and optimization on the passenger flow data collected by the monitoring camera. In addition, based on the difference between the passenger flow inflow and outflow, a filter can be constructed. The change presented by the filter is set as an error, and the change is fitted, which is the error function of the passenger flow change. The optimized passenger flow can be obtained by eliminating the error function. Therefore, it can be used to simulate situation evolution, cluster and calculate spatiotemporal hotspots, and analyze personnel flow.

[0023] During implementation, the personnel data prediction module can be used to predict the passenger flow in the area and the passenger flow in the surrounding areas in future time periods.

[0024] It should be noted that the fixed period mentioned in the present invention needs to be determined according to the specific project. For example, 10 minutes can be expressed as the camera detecting the number of people entering and leaving the photographed area from 10:00 to 10:10 on a certain date, and calculating the number of people entering and leaving the photographed area again from 10:10 to 10:20. The specific fixed period can be defined and divided according to the statistical needs of the user, and will not be repeated here.

[0025] In combination with a preferred embodiment of the present invention, the system management module includes a data acquisition submodule, a hardware monitoring submodule, a data missing processing submodule, and a data storage submodule that communicate with each other. Specifically: the data acquisition submodule is used to obtain the passenger flow change data transmitted at a fixed period by the ports of the entrance and exit monitoring cameras in the area. The hardware monitoring submodule is used to use an image recognition algorithm to perform abnormality detection and report abnormalities on each entrance and exit monitoring camera device in a fixed period. The image recognition algorithm detects the camera image, and if all units in the image are a single color, it is judged that an abnormality has occurred. At the same time, the abnormality reporting process adopted is that after the system management module finds that the camera image is lost, it reports it to the operation and maintenance unit corresponding to the camera. Taking into account the convenience of implementing maintenance, the operation and maintenance unit is the maintenance company or individual to which the camera belongs.

[0026] During the implementation, the data missing processing submodule is used to use the graph neural network algorithm (GNN) and the multivariate time series interpolation algorithm (SAITS) based on the self-attention mechanism to recover data when the surveillance camera has equipment abnormalities. The SAITS used in the present invention is the current SOTA method of multivariate time series interpolation. In addition, after considering the geographical location of the cameras in the same venue, the graph neural network algorithm (GNN) is introduced to build a network for the camera. At the same time, the data storage submodule is used to store passenger flow change data transmitted in a fixed period, equipment abnormality records within a fixed period, abnormal records, including data loss, and camera screen loss.

[0027] Further, the statistical analysis module includes a data optimization submodule, a data statistics comparison submodule, a passenger flow situation evolution submodule, a spatiotemporal hotspot clustering analysis submodule, and a personnel flow analysis submodule that communicate with each other. Specifically, the data optimization submodule calculates the passenger flow change in the area through a centralized fitting mean algorithm, fits the surveillance camera image algorithm, and eliminates it through an error accumulation function, thereby improving the accuracy of the passenger flow data. At the same time, relying on the data statistics comparison submodule, the acquired passenger flow data is statistically analyzed. The passenger flow data involved are the inflow and outflow of each entrance and exit collected by the camera. In addition, the passenger flow situation evolution submodule simulates the passenger flow quantity situation in the area through a smoothing convolution algorithm. The smoothing convolution algorithm used in the present invention is a conventional general algorithm, and the simulation process is smoothing convolution. In other words, through smoothing convolution, the features (such as ups and downs within a day) are erased to reflect the trend. The result of using smoothing convolution can show the overall changes in passenger flow, such as more today than yesterday. During the implementation period, the spatiotemporal hotspot clustering analysis submodule adopted by the present invention uses a multivariate time series clustering algorithm to count the spatiotemporal distribution of passenger flow peaks and troughs in the region, and obtain the spatiotemporal hotspot clustering analysis results. The spatiotemporal hotspot clustering adopted is to cluster the personnel data within the corresponding time and point to meet the change processing of personnel data obtained by the cameras installed at each entrance and exit in the region. Furthermore, the personnel flow analysis submodule can be used to obtain the personnel flow situation in the region in different time periods through the spatiotemporal hotspot clustering analysis results. Considering the convenience of data processing, the data statistics comparison submodule can perform year-on-year and month-on-month data statistics, and can realize daily or monthly reports.

[0028] In view of the actual implementation, the personnel data prediction module adopted by the present invention includes a multivariate spatiotemporal sequence prediction algorithm submodule and a surrounding passenger flow prediction submodule that communicate with each other. Specifically, the multivariate spatiotemporal sequence prediction submodule models the proximity, periodicity, and trend of passenger flow data through a deep convolution residual network model, introduces external information through a fully connected layer, and obtains passenger flow data in the future time period. During the implementation period, the proximity adopted is that the passenger flow at the set time is highly correlated with the passenger flow at the adjacent time. The periodicity includes the time period at the current time, the time period at this moment yesterday, the time period at this moment last week, and the time period at this moment last month. The trend modeling is the modeling obtained by the smoothing convolution algorithm. At the same time, the surrounding passenger flow prediction submodule can be used to establish a generalized linear model by combining the passenger flow data of each entrance and exit of the historical area with the number of people in the surrounding area, and predict the passenger flow situation in the surrounding area through the passenger flow situation in the area. Relying on the generalized linear model, a generalized linear relationship can be established based on the positive correlation between the in-and-out passenger flow of the current area and the surrounding area, and the number of people in the surrounding area is obtained. The correlation fitting is performed, and the fitting function is used for prediction. Moreover, the current area can be divided according to the specific location during the implementation of the present invention, which may include living areas, surrounding factory areas, etc., which will not be repeated here. Furthermore, considering that external information may affect passenger flow data, the external information involved in the present invention includes other modeling information such as external weather information and holiday information to improve the accuracy of the final data.

[0029] In order to better implement the present invention, a method for comprehensive management of personnel in a region based on big data is provided, which is characterized by comprising the following steps: Step 1: Obtain the passenger flow data transmitted by each surveillance camera in the area in real time and monitor the working status of the camera.

[0030] Step 2: When the camera has equipment failure, resulting in missing passenger flow data, the graph neural network algorithm and the time series interpolation model based on the self-attention mechanism are used to recover the data and obtain complete passenger flow data. During the implementation period, the time series interpolation model used was constructed using the SAITS technology. SAITS technology is Self-Attention-based Imputation for Time Series, which is a prior art and will not be described here. The time series interpolation model constructed in this way can use the geographic location information of each camera to build a network, and establish a graph neural network based on the weighted adjacency matrix of the location information to accelerate model training. At the same time, in the early stage of construction, the amount of passenger flow data obtained by the camera is not very large. The camera's geographic location is used as a priori data to provide a reference for the correlation of passenger flow data of each camera, which speeds up the construction of the time series interpolation model. Afterwards, based on the spatiotemporal laws of the passenger flow sequence data of each camera, the missing data is simulated and the data is recovered.

[0031] Step three, for the incomplete and fixed time period passenger flow data, the passenger flow difference is calculated, the error in the passenger flow data is fitted by the centralized fitting mean, and it is eliminated to obtain accurate passenger flow data.

[0032] Step 4: Count the accurate passenger flow in and out data to obtain the passenger flow changes. During this period, the year-on-year and month-on-month changes in passenger flow in the region can be compared from a time dimension.

[0033] Step 5: For accurate passenger flow data, use the convolution smoothing algorithm to obtain the overall trend of passenger flow data. Specifically, define a mean filter as the convolution kernel of unit time (hour, day or month), and slide the convolution kernel on the passenger flow sequence data. After that, calculate the new value of the center pixel (that is, the mean). The curve constructed by the new value of the center pixel reduces the impact of random fluctuations, highlights the long-term trend, and makes it easier to observe the trend of passenger flow.

[0034] Step six, perform multivariate time series clustering on the complete passenger flow data to obtain the spatiotemporal distribution of passenger flow peaks and troughs in the region. During implementation, the k-shape algorithm can be used to calculate the sequence contour coefficient for the existing complete passenger flow sequence data, find the number of clusters k when the contour coefficient is the largest, and use the optimal number of clusters to obtain the sequence clustering label. Therefore, multivariate time series clustering can help observe the spatiotemporal points with similar passenger flow changes. For example, in terms of time, set the peak point on weekdays and the peak point on holidays. In terms of space, if the entrance and exit 1 is consistent with the entrance and exit 2 at time t1, it can be adjusted and managed.

[0035] Step seven, analyze the temporal and spatial distribution of peak and trough passenger flow in the area to obtain the personnel flow situation in the area.

[0036] Step 8: For the complete passenger flow data, use the deep convolutional residual network model to predict the passenger flow at each entrance and exit in the future time period.

[0037] Step nine, use the historical accurate passenger flow in and out data and relevant passenger flow data outside the region to establish a generalized linear model to predict relevant passenger flow data outside the region. During the implementation period, relevant passenger flow data outside the region are fully authorized by all parties, and the use and processing comply with relevant national and regional laws, regulations and standards to improve data accuracy. Among them, the generalized linear model is a prior art, and the conversion model of its independent variable and dependent variable should be determined according to the specific data, which will not be repeated here. In addition, there are many ways to obtain relevant passenger flow data outside the region, which can be achieved by conventional methods in the prior art, which will not be repeated here.

[0038] The simple implementation process of the present invention is as follows: 1. Obtain the original data of people entering and leaving the camera, and there is data missing.

[0039] 2. Use interpolation algorithms to repair and adjust data.

[0040] 3. Use the repaired data for analysis, including: a) Statistical analysis (month-on-month and year-on-year comparisons, etc.).

[0041] b) Situation simulation.

[0042] c) Judging peaks and troughs.

[0043] d) Simulate the number of people in the periphery.

[0044] It can be seen from the above textual description and the accompanying drawings that the present invention has the following advantages: 1. In the event of a camera failure, an algorithm can be used to make a data estimate to facilitate tracking of data changes of people in the area.

[0045] 2. It can import algorithms related to existing image processing methods to reduce the difficulty of implementation.

[0046] 3. Provides a modular implementation layout to facilitate users to implement layout according to the actual situation of the current area.

[0047] In addition, the indicated orientations or positional relationships described in the present invention are all based on the orientations or positional relationships shown in the accompanying drawings and are only for the convenience of describing the present invention and simplifying the description. They do not indicate or imply that the device or structure referred to must have a specific orientation or be operated with a specific orientation structure. Therefore, they cannot be understood as limitations on the present invention.

[0048] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. It should be pointed out that a person skilled in the art can make several improvements and modifications without departing from the technical principles of the present invention, and these improvements and modifications should also be regarded as within the scope of protection of the present invention.

Claims

1. A comprehensive regional personnel management system based on big data, including cameras installed at each entrance and exit of the region, characterized by: It includes a system management module, a statistical analysis module, and a personnel data prediction module that communicate with each other. The system management module is used to obtain passenger flow data in the area, pre-process and store the passenger flow data, monitor and maintain the hardware, and process data loss caused by hardware failure; The process of preprocessing and storing the passenger flow data is that the system management module receives and stores the passenger flow inflow and outflow transmitted by the camera within a fixed period, and marks if the data is missing; The monitoring and operation process is to detect whether the camera has a picture at the beginning and end of a fixed period. If there is no picture, it is considered that the camera has a fault and report it to the system management module, and the system management module supplements the lost data through algorithm simulation; The statistical analysis module is used to compare and optimize the passenger flow data collected by the surveillance camera, and to construct a filter based on the difference between the passenger flow inflow and outflow. The change amount presented by the filter is set as the error. Fitting the variation is the error function of passenger flow variation. Eliminating the error function can get the optimized passenger flow. The personnel data prediction module is used to predict the passenger flow in the area and the passenger flow in the surrounding areas in the future time period.

2. The regional comprehensive personnel management system based on big data according to claim 1 is characterized by: The system management module includes a data acquisition submodule, a hardware monitoring submodule, a data loss processing submodule, and a data storage submodule that communicate with each other. The data acquisition submodule is used to acquire passenger flow change data transmitted at fixed periods by the surveillance camera ports at each entrance and exit in the area; The hardware monitoring submodule is used to detect abnormalities of each entrance and exit monitoring camera device at a fixed period using an image recognition algorithm and report abnormalities; The image recognition algorithm detects the camera image and determines that an abnormality occurs if all cells in the image are of a single color; The abnormal report is that after the system management module finds that the camera image is lost, it reports it to the operation and maintenance unit corresponding to the camera; The data missing processing submodule is used to utilize the graph neural network algorithm and the multivariate time series interpolation algorithm based on the self-attention mechanism to recover data when a device abnormality occurs in the surveillance camera; The data storage submodule is used to store passenger flow change data transmitted in a fixed period and equipment abnormality records within a fixed period, wherein the abnormality records include data loss and camera image loss.

3. The regional comprehensive personnel management system based on big data according to claim 1 is characterized by: The statistical analysis module includes a data optimization submodule, a data statistical comparison submodule, a passenger flow situation evolution submodule, a spatiotemporal hotspot clustering analysis submodule, and a personnel flow analysis submodule. The data optimization submodule calculates the passenger flow change in the area through a centralized fitting mean algorithm, fits the surveillance camera image algorithm, and eliminates it through an error accumulation function, thereby improving the accuracy of passenger flow data; The data statistics comparison submodule performs data statistics on the passenger flow data, where the passenger flow data is the inflow and outflow collected by the camera; The passenger flow situation evolution submodule simulates the passenger flow quantity situation in the area through a smoothing convolution algorithm; The spatiotemporal hotspot clustering analysis submodule uses a multivariate time series clustering algorithm to count the spatiotemporal distribution of peak and trough passenger flow in the region and obtain spatiotemporal hotspot clustering analysis results. The spatiotemporal hotspot clustering is to cluster the personnel data within the corresponding time and point; The personnel flow analysis submodule obtains the personnel flow situation in the area in different time periods through the results of spatiotemporal hot spot clustering analysis.

4. The regional comprehensive personnel management system based on big data according to claim 3 is characterized by: The data statistics comparison submodule can perform year-on-year and month-on-month data statistics, and can realize daily or monthly reports.

5. The regional comprehensive personnel management system based on big data according to claim 1 is characterized by: The personnel data prediction module includes a multivariate spatiotemporal sequence prediction algorithm submodule and a surrounding passenger flow prediction submodule that communicate with each other. The multivariate spatiotemporal sequence prediction submodule uses a deep convolutional residual network model to model the proximity, periodicity, and trend of passenger flow data, and introduces external information through a fully connected layer to obtain passenger flow data in the future time period. The proximity is that the passenger flow at a set time is highly different from the passenger flow at adjacent time; The periodicity includes the time period of the current moment, the time period of yesterday at this moment, the time period of last week at this moment, and the time period of last month at this moment; The trend modeling is a modeling obtained by a smoothing convolution algorithm; The surrounding passenger flow prediction submodule uses the passenger flow data of each entrance and exit of the historical area and the number of people in the surrounding area to establish a generalized linear model, and predicts the passenger flow situation in the surrounding area based on the passenger flow situation in the area. The generalized linear model establishes a generalized linear relationship based on the positive correlation between the passenger flow in and out of the current area and the surrounding area, obtains the headcount of the surrounding area and performs correlation fitting with the camera passenger flow, and uses the fitting function to make predictions.

6. The regional comprehensive personnel management system based on big data according to claim 5 is characterized by: The external information includes external weather information and holiday information.

7. A comprehensive regional personnel management method based on big data, characterized by The following steps are involved: Step 1: Obtain passenger flow data transmitted by each surveillance camera in the area in real time and monitor the working status of the camera; Step 2: When a camera fails, resulting in missing passenger flow data, the graph neural network algorithm and the time series interpolation model based on the self-attention mechanism are used to recover the data and obtain complete passenger flow data. Step 3: Calculate the passenger flow in and out difference of the passenger flow data, fit the error in the passenger flow data by the central fitting mean, and eliminate it to obtain accurate passenger flow in and out data; Step 4: Count the accurate passenger flow in and out data to obtain the passenger flow change; Step 5: Use the smoothing convolution algorithm to obtain the overall situation of the accurate passenger flow in and out data; Step 6: Perform multivariate time series clustering on the complete passenger flow data to obtain the temporal and spatial distribution of passenger flow peaks and troughs in the region; Step 7: Analyze the temporal and spatial distribution of peak and trough passenger flow in the region to obtain the flow of people in the region; Step 8: For the complete passenger flow data, use the deep convolutional residual network model to predict the passenger flow in and out of each entrance and exit in the future period; Step nine: Use historical accurate passenger flow in and out data and relevant passenger flow data outside the region to establish a generalized linear model to predict relevant passenger flow data outside the region.

Images