A library space service performance optimization method and system based on big data

Abstract

The application relates to a library space service efficiency optimization method and system based on big data. The method comprises the following steps: acquiring real-time space occupancy rate data of each functional area in a library and real-time reader flow data of each connecting channel; based on the real-time space occupancy rate data and the real-time reader flow data, each functional area is divided to obtain a high-demand area set and a potential conversion area set; based on the high-demand area set, the real-time reader flow data and a library space topology graph, the safety of each functional area in the potential conversion area set is evaluated to obtain a safety evaluation index, and based on the safety evaluation index, a functional area is selected from the potential conversion area set to form a target conversion area set; based on the real-time reader flow data and the library space topology graph, paths from the functional areas in the high-demand area set to the functional areas in the target conversion area set are planned to obtain a set of dredging paths, and a space layout optimization instruction is generated. The method can improve the library space service efficiency.

Description

Technical Field

[0001] This invention belongs to the field of smart library construction technology, and in particular relates to a method and system for optimizing library space service efficiency based on big data. Background Technology

[0002] With the continuous development of smart library construction technology, big data analysis and intelligent sensing technology have been gradually introduced into the field of library space management. These technologies make it possible to monitor the density of people and the status of space occupancy in the library in a comprehensive and real-time manner, thereby promoting the evolution of library space services from traditional static planning to data-driven dynamic regulation.

[0003] The current spatial layout of libraries generally adopts a static layout method based on historical experience, which involves functional zoning and long-term maintenance. In this traditional method, administrators divide the space into fixed areas such as quiet reading areas, seminar rooms, and leisure reading areas according to the functional positioning of the library when it was first built. The physical boundaries and functional attributes of each area remain unchanged for a long period of time. In daily operation, the use of space is mainly perceived by librarians through patrols and reader feedback, and minor adjustments are made based on fixed rules or experience. For example, reading seats may be added temporarily during exam weeks or opening hours may be extended, but the overall layout framework remains unchanged.

[0004] However, this static layout method is difficult to cope with the instantaneous fluctuations in reader traffic and activity types. For example, during exam weeks or large academic events, the demand for seminar rooms will increase sharply, forming a peak in reservations. The static layout cannot quickly convert idle document reading areas or underutilized leisure areas into seminar spaces, resulting in the contradiction that while the overall space resources are tight, there are also some areas that are idle. The underlying reason is the lack of real-time and accurate perception of the usage status of the entire library space, resulting in insufficient space service efficiency. Summary of the Invention

[0005] Therefore, it is necessary to provide a big data-based method for optimizing library space service efficiency to address the aforementioned technical problems.

[0006] Firstly, this application provides a method for optimizing the efficiency of library space services based on big data, including:

[0007] Obtain real-time space occupancy data for each functional area in the library and real-time reader flow data for each connecting passageway;

[0008] Based on real-time space occupancy data and real-time reader traffic data, each functional area is divided to obtain a set of high-demand areas and a set of potential conversion areas;

[0009] Based on the high-demand area set, real-time reader traffic data, and a pre-defined library space topology map, a security assessment is conducted on each functional area in the potential conversion area set to obtain a security assessment index. Based on the security assessment index, functional areas are selected from the potential conversion area set to form a target conversion area set.

[0010] Based on real-time reader traffic data and library spatial topology map, a path is planned from the high-demand area to the target conversion area, resulting in a set of diversion paths. Based on the target conversion area set and the diversion path set, spatial layout optimization instructions are generated.

[0011] Furthermore, the real-time space occupancy data was obtained through the following methods:

[0012] Obtain network address parameters and real-time video data for each functional area;

[0013] Target detection is performed on real-time video data to obtain the number of visual recognition personnel in each functional area;

[0014] Input the network address parameters into a preset linear regression model related to the number of people associated with the network address to obtain the number of indirectly identified individuals;

[0015] The corrected real-time number of personnel in each functional area is obtained by weighted summing of the number of personnel indirectly identified and the number of personnel visually identified.

[0016] Based on the revised real-time personnel numbers and the design capacity parameters of each functional area, the real-time space occupancy rate data of each functional area is calculated.

[0017] Furthermore, based on real-time space occupancy data and real-time reader traffic data, the functional areas are divided to obtain a set of high-demand areas and a set of potential conversion areas, including:

[0018] Real-time space occupancy data and real-time reader traffic data are input into the operation mode classification model to obtain the current operation mode label. The operation mode classification model is constructed based on support vector machine.

[0019] Select functional areas whose real-time space occupancy rate data is greater than the preset first occupancy rate threshold to form a high-demand area set;

[0020] Select functional areas whose real-time space occupancy rate data is less than the preset second occupancy rate threshold to form a candidate area set;

[0021] For each functional area in the candidate area set, based on the preset adjacency relationship flag set, the adjacency relationship between the functional area and any functional area in the high-demand area set is determined, and the high-demand adjacency relationship flag value of the functional area is obtained.

[0022] For each functional area in the candidate area set, based on the demand prediction model, the real-time space occupancy rate data, current operation mode label, high demand adjacent relationship identifier value and preset historical occupancy rate data are processed to obtain the future occupancy rate prediction data of the functional area.

[0023] From the candidate region set, functional areas whose future occupancy rate predictions are all lower than the second occupancy rate threshold are selected to form a potential conversion region set.

[0024] Furthermore, the demand forecasting model was obtained through the following method:

[0025] Acquire historical operational data; historical operational data includes the historical occupancy rate sequence of functional areas, historical operational mode labels, historical high-demand adjacent relationship identifiers, and actual occupancy rate data of functional areas;

[0026] Using the historical occupancy rate sequence of functional areas, historical operation mode labels, and historical high-demand adjacent relationship historical identifiers as feature values, and the actual occupancy rate data of functional areas as output labels, the historical operation data are reorganized to obtain the training sample set;

[0027] Decision tree models are constructed by randomly selecting feature values ​​and combinations, and a preliminary demand prediction model is built based on multiple decision tree models.

[0028] The initial demand prediction model is iteratively trained using the training sample set until the preset iterative training requirements are met, resulting in a well-trained demand prediction model.

[0029] Furthermore, the graph nodes in the library space topology map include functional area nodes and passageway nodes. The node characteristic of functional area nodes is the inherent risk coefficient. The connecting edges in the library space topology map are used to represent the connection relationships between the graph nodes, and the edge weight of the connecting edge is the basic travel distance. Based on the high-demand area set, real-time reader traffic data, and the preset library space topology map, a security assessment is conducted on each functional area in the potential conversion area set to obtain a security assessment index, including:

[0030] For each functional area in the potential conversion area set, the functional area node corresponding to the functional area in the library spatial topology map is determined as a potential graph node, and the node features of the potential graph node are extracted to obtain the inherent risk coefficient of the functional area.

[0031] For each functional area in the library, based on the library's spatial topology map, the connecting channels connected to the functional area are determined, forming a set of adjacent connecting channels for the functional area. The real-time reader flow data of each connecting channel in the set of adjacent connecting channels are accumulated to obtain the real-time flow pressure value of the functional area.

[0032] For each functional area in the high-demand area, the functional area node corresponding to the functional area in the library spatial topology map is determined as the high-demand graph node, and the connecting edges connecting the high-demand graph nodes are extracted to obtain the connecting edge set of the high-demand graph nodes.

[0033] The bottleneck edge set of high-demand graph nodes is obtained by filtering the connection edge set based on the preset global bottleneck edge set.

[0034] For each functional area in high-demand regions, traverse each connection edge in the bottleneck edge set, extract the graph node connected to the other end of the connection edge, and form a real-time bottleneck association node set.

[0035] For each functional area in the potential conversion area set, a security assessment index is calculated based on the inherent risk coefficient, real-time traffic pressure value, real-time bottleneck associated node set, and library spatial topology map. The expression for the security assessment index is as follows:

[0036]

[0037] in, It is an index of any functional area in the potential transformation area set. For the first Safety assessment index of each functional area It is the first The inherent risk coefficient of each functional area It is the maximum value of the inherent risk coefficient in the library's spatial topology map. It is the first Real-time traffic pressure values ​​for each functional area It is the maximum real-time traffic pressure value of all functional areas. It is an indicator function. It is the first The potential graph nodes corresponding to each functional area in the library's spatial topology map It is a real-time bottleneck-related node set. It is the inherent risk coefficient weight. It's a traffic value weight. It is a high-demand connection weight.

[0038] Furthermore, based on real-time reader traffic data and the library's spatial topology map, paths are planned from high-demand concentrated functional areas to target conversion concentrated functional areas, resulting in a set of diversion paths, including:

[0039] For each connecting edge in the library's spatial topology graph, the dynamic passage cost parameter of the connecting edge is calculated based on the edge weight and real-time reader traffic data.

[0040] Based on the dynamic spatial topology graph, starting from the high-demand graph node and ending at the functional area graph node corresponding to the target transformation area, the connection path between the starting point and the ending point is traversed, and the connection edges included in the connection path are extracted to form a connection path set.

[0041] For each connection path in the connection path set, the total dynamic passage cost parameter of the connection path is obtained based on the dynamic passage cost parameter of the connection edge.

[0042] Select the connection path corresponding to the minimum total dynamic traffic cost parameter to form a set of diversion paths.

[0043] Secondly, this application also provides a library space service efficiency optimization system based on big data, including:

[0044] The data acquisition module is used to acquire real-time space occupancy data of each functional area in the library and real-time reader flow data of each connecting passage.

[0045] The preliminary area division module is used to divide the functional areas based on real-time space occupancy data and real-time reader traffic data, to obtain a set of high-demand areas and a set of potential conversion areas;

[0046] The final area division module is used to conduct security assessments on each functional area in the potential conversion area set based on the high-demand area set, real-time reader traffic data, and a preset library space topology map, to obtain a security assessment index, and to select functional areas from the potential conversion area set based on the security assessment index to form the target conversion area set.

[0047] The optimization instruction generation module is used to plan the path from the high-demand concentrated functional area to the target conversion concentrated functional area based on real-time reader traffic data and library spatial topology map, obtain the set of diversion paths, and generate spatial layout optimization instructions based on the target conversion area set and the diversion path set.

[0048] Thirdly, this application also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement any of the big data-based library space service efficiency optimization methods described in the first aspect of this application.

[0049] Fourthly, this application also provides a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements any of the big data-based library space service efficiency optimization methods described in the first aspect of this application.

[0050] The aforementioned method and system for optimizing library space service efficiency based on big data acquires real-time space occupancy data for each functional area of ​​the library and real-time reader flow data for each connecting passageway. Based on the real-time space occupancy data and real-time reader flow data, the functional areas are divided into a high-demand area set and a potential conversion area set. Based on the high-demand area set, real-time reader flow data, and a pre-defined library space topology map, a security assessment is performed on each functional area in the potential conversion area set to obtain a security assessment index. Based on the security assessment index, functional areas are selected from the potential conversion area set to form a target conversion area set. Based on the real-time reader flow data and the library space topology map, paths are planned from the functional areas in the high-demand area set to the functional areas in the target conversion area set to obtain a set of guidance paths. Based on the target conversion area set and the guidance path set, space layout optimization instructions are generated. This achieves a complete closed loop from real-time perception and intelligent decision-making to dynamic execution, effectively solving the problem that traditional static layouts cannot cope with instantaneous demand fluctuations. While ensuring reader safety, it improves space resource utilization and service response efficiency. Attached Figure Description

[0051] To more clearly illustrate the technical solutions in the embodiments or related technologies of this application, the accompanying drawings used in the description of the embodiments or related technologies will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0052] Figure 1 A flowchart illustrating a method for optimizing the efficiency of library space services based on big data, provided as an embodiment of this application;

[0053] Figure 2 This is a schematic diagram of the structure of a big data-based library space service efficiency optimization system provided in one embodiment of this application. Detailed Implementation

[0054] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0055] In one embodiment, such as Figure 1As shown, a method for optimizing library space service efficiency based on big data is provided. This embodiment illustrates the application of this method to a management terminal. It is understood that this method can also be applied to a server, and to a system including both a management terminal and a server, and is implemented through the interaction between the management terminal and the server. In this embodiment, the method includes the following S101-S104, wherein:

[0056] S101, obtain real-time space occupancy data of each functional area in the library and real-time reader flow data of each connecting passage.

[0057] Specifically, the management terminal acquires real-time space occupancy data for each functional area in the library and real-time reader flow data for each connecting passageway. A functional area refers to an independent service unit within the library, defined by physical space, such as a quiet reading area, seminar room, leisure reading area, and document reading area. Each functional area has a unique index to represent it. Real-time space occupancy data is used to characterize the population density of any given functional area at any given moment. The functional area, in fact, the mathematical form of its real-time space occupancy data can be expressed as: , indicating the first Each functional area at the current moment Space occupancy rate is defined as the ratio of the current real-time number of people in a functional area to its designed capacity. The range of values ​​is A value greater than 1 indicates overload. Real-time space occupancy data is obtained by collecting raw signals from devices such as Wi-Fi probes, millimeter-wave radar, or high-definition cameras deployed in various functional areas and then calculating the data. Connecting passages refer to the physical pathways within the library that connect different functional areas, including corridors, stairwells, entrances, and exits. Each connecting passage also has a unique index. Real-time reader flow data is used to characterize the number of readers passing through any connecting passage per unit time. For any index... The connection channel, whose real-time reader traffic data can be mathematically represented as: , indicating the first Each connection channel at the current moment Real-time traffic is measured in people per minute. Real-time reader traffic data can be collected using infrared counters, video analytics devices, or Wi-Fi probes deployed at the connection points.

[0058] S102, based on real-time space occupancy data and real-time reader traffic data, divides each functional area to obtain a set of high-demand areas and a set of potential conversion areas.

[0059] Specifically, the management terminal uses real-time space occupancy data from each functional area. Real-time reader traffic data for each connection channel The library's entire functional area is dynamically divided into two disjoint sets. The high-demand area set can be denoted as... This represents the areas that currently require relocation or special attention; its mathematical form is a set of functional area indices. The set of potential transformation regions can be denoted as... This represents a region with low utilization at present that may be converted to other functional uses in the future; its mathematical form is a set of functional area indices. The division can be based on preset threshold rules or a classification model that incorporates real-time traffic data, in order to provide target areas for subsequent dynamic adjustment of spatial resources.

[0060] S103, based on the high-demand area set, real-time reader traffic data and the preset library space topology map, conducts a security assessment on each functional area in the potential conversion area set, obtains a security assessment index, and selects functional areas from the potential conversion area set based on the security assessment index to form a target conversion area set.

[0061] Specifically, the management terminal calls up a preset library space topology map and, based on a set of high-demand areas... For the set of potential conversion regions Risk assessments are conducted for each functional area within the library. The pre-defined library spatial topology map is a graph-structured data representation of the library's physical spatial structure, containing at least the following information: connectivity between all functional areas and connecting passageways, and the static attributes of each connection (such as physical travel distance). For any index in the potential conversion area set... Based on real-time reader traffic data and the likelihood of path conflicts with functional areas in high-demand regions, the management terminal calculates a quantitative security assessment index. Safety assessment index This reflects the potential risks of traffic flow intersections or congestion when a functional area is temporarily converted to other uses; a higher index value indicates a greater risk. Subsequently, the management terminal, based on a preset risk threshold, selects functional areas with a safety assessment index lower than the preset safety threshold from the potential conversion area set as candidate functional areas for conversion. These candidate functional areas are then sorted according to their safety assessment index from lowest to highest. Finally, based on a preset conversion quantity, functional areas are selected from the candidate functional areas to form the target conversion area set. The target conversion region set is the region whose functional attributes need to be changed. The preset security threshold and preset conversion number can be set according to actual work requirements; this embodiment does not further limit the setting of these preset security threshold and preset conversion number.

[0062] S104, based on real-time reader traffic data and library space topology map, plans the path from high-demand concentrated functional area to target conversion concentrated functional area, obtains a set of diversion paths, and generates space layout optimization instructions based on the target conversion area set and diversion path set.

[0063] Specifically, the management terminal utilizes real-time reader traffic data. Based on the library's spatial topology map, optimal reader flow paths are planned between each functional area in the high-demand concentration zone and each functional area in the target transition zone zone. The planning process considers the physical length of the paths and real-time congestion conditions, ensuring that the paths guiding readers from congested areas to newly opened areas are both short and efficient. All planned optimal paths constitute a set of flow paths. Each path typically consists of a series of continuous connecting channels and functional areas. Finally, the management terminal converts the target conversion area set. and the set of diversion paths Generate spatial layout optimization instructions to direct adjustments to the library's spatial layout and services. For example, spatial layout optimization instructions may include two parts: first, instructions to adjust the functional attributes of the target transformation area, such as changing the area's lighting through an intelligent lighting system to indicate the functional change, or adjusting the physical space through movable partitions; second, dynamic guidance instructions for readers, such as showing readers the optimal evacuation path through mobile app push notifications, electronic screen displays, etc.

[0064] This embodiment provides a big data-based method for optimizing library space service efficiency. By acquiring real-time space occupancy data and real-time reader traffic data, it dynamically divides functional areas to obtain a set of high-demand areas and a set of potential conversion areas. Combined with a pre-set library space topology map, a safety assessment is performed to determine safe conversion target areas, resulting in a set of target conversion areas. Real-time evacuation routes are then planned, resulting in a set of evacuation routes, and finally, space layout optimization instructions are generated. This achieves a complete closed loop from real-time perception and intelligent decision-making to dynamic execution, effectively solving the problem that traditional static layouts cannot cope with instantaneous demand fluctuations. While ensuring reader safety, it improves space resource utilization and service response efficiency.

[0065] In one embodiment, real-time space occupancy data is obtained through the following method:

[0066] S201, obtain network address parameters and real-time video data for each functional area.

[0067] Specifically, the management terminal obtains network address parameters and real-time video data from each functional area. For any index in the library... The network address parameters of a functional area refer to the network address information of the reader's mobile device (such as a mobile phone or tablet) captured by the wireless network probe within that functional area. This includes at least the device's MAC address, and can be mathematically represented as a set. Each record Corresponding to a detected device, It is the first The network address of a device (usually a MAC address, which can be hashed to protect privacy). It is the first The signal strength of each device is used to help determine whether the device is indeed located within the functional area. It is the first The detection timestamp of each device. Real-time video data refers to continuous video frames captured by high-definition cameras deployed within the functional area. One or more cameras can be deployed in a functional area to ensure coverage without blind spots. Its mathematical form can be expressed as: ,in For the first Each camera at the current moment Each captured image frame is a three-dimensional tensor with a height dimension. width The number of channels (RGB three channels) records the real-time visual scene information of this functional area. Network address parameters and real-time video data can be acquired through wireless network probes and video capture devices.

[0068] S202 performs target detection on real-time video data to obtain the number of visual recognition personnel in each functional area.

[0069] Specifically, the management terminal inputs real-time video frames from each functional area into a pre-trained target detection model. This model identifies and counts the number of people in each frame, thus determining the number of visually recognized people in each functional area. The pre-trained target detection model can be built using a deep learning-based detector (such as the YOLO series or SSD), and its training process utilizes a large amount of historical image data annotated with human locations. It can output the bounding box of each detected pedestrian in the image. For the first... The first functional area Each camera at the current moment Images acquired The object detection model outputs the number of people detected in the image. . No. Each functional area is covered by multiple cameras. The management terminal also needs to perform deduplication and fusion processing on the number of people detected by these cameras (e.g., through spatial overlap analysis) to obtain the final visual recognition number of people in that functional area. Number of visual recognition personnel It is a non-negative integer, representing the first value estimated based on visual information. Real-time number of people in each functional area. Optionally, if no cameras are deployed in a functional area, the number of visual recognition personnel in that area is considered to be 0.

[0070] S203, input the network address parameters into the preset linear regression model related to the number of network address users to obtain the number of indirectly identified personnel.

[0071] Specifically, for each functional area, the management terminal sets its corresponding network address parameters. Preprocessing is performed to extract feature values ​​for the regression model. These feature values ​​are then input into a pre-defined linear regression model related to the number of people per network address to obtain the estimated number of indirectly identified personnel in the functional area based on network address information. For the first... Each functional area, the management terminal first... The records in the database are filtered and deduplicated: only records with signal strength higher than a preset threshold and timestamps are retained. closest to the current moment The records are then tallied to determine the number of unique network addresses for that functional area at the current moment. The preset linear regression model related to the number of network addresses is a mathematical mapping model pre-trained using historical data. The training process involves the management terminal collecting the number of network addresses in each functional area of ​​the library during a historical period. The actual number of people obtained through manual statistics or other high-precision methods during the corresponding time period. Multiple training samples are formed. The least squares method was used to fit the univariate linear regression equation. Solve for the coefficients and This makes the predicted value Compared with the true value The sum of squared errors between them is minimized. After training is complete, the management terminal will assign the coefficients... and The network address-user correlation linear regression model is solidified and formed. During real-time execution, for the... Each functional area allows the management terminal to extract the number of unique network addresses at the current moment. The number of indirectly identified individuals is calculated from the input model. Indirectly identified number of personnel It is a non-negative real number, representing the first value estimated solely based on network address information. Real-time user count for each functional area. Optionally, if a functional area has no network probes deployed or the number of network addresses collected is zero, then... It can be set to 0 directly.

[0072] S204, the number of indirectly identified personnel and the number of visually identified personnel are weighted and summed to obtain the corrected real-time number of personnel in each functional area.

[0073] Specifically, for any index as In the management area, the management terminal uses visual recognition to count the number of personnel. and the number of indirectly identified personnel By performing weighted fusion, the corrected real-time number of personnel in this functional area can be obtained. The formula for calculating the weighted sum is as follows: ,in For the first Each functional area at the current moment The corrected real-time personnel count. It is the fusion weighting coefficient, and its value range is... This weighting coefficient can be adjusted based on the reliability of each modality of data in the data processing. It can be preset to a fixed value or calculated in real time through an additional confidence assessment model.

[0074] S205 calculates the real-time space occupancy rate data of each functional area based on the corrected real-time personnel number and the design capacity parameters of each functional area.

[0075] Specifically, the management terminal uses the correction of real-time personnel counts. Combined with the pre-stored design capacity parameters of each functional area The real-time space occupancy rate of each functional area is calculated. The pre-stored design capacity parameters for each functional area are also used. This refers to the maximum number of people that can be accommodated in each functional area during the planning and design phase. This number can be calculated based on actual work needs. For example, if a reading room has 100 seats, then... For any index, The management area, in fact, has a high space occupancy rate. The calculation formula is: This ratio is a dimensionless real number, and its value range is usually [range missing]. When the value exceeds 1, it indicates that the area is in an overloaded state.

[0076] This embodiment provides a method for optimizing library space service efficiency based on big data. By acquiring network address parameters and real-time video data, the number of visually identified personnel and the number of indirectly identified personnel are obtained through target detection and linear regression models, respectively. Weighted fusion is then used to obtain a corrected real-time number of personnel. Finally, the real-time space occupancy rate data is obtained by combining the design capacity calculation. This method effectively integrates and improves the accuracy of multi-source heterogeneous sensing data, overcomes the blind spots or noise problems of single sensors, and provides reliable and accurate input data for subsequent space optimization decisions.

[0077] In one embodiment, based on real-time space occupancy data and real-time reader traffic data, the functional areas are divided to obtain a set of high-demand areas and a set of potential conversion areas, including:

[0078] S301. Input the real-time space occupancy rate data and real-time reader traffic data into the operation mode classification model to obtain the current operation mode label. The operation mode classification model is constructed based on support vector machine.

[0079] Specifically, the real-time space occupancy data of each functional area of ​​the management terminal are combined to obtain a real-time space occupancy vector. The real-time reader traffic data from each connection channel is combined to obtain a real-time reader traffic vector. Combining; and also analyzing the current moment. Get the time stamp for that day Weekly Mark and holiday signs Together, they form a high-dimensional feature vector. Among them, week markers Used to represent the current time Which day of the week it falls on, and what are the holiday signs? Used to represent the current time Is it a holiday? The management terminal will display the high-dimensional feature vector. The data is input into a pre-trained operation mode classification model. This model is a multi-classifier built on the Support Vector Machine (SVM) algorithm. Its training process is as follows: The management terminal collects historical library operation records, including the library's space occupancy rate, reader traffic data, and the corresponding day, week, and holiday markers for each record. Then, each historical record is labeled with an operation mode category tag (e.g., "Exam Week Peak Mode," "Regular Weekday Mode," "Weekend Leisure Mode," etc.). This category tag can be set based on domain expert annotations or preset rules (e.g., exam week, winter / summer vacation, weekday, weekend, etc.). Then, the management terminal uses a radial basis function kernel. Historical operation records are mapped to a high-dimensional feature space. The optimal classification hyperplane is found by solving a quadratic programming problem, thereby determining the support vectors and model parameters. After training, a classification model with fixed operation mode parameters is obtained. During real-time operation, the management terminal transmits the high-dimensional feature vectors... The input is fed into the operation mode classification model, which then outputs the current operation mode label. ,in The total number of predefined pattern categories, used to characterize the current overall usage scenario types of the library.

[0080] S302, Select functional areas whose real-time space occupancy rate data is greater than the preset first occupancy rate threshold to form a high-demand area set.

[0081] Specifically, the management terminal traverses the real-time space occupancy data of all functional areas. (in ), This represents the total number of functional zones, and the real-time space occupancy data for each functional zone. Compared with the preset first occupancy threshold A comparison is made. The preset first occupancy threshold is used. This is a pre-set constant that can be determined based on the library's operational experience or historical data statistical analysis. For example, it could be set to 0.8, indicating that a functional area is considered a high-demand area when its occupancy rate reaches 80% or more. For conditions met... The function area, index it Add to collection After traversal, the set of high-demand regions is obtained. .

[0082] S303: Select functional areas whose real-time space occupancy rate data is less than the preset second occupancy rate threshold to form a candidate area set.

[0083] Specifically, the management terminal again scans the real-time space occupancy rate of all functional areas. Real-time space occupancy data for each functional area Compared with the preset second occupancy threshold A comparison is made. The preset second occupancy threshold is used. This setting can be adjusted based on actual operational needs. For example, it can be set to 0.3, indicating that areas with a functional area occupancy rate below 30% are considered low-utilization areas and can be considered potential conversion candidates. For those meeting the conditions... The function area, index it Add to collection After traversing the region, a set of candidate regions is obtained. .

[0084] S304. For each functional area in the candidate area set, based on the preset adjacency relationship flag set, determine the adjacency relationship between the functional area and any functional area in the high-demand area set, and obtain the high-demand adjacency relationship flag value of the functional area.

[0085] Specifically, for candidate region sets Each functional area The management terminal determines whether it is associated with a high-demand area. Any functional area within the system is physically adjacent. The preset set of adjacency markers is... binary matrix , It is the total number of functional areas, where any element Indicates functional area and functional areas They are physically adjacent (e.g., connected by a door or corridor). Table Function Area and functional areas Physically isolated. A pre-defined set of adjacency markers can be constructed based on the library's architectural floor plan. For the candidate area set... Any functional area The management terminal traverses high-demand areas. Each functional area Preset set of adjacency relationship markers Corresponding element in If there is at least one Make Then the candidate region is determined. High-demand adjacency identifier value This indicates that it is adjacent to high-demand areas; otherwise, its high-demand adjacency indicator value is... .

[0086] S305, for each functional area in the candidate area set, based on the demand prediction model, processes the real-time space occupancy rate data, the current operation mode label, the high demand adjacent relationship identifier value and the preset historical occupancy rate data to obtain the future occupancy rate prediction data of the functional area.

[0087] Specifically, the management terminal targets the candidate region set Each functional area Construct real-time feature vectors for prediction. Among them, the real-time feature vector At least include: real-time space occupancy data Current operating mode label High-demand adjacency relationship identifier value The preset historical occupancy data is a pre-stored data structure that covers the historical operation records of all functional areas. Its mathematical form can be represented as a three-dimensional array. ,in This represents the total number of functional zones. The number of time points within a day (e.g., sampled by minute or hour). For the historical number of days. For the [number]th Each functional area, the management terminal according to the current time The corresponding specific time (such as hours and minutes), from Extract the first The functional areas in the past The occupancy rate at the same time of day is calculated, and the average value is also calculated. These are aggregated values ​​representing historical features. The management terminal combines these features into a real-time feature vector. Then, the data is input into a pre-trained demand prediction model. This model, built on a random forest regression algorithm, can output the predicted time points for the functional area based on the input features. The predicted occupancy rate (e.g., the occupancy rate predicted for the first to the Kth prediction steps) is denoted as... These forecasts together constitute the projected future occupancy rate data for this functional area.

[0088] S306. From the candidate region set, select functional areas whose future occupancy rate predictions are all lower than the second occupancy rate threshold to form a potential conversion region set.

[0089] Specifically, the management terminal traverses the candidate region set Each functional area Examine the future occupancy forecast data for this functional area. For each future time point... All have If all predicted values ​​do not exceed the second occupancy threshold, it indicates that the area will remain underutilized for the next hour and has the potential for temporary conversion. Index the functional areas that meet this condition. Add to collection The final set of potential conversion regions is obtained. .

[0090] This embodiment provides a big data-based method for optimizing library space service efficiency. It identifies the current scenario type using an operational mode classification model, obtains the current operational mode label, and quickly filters out high-demand areas using a first occupancy threshold. Combining adjacency relationship identifiers and historical data, a demand prediction model is used to predict future occupancy rates, ultimately yielding a high-demand area set and a potential conversion area set. This achieves dynamic perception and forward-looking prediction of library space demand, providing precise target areas for subsequent security assessments and resource allocation, thereby improving the scientific rigor and timeliness of space optimization.

[0091] In one embodiment, the demand forecasting model is obtained through the following method:

[0092] S401, Obtain historical operation data; historical operation data includes the historical occupancy rate sequence of functional areas, historical operation mode labels, historical high demand adjacent relationship identifiers, and actual occupancy rate data of functional areas.

[0093] Specifically, the management terminal acquires the library's historical operational data. Each historical operational data point corresponds to a historical moment. and a functional area Historical operational data includes that historical moment. The first in the library The data includes the historical occupancy rate sequence, historical operation mode labels, historical high-demand adjacency relationship identifiers, and actual occupancy rate data for each functional area. The historical occupancy rate sequence refers to the sequence of functions. Each functional area at a historical moment The space occupancy data for several consecutive moments prior to this can be represented as six occupancy rates at 10-minute intervals over the past hour, which can be denoted as... ,in The sampling interval is defined as follows. Historical operation mode labels refer to historical moments identified using the same classification method as in step S301. The Middle The operating mode category is marked in each functional area. Historical high-demand adjacency identifier values ​​refer to historical moments. Next, the Each functional area and this historical moment High-demand areas The high-demand adjacency identifier value for any functional area can be denoted as: The calculation method is consistent with step S304, that is, it is based on a preset set of adjacent relationship markers. The actual occupancy rate data for functional areas refers to historical data. Then, within a preset time period (e.g., the next 10 minutes, 20 minutes, up to 60 minutes), the... The actual occupancy rate of each functional area can be denoted as: .

[0094] S402 uses the historical occupancy rate sequence of functional areas, historical operation mode labels, and historical high-demand adjacent relationship historical identifiers as feature values, and the actual occupancy rate data of functional areas as output labels to reorganize the historical operation data to obtain the training sample set.

[0095] Specifically, the management terminal organizes historical operational data into a sample set suitable for training machine learning models. For each historical operational data point, an independent training sample is constructed. The feature values ​​of this training sample are obtained through the following method: [The text abruptly ends here, likely due to an incomplete sentence or missing information.] Expand into a one-dimensional vector, and then combine this one-dimensional vector with the historical running mode labels. Historical high-demand adjacent relationship identifier value The feature values ​​of the training sample are obtained by concatenating (scalar) values. The output label for this training sample is then set to... ,in This represents the number of future time points to predict. All training samples from historical moments are combined to form a training sample set.

[0096] S403, randomly select feature values ​​and combinations to construct decision tree models respectively, and construct a preliminary demand prediction model based on multiple decision tree models.

[0097] Specifically, the management terminal constructs a preliminary demand prediction model based on the training sample set using the random forest algorithm. Random forest is an ensemble learning method whose core idea is to construct multiple decision trees and improve the model's accuracy and stability by aggregating the prediction results of all trees. The construction process is as follows: First, multiple subsets of samples are randomly selected with replacement from the training sample set using bootstrap sampling. Each subset has the same size as the original training sample set, but slightly different sample content, thus ensuring the diversity of training data for each decision tree. For each subset, the terminal constructs a decision tree. When constructing each decision tree, at each node split, a subset of feature values ​​(e.g., the square root of the total number of features) are randomly selected as candidate splitting features. Then, the optimal split point is calculated based on these candidate features to maximize the purity of the child nodes after splitting (usually using mean squared error as the splitting criterion). In this way, multiple decision trees are generated, each learning different patterns from different data subsets and feature subsets. The collection of all decision trees constitutes the preliminary demand prediction model. The key feature of this model is that for a new input sample, each decision tree will independently output a predicted value, and the final prediction result of the model is the mean of the output values ​​of all decision trees (for regression tasks), thereby achieving an accurate estimate of future occupancy rates.

[0098] S404. The initial demand prediction model is iteratively trained using the training sample set until the preset iterative training requirements are met, resulting in a well-trained demand prediction model.

[0099] Specifically, the management terminal uses the training sample set to train and optimize the constructed preliminary demand prediction model. First, the management terminal trains the preliminary model using the training sample set, allowing the model to learn the mapping relationship between training sample feature values ​​and output labels, and calculates the model's prediction error on the training sample set, which can be the root mean square error (RMSE) or the mean absolute error (MAE), to evaluate the model's generalization ability. If the prediction error exceeds a preset accuracy threshold, the management terminal adjusts the hyperparameters of the random forest, such as the number of decision trees. The maximum depth of each tree, the minimum number of samples required for node splitting, etc., are then determined. Step S403 is then re-executed to construct a new preliminary model, which is then trained and validated again. This process is repeated until the performance of the preliminary demand prediction model on the training sample set meets the preset iterative training requirements, resulting in a trained demand prediction model whose parameters (the structure of each decision tree and the splitting threshold) are fixed and can predict the future occupancy rate of the current functional area in real time. For example, the preset iterative training requirements can be that the prediction error is less than a preset accuracy threshold or that the prediction error does not change after multiple consecutive iterations; these requirements can also be set according to actual work needs.

[0100] This embodiment provides a big data-based method for optimizing library space service efficiency. It acquires historical operational data and constructs a training sample set containing feature values ​​and output labels. A preliminary demand prediction model, composed of multiple decision trees, is built using a random forest algorithm. The model's performance is then optimized through iterative training and validation, ultimately resulting in a well-trained demand prediction model. By fully utilizing the spatiotemporal patterns in historical data, it can accurately predict future occupancy rates, providing a reliable basis for dynamic space allocation and thus enhancing the foresight and scientific rigor of library space optimization.

[0101] In one embodiment, the graph nodes in the library space topology map include functional area nodes and passageway nodes. The node characteristic of the functional area nodes is an inherent risk coefficient. The connecting edges in the library space topology map are used to characterize the connection relationships between the graph nodes, and the edge weight of the connecting edge is the basic travel distance. Based on the high-demand area set, real-time reader traffic data, and the preset library space topology map, a security assessment is performed on each functional area in the potential conversion area set to obtain a security assessment index, including:

[0102] S501, for each functional area in the potential conversion area set, the functional area node corresponding to the functional area in the library spatial topology map is determined as a potential graph node, and the node features of the potential graph node are extracted to obtain the inherent risk coefficient of the functional area.

[0103] Specifically, the management terminal first acquires a preset library space topology map. This preset library space topology map is a graph-structured data used to describe the physical spatial structure of the library, and its mathematical form can be represented as a graph. ,in For a set of nodes, This is the set of connecting edges. The set of nodes. It includes two types of nodes: functional area nodes. (Corresponding to each functional area in the library, each functional area node has a unique functional area index) and passageway nodes. (Corresponding to each connecting passage in the library, such as corridor intersections, stairwells, entrances, etc.). Each functional area node The node characteristics include at least the inherent risk coefficient of the functional area. The inherent risk coefficient is a static, pre-calculated value. Its calculation method can be as follows: based on historical data, statistically analyze the average traffic flow of all adjacent channel nodes during typical peak hours, and take the maximum value or weighted average of these average traffic flows as the inherent risk coefficient of the functional area. This coefficient characterizes the potential traffic pressure that the functional area will cause to surrounding channels under normal circumstances. (Connection edge set) Each edge in This indicates that there is a physical channel connection between the two nodes, and each edge has its weight pre-stored. The edge weight is based on the travel distance from node 1. To the node The shortest walking distance, in meters, can be measured based on library building plans. This is for a set of potential conversion areas. Each functional area The management terminal locates the corresponding functional area node in the library's spatial topology map using its unique functional area index. This functional area node Identify a potential graph node and directly read its inherent risk coefficient from its node characteristics. The inherent risk coefficient of this functional area is obtained. .

[0104] S502: For each functional area in the library, based on the library's spatial topology map, determine the connecting channels that connect to the functional area, form a set of adjacent connecting channels for the functional area, and accumulate the real-time reader flow data of each connecting channel in the set of adjacent connecting channels to obtain the real-time flow pressure value of the functional area.

[0105] Specifically, for each functional area The management terminal traverses all connection edges to find all nodes connected to the functional area. Connected channel nodes are denoted as the set of adjacent connected channels. Then, the management terminal obtains the current time. Real-time reader traffic data for the connected channels corresponding to these channel nodes The flow values ​​of all channel nodes within the set are summed to obtain the real-time flow pressure value for that functional area. This value reflects the overall congestion level of the surrounding passageways of this functional area at the current moment.

[0106] S503: For each functional area in the high-demand area cluster, the functional area node corresponding to the functional area in the library spatial topology map is determined as the high-demand graph node, and the connecting edges connecting the high-demand graph nodes are extracted to obtain the connecting edge set of the high-demand graph nodes.

[0107] Specifically, targeting high-demand areas Each functional area The management terminal indexes the library's spatial topology map through its function area. Locate the corresponding functional area node in the graph and designate that node as a high-demand graph node. Subsequently, the management terminal retrieves the edge set from the topology graph. Extract all nodes in the high-demand graph Connecting edges, i.e., all edges connected by... The set of edges connecting the endpoints is denoted as a node in the high-demand graph. Connection edge set .

[0108] S504: Based on the preset global bottleneck edge set, the connection edge set is filtered to obtain the bottleneck edge set of high-demand graph nodes.

[0109] Specifically, the management terminal calls a preset global bottleneck edge set. Preset global bottleneck edge set This is a predefined set of edges containing key connections in the library's spatial topology that are prone to congestion, such as connecting passageways between the main staircase and floors, and narrow corridors leading to popular areas. It also includes a predefined set of global bottleneck edges. This can be pre-defined based on the library's architectural structure, historical traffic data, and daily operational experience. For each high-demand graph node... Connection edge set The management terminal takes its set of global bottleneck edges. The intersection of these nodes yields the nodes of the high-demand graph. Bottleneck edge set ,Right now .

[0110] S505 focuses on each functional area in high-demand regions, traverses each connecting edge in the bottleneck edge set, extracts the graph nodes connected to the other end of the connecting edge, and forms a real-time bottleneck association node set.

[0111] Specifically, targeting high-demand areas Each functional area The management terminal traverses the bottleneck edge set of its corresponding high-demand graph nodes. For each connecting edge in the graph, extract the other graph node of that connecting edge (i.e., the node connected to the graph). (another connected node), and collect all the extracted other end graph nodes to form a functional area. Local real-time bottleneck associated node set Subsequently, the local real-time bottleneck association nodes of all functional areas in the high-demand region are merged to obtain the global real-time bottleneck association node set. .

[0112] S506, for each functional area in the potential conversion area set, calculates the security assessment index of the functional area based on the inherent risk coefficient, real-time traffic pressure value, real-time bottleneck associated node set, and library spatial topology map. The expression for the security assessment index is:

[0113]

[0114] in, It is an index of any functional area in the potential transformation area set. For the first Safety assessment index of each functional area It is the first The inherent risk coefficient of each functional area It is the maximum value of the inherent risk coefficient in the library's spatial topology map. It is the first Real-time traffic pressure values ​​for each functional area It is the maximum real-time traffic pressure value of all functional areas. It is an indicator function. It is the first The potential graph nodes corresponding to each functional area in the library's spatial topology map It is a real-time bottleneck-related node set. It is the inherent risk coefficient weight. It's a traffic value weight. It is a high-demand connection weight.

[0115] Specifically, for the set of potential conversion regions Each functional area The management terminal calculates the functional area by combining its inherent risk coefficient, real-time traffic pressure value, real-time bottleneck associated node set, and library spatial topology map. Safety assessment index In the calculation formula, the first term Contribution to the inherent risk of normalization, among which This can be obtained from step S501. The first term represents the maximum inherent risk coefficient of all functional areas in the library, which can be obtained by traversing the node characteristics of each functional area in the library's spatial topology map. This term reflects the inherent traffic sensitivity of the area; the second term... Contribution to normalized real-time traffic pressure, of which It can be calculated by step S502. This is the maximum real-time traffic pressure value among all functional areas at the current moment (calculated in real time). This item reflects the real-time congestion level of the area's surroundings at the current moment; the third item... Contributing to bottleneck correlation, among which This is the graph node corresponding to this functional area. It is the real-time bottleneck associated node set obtained in step S505, and the indicator function. when belong The value is 1 if the condition is met, and 0 otherwise. This item reflects the risk that, after converting this area to a new function, readers will flow from high-demand areas to this area through the bottleneck channel. (Inherent risk coefficient weight) Traffic value weight High-demand connection weight The preset weighting coefficients satisfy... It can be adjusted based on actual operational experience.

[0116] This embodiment provides a big data-based method for optimizing library space service efficiency. It extracts the inherent risk coefficients of functional areas through a spatial topology map, calculates real-time flow pressure values ​​by combining real-time reader flow, and constructs a real-time bottleneck association node set based on a global bottleneck edge set. Finally, it integrates the above multi-dimensional information to calculate a quantitative safety assessment index, realizing a comprehensive and dynamic assessment of the safety risks of potential conversion areas. It can accurately identify high-risk areas that may cause traffic conflicts after conversion, even if they are currently vacant. This effectively avoids safety hazards before space reorganization and ensures the order of reader passage and the stability of library services.

[0117] In one embodiment, based on real-time reader traffic data and a library spatial topology map, a path is planned from the high-demand concentrated functional area to the target conversion concentrated functional area, resulting in a set of diversion paths, including:

[0118] S601 calculates the dynamic passage cost parameters of each connecting edge in the library spatial topology graph based on the edge weight and real-time reader traffic data.

[0119] Specifically, the management terminal traverses the library's spatial topology map. Each connecting edge in For each connecting edge The management terminal obtains the graph nodes at both ends of the connection edge. and At the present moment Real-time reader traffic data and It should be noted that when one of the graph nodes is a functional area node, the real-time reader traffic of that functional area node is set to 0; when the node is a channel node, This refers to the real-time reader traffic data of the connected channel corresponding to the channel node. The management terminal calculates the dynamic passage cost parameter of the connected edge according to the following formula. : ,in, The pre-set congestion penalty coefficient is used to adjust the degree of impact of congestion on travel costs. It is a positive real number and can be set based on actual operational experience. This represents the maximum real-time reader traffic across all channel nodes at the current moment, used to normalize the average traffic. This is the edge weight of the connecting edge. The physical meaning of this formula is: basic travel distance. It is the main part of the path cost, while the congestion penalty item The base distance is weighted according to the real-time congestion level of the nodes at both ends of the edge. The greater the flow at both ends, the more congested the edge is, and the higher the dynamic passage cost.

[0120] S602, based on the dynamic spatial topology graph, takes the high-demand graph node as the starting point and the functional area graph node corresponding to the target transformation area as the ending point, traverses the connection path between the starting point and the ending point, and extracts the connection edges included in the connection path to form a connection path set.

[0121] Specifically, the management terminal determines the set of starting and ending points for route planning: the starting point is the set of high-demand areas. All corresponding high-demand graph nodes ( The endpoint is the target transformation region set. Function area diagram nodes corresponding to each functional area ( For each pair of starting points and the end point The management terminal is in the dynamic spatial topology map The process involves enumerating or searching for all possible consecutive node sequences (i.e., connection paths) connecting the two points. Each connection path consists of a series of edges that connect the beginning and end of the path. The management terminal extracts all the edges contained in each connection path, forming the edge sequence of that connection path, and collects all the edge sequences of all connection paths to form the set of connection paths between the start and end points. .

[0122] S603: For each connection path in the connection path set, the total dynamic passage cost parameter of the connection path is obtained based on the dynamic passage cost parameter of the connection edge.

[0123] Specifically, for the set of connection paths Each connection path in The connection path consists of a series of continuous connecting edges. composition( (Number of edges contained in the path). Dynamic passage cost parameter for each connected edge in the management terminal. (in Indicates the first in the path The total dynamic toll cost parameter of the connected path is obtained by summing the sums of the edges (each edge). The calculation formula is: This total cost parameter comprehensively reflects the high-demand areas along this connection path from the starting point. Reach the destination target transition area The total cost that needs to be paid.

[0124] S604 Select the connection path corresponding to the minimum total dynamic traffic cost parameter to form a set of diversion paths.

[0125] Specifically, for each pair of high-demand graph nodes at the starting point and endpoint target functional area map nodes The management terminal accesses its corresponding set of connection paths. In the middle, select the total dynamic toll cost parameter. The path with the shortest connection is taken as the optimal routing path between the origin and the destination, denoted as . The management terminal collects all such optimal paths, forming the final set of diversion paths. Each connecting path in this set corresponds to an optimal route from a congested area to a destination transition area, effectively guiding readers to avoid congestion and quickly reach their destination. For example, these optimal paths can also be efficiently found using shortest path search algorithms such as Dijkstra's algorithm or A* algorithm.

[0126] This embodiment provides a big data-based method for optimizing library space service efficiency. By combining basic travel distance and real-time reader flow data to calculate dynamic travel cost parameters, and based on this, selecting the optimal path with the lowest total cost for each pair of high-demand areas and target transition areas, a comprehensive and dynamic set of guidance paths is constructed. By integrating static spatial structure with real-time congestion information, the generated guidance paths not only meet physical accessibility requirements but also dynamically avoid congested nodes, thereby effectively improving reader flow efficiency, preventing secondary congestion caused by improper path planning, and enhancing the feasibility of space reorganization schemes and the reader service experience.

[0127] The aforementioned method for optimizing library space service efficiency based on big data involves acquiring real-time space occupancy data for each functional area of ​​the library and real-time reader flow data for each connecting passageway. Based on these data, the functional areas are divided into a high-demand area set and a potential conversion area set. Using the high-demand area set, real-time reader flow data, and a pre-defined library space topology map, a security assessment is conducted on each functional area within the potential conversion area set to obtain a security assessment index. Based on this index, functional areas are selected from the potential conversion area set to form a target conversion area set. Using the real-time reader flow data and the library space topology map, paths are planned from the high-demand area set to the target conversion area set, resulting in a set of evacuation paths. Finally, based on the target conversion area set and the evacuation path set, a space layout optimization instruction is generated. This achieves a complete closed loop from real-time perception and intelligent decision-making to dynamic execution, effectively solving the problem that traditional static layouts cannot cope with instantaneous demand fluctuations. While ensuring reader safety, it improves space resource utilization and service response efficiency.

[0128] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0129] Based on the same inventive concept, this application also provides a big data-based library space service efficiency optimization system for implementing the aforementioned big data-based library space service efficiency optimization method. The solution provided by this system is similar to the implementation scheme described in the above method. Therefore, the specific limitations of one or more big data-based library space service efficiency optimization system embodiments provided below can be found in the limitations of the big data-based library space service efficiency optimization method described above, and will not be repeated here.

[0130] In one exemplary embodiment, such as Figure 2 As shown, a system 200 for optimizing library space service efficiency based on big data is provided, comprising:

[0131] The data acquisition module 201 is used to acquire real-time space occupancy data of each functional area in the library and real-time reader flow data of each connecting channel;

[0132] The preliminary area division module 202 is used to divide the functional areas based on real-time space occupancy rate data and real-time reader traffic data to obtain a set of high-demand areas and a set of potential conversion areas;

[0133] The final area division module 203 is used to conduct security assessments on each functional area in the potential conversion area set based on the high-demand area set, real-time reader traffic data, and a preset library space topology map, to obtain a security assessment index, and to select functional areas from the potential conversion area set based on the security assessment index to form a target conversion area set.

[0134] The optimization instruction generation module 204 is used to plan the path from the high-demand area to the target conversion area based on real-time reader traffic data and library space topology map, obtain a set of diversion paths, and generate space layout optimization instructions based on the target conversion area set and the diversion path set.

[0135] Furthermore, the system also includes a occupancy rate data generation module, which can be used for:

[0136] Obtain network address parameters and real-time video data for each functional area;

[0137] Target detection is performed on real-time video data to obtain the number of visual recognition personnel in each functional area;

[0138] Input the network address parameters into a preset linear regression model related to the number of people associated with the network address to obtain the number of indirectly identified individuals;

[0139] The corrected real-time number of personnel in each functional area is obtained by weighted summing of the number of personnel indirectly identified and the number of personnel visually identified.

[0140] Based on the revised real-time personnel numbers and the design capacity parameters of each functional area, the real-time space occupancy rate data of each functional area is calculated.

[0141] Furthermore, the preliminary region division module can also be used for:

[0142] Real-time space occupancy data and real-time reader traffic data are input into the operation mode classification model to obtain the current operation mode label. The operation mode classification model is constructed based on support vector machine.

[0143] Select functional areas whose real-time space occupancy rate data is greater than the preset first occupancy rate threshold to form a high-demand area set;

[0144] Select functional areas whose real-time space occupancy rate data is less than the preset second occupancy rate threshold to form a candidate area set;

[0145] For each functional area in the candidate area set, based on the preset adjacency relationship flag set, the adjacency relationship between the functional area and any functional area in the high-demand area set is determined, and the high-demand adjacency relationship flag value of the functional area is obtained.

[0146] For each functional area in the candidate area set, based on the demand prediction model, the real-time space occupancy rate data, current operation mode label, high demand adjacent relationship identifier value and preset historical occupancy rate data are processed to obtain the future occupancy rate prediction data of the functional area.

[0147] From the candidate region set, functional areas whose future occupancy rate predictions are all lower than the second occupancy rate threshold are selected to form a potential conversion region set.

[0148] Furthermore, the system also includes a model building module, which can be used for:

[0149] Acquire historical operational data; historical operational data includes the historical occupancy rate sequence of functional areas, historical operational mode labels, historical high-demand adjacent relationship identifiers, and actual occupancy rate data of functional areas;

[0150] Using the historical occupancy rate sequence of functional areas, historical operation mode labels, and historical high-demand adjacent relationship historical identifiers as feature values, and the actual occupancy rate data of functional areas as output labels, the historical operation data are reorganized to obtain the training sample set;

[0151] Decision tree models are constructed by randomly selecting feature values ​​and combinations, and a preliminary demand prediction model is built based on multiple decision tree models.

[0152] The initial demand prediction model is iteratively trained using the training sample set until the preset iterative training requirements are met, resulting in a well-trained demand prediction model.

[0153] Furthermore, the graph nodes in the library spatial topology graph include functional area nodes and passageway nodes. The node characteristic of functional area nodes is the inherent risk coefficient. The connecting edges in the library spatial topology graph are used to represent the connection relationships between the graph nodes, and the edge weight of the connecting edge is the basic travel distance. The final area division module can also be used for:

[0154] For each functional area in the potential conversion area set, the functional area node corresponding to the functional area in the library spatial topology map is determined as a potential graph node, and the node features of the potential graph node are extracted to obtain the inherent risk coefficient of the functional area.

[0155] For each functional area in the library, based on the library's spatial topology map, the connecting channels connected to the functional area are determined, forming a set of adjacent connecting channels for the functional area. The real-time reader flow data of each connecting channel in the set of adjacent connecting channels are accumulated to obtain the real-time flow pressure value of the functional area.

[0156] For each functional area in the high-demand area, the functional area node corresponding to the functional area in the library spatial topology map is determined as the high-demand graph node, and the connecting edges connecting the high-demand graph nodes are extracted to obtain the connecting edge set of the high-demand graph nodes.

[0157] The bottleneck edge set of high-demand graph nodes is obtained by filtering the connection edge set based on the preset global bottleneck edge set.

[0158] For each functional area in high-demand regions, traverse each connection edge in the bottleneck edge set, extract the graph node connected to the other end of the connection edge, and form a real-time bottleneck association node set.

[0159] For each functional area in the potential conversion area set, a security assessment index is calculated based on the inherent risk coefficient, real-time traffic pressure value, real-time bottleneck associated node set, and library spatial topology map. The expression for the security assessment index is as follows:

[0160]

[0161] in, It is an index of any functional area in the potential transformation area set. For the first Safety assessment index of each functional area It is the first The inherent risk coefficient of each functional area It is the maximum value of the inherent risk coefficient in the library's spatial topology map. It is the first Real-time traffic pressure values ​​for each functional area It is the maximum real-time traffic pressure value of all functional areas. It is an indicator function. It is the first The potential graph nodes corresponding to each functional area in the library's spatial topology map It is a real-time bottleneck-related node set. It is the inherent risk coefficient weight. It's a traffic value weight. It is a high-demand connection weight.

[0162] Furthermore, the optimized instruction generation module can also be used for:

[0163] For each connecting edge in the library's spatial topology graph, the dynamic passage cost parameter of the connecting edge is calculated based on the edge weight and real-time reader traffic data.

[0164] Based on the dynamic spatial topology graph, starting from the high-demand graph node and ending at the functional area graph node corresponding to the target transformation area, the connection path between the starting point and the ending point is traversed, and the connection edges included in the connection path are extracted to form a connection path set.

[0165] For each connection path in the connection path set, the total dynamic passage cost parameter of the connection path is obtained based on the dynamic passage cost parameter of the connection edge.

[0166] Select the connection path corresponding to the minimum total dynamic traffic cost parameter to form a set of diversion paths.

[0167] In one embodiment, a computer device is provided, including a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the previously described method for optimizing the efficiency of library space services based on big data.

[0168] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the steps in the above method embodiments.

[0169] For the device embodiments, since they basically correspond to the method embodiments, the relevant parts can be referred to in the description of the method embodiments. The device embodiments described above are merely illustrative. The components described as separate parts may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this disclosure according to actual needs. Those skilled in the art can understand and implement this without creative effort.

[0170] The above-described embodiments are merely illustrative of several implementation methods of the embodiments of this application, and their descriptions are relatively specific and detailed. However, they should not be construed as limiting the scope of the patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the embodiments of this application, and these modifications and improvements all fall within the protection scope of the embodiments of this application.

Claims

1. A method for optimizing the efficiency of library space services based on big data, characterized in that, The method includes: Obtain real-time space occupancy data for each functional area in the library and real-time reader flow data for each connecting passageway; Based on the real-time space occupancy rate data and the real-time reader traffic data, each of the functional areas is divided to obtain a set of high-demand areas and a set of potential conversion areas; Based on the high-demand area set, the real-time reader traffic data, and the preset library space topology map, a security assessment is performed on each functional area in the potential conversion area set to obtain a security assessment index. Based on the security assessment index, the functional areas are selected from the potential conversion area set to form a target conversion area set. Based on the real-time reader traffic data and the library space topology map, a path is planned from the functional area concentrated in the high-demand area to the functional area concentrated in the target conversion area, resulting in a set of diversion paths. Based on the set of target conversion areas and the set of diversion paths, a space layout optimization instruction is generated.

2. The method according to claim 1, characterized in that, The real-time space occupancy rate data was obtained through the following method: Obtain the network address parameters and real-time video data of each of the aforementioned functional areas; Target detection is performed on the real-time video data to obtain the number of visually recognized personnel in each of the functional areas; The network address parameters are input into a preset linear regression model related to the number of people associated with the network address to obtain the number of indirectly identified personnel. The corrected real-time number of personnel in each functional area is obtained by weighted summing of the number of indirectly identified personnel and the number of visually identified personnel. Based on the corrected real-time personnel count and the design capacity parameters of each functional area, the real-time space occupancy rate data of each functional area is calculated.

3. The method according to claim 1, characterized in that, Based on the real-time space occupancy data and the real-time reader traffic data, the functional areas are divided to obtain a high-demand area set and a potential conversion area set, including: The real-time space occupancy rate data and the real-time reader traffic data are input into the operation mode classification model to obtain the current operation mode label. The operation mode classification model is constructed based on support vector machine. The functional areas whose real-time space occupancy rate data is greater than a preset first occupancy rate threshold are selected to form the high-demand area set. Select the functional areas whose real-time space occupancy rate data is less than a preset second occupancy rate threshold to form a candidate area set; For each functional area in the candidate region set, based on a preset set of adjacency relationship markers, the adjacency relationship between the functional area and any functional area in the high-demand region set is determined, and the high-demand adjacency relationship marker value of the functional area is obtained. For each functional area in the candidate area set, based on the demand prediction model, the real-time space occupancy rate data, the current operating mode label, the high-demand adjacent relationship identifier value, and the preset historical occupancy rate data are processed to obtain the future occupancy rate prediction data of the functional area. From the candidate region set, the functional areas whose future occupancy rate prediction values ​​are all lower than the second occupancy rate threshold are selected to form the potential conversion region set.

4. The method according to claim 3, characterized in that, The demand forecasting model was obtained through the following method: Acquire historical operational data; the historical operational data includes the historical occupancy rate sequence of the functional area, historical operational mode labels, historical high-demand adjacent relationship identifiers, and the actual occupancy rate data of the functional area. Using the historical occupancy rate sequence of the functional area, the historical operation mode label, and the historical high demand adjacency relationship historical identifier as feature values, and the actual occupancy rate data of the functional area as output label, the historical operation data is reorganized to obtain a training sample set; Randomly select the feature values ​​and combinations to construct decision tree models respectively, and construct a preliminary demand prediction model based on multiple decision tree models; The initial demand prediction model is iteratively trained using the training sample set until the preset iterative training requirements are met, resulting in a well-trained demand prediction model.

5. The method according to claim 1, characterized in that, The graph nodes in the library space topology map include functional area nodes and passageway nodes. The node characteristic of the functional area nodes is an inherent risk coefficient. The connecting edges in the library space topology map are used to characterize the connection relationships between the graph nodes, and the edge weight of the connecting edge is the basic travel distance. Based on the high-demand area set, the real-time reader traffic data, and the preset library space topology map, a security assessment is performed on each functional area in the potential conversion area set to obtain a security assessment index, including: For each functional area in the potential conversion area set, the functional area node corresponding to the functional area in the library space topology map is determined as a potential graph node, and the node features of the potential graph node are extracted to obtain the inherent risk coefficient of the functional area. For each functional area in the library, based on the library spatial topology map, each connection channel connected to the functional area is determined to form an adjacent connection channel set for the functional area. The real-time reader flow data of each connection channel in the adjacent connection channel set is accumulated to obtain the real-time flow pressure value of the functional area. For each functional area in the high-demand area cluster, the functional area node corresponding to the functional area in the library space topology map is determined as a high-demand graph node, and the connecting edges connecting the high-demand graph nodes are extracted to obtain the connecting edge set of the high-demand graph nodes. The bottleneck edge set of the high-demand graph node is obtained by filtering the set of connection edges based on a preset global bottleneck edge set. For each functional area in the high-demand area set, traverse each connection edge in the bottleneck edge set, extract the graph node connected to the other end of the connection edge, and form a real-time bottleneck association node set. For each functional area in the potential conversion area set, based on the inherent risk coefficient, the real-time traffic pressure value, the real-time bottleneck associated node set, and the library spatial topology map, the security assessment index of the functional area is calculated, wherein the expression of the security assessment index is: in, It is an index of any functional area in the potential transformation area set. For the first Safety assessment index of each functional area It is the first The inherent risk coefficient of each functional area It is the maximum value of the inherent risk coefficient in the library's spatial topology map. It is the first Real-time traffic pressure values ​​for each functional area It is the maximum real-time traffic pressure value of all functional areas. It is an indicator function. It is the first The potential graph nodes corresponding to each functional area in the library's spatial topology map It is a real-time bottleneck-related node set. It is the inherent risk coefficient weight. It's a traffic value weight. It is a high-demand connection weight.

6. The method according to claim 5, characterized in that, Based on the real-time reader traffic data and the library spatial topology map, a path is planned from the functional area concentrated in the high-demand area to the functional area concentrated in the target conversion area, resulting in a set of diversion paths, including: For each connecting edge in the library space topology graph, the dynamic passage cost parameter of the connecting edge is calculated based on the edge weight and the real-time reader traffic data. Based on the dynamic spatial topology graph, starting from the high-demand graph node and ending at the functional area graph node corresponding to the functional area in the target conversion region set, the connection path between the starting point and the ending point is traversed, and the connection edges included in the connection path are extracted to form a connection path set. For each of the connection paths in the connection path set, the total dynamic travel cost parameter of the connection path is obtained based on the dynamic travel cost parameter of the connection edge. The connection path corresponding to the minimum total dynamic traffic cost parameter is selected to form the diversion path set.

7. A library space service efficiency optimization system based on big data, characterized in that, The system includes: The data acquisition module is used to acquire real-time space occupancy data of each functional area in the library and real-time reader flow data of each connecting passage. The preliminary area division module is used to divide each of the functional areas based on the real-time space occupancy rate data and the real-time reader traffic data to obtain a set of high-demand areas and a set of potential conversion areas; The final area division module is used to conduct a security assessment of each functional area in the potential conversion area set based on the high-demand area set, the real-time reader traffic data, and the preset library space topology map, to obtain a security assessment index, and to select the functional areas from the potential conversion area set based on the security assessment index to form a target conversion area set; The optimization instruction generation module is used to plan a path from the functional area concentrated in the high-demand area to the functional area concentrated in the target conversion area based on the real-time reader traffic data and the library space topology map, obtain a set of diversion paths, and generate a space layout optimization instruction based on the target conversion area set and the diversion path set.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 6.

Images