Library space resource dynamic scheduling control method based on internet of things

Abstract

The present application relates to the technical field of intelligent control, and especially relates to a library space resource dynamic scheduling control method based on the Internet of Things. The method comprises the following steps: collecting global monitoring classification data of the library, wherein the global monitoring classification data of the library comprises space resource data, personnel data, environmental data and space function resource operation data of the library; analyzing space flow distribution characteristics through the space resource data and the personnel data to generate space flow distribution characteristic data; analyzing space response correlation characteristics of the environmental data and the space function resource operation data based on the space flow distribution characteristic data to generate space multi-source response correlation characteristic data; and analyzing library space resource dynamic scheduling control parameters based on the space multi-source response correlation characteristic data to generate library space resource dynamic scheduling control data. The present application realizes intelligent scheduling control distribution of the library space.

Description

Technical Field

[0001] This invention relates to the field of intelligent control technology, and in particular to a method for dynamic scheduling and control of library space resources based on the Internet of Things. Background Technology

[0002] With the continuous development of information technology and intelligent management, modern libraries have gradually transformed from traditional static borrowing spaces into comprehensive public spaces integrating learning, discussion, exhibition, and exchange. Libraries possess a rich variety of internal spatial resources, including reading areas, study areas, stacks, multimedia learning areas, discussion areas, and smart service areas. The rapid popularization of IoT technology has led to the deployment of numerous sensor devices in libraries, such as infrared sensors, environmental monitoring nodes, personnel positioning devices, seat occupancy detection modules, and energy consumption monitoring terminals. These devices enable real-time perception and data collection of personnel behavior, environmental conditions, and equipment operation, providing a rich data foundation for spatial resource management. With the diversification of user needs and the dynamic changes in pedestrian flow, efficient utilization of library space resources and a balance of comfort can be achieved without increasing hardware investment. However, existing technologies that use machine learning or statistical analysis to predict library visitor flow often only consider single-dimensional changes in visitor flow or the usage of local spaces. They lack the ability to model the interaction between visitor flow, environment, and the operational status of functional resources, and have obvious lag and low accuracy problems. They are unable to cope with real-time changes in visitor flow dynamics, resulting in delayed resource allocation and a decline in user experience. Summary of the Invention

[0003] Based on this, the present invention provides a dynamic scheduling and control method for library space resources based on the Internet of Things to solve at least one of the above-mentioned technical problems.

[0004] To achieve the above objectives, a dynamic scheduling and control method for library space resources based on the Internet of Things (IoT) includes the following steps:

[0005] Step S1: Perform global monitoring and classification processing of the library through the library's IoT monitoring equipment to generate global monitoring classification data of the library, wherein the global monitoring classification data of the library includes spatial resource data, personnel data, environmental data and spatial functional resource operation data of the library;

[0006] Step S2: Analyze the spatial pedestrian flow distribution characteristics using spatial resource data and personnel data to generate spatial pedestrian flow distribution characteristic data;

[0007] Step S3: Based on spatial pedestrian flow distribution characteristic data, perform spatial response correlation characteristic analysis on environmental data and spatial functional resource operation data to generate spatial multi-source response correlation characteristic data;

[0008] Step S4: Establish a demand prediction and optimization mapping relationship for spatial resource regulation based on spatial multi-source response correlation feature data, and generate an optimized spatial resource regulation demand prediction model; use the optimized spatial resource regulation demand prediction model to analyze the library spatial resource dynamic scheduling and control parameters of spatial functional resource operation data, and generate library spatial resource dynamic scheduling and control data; execute intelligent spatial resource dynamic scheduling and control operations based on the library spatial resource dynamic scheduling and control data.

[0009] Furthermore, the space function resource operation data mentioned in step S1 can be used to perform scheduling and control operations on the space resource data.

[0010] Furthermore, step S1 includes the following steps:

[0011] Step S11: Perform global monitoring and processing of the library through the library's IoT monitoring equipment to generate global monitoring data of the library;

[0012] Step S12: Perform monitoring and sensing dimension analysis based on the IoT monitoring device to obtain monitoring and sensing dimension data, and perform sensing dimension correction processing on the library global monitoring data through the monitoring and sensing dimension data to generate corrected library global monitoring data.

[0013] Step S13: Perform time-series and spatial synchronization mapping processing on the calibration library global monitoring data to generate synchronized library global monitoring data;

[0014] Step S14: Perform monitoring data classification processing on the synchronized library global monitoring data to generate library global monitoring classification data.

[0015] Furthermore, step S2 includes the following steps:

[0016] Step S21: Perform spatial resource characteristic analysis and processing on the spatial resource data to generate spatial resource characteristic data;

[0017] Step S22: Perform library spatial resource modeling based on spatial resource characteristic data to generate a library spatial resource model;

[0018] Step S23: Map personnel data to the library space model to perform personnel perception space mapping processing, and generate spatial personnel perception data;

[0019] Step S24: Perform local spatial and short-term temporal occupancy distribution tensor analysis based on spatial personnel perception data to generate spatiotemporal personnel tensor data;

[0020] Step S25: Perform temporal intrinsic correlation analysis of regional personnel behavior based on spatiotemporal personnel tensor data to generate temporal intrinsic correlation data of regional personnel behavior, and perform dynamic perception analysis of regional personnel behavior through the temporal intrinsic correlation data of regional personnel behavior to generate dynamic perception data of regional personnel behavior.

[0021] Step S26: Analyze the spatial pedestrian flow distribution characteristics based on the dynamic perception data of regional personnel behavior to generate spatial pedestrian flow distribution characteristic data.

[0022] Furthermore, step S24 includes the following steps:

[0023] Multi-scale convolutional local feature analysis is performed on spatial personnel perception data to generate local spatial personnel perception feature data.

[0024] Perform short-term temporal feature analysis on space personnel perception data to generate short-term temporal feature data of personnel perception.

[0025] Based on local spatial personnel perception feature data and short-term temporal feature data of personnel perception, a local spatial and short-term temporal occupancy distribution personnel perception tensor analysis is performed to generate spatiotemporal personnel tensor data.

[0026] Furthermore, step S26 includes the following steps:

[0027] The spatial flow and migration probability analysis is performed on the dynamic perception data of regional personnel behavior to generate spatial personnel flow and migration probability data. Based on the spatial personnel flow and migration probability data, spatial flow vector field analysis is performed to generate spatial flow vector field data.

[0028] Based on the dynamic perception data of regional personnel behavior, regional personnel stay characteristics are analyzed to generate regional personnel stay characteristic data.

[0029] Based on spatial pedestrian flow vector field data and regional personnel dwell characteristic data, spatial pedestrian flow distribution characteristic analysis is performed to generate spatial pedestrian flow distribution characteristic data.

[0030] Furthermore, step S3 includes the following steps:

[0031] Step S31: Analyze the environmental status of the impact of pedestrian flow on the environmental data using spatial pedestrian flow distribution characteristic data, generate pedestrian flow impact environmental status data, and analyze the spatial environmental response characteristics of pedestrian flow disturbance based on the pedestrian flow impact environmental status data, generate pedestrian flow disturbance spatial environmental response characteristic data.

[0032] Step S32: Based on the spatial functional resource operation data and spatial pedestrian flow distribution characteristic data, perform a two-way correlation characteristic analysis between pedestrian flow and functional resource operation to generate spatial pedestrian flow-functional resource operation related data;

[0033] Step S33: Based on the spatial functional resource operation data and the spatial environment response characteristic data of human flow disturbance, perform a two-way correlation characteristic analysis of the operation of the environment and functional resources to generate spatial environment-functional resource operation related data;

[0034] Step S34: Based on the spatial human flow-functional resource operation related data and the spatial environment-functional resource operation related data, conduct spatial functional resource operation regulation and response analysis to generate spatial functional resource operation regulation and response data;

[0035] Step S35: Perform spatial response correlation characteristic analysis on the spatial environment response characteristic data of pedestrian flow disturbance and the spatial function resource operation and regulation response data to generate spatial multi-source response correlation characteristic data.

[0036] Furthermore, step S4 includes the following steps:

[0037] Step S41: Perform temporal extension analysis on the spatial multi-source response correlation feature data to generate temporal extension data of spatial multi-source response correlation features;

[0038] Step S42: Perform spatial multi-source response trend feature analysis based on the temporal extension data of spatial multi-source response correlation features to generate spatial multi-source response trend feature data;

[0039] Step S43: Analyze the operational constraints of spatial functions and resources based on spatial resource data and operational data of spatial functions and resources, and generate operational constraint data of spatial functions and resources.

[0040] Step S44: Establish a demand forecasting mapping relationship for spatial resource regulation by using spatial multi-source response trend characteristic data and spatial functional resource operation constraint characteristic data, so as to obtain a spatial resource regulation demand forecasting model;

[0041] Step S45: Design multi-objective optimization indicators for library space based on the preset library space resource optimization demand data and the spatial resource regulation demand prediction model, and use the multi-objective optimization indicators for library space to perform multi-objective optimization allocation of model parameters for the spatial resource regulation demand prediction model to obtain the optimized spatial resource regulation demand prediction model.

[0042] Step S46: Utilize the optimized spatial resource regulation demand prediction model to analyze the library spatial resource dynamic scheduling and control parameters based on the spatial function resource operation data, and generate library spatial resource dynamic scheduling and control data;

[0043] Step S47: Execute intelligent spatial resource dynamic scheduling and control operations based on library spatial resource dynamic scheduling and control data.

[0044] Furthermore, step S41 includes the following steps:

[0045] Based on the preset gated recursive unit algorithm, a preliminary temporal extension analysis of spatial multi-source response correlation features is performed on the spatial multi-source response correlation feature data to generate preliminary temporal extension data of spatial multi-source response correlation features.

[0046] Based on the spatial multi-source response correlation feature data, an analysis of the impact of adjacent nodes on spatial multi-source response is performed to generate spatial multi-source response adjacent node impact data;

[0047] By correcting the temporal extension deviation of the influence of spirit nodes on the preliminary temporal extension data of spatial multi-source response correlation features using the influence data of adjacent nodes of spatial multi-source response, the temporal extension data of spatial multi-source response correlation features is generated.

[0048] Furthermore, step S43 includes the following steps:

[0049] Based on the spatial function resource operation data, the spatial resource regulation characteristics of the spatial resource operation are analyzed to generate spatial resource regulation characteristic data of the functional operation.

[0050] Based on the functional operation space resource regulation characteristic data, analyze the functional operation space resource regulation mode and generate functional operation space resource regulation mode data;

[0051] Based on the spatial function resource operation data and the spatial resource regulation mode data of the function operation, multi-dimensional constraint inversion processing of spatial function resource operation is performed to generate multi-dimensional constraint inversion data of spatial function resource operation.

[0052] The spatial function resource operation constraint characteristics are analyzed by performing multi-dimensional constraint inversion data on spatial function resource operation, and spatial function resource operation constraint characteristic data is generated.

[0053] The beneficial effects of this application are as follows: By collecting, correcting, and classifying library-wide monitoring data at multiple levels, this invention enables comprehensive perception and dynamic control of library space resources, personnel activities, environmental conditions, and functional resource operation. This step, through the introduction of a multi-dimensional sensing system of IoT monitoring devices, achieves simultaneous collection of heterogeneous data from multiple sources, such as temperature and humidity, light intensity, noise, air quality, equipment energy consumption, and personnel density. Through sensor dimension correction and spatiotemporal synchronization mapping, the spatial consistency and temporal accuracy of the monitoring data are significantly improved. The corrected global monitoring data is then classified into four categories: space resource data, personnel data, environmental data, and functional resource operation data. This provides a high-quality, computable data foundation for subsequent spatial flow analysis and resource regulation. Simultaneously, the spatial functional resource operation data can directly execute scheduling and control operations on the space resource data, enabling adaptive adjustment of library resources such as lighting, air conditioning, and seating systems, thereby improving the utilization efficiency of library space resources, environmental comfort, and energy efficiency. Through correlation modeling and in-depth analysis of space resource data and personnel data, refined modeling and dynamic characterization of the internal flow distribution of the library are achieved. By analyzing and modeling spatial resource characteristics, a digital resource model of the library space was constructed, providing a spatial framework for personnel perception and occupancy distribution analysis. Subsequently, personnel data was mapped to the spatial model to generate multi-dimensional personnel perception data. Spatiotemporal personnel tensor data was generated through a combination of multi-scale convolution and short-term temporal series analysis, thereby capturing local space occupancy patterns and short-term pedestrian flow trends. Through temporal correlation analysis and dynamic perception processing of regional personnel behavior, the behavioral migration characteristics and time-dependent patterns of personnel between different functional areas can be revealed. This allows for the quantification and predictability of pedestrian flow distribution based on spatial pedestrian flow vector fields and dwell characteristics. The system can identify potential high-density areas and spatial congestion risks in advance, enabling intelligent response control such as pedestrian flow guidance, lighting adjustment, seat allocation, and environmental optimization, providing strong predictive basis for subsequent spatial resource regulation and comfort optimization. By deeply exploring the multi-dimensional coupling relationship between pedestrian flow, environment, and functional resource operation within the library based on spatial pedestrian flow distribution characteristic data, the accuracy and intelligence level of spatial response regulation are significantly improved. Analysis of the impact of pedestrian flow distribution on environmental conditions can identify environmental disturbance characteristics such as temperature and humidity changes, noise increases, and air quality declines caused by dense pedestrian traffic, and establish a quantitative correlation model between pedestrian flow disturbances and environmental responses. Through bidirectional correlation analysis between pedestrian flow distribution characteristics and spatial functional resource operation data, the dynamic matching relationship between functional resources such as lighting, air conditioning, and seating management systems and pedestrian flow can be identified, enabling the system to understand the dynamic chain between "pedestrian behavior, environmental changes, and resource responses."By comprehensively analyzing the interaction characteristics of environmental and functional operation, the system can deduce the real-time regulation requirements of spatial functional resources. Furthermore, through multi-source data fusion, it generates spatial multi-source response correlation characteristic data, providing high-quality input for subsequent scheduling and prediction models. This transforms library management from traditional static facility management to a proactive space management mechanism centered on pedestrian flow response, achieving dynamic balance regulation of library space comfort, energy efficiency, and utilization. A high-precision spatial resource regulation demand prediction model is constructed based on spatial multi-source response correlation characteristics, enabling forward-looking optimized scheduling and intelligent control of library space resources. Through temporal extension and trend analysis, the system captures the evolutionary patterns of spatial pedestrian flow, environmental, and functional resource operation status over time. Combined with gated recursive unit algorithms and adjacent node influence correction techniques, it effectively overcomes the time lag and nonlinear interference problems in multi-source dynamic data, thus obtaining more stable and reliable trend characteristic data. Through spatial functional resource operation constraint characteristic analysis and multi-dimensional constraint inversion processing, the system can identify functional coupling and mutual exclusion relationships between different spatial resources, forming a multi-dimensional constraint model for resource regulation, ensuring the feasibility and optimal energy efficiency of the prediction results in actual implementation. Based on a demand forecasting model established using spatial multi-source response trend characteristic data and constraint characteristic data, combined with multi-objective optimization index design and model parameter optimization allocation, the system can achieve precise scheduling of multi-dimensional spatial resources such as lighting, air conditioning, seating utilization, and functional configuration of learning and reading areas. Through dynamic scheduling control parameter analysis and execution, the system can adjust the configuration of internal library space resources in real time, achieving coordinated optimization of energy conservation, comfort improvement, and intelligent operation. This effectively enhances the proactive control capability of library space operation and provides strong technical support for the refined, green, and adaptive management of smart libraries.

[0054] Therefore, the IoT-based dynamic scheduling and control method for library space resources of this invention achieves collaborative perception and intelligent analysis of the library's internal pedestrian flow status, spatial environment parameters, and functional resource operation status by constructing a multi-dimensional integrated dynamic response scheduling model for pedestrian flow. Compared with existing pedestrian flow analysis technologies that rely solely on single-dimensional prediction, this method, within a unified IoT data framework, comprehensively utilizes environmental monitoring data, equipment energy consumption data, and personnel behavior data to establish a dynamic coupling model among pedestrian flow, environment, and resources, thereby significantly improving the accuracy of prediction and the real-time nature of response. By extracting features and recognizing patterns from multi-source data, it achieves adaptive regulation and energy efficiency optimization of spatial resource distribution. It can adjust the allocation of lighting, air conditioning, and reading resources in real time according to pedestrian flow fluctuations, reducing energy consumption while improving user comfort and resource utilization. This not only overcomes the problems of prediction lag and slow regulation in existing technologies but also achieves efficient matching of library space resources and pedestrian flow distribution, significantly improving overall operational efficiency and user experience. Attached Figure Description

[0055] Figure 1 This is a flowchart illustrating the steps of a dynamic scheduling and control method for library space resources based on the Internet of Things according to the present invention.

[0056] Figure 2 for Figure 1 A detailed flowchart illustrating the implementation steps of step S4.

[0057] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0058] The technical method of the present invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0059] Furthermore, the accompanying drawings are merely illustrative of the invention and are not necessarily drawn to scale. The same reference numerals in the drawings denote the same or similar parts, and therefore repeated descriptions of them will be omitted. Some block diagrams shown in the drawings are functional entities and do not necessarily correspond to physically or logically independent entities. Functional entities may be implemented in software, or in one or more hardware modules or integrated circuits, or in different network and / or processor methods and / or microcontroller methods. The term "and / or" as used herein includes any and all combinations of one or more of the associated items listed.

[0060] To achieve the above objectives, please refer to Figures 1 to 2 This invention provides a method for dynamic scheduling and control of library space resources based on the Internet of Things. In the embodiments of this invention, please refer to... Figure 1 The diagram shown is a flowchart illustrating the steps of a dynamic scheduling and control method for library space resources based on the Internet of Things (IoT) according to the present invention. The method includes the following steps:

[0061] Step S1: Perform global monitoring and classification processing of the library through the library's IoT monitoring equipment to generate global monitoring classification data of the library, wherein the global monitoring classification data of the library includes spatial resource data, personnel data, environmental data and spatial functional resource operation data of the library;

[0062] In this embodiment of the invention, when performing global monitoring and classification of the library, an IoT monitoring network covering all functional areas of the library is first deployed. This network includes four types of core monitoring devices: Space resource monitoring devices employ ultrasonic sensors and RFID tag readers. Ultrasonic sensors are installed directly above each reading seat to ensure accurate detection of seat occupancy. RFID tag readers are deployed near the entrances and exits of each seminar room and storage areas to identify the usage status of seminar rooms and the occupancy of storage space. Personnel data monitoring devices use binocular vision cameras and infrared array sensors. Binocular vision cameras are installed on the top of each floor entrance to cover the entire entrance area to count the number of people entering and exiting. Infrared array sensors are embedded in the desktops of each reading area and the floor of the corridor to capture the real-time location and movement trends of personnel. Environmental data monitoring devices are multi-parameter sensing modules integrating temperature, humidity, TVOC concentration, and PM2.5 concentration detection functions. These modules are installed near the ceiling ventilation openings in each area to ensure that the detection data reflects the overall environmental status of the area. Space function resource operation data monitoring devices are intelligent control modules that are directly connected to the controllers of the air conditioning, lighting, and fresh air systems via an industrial bus to obtain real-time equipment operating parameters. All monitoring devices collect raw data at fixed time intervals. During collection, timestamp synchronization technology ensures data from different devices remains consistent across time, preventing discrepancies caused by data latency. After raw data collection, classification rules are established based on the data representation object and source: data related to seating, seminar rooms, and storage spaces collected by ultrasonic sensors and RFID tag readers are classified as spatial resource data; data on the number, location, and movement of personnel collected by binocular vision cameras and infrared array sensors are classified as personnel data; temperature, humidity, and air quality data collected by multi-parameter sensing modules are classified as environmental data; and equipment operating power, on / off status, and adjustment parameters collected by intelligent control modules are classified as spatial function resource operation data. After classification, the data is cleaned and processed. An outlier determination mechanism based on historical data fluctuation range is used; data exceeding this range is marked as anomalies. Linear interpolation is used to fill in the outlier positions with adjacent normal data, ultimately generating the library's global monitoring classification data.

[0063] Step S2: Analyze the spatial pedestrian flow distribution characteristics using spatial resource data and personnel data to generate spatial pedestrian flow distribution characteristic data;

[0064] In this embodiment of the invention, spatial resource data is structured and analyzed to extract key information such as area division, total number of seats, and functional area types (e.g., reading area, discussion area, leisure area). A graphical modeling technique is used to construct a library spatial topology model. During the modeling process, each floor is converted into a digital map according to the actual floor plan of the library. The boundaries of each functional area, the specific locations of seats, and the distribution paths of passageways are marked on the map to ensure that the model accurately maps the physical space of the library. Next, the personnel data is processed in depth to extract information such as personnel entry time, exit time, and real-time location coordinates. Coordinate matching technology is used to align the real-time location coordinates of personnel with the coordinate system in the spatial topology model, accurately mapping the personnel positions to the corresponding physical areas in the model, forming a dynamic distribution view of personnel within the space. Subsequently, spatiotemporal data analysis methods were employed to mine the mapped personnel distribution data from multiple dimensions: Spatially, the real-time number of people in each functional area was counted, and personnel density was calculated based on the area, reflecting the degree of personnel gathering in different areas; Temporally, the dwell time of people in each area was calculated by the difference between the entry time and the current time, analyzing the dwell preferences of people in different areas; Motionally, the movement trajectory of people was tracked by fitting the position coordinates of people at continuous time points, tracking the movement paths of people between different areas. Simultaneously, tensor analysis technology was introduced to fuse the personnel density in the spatial dimension, the dwell time in the temporal dimension, and the trajectory features in the motion dimension into a spatiotemporal personnel tensor. Tensor decomposition algorithms were used to process this tensor, extracting core features, including personnel gathering hotspots (areas with consistently higher personnel density than other areas), peak personnel flow periods (the periods with the most personnel movements per unit time), and main personnel flow paths (paths with the highest overlap in personnel movement trajectories). These core features were then combined to generate comprehensive spatial personnel flow distribution characteristic data reflecting the distribution patterns and flow trends of people within the library.

[0065] Step S3: Based on spatial pedestrian flow distribution characteristic data, perform spatial response correlation characteristic analysis on environmental data and spatial functional resource operation data to generate spatial multi-source response correlation characteristic data;

[0066] In this embodiment of the invention, when performing spatial response correlation feature analysis of pedestrian flow distribution characteristics, a multi-dimensional correlation analysis framework is built using spatial pedestrian flow distribution feature data as the core benchmark. At the environmental data correlation analysis level, the Pearson correlation analysis algorithm is used to calculate the correlation coefficients between pedestrian density, pedestrian dwell time, and corresponding regional temperature, humidity, TVOC concentration, and PM2.5 concentration. The magnitude of the correlation coefficient is used to determine the strength of the correlation between pedestrian flow distribution and environmental parameters; the closer the absolute value of the coefficient is to 1, the stronger the correlation. Simultaneously, the time points when significant changes occur in pedestrian density (e.g., increases or decreases exceeding 20%) are recorded, and the time intervals between subsequent changes in environmental parameters are tracked to determine the time lag of the impact of pedestrian flow changes on the environment. This clarifies the response patterns of environmental parameters under pedestrian flow disturbances and generates spatial environmental response feature data for pedestrian flow disturbances. At the level of spatial functional resource operation data correlation analysis, a dual correlation analysis is implemented: The first analysis examines the correlation between pedestrian flow distribution and the operational status of functional resources, statistically analyzing the specific changes in air conditioning operating power, lighting activation, and fresh air system airflow within different pedestrian density ranges, establishing a correspondence between pedestrian density and resource operation parameters, such as the average change in air conditioning operating power for every 10% increase in pedestrian density; the second analysis examines the correlation between environmental parameters and the operational status of functional resources, setting comfort thresholds for environmental parameters (e.g., temperature 22-26℃, humidity 40%-60%), and recording the adjustment trends of air conditioning and fresh air system operating parameters when environmental parameters exceed these thresholds, such as the percentage increase in air conditioning cooling power when the temperature is above 26℃. After obtaining the interaction mechanism among pedestrian flow, environment, and functional resources through dual correlation analysis, principal component analysis is used to fuse the spatial environmental response characteristic data of pedestrian flow disturbance with the functional resource operation correlation characteristic data, eliminating redundant information and retaining features that reflect the core correlation among the three, generating spatial multi-source response correlation characteristic data.

[0067] Step S4: Establish a demand prediction and optimization mapping relationship for spatial resource regulation based on spatial multi-source response correlation feature data, and generate an optimized spatial resource regulation demand prediction model; use the optimized spatial resource regulation demand prediction model to analyze the library spatial resource dynamic scheduling and control parameters of spatial functional resource operation data, and generate library spatial resource dynamic scheduling and control data; execute intelligent spatial resource dynamic scheduling and control operations based on the library spatial resource dynamic scheduling and control data.

[0068] In this embodiment of the invention, spatial multi-source response correlation feature data undergoes temporal extension processing. A gated recursive unit algorithm is employed, using correlation feature data from historical time periods as training samples to learn the patterns of data change over time. Based on these patterns, the changing trends of correlation feature data within a future preset time period are predicted, generating temporal extension data of spatial multi-source response correlation features. Next, combining resource capacity limitations (such as maximum personnel capacity and total number of seats in each area), functional area usage constraints (such as time limits for seminar room use and quiet zone requirements) from spatial resource data, and equipment operating parameter thresholds (such as maximum air conditioning power and maximum lighting energy consumption limits) from spatial functional resource operation data, spatial resource regulation constraints are established, clarifying the inviolable boundary ranges during scheduling. Based on the temporal extension data of spatial multi-source response correlation features and the regulation constraints, a gradient boosting tree algorithm is used to construct a spatial resource regulation demand prediction model. During the model training phase, historical scheduling data is used as samples, with historical correlation feature data as input and the corresponding actual resource regulation demand as output. By iteratively adjusting the model parameters, the deviation between the model prediction results and actual demand is controlled within a preset range. Subsequently, a weighted summation multi-objective optimization algorithm was introduced, with the optimization objectives of maximizing resource utilization, minimizing energy consumption, and maximizing user comfort. Weights were assigned to each objective based on the library's operational priorities. This algorithm was then used to optimize the parameters of the spatial resource regulation demand prediction model, determining the optimal weights for each predictive factor and generating an optimized spatial resource regulation demand prediction model. Spatial functional resource operation data was input into this optimization model. By comparing current resource operation parameters with predicted resource regulation demands, the model calculated the necessary adjustments to scheduling and control parameters, such as air conditioning operating power, number of lights turned on, fresh air system airflow, and seat allocation schemes, forming dynamic scheduling and control data for library spatial resources. Finally, the scheduling and control data was transmitted to the control terminals of each functional resource via an industrial bus. The control terminals automatically adjusted the equipment operating status based on the data instructions and simultaneously published seat vacancy information and recommended areas through the display systems on each floor of the library, guiding people to vacant areas and completing the intelligent dynamic scheduling and control operation of spatial resources.

[0069] Furthermore, the space function resource operation data mentioned in step S1 can be used to perform scheduling and control operations on the space resource data.

[0070] Furthermore, step S1 includes the following steps:

[0071] Step S11: Perform global monitoring and processing of the library through the library's IoT monitoring equipment to generate global monitoring data of the library;

[0072] In this embodiment of the invention, when performing global library monitoring, the deployment of IoT monitoring devices across the entire area is first completed: Space resource monitoring uses ultrasonic sensors and RFID tag readers. Ultrasonic sensors are installed directly above each reading seat to detect whether the seat is occupied. RFID tag readers are deployed at the entrances and exits of each seminar room and next to storage shelves to identify the door control status of seminar rooms and the occupancy status of storage units. Personnel monitoring uses binocular vision cameras and infrared array sensors. Binocular vision cameras are installed at the top of each floor entrance to cover the entire field of view to count the number of people entering and exiting. Infrared array sensors are embedded in the desktops of each reading area and the floor of the corridor to capture the real-time location coordinates of personnel. Environmental monitoring uses a multi-parameter sensing module integrating temperature, humidity, TVOC concentration, and PM2.5 concentration, installed below the ceiling vents in each area to ensure that the detection data reflects the overall environmental status of the area. Space function resource operation monitoring uses an intelligent control module, which is directly connected to the controllers of the air conditioning, lighting, and fresh air systems via an industrial bus to obtain real-time equipment operating power, on / off status, and adjustment parameters. All devices start data acquisition at fixed time intervals. During the acquisition process, hardware clock synchronization technology ensures that the acquisition actions of each device are triggered simultaneously. The raw data collected is transmitted to the local data processing node in real time. After preliminary formatting (unifying data field names and data types), it is integrated to form global library monitoring data covering all areas and all monitoring dimensions of the library.

[0073] Step S12: Perform monitoring and sensing dimension analysis based on the IoT monitoring device to obtain monitoring and sensing dimension data, and perform sensing dimension correction processing on the library global monitoring data through the monitoring and sensing dimension data to generate corrected library global monitoring data.

[0074] In this embodiment of the invention, the monitoring sensing dimensions are divided according to the functional attributes of the IoT monitoring devices: ultrasonic sensors and RFID tag readers are classified as "spatial resource sensing dimension", binocular vision cameras and infrared array sensors are classified as "personnel sensing dimension", multi-parameter sensing modules are classified as "environmental sensing dimension", and intelligent control modules are classified as "functional resource sensing dimension". By analyzing the detection principles (such as ultrasonic sensors based on echo ranging and infrared array sensors based on thermal radiation sensing) and error sources (such as ultrasonic interference due to obstruction and infrared interference due to ambient light), monitoring sensing dimension data containing the error range and correction requirements of each dimension are generated. Subsequently, based on this data, the library's overall monitoring data was corrected in several dimensions: For the spatial resource sensing dimension, a Kalman filter algorithm was used to filter the seat occupancy signals collected by the ultrasonic sensors, removing misjudgments caused by obstruction; for the personnel sensing dimension, cross-validation was used, comparing the number of people counted by the binocular cameras with the number of personnel positions captured by the infrared array sensors. If the difference exceeded a preset range, the binocular camera data was used (binocular vision has higher accuracy due to stereo imaging); for the environmental sensing dimension, a sliding window averaging method was used to smooth the real-time data collected by the multi-parameter sensing modules, eliminating fluctuations in environmental parameters caused by instantaneous airflow; for the functional resource sensing dimension, benchmark comparison was used, comparing the equipment operating parameters collected by the intelligent control module with the equipment's rated parameter range, eliminating abnormal data exceeding the rated range (such as fault data where the air conditioner power suddenly drops to 0). After these dimensional corrections, corrected library overall monitoring data with accuracy meeting the requirements of subsequent analysis was generated.

[0075] Step S13: Perform time-series and spatial synchronization mapping processing on the calibration library global monitoring data to generate synchronized library global monitoring data;

[0076] In this embodiment of the invention, the collection timestamps of each data point in the global monitoring data of the library are extracted and corrected. Using the timestamp of the highest-precision device (binocular vision camera, timestamp accuracy down to milliseconds) as a benchmark, the timestamps of data from other devices are compared. If there is a time deviation (e.g., the timestamp of the multi-parameter sensor module lags by 1 second), linear interpolation is used to supplement data points within the deviation time period, ensuring that the data collected by different devices correspond one-to-one on the time axis and eliminating the temporal misalignment caused by data transmission delay. Next, spatial synchronization mapping is performed: a digital spatial topology model is established based on the library's physical spatial layout. The model is marked with the installation location coordinates of each monitoring device (e.g., "ultrasonic sensor above reading seat 1-01 in Area A, first floor, coordinates X100 / Y50"). The device identifiers of each device in the correction data are extracted, and their installation location coordinates are associated with the device identifiers. The corresponding monitoring data is mapped to the corresponding location in the spatial topology model, achieving a precise correspondence between "data and physical spatial location." For example, the temperature and humidity data collected by a multi-parameter sensor module is mapped to the location of the module in the "leisure area in Area B, second floor" in the model. After time synchronization and spatial mapping integration, synchronized global library monitoring data with both accurate time and spatial attributes is generated.

[0077] Step S14: Perform monitoring data classification processing on the synchronized library global monitoring data to generate library global monitoring classification data.

[0078] In this embodiment of the invention, classification rules are established based on preset categories (spatial resource data, personnel data, environmental data, and spatial function resource operation data) of the library's global monitoring classification data: spatial resource data must include the fields of "area identifier, resource type (seat / discussion room / storage unit), total number of resources, number of occupied, and occupancy status"; personnel data must include the fields of "area identifier, real-time number of people, personnel location coordinates, number of people entering, and number of people leaving"; environmental data must include the fields of "area identifier, temperature, humidity, TVOC concentration, and PM2.5 concentration"; and spatial function resource operation data must include the fields of "area identifier, equipment type (air conditioning / lighting / fresh air), operating power, on / off status, and adjustment parameters (such as air conditioning temperature setting value)". Subsequently, the synchronized library global monitoring data underwent field matching and data segmentation: each synchronized data entry was traversed, and its category was determined based on field content. For example, data containing the "seat occupancy status" field was categorized into spatial resource data, data containing the "personnel location coordinates" field was categorized into personnel data, data containing the "TVOC concentration" field was categorized into environmental data, and data containing the "equipment operating power" field was categorized into spatial function resource operation data. During the segmentation process, redundant fields that were repeated across categories (such as "area identifier," which is common to all categories) were retained to ensure the independence of data in each category. Invalid fields (such as the "PM2.5 concentration" field in spatial resource data) were deleted, ultimately resulting in four categories of library global monitoring classification data that are structurally independent, have complete fields, and are accurate.

[0079] Furthermore, step S2 includes the following steps:

[0080] Step S21: Perform spatial resource characteristic analysis and processing on the spatial resource data to generate spatial resource characteristic data;

[0081] In this embodiment of the invention, when performing spatial resource characteristic analysis on spatial resource data, a multi-dimensional feature extraction algorithm is used to decompose and quantify the core characteristic dimensions from the spatial resource data. First, the analysis dimensions are defined as follows: Area function type, space capacity, equipment configuration status, and usage constraints. Area function type analysis involves matching area identifiers (e.g., "Reading Area-A1", "Seminar Room-B3") in spatial resource data with a pre-defined functional classification system to clarify the unique functional attributes of each area. Simultaneously, usage scenario data from IoT devices (e.g., whether seminar rooms are connected to video conferencing equipment) is used to verify the accuracy of the function type. Space capacity analysis is based on the actual area data of the area and the library's space design standards (average area per person ≥ 1.5㎡). The theoretical capacity is calculated by "area area ÷ standard average area per person," and then adjusted based on the actual seating layout (e.g., number of fixed seats, number of movable seats) to ensure the capacity data aligns with actual usage scenarios. Equipment configuration status analysis retrieves operational status data from IoT monitoring devices for space-bound functional equipment (air conditioning, lighting, charging sockets), statistically analyzing the equipment integrity rate and equipment type proportion in each area to form equipment configuration characteristics. Usage constraint analysis combines library management rules (e.g., seminar room single-use duration ≤ 3 minutes). (For example, group discussions are prohibited in quiet areas for hours). The constraints are transformed into quantifiable parameters (such as duration thresholds and noise limits). Throughout the analysis, the feature extraction algorithm sets the sliding window size to 10 minutes to ensure real-time capture of characteristic data changes. At the same time, data normalization is used to uniformly map characteristic data of different dimensions to the [0,1] interval, ultimately generating spatial resource characteristic data containing parameters such as regional function type, actual capacity, equipment configuration integrity rate, usage duration threshold, and noise limit.

[0082] Step S22: Perform library spatial resource modeling based on spatial resource characteristic data to generate a library spatial resource model;

[0083] In this embodiment of the invention, when modeling library spatial resources based on spatial resource characteristic data, three-dimensional modeling software is used to build a digital spatial model, and the modeling process strictly follows the three-step process of "data mapping - accuracy calibration - dynamic association". First, import the CAD floor plans of each floor of the library and construct a basic spatial framework at a 1:1 scale. This framework includes physical structures such as walls, doors, windows, and passageways, with dimensional errors controlled within ±2mm. Then, map the spatial resource characteristic data to the basic framework according to area identifiers. Specifically, embed function type labels (e.g., blue for reading areas, green for seminar rooms) into the corresponding area model, label actual capacity values, and set capacity warning visualization trigger conditions (the area model edge turns yellow when capacity reaches 90%). Associate equipment configuration data (e.g., use icons to mark the locations of air conditioners and lighting in the model; icons turn red when equipment malfunctions), and write usage constraint parameters (e.g., display a "≤3 hours" duration indicator next to the seminar room model). Next, establish a spatial coordinate system, using a Cartesian coordinate system to assign unique three-dimensional coordinates to each area (X-axis represents horizontal, Y-axis represents vertical, and Z-axis represents floor height), with coordinate accuracy down to the centimeter level to ensure that the position matching error during subsequent personnel data mapping is ≤5cm. Finally, set up a dynamic model update mechanism via API. The interface connects to the spatial resource characteristic data repository. When the characteristic data changes (such as equipment failure or capacity adjustment), the model completes data synchronization and update within 10 seconds. It also supports access from multiple terminals (such as management computers and lobby displays), and finally generates a library spatial resource model that accurately reflects the library's spatial physical structure, resource characteristics and dynamic changes.

[0084] Step S23: Map personnel data to the library space model to perform personnel perception space mapping processing, and generate spatial personnel perception data;

[0085] In this embodiment of the invention, when mapping personnel data to the library space model for personnel perception space mapping processing, a coordinate matching algorithm is used to achieve accurate association between data and model. The whole process is based on the core logic of "spatiotemporal synchronization-coordinate alignment-dynamic marking". First, the personnel data is preprocessed. Unique identifiers (e.g., infrared sensor ID + detection timestamp), real-time location information (pixel coordinates from binocular cameras, area coordinates from infrared array sensors), and motion status (stationary / moving) are extracted from the data. A coordinate transformation algorithm converts the location coordinates from different sensors to the Cartesian coordinate system of the library's spatial resource model. During the transformation, quadratic interpolation is used to correct coordinate deviations, ensuring that the deviation between the transformed coordinates and the model coordinates is ≤3cm. Next, a timestamp is attached to each personnel data entry, with millisecond precision to avoid mapping misalignment due to data latency. Then, the processed personnel data is associated with the spatial resource model according to "identifier-coordinate-timestamp-motion status," generating visual markers at the corresponding coordinate positions in the model (e.g., red dots for stationary personnel, blue arrows for moving personnel, with the arrow direction representing the direction of movement). The duration of each person's stay is also marked (accumulated from the timestamp of their entry into the coordinate area). If multiple people are present in the same area, the model automatically adjusts the marker density (personnel density ≥ 0.5). The person / ㎡ marker is displayed in a semi-transparent overlay to avoid visual obstruction; the final result is spatial personnel perception data that includes real-time personnel location, movement status, dwell time and area personnel density statistics.

[0086] Step S24: Perform local spatial and short-term temporal occupancy distribution tensor analysis based on spatial personnel perception data to generate spatiotemporal personnel tensor data;

[0087] In this embodiment of the invention, when performing local spatial and short-term temporal occupancy distribution tensor analysis based on spatial personnel perception data, a multi-dimensional tensor fusion technique is employed to construct a three-dimensional tensor of "space-time-feature" and perform in-depth analysis. First, local spatial units are divided. Based on the functional areas in the library's spatial resource model, each area is subdivided into 1m×1m local spatial grids, with each grid serving as the basic unit of the tensor's spatial dimension. Then, the short-term temporal range is determined, selecting spatial personnel perception data from the past hour as the analysis sample. Temporal units are divided into time slices of 5 minutes each, forming the temporal dimension of the tensor. Next, personnel features corresponding to each "local spatial grid-time slice" are extracted, including the number of personnel within the grid, average dwell time, and the proportion of personnel in motion (percentage of stationary personnel / percentage of moving personnel), constituting the feature dimension of the tensor. A multi-scale convolution algorithm is used to extract local spatial features. Two types of convolution kernels, 3×3 and 5×5, are set to capture local crowd aggregation features within a range of 1-3m and 3-5m, respectively. During the convolution process, zero-padding technology is used to keep the feature map size consistent with the original grid size. A time series sliding window algorithm is used to process short-term temporal features. The window step size is set to 1 time slice (5 minutes). The rate of change of personnel features within each window is calculated to reflect the dynamic changes of personnel in the temporal dimension. Local spatial features, short-term temporal features and personnel features are fused according to dimensions to construct a personnel perception tensor with the dimension of "number of local spatial grids × number of temporal slices × number of features". The number of local spatial grids is determined according to the actual area of ​​the library (e.g., a 1000㎡ library is divided into 1000 1m×1m grids), the number of temporal slices is 12 (1 hour ÷ 5 minutes), and the number of features is 3 (number of people, average stay time, and percentage of movement state). Finally, the tensor decomposition algorithm is used to reduce the dimensionality of the constructed tensor. The number of iterations is set to 50 and the convergence threshold is 1e-5. The core occupancy distribution features in the tensor are extracted to generate spatiotemporal personnel tensor data containing local spatial personnel gathering intensity and short-term temporal personnel change trends.

[0088] Step S25: Perform temporal intrinsic correlation analysis of regional personnel behavior based on spatiotemporal personnel tensor data to generate temporal intrinsic correlation data of regional personnel behavior, and perform dynamic perception analysis of regional personnel behavior through the temporal intrinsic correlation data of regional personnel behavior to generate dynamic perception data of regional personnel behavior.

[0089] In this embodiment of the invention, when performing temporal intrinsic correlation analysis of regional personnel behavior based on spatiotemporal personnel tensor data, the Pearson correlation coefficient algorithm is used to calculate the correlation between personnel behavior indicators and the time dimension. The core analysis indicators include the number of people in the region, average stay duration, and personnel flow frequency. First, the number of people, average stay duration, and flow frequency data for each region every 5 minutes (consistent with the time series slice) are extracted from the spatiotemporal personnel tensor data to form three time series. A time series curve is constructed with the time slice as the horizontal axis and the value of each indicator as the vertical axis. The Pearson correlation coefficient algorithm is used to calculate the correlation coefficient between each curve and the time axis. The correlation coefficient ranges from [-1, 1]. When the absolute value of the coefficient is ≥ 0.7, it is considered a strong correlation; when the absolute value is ≤ 0.3 and < 0.7, it is considered a moderate correlation; and when the absolute value is < 0.3, it is considered a weak correlation. The correlation coefficient clarifies the pattern of personnel behavior changes over time (e.g., the number of people from 9:00 to 11:00 AM shows a strong positive correlation with time, indicating that the number of people continues to increase during this period), generating temporal intrinsic correlation data of regional personnel behavior. Subsequently, dynamic perception analysis of regional personnel behavior was conducted using this correlation data. A behavior pattern recognition algorithm was employed, and behavior pattern determination rules were set based on the correlation data: when there was a strong positive correlation between the number of people, a moderate correlation between the frequency of movement and the duration of stay, it was determined to be a "personnel gathering pattern"; when there was a weak correlation between the number of people, a strong positive correlation between the frequency of movement and the duration of stay, it was determined to be a "personnel movement pattern"; and when there was a moderate correlation between the number of people, a weak correlation between the frequency of movement and the duration of stay, it was determined to be a "personnel stay pattern". The algorithm set a pattern matching threshold of 85%, meaning that when the data for a certain period matched the determination rules by ≥85%, it was identified as the corresponding behavior pattern. Simultaneously, the duration and occurrence time of each pattern were statistically analyzed to form a dynamic profile of regional personnel behavior. Finally, dynamic perception data of regional personnel behavior was generated, including the behavior pattern type, pattern duration, pattern occurrence time, and characteristics of personnel associated with the pattern (such as the average personnel density in the gathering pattern).

[0090] Step S26: Analyze the spatial pedestrian flow distribution characteristics based on the dynamic perception data of regional personnel behavior to generate spatial pedestrian flow distribution characteristic data.

[0091] In this embodiment of the invention, when analyzing the spatial flow distribution characteristics based on the dynamic perception data of regional personnel behavior, a multi-source feature fusion algorithm is used to extract core features from three dimensions: "flow trend - dwell characteristics - spatial association". First, the spatial personnel flow migration characteristics are analyzed. Based on the "personnel flow pattern" data in the dynamic perception data of regional personnel behavior, a Markov chain model is used to calculate the migration probability between regions: the number of people who migrated from region A to region B in the past hour is counted, and divided by the total number of people leaving region A to obtain the migration probability from A to B; all region combinations are traversed to construct a migration probability matrix between regions; based on this matrix, a vector field generation algorithm is used to generate spatial personnel flow vector field data, with regions as nodes, migration probability as vector length, and migration direction as vector direction. The resolution of the vector field is consistent with the local spatial grid of the library spatial resource model (1m×1m), and the vector display threshold is set to a migration probability ≥5% (vectors below this threshold are not displayed). Secondly, the characteristics of people's stay in the region were analyzed. Based on the "people's stay pattern" data, the average stay time, standard deviation of stay time, and proportion of people with long stays (the proportion of people with a stay time of ≥60 minutes) of people in each region were statistically analyzed to form regional people's stay characteristic data. The average stay time was calculated using a weighted average method (with the number of people as the weight), and the standard deviation was calculated using the sample standard deviation formula to ensure that the data can reflect the differences in stay preferences in different regions. Finally, a weighted feature fusion algorithm was used to fuse the spatial pedestrian flow vector field data and the regional personnel dwelling characteristic data. According to the library's space scheduling needs, weights were assigned to the two types of data (flow trend weight 0.6, dwelling characteristic weight 0.4). Through weighted summation, the multi-dimensional features were integrated into a unified spatial pedestrian flow distribution characteristic index. The index covers the pedestrian gathering level of each area (divided into high / medium / low levels based on migration probability and personnel density), the dominant pedestrian flow trend (such as "migration from the entrance area to the reading area"), and the dwelling hotspot areas (the top 5 areas with the longest average dwell time). At the same time, the feature data was standardized and all indicators were mapped to the [0,1] interval for easy use in subsequent steps. Finally, spatial pedestrian flow distribution characteristic data that can comprehensively reflect the flow pattern, dwelling preferences, and spatial gathering status of pedestrians in the library was generated.

[0092] Furthermore, step S24 includes the following steps:

[0093] Multi-scale convolutional local feature analysis is performed on spatial personnel perception data to generate local spatial personnel perception feature data.

[0094] Perform short-term temporal feature analysis on space personnel perception data to generate short-term temporal feature data of personnel perception.

[0095] Based on local spatial personnel perception feature data and short-term temporal feature data of personnel perception, a local spatial and short-term temporal occupancy distribution personnel perception tensor analysis is performed to generate spatiotemporal personnel tensor data.

[0096] In this embodiment of the invention, when performing multi-scale convolutional local feature analysis on spatial personnel perception data, a three-dimensional convolutional neural network architecture is adopted. The local grid (1m × 1m × floor height) of the library spatial resource model is used as the basic analysis unit, and two fixed-scale convolutional kernels are set to achieve hierarchical feature extraction. First, the spatial personnel perception data is divided into local grids, with each grid corresponding to a set of personnel data (number of personnel, location coordinates, and movement status). Before data input, normalization processing is performed, mapping the number of personnel to the [0,20] interval (corresponding to the maximum number of people a single grid can hold), and the location coordinates are converted to relative coordinates within the grid ([0,1] interval). Two convolutional layers are used: the first layer uses a 3×3×1 kernel (spatial horizontal × spatial vertical × floor dimension) with a stride of 1 and "same" padding to capture local clustering features within the grid and one adjacent grid (e.g., small groups of people clustering together); the second layer uses a 5×5×1 kernel with a stride of 1 and "same" padding to capture medium-range population distribution features across three adjacent grids (e.g., overall population density in the reading area). A ReLU activation function is applied after each convolution to suppress invalid feature output. A batch normalization layer is also used to control the mean of the feature data to 0±0.1 and the variance to 1±0.05, preventing gradient vanishing. During feature extraction, for high-density areas with a population density ≥10 people / grid, an additional convolution operation is performed to enhance feature details; for sparse areas with a population density ≤2 people / grid, the convolution operation is simplified to reduce computation. The final output is local spatial personnel perception feature data, with each grid corresponding to a 256-dimensional feature vector (including sub-features such as clustering intensity, personnel distribution uniformity, and concentration of movement direction). The feature extraction accuracy is verified by comparing the actual personnel distribution with the feature prediction results. For short-term temporal feature analysis of the spatial personnel perception data, a combination of sliding window temporal statistics and difference analysis is used. The short-term temporal analysis period is determined to be 1 hour, with temporal windows divided into 5-minute units, generating a total of 12 continuous temporal windows. First, the core temporal indicators within each temporal window are extracted from the spatial personnel perception data: the average number of personnel in each local grid within the window, the maximum / minimum number of personnel, the average duration of personnel stay, and the number of times personnel enter / leave. A sliding window weighted statistical approach is used for each indicator, assigning higher weights to data closer to the current time within the window (weight coefficients increase linearly, with a weight of 0.5 for the first moment and 1.0 for the last moment), ensuring that recent data has a more significant impact on the features. Subsequently, time-series difference analysis was performed to calculate the change in each indicator between two adjacent time-series windows (e.g., the average number of people in the later window minus the average number of people in the previous window) and the rate of change (change / indicator value in the previous window), thereby quantifying the short-term dynamic trend of personnel distribution (e.g., the number of people is increasing, decreasing, or stable).Simultaneously, a time-series anomaly detection mechanism is set up. When the absolute value of the change rate of a certain indicator exceeds 30%, secondary verification is triggered (comparing with the original data collected by IoT devices) to eliminate misjudgments caused by data anomalies. The final generated short-term time-series characteristic data of personnel perception is based on local grids, with each grid corresponding to a combination of "indicator statistical value + change rate" data for 12 time-series windows. When performing tensor analysis based on the local spatial personnel perception characteristic data and the short-term time-series characteristic data of personnel perception, a high-order tensor construction and decomposition algorithm is used to clarify the tensor dimension as "local spatial dimension × short-term time-series dimension × feature dimension". First, the parameters for each dimension are determined: the local spatial dimension corresponds to the total number of local grids divided in the library (e.g., 1000 1m×1m grids, dimension size 1000); the short-term time series dimension corresponds to 12 5-minute time series windows (dimensional size 12); the feature dimension integrates the core sub-features of the two types of features, including the clustering intensity and distribution uniformity of the local space and the average number of people and the rate of change of the short-term time series, totaling 4 sub-features (dimensional size 4), forming a 1000×12×4 third-order tensor structure. During the tensor construction process, precise association between the two types of features is achieved through index matching: using "local grid number + time series window number" as a dual index, the sub-features corresponding to the grids and time series windows in the local spatial features, and the sub-features with the same index in the short-term time series features, are filled into different positions under the "feature dimension" of the tensor, ensuring that each tensor element (i,j,k) uniquely corresponds to the data of "the i-th grid, the j-th time series window, and the k-th sub-feature". After tensor construction, the Tucker decomposition algorithm is used for optimization. The decomposition rank is set to (50, 6, 2), the number of iterations is 50, and the convergence threshold is 1e-5. Redundant features (such as highly correlated sub-features like "clustering intensity" and "average number of people") are removed through decomposition, while retaining core correlation information. After decomposition, the tensor elements are normalized to map the values ​​to the [0, 1] interval. The generated spatiotemporal personnel tensor data supports querying by any dimension slice (such as querying the clustering intensity feature of all grids in the 3rd time window).

[0097] Furthermore, step S26 includes the following steps:

[0098] The spatial flow and migration probability analysis is performed on the dynamic perception data of regional personnel behavior to generate spatial personnel flow and migration probability data. Based on the spatial personnel flow and migration probability data, spatial flow vector field analysis is performed to generate spatial flow vector field data.

[0099] Based on the dynamic perception data of regional personnel behavior, regional personnel stay characteristics are analyzed to generate regional personnel stay characteristic data.

[0100] Based on spatial pedestrian flow vector field data and regional personnel dwell characteristic data, spatial pedestrian flow distribution characteristic analysis is performed to generate spatial pedestrian flow distribution characteristic data.

[0101] In this embodiment of the invention, when performing spatial personnel migration probability analysis on the dynamic perception data of regional personnel behavior, an inter-regional migration counting and probability calculation model is adopted. Pre-defined functional areas of the library (such as reading area A, reading area B, discussion area C, and rest area D) are used as migration analysis units, and the statistical period is set to 15 minutes. First, the "number of people moving in" and "total number of people in the area" for each area within each statistical period are extracted from the dynamic perception data of regional personnel behavior: the number of people moving in is determined by tracking unique personnel identifiers (such as the thermal imaging feature code detected by an infrared sensor), representing the number of people entering the target area from other areas within the statistical period; the total number of people in the area is the actual number of people in the target area at the end of the period. The migration probability calculation uses the formula "number of people moving in to the target area ÷ total number of people moving out of the source area," where the total number of people moving out of the source area is the sum of the number of people moving to all other areas within that area within the period. Each area combination (such as A to B, B to A) corresponds to a unique migration probability value. The probability calculation result is retained to 3 decimal places, and the sum of the probabilities of all outgoing directions from the same source area is forcibly normalized to 1. Extremely low probability values ​​(≤0.02) are marked as "accidental migration" and recorded separately to avoid interfering with the overall migration pattern analysis. This results in spatial population flow migration probability data containing migration probabilities from all regions. When performing spatial population flow vector field analysis based on this data, a vector field generation algorithm is used. Functional areas are used as vector starting points, migration direction is the vector direction, and migration probability is the vector length (probability value × 10 is the visualization length coefficient). The library space is divided into 5m × 5m vector field grids. Vectors within each grid are integrated with migration data from adjacent areas using an interpolation algorithm to ensure a continuous and smooth vector field. A vector display threshold of 0.05 is set; vectors below this threshold are not displayed. Color gradients are used to distinguish vector intensity (blue to red corresponds to probabilities of 0.05 to 0.5). This ultimately generates spatial population flow vector field data that intuitively reflects the migration direction and intensity between regions. When analyzing the dwell time characteristics of people in a region based on the dynamic perception data of regional personnel behavior, a combination of time-series statistics and distribution fitting is used to determine the dwell time characteristic analysis dimensions, including four categories: average dwell time, standard deviation of dwell time, proportion of people with long dwell times, and dwell time distribution type. The analysis cycle is consistent with the update cycle of the dynamic perception data of regional personnel behavior (once every 5 minutes). First, the "entry timestamp" and "current timestamp" (for people who have not left) or "departure timestamp" (for people who have left) of each person in the dynamic perception data of regional personnel behavior are extracted. The dwell time of a single person is calculated: for people who have left, it is "departure timestamp - entry timestamp", and for people who have not left, it is "current timestamp - entry timestamp". The unit of dwell time is uniformly set to minutes, accurate to 1 minute.The average stay duration is calculated using a weighted average method, with the proportion of each person's stay duration within the analysis period as the weight (e.g., if a person stays for 3 minutes within a period, and the period duration is 5 minutes, the weight is 0.6), to avoid excessive influence of short-staying individuals on the average. The standard deviation of stay duration is calculated using the sample standard deviation formula, reflecting the dispersion of stay duration. A standard deviation ≤10 minutes is considered "stable stay," while >20 minutes is considered "large variation in stay." The proportion of long-staying individuals is defined as "the number of people with a stay duration ≥60 minutes ÷ the total number of people in the area," and a proportion ≥0.3 is marked as "areas dominated by long stays." The distribution type of stay duration is determined by fitting a gamma distribution (stay duration is mostly positively skewed), calculating the shape and scale parameters of the distribution, and verifying the goodness of fit using a chi-square test; R² ≥0.8 is considered a valid fit. Abnormal stay data encountered during the analysis (e.g., stay duration >240 minutes, exceeding library opening hours) is automatically verified by linking to access control system data (e.g., whether the person is a staff member), and is removed after confirmation of abnormal data, with a removal rate ≤3%. The final data generated includes regional population dwell time characteristics, including average dwell time (±2 minutes error), standard deviation of dwell time, proportion of long-staying individuals (retaining 2 decimal places), and distribution shape / scale parameters. When analyzing spatial population distribution characteristics based on spatial population vector field data and regional population dwell time characteristics data, a weighted feature fusion algorithm is used, with clearly defined fusion weights: spatial population vector field data accounts for 0.6 (reflecting flow trends), and regional population dwell time characteristics data accounts for 0.4 (reflecting dwell characteristics). The weight allocation is determined based on the library's spatial scheduling priority (flow efficiency takes precedence over dwell time analysis). The two types of data are standardized, mapping the migration probability (0 to 1) of the vector field data to the average dwell time (0 to 180 minutes) and the proportion of long-staying individuals (0 to 1) of the dwell time characteristics data to the [0,1] interval. The mapping formula is "(original value - minimum value) ÷ (maximum value - minimum value)", where the maximum average dwell time is set to 180 minutes (the longest recommended dwell time for a single visit in the library).A multi-dimensional index of pedestrian flow distribution characteristics is constructed: The first index is "spatial clustering level", which is calculated by weighting the mean migration probability (flow dimension) and the mean dwell time (dwelling dimension) of the vector field, with the formula "0.6 × mean migration probability + 0.4 × mean dwell time". Based on the results, it is divided into three levels: "low clustering (0 to 0.3), medium clustering (0.3 to 0.7), and high clustering (0.7 to 1)". The second index is "dominant pedestrian flow trend", which extracts the migration direction with the highest probability in the vector field (such as "from reading area A to discussion area C") and combines it with the information of long-stay dominant areas in the dwelling characteristics to determine the overall pedestrian flow trend (such as "migration to high-dwelling comfort area"). The third index is "urgency of resource demand". When the high-clustering area has a long-staying ratio ≥ 0.3, the urgency is set to "high", and resources need to be prioritized (such as increasing seats and increasing air conditioning power). Other cases are set to "medium" or "low". An anomaly detection mechanism is set up during the feature fusion process. When the difference between the two types of data after standardization is greater than 0.5 (e.g., the flow dimension shows low aggregation but the dwell dimension shows high aggregation), the original data is backtracked and recalculated to ensure the consistency of the fusion and generate spatial pedestrian flow distribution feature data.

[0102] Furthermore, step S3 includes the following steps:

[0103] Step S31: Analyze the environmental status of the impact of pedestrian flow on the environmental data using spatial pedestrian flow distribution characteristic data, generate pedestrian flow impact environmental status data, and analyze the spatial environmental response characteristics of pedestrian flow disturbance based on the pedestrian flow impact environmental status data, generate pedestrian flow disturbance spatial environmental response characteristic data.

[0104] In this embodiment of the invention, when analyzing environmental data using spatial pedestrian flow distribution characteristic data, a two-step analysis method of "multi-dimensional correlation - disturbance response quantification" is adopted. The core configuration parameters include a 10-minute analysis time window, a correlation threshold of 0.6, and a disturbance response sampling interval of 30 seconds. First, an environmental state analysis of the impact of pedestrian flow is conducted: two core indicators—real-time pedestrian density and pedestrian flow rate—are extracted from the spatial pedestrian flow distribution characteristic data for each area. Four parameters—temperature, humidity, TVOC concentration, and PM2.5 concentration—are extracted from the environmental data for the corresponding areas. The Pearson correlation analysis algorithm is used to calculate the correlation coefficient between each type of pedestrian flow indicator and each type of environmental parameter. The algorithm's sliding window step size is set to match the IoT device's data collection frequency (30 seconds / time) to ensure real-time capture of correlation changes. When the absolute value of the correlation coefficient is ≥0.6, it is determined to be a strong correlation, and this group of pedestrian flow-environment correlation relationships is recorded. Simultaneously, the average environmental parameters corresponding to different pedestrian flow density intervals are statistically analyzed to form a "pedestrian density-environmental parameter" correspondence table, generating data on the impact of pedestrian flow on the environmental state. Next, a spatial environmental response characteristic analysis of pedestrian disturbances was conducted: Periods with pedestrian density changes ≥20% were selected as "disturbance events," and the time of occurrence of the disturbance was recorded. Subsequently, the changes in environmental parameters after this period were tracked. Response sensitivity was calculated, and the time it took for environmental parameters to reach a new stable state (i.e., response lag time) was statistically analyzed. The sensitivity calculation error was controlled within ±0.05, and the lag time was recorded with second-level accuracy. The generated spatial environmental response characteristic data of pedestrian disturbances includes four core parameters: strongly correlated pedestrian-environmental indicator pairs in each region, mean environmental parameters for different density intervals, response sensitivity, and lag time.

[0105] Step S32: Based on the spatial functional resource operation data and spatial pedestrian flow distribution characteristic data, perform a two-way correlation characteristic analysis between pedestrian flow and functional resource operation to generate spatial pedestrian flow-functional resource operation related data;

[0106] In this embodiment of the invention, a cross-correlation analysis algorithm is used, and the configuration parameters include a time lag step of 1 minute (30 steps in total, covering 0-30 minutes), a correlation coefficient confidence level of 95%, and a parameter statistical interval of 5 minutes. The analysis process is divided into two dimensions: "Impact of People Flow on Resources" and "Impact of Resources on People Flow." In the "Impact of People Flow on Resources" dimension, the population density and dwell time of each area are extracted from spatial population distribution data. Similarly, the air conditioning power, number of lights on, and fresh air system volume of the corresponding areas are extracted from spatial functional resource operation data. The mean change rate of the three resource parameters under different population densities is calculated (e.g., the mean change rate of air conditioning power when population density increases from 0.2 people / m² to 0.5 people / m²). Simultaneously, cross-correlation analysis is used to determine the time lag between population flow changes and resource parameter changes. In the "Impact of Resources on People Flow" dimension, the dwell time and population attraction (the ratio of new people in the area to surrounding areas) of different resource operation parameter intervals are statistically analyzed. The increase in population attraction after resource parameter optimization is calculated to ensure that the data reflects the reverse effect of resource operation on population distribution. Throughout the analysis, outlier data (such as parameter mutations caused by resource failures) were removed using the 3σ criterion. The resulting spatial pedestrian flow-functional resource operation related data includes a two-way correlation coefficient matrix (rows: pedestrian flow indicators, columns: resource parameters), the mean and rate of change of resource parameters in different pedestrian flow intervals, and the influence coefficient of resource parameters on pedestrian attraction. The correlation coefficient matrix has a dimension of 4×3 (4 types of pedestrian flow indicators: density, dwell time, flow rate, and aggregation; 3 types of resource parameters: air conditioning power, number of lights, and fresh air volume).

[0107] Step S33: Based on the spatial functional resource operation data and the spatial environment response characteristic data of human flow disturbance, perform a two-way correlation characteristic analysis of the operation of the environment and functional resources to generate spatial environment-functional resource operation related data;

[0108] In this embodiment of the invention, a multiple linear regression algorithm combined with threshold-triggered analysis is employed. The configured parameters include 100 iterations of the regression model, a convergence threshold of 1e-5, a comfort threshold for environmental parameters, and a threshold determination interval of 5 minutes. The analysis is divided into two directions: "environment driving resources" and "resource regulating the environment." In the "environment driving resources" direction, temperature, humidity, and TVOC concentration from environmental data are used as independent variables, while air conditioning operating parameters and fresh air system airflow from spatial functional resource operation data are used as dependent variables. A multiple linear regression model is constructed, and the model parameters are optimized through 100 iterations to obtain the influence coefficients of each environmental parameter on resource parameters. Simultaneously, the frequency of automatic adjustment of resource parameters when environmental parameters exceed the comfort threshold is statistically analyzed. In the "resource regulating the environment" direction, the changes in environmental parameters after resource parameter adjustments are recorded, the regulation efficiency is calculated, and the regulation effects of different resource combinations on environmental parameters are analyzed, comparing the efficiency differences of single resource operation. During the analysis, the goodness of fit of the regression model was verified; a moving average filter (window size 5) was used on the regulation efficiency data to eliminate instantaneous fluctuations, and the generated spatial environment-functional resource operation related data included a multiple linear regression coefficient matrix, the frequency of resource adjustment triggered by environmental thresholds, resource regulation efficiency, and a comparison table of the regulation effects of different resource combinations.

[0109] Step S34: Based on the spatial human flow-functional resource operation related data and the spatial environment-functional resource operation related data, conduct spatial functional resource operation regulation and response analysis to generate spatial functional resource operation regulation and response data;

[0110] In this embodiment of the invention, when performing spatial functional resource operation control response analysis, a demand priority assessment algorithm combined with deviation quantification analysis is adopted. Configurable parameters include priority weights, control deviation thresholds, demand level classification, and assessment cycles. Data related to spatial pedestrian flow and functional resource operation are integrated with data related to spatial environment and functional resource operation to extract core correlation indicators: "Pedestrian density - air conditioning power" and "Pedestrian dwell time - number of lights turned on" correlation coefficients are obtained from the spatial pedestrian flow and functional resource operation data; "Temperature - air conditioning power" adjustment coefficients and "TVOC - fresh air volume" adjustment coefficients are obtained from the spatial environment and functional resource operation data. Next, the deviation between the current resource operation parameters and the ideal parameters is calculated, and then a comprehensive demand score is calculated based on the priority weights, while simultaneously determining the control direction and control magnitude. During the analysis, control items with a demand level of "high" are marked with an "urgent execution" tag to ensure priority response; for multi-resource collaborative control demands, the control ratio of each resource is clarified. The generated spatial functional resource operation control response data includes the resource control object, ideal parameters, current deviation, control direction, control magnitude, demand level, and execution priority for each region.

[0111] Step S35: Perform spatial response correlation characteristic analysis on the spatial environment response characteristic data of pedestrian flow disturbance and the spatial function resource operation and regulation response data to generate spatial multi-source response correlation characteristic data.

[0112] In this embodiment of the invention, when performing spatial response correlation feature analysis of pedestrian flow distribution characteristics, Principal Component Analysis (PCA) combined with a correlation matrix is ​​used. The PCA principal component retention rate is set at 90%, the correlation calculation uses cosine similarity, the matrix dimension equals the number of library areas (e.g., a 20×20 matrix for 20 areas), and the feature update cycle is 10 minutes. First, pedestrian flow disturbance spatial environment response feature data and spatial functional resource operation control response data are integrated to determine the analysis dimensions: pedestrian flow disturbance intensity (pedestrian density change amplitude), environmental response sensitivity, and response lag time are selected from the pedestrian flow disturbance spatial environment response feature data; control response level, control amplitude, and execution priority are selected from the spatial functional resource operation control response data, for a total of 6 core feature dimensions. Then, PCA is used to reduce the dimensionality of the 6-dimensional feature data. Two to three principal components are extracted with a 90% principal component retention rate to eliminate redundant information between features. Simultaneously, the contribution of each original feature to the principal components is calculated to ensure that the principal components reflect more than 90% of the original data information. Then, a correlation matrix is ​​constructed: using each area of ​​the library as the rows and columns of the matrix, the cosine similarity of "human flow disturbance - environmental response - regulatory response" between different areas is calculated (the similarity value ranges from [0,1], with the closer the value is to 1, the stronger the correlation). For example, if human flow disturbance in area A triggers an environmental response in area B, and the regulatory responses of the two are similar, then the correlation between A and B is ≥0.7. At the same time, the temporal correlation of human flow disturbance, environmental response, and regulatory response within the same area is analyzed. The temporal order of environmental response and regulatory response after human flow disturbance occurs is recorded to clarify the causal relationship among the three. The generated spatial multi-source response correlation feature data includes PCA principal component feature vectors, inter-regional correlation matrix, human flow-environment-regulation temporal correlation table, and the contribution of each feature to the principal component.

[0113] Furthermore, as an embodiment of the present invention, reference is made to... Figure 2 As shown, Figure 1 A detailed flowchart of step S4 is shown below. In this embodiment, step S4 includes the following steps:

[0114] Step S41: Perform temporal extension analysis on the spatial multi-source response correlation feature data to generate temporal extension data of spatial multi-source response correlation features;

[0115] In this embodiment of the invention, when performing temporal extension analysis on spatial multi-source response correlation feature data, a gated recurrent unit (GRU) algorithm combined with adjacent node influence correction is adopted. The time window length is set to 60 minutes, the number of GRU hidden layer nodes is 64, the number of iterations is 100, the learning rate is 0.01, and the adjacent influence weight coefficient is 0.3. First, the spatial multi-source response correlation feature data (including PCA principal components and regional correlation matrix) is preprocessed. After being sorted by timestamp, the data is divided into a training set (80% historical data) and a validation set (20% historical data). The data is standardized to the [-1,1] interval to improve the model convergence speed. The GRU algorithm captures the temporal dependencies of feature data within the past 60 minutes (such as the time lag pattern of environmental response in adjacent areas after a disturbance of pedestrian flow in a certain area) to preliminarily predict the changes in feature data in the next 30 minutes, generating preliminary temporal extension data. Subsequently, an adjacent node impact analysis was introduced: based on the regional correlation matrix, adjacent regions with a correlation degree ≥ 0.6 were screened, and the impact value of the current characteristic data of these regions on the target region extension data was calculated. This impact value was used to correct the deviation in the initial extension data (such as correcting the environmental response prediction value of the target region when the flow of people in adjacent regions increases suddenly), and spatial multi-source response correlation characteristic time series extension data containing the principal component characteristics and correlation degree change trends of each region in the next 30 minutes was generated.

[0116] Step S42: Perform spatial multi-source response trend feature analysis based on the temporal extension data of spatial multi-source response correlation features to generate spatial multi-source response trend feature data;

[0117] In this embodiment of the invention, when performing spatial multi-source response trend feature analysis based on time-series extended data, the STL time series decomposition algorithm combined with trend slope calculation is used. The spatial multi-source response correlation feature time-series extended data (including regional features for the next 30 minutes) is split by region, generating an independent time series for each region (e.g., the "human flow disturbance intensity-time" series and the "environmental response sensitivity-time" series for region A). The STL algorithm decomposes each time series into a trend term, a seasonal term, and a residual term, removing the residual term (random fluctuations) and the seasonal term (periodic repetitive changes), retaining the trend term as the core analysis object. Subsequently, the linear regression slope of the trend term is calculated: with time as the x-axis (unit: minutes) and the trend term value as the y-axis, a straight line is fitted using the least squares method to obtain the slope value (a positive slope indicates an increase in the feature value, and a negative slope indicates a decrease). Trend types are determined based on slope thresholds: a slope ≥ 0.1 indicates a "significant upward trend" (e.g., a continuous increase in air conditioning demand in a region), a slope ≤ -0.1 indicates a "significant downward trend" (e.g., a gradual decrease in pedestrian traffic disturbance in a region), and an absolute slope < 0.1 indicates a "stable trend." Simultaneously, the duration and peak of the trend are statistically analyzed to generate spatial multi-source response trend characteristic data that includes the characteristic trend type, trend slope, duration, and peak time for each region.

[0118] Step S43: Analyze the operational constraints of spatial functions and resources based on spatial resource data and operational data of spatial functions and resources, and generate operational constraint data of spatial functions and resources.

[0119] In this embodiment of the invention, a constraint inversion algorithm combined with multi-dimensional threshold definition is employed. Configuration parameters include constraint weight allocation (resource capacity constraint 0.4, equipment threshold constraint 0.3, management rule constraint 0.3), allowable threshold deviation range ±5%, and constraint verification interval of 15 minutes. First, basic constraint parameters are extracted from spatial resource data: resource capacity constraints and functional type constraints; equipment threshold constraints are extracted from spatial functional resource operation data. The constraint inversion algorithm transforms these basic parameters into quantifiable constraints and calculates the deviation between the current operating state and the constraints. For conflicting constraints (e.g., the number of people in a region is not exceeded but the air conditioning power reaches the threshold), higher-weight constraints are prioritized (prioritizing reducing air conditioning power to meet the equipment threshold). Simultaneously, a constraint characteristic library is established, recording the trigger frequency of each constraint condition (e.g., the average daily number of times air conditioning power exceeds the threshold), conflict solutions, and generating spatial functional resource operation constraint characteristic data containing the capacity upper limit of each region, equipment operation threshold, management rule restrictions, and constraint conflict priority.

[0120] Step S44: Establish a demand forecasting mapping relationship for spatial resource regulation by using spatial multi-source response trend characteristic data and spatial functional resource operation constraint characteristic data, so as to obtain a spatial resource regulation demand forecasting model;

[0121] In this embodiment of the invention, when establishing a spatial resource regulation demand prediction model, a multiple linear regression combined with the gradient boosting tree (GBRT) algorithm is used. The GBRT decision tree count is set to 100, the maximum tree depth to 5, the learning rate to 0.05, and the regression model goodness-of-fit target value to be ≥0.85. First, the model input and output are determined: the input consists of spatial multi-source response trend characteristic data (trend type, slope, peak time) and spatial functional resource operation constraint characteristic data (capacity upper limit, equipment threshold), totaling 8 feature dimensions (such as "trend slope", "capacity deviation", and "equipment threshold margin"); the output consists of spatial resource regulation demand parameters (such as air conditioning power adjustment, number of lights turned on, and fresh air volume adjustment value). The input data from the past three months and the corresponding actual regulation demand data are combined to form a training set. First, a basic prediction model is constructed using multiple linear regression to initially fit the linear relationship between the input and output. Then, the GBRT algorithm is used to optimize the nonlinear part of the model (such as the nonlinear growth law of regulation demand when the flow of people suddenly increases). The prediction bias is corrected iteratively through 100 decision trees, with each tree learning the residual of the previous tree and adjusting its weights. During model training, 5-fold cross-validation is used to avoid overfitting. The resulting spatial resource regulation demand prediction model can predict the input trend characteristics and constraints.

[0122] Step S45: Design multi-objective optimization indicators for library space based on the preset library space resource optimization demand data and the spatial resource regulation demand prediction model, and use the multi-objective optimization indicators for library space to perform multi-objective optimization allocation of model parameters for the spatial resource regulation demand prediction model to obtain the optimized spatial resource regulation demand prediction model.

[0123] In this embodiment of the invention, when designing multi-objective optimization indices and optimizing the prediction model, a non-dominated sorting genetic algorithm (NSGA-II) is used. The NSGA-II configuration parameters include a population size of 100, 50 iterations, a crossover probability of 0.8, and a mutation probability of 0.1, with preset optimization target weights (resource utilization 0.4, energy consumption 0.3, and user comfort 0.3). Multi-objective optimization indices are determined based on library operation needs: resource utilization, energy consumption, and user comfort. The parameters of the spatial resource regulation demand prediction model (such as GRU weights and GBRT tree depth) are used as optimization variables, with "maximizing utilization, minimizing energy consumption, and maximizing comfort" as the objective function to construct a multi-objective optimization problem. The NSGA-II algorithm generates 100 sets of model parameter combinations through population initialization and performs selection, crossover, and mutation operations through 50 iterations: the selection operation retains the parameter combination with the better objective function value, the crossover operation integrates the advantageous features of the two sets of parameters, and the mutation operation randomly adjusts the parameters to avoid local optima. After the iteration is completed, parameter combinations that meet the weight requirements are selected from the non-dominated solution set to update the parameters of the spatial resource regulation demand prediction model, thus obtaining the optimized model.

[0124] Step S46: Utilize the optimized spatial resource regulation demand prediction model to analyze the library spatial resource dynamic scheduling and control parameters based on the spatial function resource operation data, and generate library spatial resource dynamic scheduling and control data;

[0125] In this embodiment of the invention, when analyzing dynamic scheduling control parameters using an optimization model, a parameter matching algorithm combined with constraint verification is employed. Real-time spatial functional resource operation data (current air conditioning power, lighting status, and fresh air volume) are input into the optimized spatial resource regulation demand prediction model, and the model outputs "ideal regulation demand parameters" for each area. The parameter matching algorithm compares the ideal parameters with the current operating parameters, calculates the adjustment difference, and verifies the rationality of the adjustment difference based on the spatial functional resource operation constraint characteristics data. If the adjustment difference violates the constraints, it is corrected according to the constraint priority (e.g., reducing to 1500W while compensating for a 20m³ / h increase in fresh air volume to maintain comfort). After successful verification, the adjustment difference is converted into specific control parameters, and the parameter execution priority is marked. Dynamic scheduling control data for library spatial resources, including equipment control parameters, execution priorities, and adjustment time windows (within the next 30 minutes) for each area, is generated.

[0126] Step S47: Execute intelligent spatial resource dynamic scheduling and control operations based on library spatial resource dynamic scheduling and control data.

[0127] In this embodiment of the invention, when performing intelligent scheduling operations based on scheduling control data, an Internet of Things (IoT) control bus (Modbus protocol) combined with closed-loop feedback monitoring is used. The dynamic scheduling control data of library space resources is broken down by equipment type and transmitted to the corresponding device controllers via the IoT control bus: The air conditioning controller receives a power adjustment command and gradually adjusts the compressor speed through the frequency converter module; the lighting controller receives an on / off command and switches lighting circuits sequentially according to execution priority; the fresh air controller receives an airflow adjustment command and adjusts the fan speed to match demand. Simultaneously, a feedback monitoring mechanism is deployed: the IoT monitoring device collects the actual operating parameters of the equipment every 5 minutes, compares them with the target parameters in the scheduling control data, calculates the execution deviation, and if the deviation is >5%, the control bus automatically sends a correction command; if the deviation is ≤5%, the execution is considered successful. Furthermore, the scheduling results are synchronized to the library display system, marking the current resource status of each area and guiding personnel to select appropriate areas.

[0128] Furthermore, step S41 includes the following steps:

[0129] Based on the preset gated recursive unit algorithm, a preliminary temporal extension analysis of spatial multi-source response correlation features is performed on the spatial multi-source response correlation feature data to generate preliminary temporal extension data of spatial multi-source response correlation features.

[0130] Based on the spatial multi-source response correlation feature data, an analysis of the impact of adjacent nodes on spatial multi-source response is performed to generate spatial multi-source response adjacent node impact data;

[0131] By correcting the temporal extension deviation of the influence of spirit nodes on the preliminary temporal extension data of spatial multi-source response correlation features using the influence data of adjacent nodes of spatial multi-source response, the temporal extension data of spatial multi-source response correlation features is generated.

[0132] In this embodiment of the invention, when performing preliminary temporal extension analysis based on a preset gated recursive unit (GRU) algorithm, the configured algorithm parameters include a time window length of 60 minutes, 64 hidden layer nodes, 100 training iterations, a learning rate of 0.01, and a data standardization interval of [-1, 1]. The spatial multi-source response correlation feature data (including PCA principal component features and regional correlation matrices) are preprocessed: sorted by timestamp order, and Min-Max standardization is used to uniformly map feature data of different dimensions to the [-1, 1] interval, eliminating the influence of dimensional differences on model training. The GRU algorithm captures the temporal dependencies of feature data through an input gate, a forget gate, and an output gate. The input gate controls the inclusion of new feature information, the forget gate filters historical feature information to be retained, and the output gate integrates effective information to generate temporal prediction results. During training, using 60 minutes of historical feature data as input, the system predicts feature changes over the next 30 minutes. After each iteration, the mean squared error (MSE) between the predicted and actual values ​​is calculated. Training stops when the MSE is less than 1e-4 for five consecutive iterations or reaches 100 iterations, generating preliminary temporal extension data of spatial multi-source response correlation features containing future principal component features of each region and trends in regional correlation. When performing adjacency node impact analysis based on the spatial multi-source response correlation feature data, a correlation threshold screening combined with influence coefficient calculation is used. Configuration parameters include the adjacency correlation threshold, influence coefficient calculation weights, and the influence range defined as physical distance. A regional correlation matrix is ​​extracted from the spatial multi-source response correlation feature data, and region pairs with correlation greater than the adjacency correlation threshold are selected (determined as adjacency nodes). Next, the influence coefficient of the adjacency nodes is calculated, with greater weight for closer proximity. Feature similarity weights are calculated using the cosine similarity of the principal component features of two regions, with greater weight for more similar features. Simultaneously, the characteristic transmission patterns of adjacent nodes are statistically analyzed. For example, after a 20% increase in pedestrian disturbance intensity in region A, the environmental response sensitivity in region B increases by 15% after 10 minutes. The transmission delay and impact magnitude are recorded. The generated spatial multi-source response adjacent node impact data includes a list of adjacent nodes for each region, corresponding impact coefficients, characteristic transmission delays, and impact magnitudes. When correcting deviations in the preliminary extended data using the spatial multi-source response adjacent node impact data, an impact value superposition algorithm combined with error feedback adjustment is used. Configurable parameters include 3 correction iterations, a correction error threshold of 5%, and the impact value superposition weight being consistent with the adjacent impact coefficient. Each region in the preliminary spatial multi-source response associated feature time-series extended data is traversed, and the adjacent node impact data for that region is retrieved: the impact value of the current characteristic data of each adjacent node on the target region is calculated.The influence values ​​of all adjacent nodes are superimposed onto the preliminary extended data of the target area to obtain the first correction value. The error between the first correction value and the historical data from the same period is compared. If the error is >5%, the influence coefficient weight is adjusted (e.g., the spatial distance weight is increased to 0.6) for a second correction, until the error is ≤5% or three iterations are completed, generating temporal extended data of spatial multi-source response correlation characteristics. When analyzing the regulatory characteristics of spatial resource data based on spatial functional resource operation data, a multi-dimensional feature extraction algorithm combined with sensitivity calculation is used. Configurable parameters include the analysis time window, sensitivity calculation error, and characteristic data update frequency. Core operating parameters are extracted from the spatial functional resource operation data: cooling / heating power and wind speed of air conditioning, number of lights turned on and brightness level, and air volume of the fresh air system. Basic attributes of the corresponding area are extracted from the spatial resource data: area area, functional type (reading area / seminar room), and design capacity. The correlation characteristics of the two types of data are analyzed by feature extraction algorithms. The lag time analysis records the time difference between changes in the state of spatial resources (such as a sudden increase in the number of people) and the start of adjustment of functional resources. For example, the air conditioner starts to improve efficiency 3 minutes after the change in the number of people, and the lag time is 3 minutes. The saturation threshold analysis determines the critical point at which the operating parameters no longer change significantly after the state of spatial resources reaches a certain value. The generated functional operation spatial resource regulation characteristic data includes the regulation sensitivity, lag time, saturation threshold of each region, and the characteristic differences of different functional types of regions.

[0133] Furthermore, step S43 includes the following steps:

[0134] Based on the spatial function resource operation data, the spatial resource regulation characteristics of the spatial resource operation are analyzed to generate spatial resource regulation characteristic data of the functional operation.

[0135] Based on the functional operation space resource regulation characteristic data, analyze the functional operation space resource regulation mode and generate functional operation space resource regulation mode data;

[0136] Based on the spatial function resource operation data and the spatial resource regulation mode data of the function operation, multi-dimensional constraint inversion processing of spatial function resource operation is performed to generate multi-dimensional constraint inversion data of spatial function resource operation.

[0137] The spatial function resource operation constraint characteristics are analyzed by performing multi-dimensional constraint inversion data on spatial function resource operation, and spatial function resource operation constraint characteristic data is generated.

[0138] In this embodiment of the invention, when analyzing control patterns based on functional operating space resource regulation characteristic data, a K-means clustering algorithm combined with pattern feature quantification is used. The number of clusters is set to 3, the number of clustering iterations to 50, the clustering error threshold to 1e-3, and the pattern recognition accuracy to ≥90%. Clustering feature dimensions are determined as follows: regulation trigger source (human flow driven / environment driven), regulation response speed, and regulation intensity change. Sensitivity, lag time, and saturation threshold from the functional operating space resource regulation characteristic data are converted into clustering feature vectors and input into the K-means algorithm: 3 cluster centers are initialized; the Euclidean distance between each sample and the cluster center is calculated; samples are assigned to the nearest cluster; the cluster center is updated to the mean of the samples in that class; this iteration is repeated until the change in cluster centers is <1e-3 or 50 iterations are completed. Finally, the clusters form the control patterns. Feature quantification is performed on each pattern to generate functional operating space resource regulation pattern data. When performing multi-dimensional constraint inversion processing based on spatial functional resource operation data and control mode data, a constraint inversion algorithm combined with threshold mapping is used, setting constraint dimension weights, 30 inversion iterations, and a constraint satisfaction rate of ≥95%. Physical thresholds for equipment are extracted from the spatial functional resource operation data, and mode constraint requirements are extracted from the functional operation spatial resource control mode data. The constraint inversion algorithm then transforms these requirements into quantifiable constraints. During the inversion process, if constraint conflicts exist (e.g., increasing air conditioning power to meet the response speed of the pedestrian-driven mode, but exceeding the equipment threshold), the higher-weighted constraint is prioritized (prioritizing power reduction to meet equipment safety), and other constraint parameters are adjusted (e.g., extending the response time to 2.5 minutes). Through 30 iterations of optimization, the generated multi-dimensional constraint inversion data for spatial functional resource operation includes equipment threshold constraints, energy consumption limits, and environmental comfort ranges under each control mode. When performing constraint characteristic analysis on multi-dimensional constraint inversion data of spatial functional resource operation, a constraint quantification algorithm combined with trigger condition definition is used to decompose the constraint conditions in the multi-dimensional constraint inversion data into monitorable characteristic indicators: constraint trigger condition analysis determines that "an early warning is triggered when the current state reaches 95% of the constraint value, and a mandatory constraint is triggered when it reaches 100%"; constraint rigidity level classification divides constraints into strong constraints, medium constraints, and weak constraints according to weight; constraint correlation characteristic analysis records the linkage relationship between different constraints. These characteristics are transformed into parameterized data through constraint quantification algorithms, and the trigger frequency of each constraint (such as the number of monthly equipment threshold early warnings) and conflict resolution success rate are statistically analyzed. The generated spatial functional resource operation constraint characteristic data includes constraint trigger early warning values, mandatory thresholds, rigidity levels, and linkage adjustment rules for each region.

[0139] Therefore, the embodiments should be considered as exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of the equivalents of the application are intended to be included within the invention.

[0140] The above description is merely a specific embodiment of the present invention, enabling those skilled in the art to understand or implement the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the present invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features of the invention herein.

Claims

1. A method for dynamic scheduling and control of library space resources based on the Internet of Things, characterized in that, Includes the following steps: Step S1: Perform global monitoring and classification processing of the library through the library's IoT monitoring equipment to generate global monitoring classification data of the library, wherein the global monitoring classification data of the library includes spatial resource data, personnel data, environmental data and spatial functional resource operation data of the library; Step S2: Analyze the spatial pedestrian flow distribution characteristics using spatial resource data and personnel data to generate spatial pedestrian flow distribution characteristic data; Step S3: Based on spatial pedestrian flow distribution characteristic data, perform spatial response correlation characteristic analysis on environmental data and spatial functional resource operation data to generate spatial multi-source response correlation characteristic data; Step S4 includes the following steps: Step S41: Perform temporal extension analysis on the spatial multi-source response correlation feature data to generate temporal extension data of spatial multi-source response correlation features; Step S42: Perform spatial multi-source response trend feature analysis based on the temporal extension data of spatial multi-source response correlation features to generate spatial multi-source response trend feature data; Step S43: Analyze the operational constraints of spatial functions and resources based on spatial resource data and operational data of spatial functions and resources, and generate operational constraint data of spatial functions and resources. Step S44: Establish a demand forecasting mapping relationship for spatial resource regulation by using spatial multi-source response trend characteristic data and spatial functional resource operation constraint characteristic data, so as to obtain a spatial resource regulation demand forecasting model; Step S45: Design multi-objective optimization indicators for library space based on the preset library space resource optimization demand data and the spatial resource regulation demand prediction model, and use the multi-objective optimization indicators for library space to perform multi-objective optimization allocation of model parameters for the spatial resource regulation demand prediction model to obtain the optimized spatial resource regulation demand prediction model. Step S46: Utilize the optimized spatial resource regulation demand prediction model to analyze the library spatial resource dynamic scheduling and control parameters based on the spatial function resource operation data, and generate library spatial resource dynamic scheduling and control data; Step S47: Execute intelligent spatial resource dynamic scheduling and control operations based on library spatial resource dynamic scheduling and control data.

2. The method for dynamic scheduling and control of library space resources based on the Internet of Things according to claim 1, characterized in that, The spatial functional resource operation data mentioned in step S1 can be used to perform scheduling and control operations on spatial resource data.

3. The method for dynamic scheduling and control of library space resources based on the Internet of Things according to claim 1, characterized in that, Step S1 includes the following steps: Step S11: Perform global monitoring and processing of the library through the library's IoT monitoring equipment to generate global monitoring data of the library; Step S12: Perform monitoring and sensing dimension analysis based on the IoT monitoring device to obtain monitoring and sensing dimension data, and perform sensing dimension correction processing on the library global monitoring data through the monitoring and sensing dimension data to generate corrected library global monitoring data. Step S13: Perform time-series and spatial synchronization mapping processing on the calibration library global monitoring data to generate synchronized library global monitoring data; Step S14: Perform monitoring data classification processing on the synchronized library global monitoring data to generate library global monitoring classification data.

4. The method for dynamic scheduling and control of library space resources based on the Internet of Things according to claim 1, characterized in that, Step S2 includes the following steps: Step S21: Perform spatial resource characteristic analysis and processing on the spatial resource data to generate spatial resource characteristic data; Step S22: Perform library spatial resource modeling based on spatial resource characteristic data to generate a library spatial resource model; Step S23: Map personnel data to the library space model to perform personnel perception space mapping processing, and generate spatial personnel perception data; Step S24: Perform local spatial and short-term temporal occupancy distribution tensor analysis based on spatial personnel perception data to generate spatiotemporal personnel tensor data; Step S25: Perform temporal intrinsic correlation analysis of regional personnel behavior based on spatiotemporal personnel tensor data to generate temporal intrinsic correlation data of regional personnel behavior, and perform dynamic perception analysis of regional personnel behavior through the temporal intrinsic correlation data of regional personnel behavior to generate dynamic perception data of regional personnel behavior. Step S26: Analyze the spatial pedestrian flow distribution characteristics based on the dynamic perception data of regional personnel behavior to generate spatial pedestrian flow distribution characteristic data.

5. The method for dynamic scheduling and control of library space resources based on the Internet of Things according to claim 4, characterized in that, Step S24 includes the following steps: Multi-scale convolutional local feature analysis is performed on spatial personnel perception data to generate local spatial personnel perception feature data. Perform short-term temporal feature analysis on space personnel perception data to generate short-term temporal feature data of personnel perception. Based on local spatial personnel perception feature data and short-term temporal feature data of personnel perception, a local spatial and short-term temporal occupancy distribution personnel perception tensor analysis is performed to generate spatiotemporal personnel tensor data.

6. The method for dynamic scheduling and control of library space resources based on the Internet of Things according to claim 4, characterized in that, Step S26 includes the following steps: The spatial flow and migration probability analysis is performed on the dynamic perception data of regional personnel behavior to generate spatial personnel flow and migration probability data. Based on the spatial personnel flow and migration probability data, spatial flow vector field analysis is performed to generate spatial flow vector field data. Based on the dynamic perception data of regional personnel behavior, regional personnel stay characteristics are analyzed to generate regional personnel stay characteristic data. Based on spatial pedestrian flow vector field data and regional personnel dwell characteristic data, spatial pedestrian flow distribution characteristic analysis is performed to generate spatial pedestrian flow distribution characteristic data.

7. The method for dynamic scheduling and control of library space resources based on the Internet of Things according to claim 1, characterized in that, Step S3 includes the following steps: Step S31: Analyze the environmental status of the impact of pedestrian flow on the environmental data using spatial pedestrian flow distribution characteristic data, generate pedestrian flow impact environmental status data, and analyze the spatial environmental response characteristics of pedestrian flow disturbance based on the pedestrian flow impact environmental status data, generate pedestrian flow disturbance spatial environmental response characteristic data. Step S32: Based on the spatial functional resource operation data and spatial pedestrian flow distribution characteristic data, perform a two-way correlation characteristic analysis between pedestrian flow and functional resource operation to generate spatial pedestrian flow-functional resource operation related data; Step S33: Based on the spatial functional resource operation data and the spatial environment response characteristic data of human flow disturbance, perform a two-way correlation characteristic analysis of the operation of the environment and functional resources to generate spatial environment-functional resource operation related data; Step S34: Based on the spatial human flow-functional resource operation related data and the spatial environment-functional resource operation related data, conduct spatial functional resource operation regulation and response analysis to generate spatial functional resource operation regulation and response data; Step S35: Perform spatial response correlation characteristic analysis on the spatial environment response characteristic data of pedestrian flow disturbance and the spatial function resource operation and regulation response data to generate spatial multi-source response correlation characteristic data.

8. The method for dynamic scheduling and control of library space resources based on the Internet of Things according to claim 1, characterized in that, Step S41 includes the following steps: Based on the preset gated recursive unit algorithm, a preliminary temporal extension analysis of spatial multi-source response correlation features is performed on the spatial multi-source response correlation feature data to generate preliminary temporal extension data of spatial multi-source response correlation features. Based on the spatial multi-source response correlation feature data, an analysis of the impact of adjacent nodes on spatial multi-source response is performed to generate spatial multi-source response adjacent node impact data; By correcting the temporal extension deviation of the influence of spirit nodes on the preliminary temporal extension data of spatial multi-source response correlation features using the influence data of adjacent nodes of spatial multi-source response, the temporal extension data of spatial multi-source response correlation features is generated.

9. The method for dynamic scheduling and control of library space resources based on the Internet of Things according to claim 1, characterized in that, Step S43 includes the following steps: Based on the spatial function resource operation data, the spatial resource regulation characteristics of the spatial resource operation are analyzed to generate spatial resource regulation characteristic data of the functional operation. Based on the functional operation space resource regulation characteristic data, analyze the functional operation space resource regulation mode and generate functional operation space resource regulation mode data; Based on the spatial function resource operation data and the spatial resource regulation mode data of the function operation, multi-dimensional constraint inversion processing of spatial function resource operation is performed to generate multi-dimensional constraint inversion data of spatial function resource operation. The spatial function resource operation constraint characteristics are analyzed by performing multi-dimensional constraint inversion data on spatial function resource operation, and spatial function resource operation constraint characteristic data is generated.

Images