A Knowledge Graph-Based Multi-Source Cross-Domain Behavioral Data Fusion and Analysis System and Method

Abstract

This invention discloses a multi-source cross-domain behavioral data fusion analysis system and method based on knowledge graphs, belonging to the field of data processing technology. Specifically, it includes: deploying gas sensors on public vehicles in the target area to periodically detect the concentration values ​​and location coordinates of pollutants along the driving path; constructing a spatiotemporal knowledge graph structure, including a static building layer, a dynamic path layer, and a fusion data layer; calculating the spatial relative relationship between public vehicles and the facades of buildings on both sides, and mapping the gas concentration values ​​to virtual monitoring points on the building facades using real-time wind direction and speed parameters to generate a building facade concentration distribution map; calculating the building gas infiltration flux based on window status to deduce the indoor pollutant concentration change process of the target building; and integrating the indoor pollutant concentration deduction results along a time axis to form a spatiotemporal evolution map of air quality in the target area. This invention achieves precise control and effective management of air quality in congested urban scenarios.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, specifically to a system and method for multi-source cross-domain behavioral data fusion and analysis based on knowledge graphs. Background Technology

[0002] Traditional air quality monitoring systems primarily rely on fixed site deployments, which reveal significant limitations when dealing with such complex scenarios. Fixed monitoring methods not only face high equipment installation and maintenance costs but also struggle to effectively cover hidden spaces such as narrow alleys and back streets in urban villages, resulting in numerous blind spots and compromising the completeness and timeliness of data collection. Furthermore, this monitoring model lacks the precise analytical capabilities to understand the diffusion paths and infiltration mechanisms of pollutants in complex built environments, making it difficult to meet the needs of refined air quality management in high-density communities.

[0003] The aforementioned situation presents numerous challenges to air quality monitoring and pollution source tracing in complex areas such as urban villages, necessitating the exploration of more targeted monitoring and analysis methods to achieve precise control and effective management of air quality in these special scenarios. Summary of the Invention

[0004] The purpose of this invention is to provide a multi-source cross-domain behavioral data fusion and analysis system and method based on knowledge graphs, and to solve the following technical problems:

[0005] Traditional air quality monitoring systems rely mainly on fixed site deployments, which exposes significant limitations when dealing with congested urban areas, lacking spatial adaptability and data validity.

[0006] The objective of this invention can be achieved through the following technical solutions:

[0007] A knowledge graph-based method for multi-source cross-domain behavioral data fusion and analysis includes the following steps:

[0008] Gas sensors are deployed on public vehicles in the target area to periodically detect the concentration values ​​and location coordinates of pollutants along the driving path, generating a time-series gas concentration dataset.

[0009] A spatiotemporal knowledge graph structure is constructed, comprising a static building layer, a dynamic path layer, and a fusion data layer. The static building layer loads the three-dimensional contour data of buildings and identifies the window positions. The dynamic path layer maps the driving trajectories of public vehicles. The fusion data layer stores the collected time-series gas concentration dataset.

[0010] Based on the real-time location of public vehicles, the current road segment in the dynamic path layer is matched, the spatial relative relationship between public vehicles and the exterior facades of buildings on both sides is calculated, and the gas concentration value is mapped to virtual monitoring points on the exterior facades of buildings in combination with real-time wind direction and speed parameters to generate a concentration distribution map of the exterior facades of buildings.

[0011] Obtain the concentration distribution map of the exterior facade of the target building, calculate the gas infiltration flux based on window type, opening size and opening status, and deduce the indoor pollutant gas concentration change process of the target building through dynamic pressure balance equation;

[0012] The indoor pollutant concentration projection results are integrated along a time axis to form a spatiotemporal evolution map of air quality in the target area, with each building as a unit.

[0013] As a further aspect of the present invention: the specific process by which the gas sensor detects polluting gas is as follows:

[0014] Parallel independent detection channels are set up to capture carbon dioxide concentration, total volatile organic compound concentration and fine particulate matter concentration respectively. Each independent detection channel shares an air inlet and uses grating wave division technology to isolate spectral interference.

[0015] When the vehicle is in a narrow alley, the low-speed detection mode is activated, and the sampling period of each gas component is extended to a multiple of the standard value. When the vehicle enters the main road, the high-speed detection mode is switched to prioritize the collection of carbon dioxide concentration values. All collected data are appended with millisecond-level timestamps, compressed by Huffman coding, and transmitted to the regional gateway via narrowband IoT.

[0016] As a further aspect of the present invention: the specific process of constructing the spatiotemporal knowledge graph structure is as follows:

[0017] Digital surface models of urban village buildings are obtained from urban planning databases, the geometric topology of building facades is extracted, and the spatial coordinates of windows are marked; the dynamic path layer integrates the fixed routes, stops, and real-time trajectories of public vehicles, discretizing the continuous trajectory into a sequence of path nodes in seconds; the fusion data layer establishes a four-dimensional spatiotemporal index matrix, with the matrix row vectors corresponding to the path node numbers and the column vectors associated with pollutant gas concentration values, temperature values, humidity values, wind speed values, and wind direction angle values;

[0018] A unique spatial hash code is generated for each building, and the spatial hash code is associated with all path nodes within a defined range around the building.

[0019] As a further aspect of the present invention: the data collection frequency of each path node in the dynamic path layer within a set time period in the past is counted, the building facade grid with a coverage frequency lower than the daily minimum requirement is marked as the target grid, the real-time location and route to be traveled of nearby public vehicles are obtained, and a supplementary monitoring path covering the target grid is generated, wherein the total mileage increment of the supplementary monitoring path does not exceed the percentage upper limit of the original route to be traveled.

[0020] The system sends route update instructions to nearby public vehicles. These instructions are embedded in the vehicle navigation system and disguised as regular traffic congestion avoidance prompts. When the target public vehicle enters the newly added route node, a high-precision detection mode is automatically triggered.

[0021] As a further aspect of the present invention: the specific process for generating the building facade concentration distribution map is as follows:

[0022] When a public vehicle enters the target road segment, the set of building facade grids within a set range around the current path node is obtained; a gas diffusion ellipse model is constructed based on the real-time wind speed and direction, with the major axis of the ellipse aligned with the wind direction and the minor axis length inversely proportional to the wind speed; the vertical distance from the center point of each facade grid to the edge of the diffusion ellipse is calculated, and the concentration allocation weight coefficient is determined based on the distance value.

[0023] Atmospheric stability correction is applied to the measured pollutant concentration values ​​of vehicles. The correction parameters include temperature gradient value and solar radiation intensity value. The corrected pollutant concentration values ​​are then distributed to each facade grid according to weight coefficients. The median concentration of the grid is aggregated once every set time interval. The concentration contour map of the building facade is generated by cubic spline interpolation.

[0024] As a further aspect of the present invention: the specific process of calculating the gas permeation flux based on the window opening size and opening state is as follows:

[0025] Obtain physical dimension data of window units on the building facade, including the width and height of the window openings; identify the structural type of the window unit; for sliding windows, measure the proportion of the window sash movement distance to the total track length and convert it into an equivalent ventilation area reduction factor.

[0026] For casement windows, real-time images of the window's open state are acquired, and the window opening angle is calculated using a geometric projection transformation algorithm. When the opening angle is less than 30 degrees, a ventilation efficiency reduction factor dominated by turbulence effect is used, and when the opening angle is greater than 30 degrees, a ventilation efficiency reduction factor dominated by laminar flow effect is used. In cases where there is damage to the window structure, the effective ventilation area compensation value is calculated based on the image edge damage characteristics.

[0027] The final effective ventilation area is obtained by multiplying the standard window area value by the reduction factor and then adding the compensation value. Based on the pollutant gas concentration gradient value at the center point of the window in the building facade concentration distribution map, combined with the component data of the real-time wind speed value in the window normal direction, the gas permeation flux value per unit time is calculated according to the principle of gas molecule diffusion dynamics.

[0028] As a further aspect of the present invention: the specific process of simulating the change in indoor pollutant gas concentration in the target building is as follows:

[0029] Using the gas permeation flux as the boundary input condition, a three-dimensional Cartesian coordinate grid space is established in the interior space of the building; the indoor pollutant gas background concentration field is initialized, and derivation equations including mass conservation equation, turbulent mixing equation and adsorption sedimentation equation are constructed.

[0030] The derivation equation is decomposed into three calculation units: convection term, diffusion term, and source term using the operator splitting method. The convection term uses an upwind scheme to calculate the concentration transfer process in the direction of gas flow. The diffusion term uses the finite difference method to solve the concentration gradient propagation process. The source term adjusts the pollution release intensity based on the abnormal gas concentration fluctuations collected from virtual monitoring points on the facade.

[0031] The concentration field of the entire grid space is synchronously updated every 30 seconds. The iterative calculation is terminated when the rate of change of the Euclidean distance of the concentration field in adjacent time steps is lower than the preset convergence threshold. The pollutant gas concentration distribution matrix of each grid cell under steady-state conditions is output as the inversion result.

[0032] As a further aspect of this invention: for the indoor pollutant gas concentration projection results of each building, calculate the percentage coefficient of the total effective ventilation area of ​​windows to the building's exterior surface area; when the coefficient is lower than a set threshold, generate a boundary condition insufficiency marker; analyze the distribution characteristics of the concentration gradient vector direction of monitoring points on adjacent facades of the building, calculate the maximum angle between vectors, and generate a data conflict marker if the maximum angle between vectors exceeds 90 degrees; calculate the indoor space volume ratio covered by pollutant gas using gas movement path simulation technology, and generate a projection blind zone marker when the volume ratio does not reach a preset standard value.

[0033] When any marker is detected, a topological anomaly warning map is dynamically rendered on the map interface of the corresponding building, and the spatial coordinates and type identifier of the anomaly are marked using a heat-coding method.

[0034] As a further aspect of the present invention: the repair operation after generating the marker is as follows:

[0035] When the boundary conditions are insufficient, the average value of the historical window permeability parameters of similar buildings is automatically used in the calculation. When the data conflict is generated, multiple public vehicles are dispatched to perform collaborative monitoring in the target area, and the concentration gradient field is reconstructed using a spatial interpolation algorithm. When the simulation blind zone is generated, indoor partition structure constraints are added to the calculation grid, and equivalent permeability channels are created at non-standard wall locations.

[0036] All repair operations generate versioned correction records. The corrected inference results are re-verified for gas motion path coverage. If the verification fails, the mesh resolution adaptive improvement mechanism is triggered until all markers are eliminated and the coverage meets the target.

[0037] This invention also includes a knowledge graph-based multi-source cross-domain behavioral data fusion and analysis system for implementing the above-described knowledge graph-based multi-source cross-domain behavioral data fusion and analysis method, comprising:

[0038] Gas sensors are used to periodically detect the concentration and location coordinates of pollutants along the driving path, generating a time-series gas concentration dataset.

[0039] The knowledge graph construction module is used to construct a spatiotemporal knowledge graph structure, which includes a static building layer, a dynamic path layer and a fusion data layer. The static building layer loads the three-dimensional outline data of buildings and identifies the window positions. The dynamic path layer maps the driving trajectory of public vehicles. The fusion data layer stores the collected time-series gas concentration dataset.

[0040] The concentration distribution generation module is used to match the current road segment in the dynamic path layer based on the real-time location of public vehicles, calculate the spatial relative relationship between public vehicles and the exterior facades of buildings on both sides, and map the gas concentration value to virtual monitoring points on the exterior surface of buildings by combining real-time wind direction and wind speed parameters, thereby generating a concentration distribution map of the exterior facade of buildings.

[0041] The pollutant gas simulation module is used to obtain the concentration distribution map of the exterior facade of the target building, calculate the gas infiltration flux based on window type, opening size and opening status, and simulate the indoor pollutant gas concentration change process of the target building through dynamic pressure balance equations.

[0042] The spatiotemporal evolution integration module is used to integrate the indoor pollutant gas concentration projection results along the time axis to form a spatiotemporal evolution map of air quality in the target area with a single building as the unit.

[0043] The beneficial effects of this invention are:

[0044] This invention utilizes gas sensors deployed in public vehicles and constructs a spatiotemporal knowledge graph to form a multi-source, cross-domain data fusion and analysis scheme, effectively overcoming the shortcomings of traditional fixed monitoring systems. This invention leverages the dynamic sampling characteristics of public vehicles to significantly reduce equipment installation and maintenance costs, avoiding the high investment required for traditional fixed sites. By matching the real-time vehicle location with a dynamic path layer and combining wind direction and speed parameters, it accurately maps gas concentration values ​​to building facades, generating concentration distribution maps. This effectively covers hidden spaces such as narrow alleys and back streets in urban villages, solving the problem of blind spots in monitoring. Based on window type, opening size, and opening status, it calculates gas permeation flux and uses dynamic pressure balance equations to extrapolate changes in indoor pollutant gas concentrations, improving the ability to analyze pollutant diffusion paths and permeation mechanisms in complex built environments, meeting the needs of refined governance in high-density communities. By statistically analyzing the data collection frequency of path nodes, it automatically generates supplementary monitoring paths and embeds them into the vehicle navigation system, ensuring the integrity and timeliness of data collection. A credibility assessment and self-repair mechanism is constructed. When insufficient boundary conditions, data conflicts, or extrapolation blind spots are detected, it automatically calls historical parameters of similar buildings, dispatches multiple vehicles for collaborative monitoring, or adjusts the computational grid to ensure the accuracy of indoor pollutant gas concentration extrapolation results, achieving precise control and effective governance of air quality in congested urban scenarios. Attached Figure Description

[0045] The invention will now be further described with reference to the accompanying drawings.

[0046] Figure 1 This is a flowchart illustrating the multi-source cross-domain behavioral data fusion and analysis method based on knowledge graphs according to the present invention.

[0047] Figure 2 This is a schematic diagram of the modules of the knowledge graph-based multi-source cross-domain behavioral data fusion and analysis system of the present invention. Detailed Implementation

[0048] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0049] Please see Figure 1 As shown, this invention is a method for multi-source cross-domain behavioral data fusion and analysis based on knowledge graphs, including the following steps:

[0050] First, gas sensors are deployed differently based on the structural characteristics and driving scenarios of different types of public vehicles. Sanitation vehicles integrate sensor modules on the sides or front bumpers to capture road dust and gases around garbage stations close to the ground; water trucks have waterproof detection equipment installed on their roofs to simultaneously collect volatile gases from the road surface during operation; buses have sensors positioned below the windows or on the roof of the passenger compartment to cover areas with high passenger flow; and delivery vehicles have lightweight equipment added to the outside of the cargo box to adapt to detection in narrow alleyways. These sensors periodically collect pollutant gas concentrations, covering indicators such as carbon dioxide, total volatile organic compounds, and fine particulate matter. Location coordinates are calculated through the fusion of GPS and base station positioning, and millisecond-level timestamps are added to generate a time-series dataset. For example, when delivery vehicles travel through urban villages, their sensors can collect gas data from back streets and alleys that are difficult to cover using traditional fixed stations.

[0051] Secondly, a spatiotemporal knowledge graph is constructed, comprising a static building layer, a dynamic path layer, and a fused data layer. The static building layer obtains the 3D outlines of buildings from the urban planning database, accurately annotating parameters such as facade geometry, window positions, and opening sizes to form a digital twin model of the building space, extracting features such as alleyway spacing and window orientation from buildings in urban villages. The dynamic path layer integrates fixed routes of public vehicles, such as sanitation vehicle operation routes, and real-time trajectories, discretizing the driving trajectories into a sequence of path nodes with second-level precision. The fused data layer establishes a four-dimensional spatiotemporal index matrix, associating path nodes with data such as pollution concentration, temperature, humidity, wind speed, and wind direction, generating a unique spatial hash code for each building, and associating it with surrounding path nodes to achieve semantic fusion of static building and dynamic monitoring data.

[0052] The third step involves spatial mapping of pollution data based on the real-time location of public vehicles. When a vehicle enters a target road segment, the system matches the current node in the dynamic path layer, calculates the spatial relationship between the vehicle and the facades of buildings on both sides, and, combined with real-time wind direction and speed parameters, uses a gas diffusion model to map the concentration values ​​detected on the vehicle to virtual monitoring points on the building facades. For example, when a water truck operates on a main road, the system distributes the concentration of exhaust pollutants on the road to the exterior windows of buildings on both sides of the road according to the wind direction; when a sanitation vehicle passes a garbage station, it maps the garbage odor data to the facades of surrounding buildings, generating a refined concentration distribution map.

[0053] The fourth step involves extrapolating indoor pollutant gas changes based on the concentration distribution on the building facade. After acquiring the concentration data of the target building's facade, the system analyzes the window type and status: for sliding windows, the equivalent ventilation area is determined by measuring the proportion of the sash blade movement distance to the total track length; for casement windows, image recognition technology is used to calculate the opening angle, and aerodynamic principles are combined to distinguish ventilation efficiency under turbulent and laminar flow conditions; for structurally damaged windows, the effective ventilation area is dynamically compensated according to the degree of damage. Based on the above parameters, and through a dynamic pressure balance equation, considering factors such as external pollutant concentration, indoor-outdoor air pressure difference, and gas diffusion rate, the changes in pollutant gas concentration inside the target building are extrapolated.

[0054] Finally, the indoor pollution projection results of each building are integrated along a timeline to form a spatiotemporal evolution map of air quality in the target area, with each building as a unit. This map presents the distribution and changing trends of pollutants in different buildings at different times in real time, intuitively showing abnormal TVOC concentrations in urban village buildings caused by the accumulation of construction waste or pollutant accumulation in commercial complexes due to ventilation system failures, providing data support and decision-making tools for precise urban air quality management.

[0055] In a preferred embodiment of the present invention, the specific process of the gas sensor detecting polluting gas is as follows:

[0056] The sensor employs a parallel independent detection channel design, with separate detection modules for carbon dioxide, total volatile organic compounds (TVOC), and fine particulate matter (PM2.5 / PM10). All channels share the same air inlet to ensure synchronous sampling and avoid timing differences affecting the detection results. Each channel integrates a grating wavelength division device at its front end, utilizing the principle of light diffraction to separate the mixed light signal according to wavelength. This allows the 4.26μm characteristic absorption peak of carbon dioxide and the 3.3μm characteristic absorption peak of TVOC to enter their respective detection units, achieving physical isolation of the spectral signals and ensuring detection accuracy.

[0057] The detection mode dynamically switches based on the vehicle's driving scenario. When the vehicle enters narrow alleys, urban villages, or other areas with poor air circulation, the system activates a low-speed detection mode, extending the sampling period for each gas component to a specific multiple of the standard value. This multiple is determined by the sensor model and energy consumption strategy to increase the gas's reaction time within the detection module, ensuring complete capture of concentration fluctuation data. For example, when a delivery vehicle travels through alleys in an urban village, the low-speed mode can effectively collect changes in TVOC concentration caused by residents cooking, renovations, or small workshop production emissions. When the vehicle enters open areas such as main roads, the system switches to a high-speed detection mode, prioritizing the collection of carbon dioxide concentration values. As a key indicator of traffic pollution, carbon dioxide is closely related to traffic flow and speed, and high-frequency collection can quickly identify regional pollution trends. In high-speed mode, the sensor reduces energy consumption in non-priority channels through dynamic power adjustment, balancing data acquisition efficiency and equipment battery life.

[0058] All collected data is appended with millisecond-level timestamps based on atomic clock calibration to ensure high-precision alignment in the time dimension. The data is then compressed using Huffman coding, an algorithm that constructs an optimal prefix coding tree by statistically analyzing symbol frequencies to reduce data redundancy and transmission load. The compressed data is then encrypted and transmitted to the regional gateway via narrowband IoT. The gateway performs preliminary data cleaning and format standardization, removing outliers to provide a standardized dataset for subsequent spatiotemporal knowledge graph construction.

[0059] In another preferred embodiment of the present invention, the specific process of constructing the spatiotemporal knowledge graph structure is as follows:

[0060] The construction of the spatiotemporal knowledge graph begins with the digital modeling of static building layers. The system obtains digital surface models of urban village buildings from an urban planning database. This model, built using LiDAR scanning and Building Information Modeling (BIM) technology, includes detailed data such as the building's three-dimensional geometric coordinates, facade materials, and window locations and dimensions. Through edge detection and feature extraction algorithms, the system automatically identifies the geometric topology of the building facade, including the boundary contours of walls and windows, the length-to-width ratio of window openings, and window types. For the annotation of window spatial coordinates, computer vision technology is used to locate window openings in the digital surface model, generating a coordinate dataset containing window sill height, window sash width, and opening direction, providing key parameters for subsequent gas permeability analysis.

[0061] The dynamic path layer integrates fixed routes, stops, and real-time trajectory data of public vehicles such as sanitation trucks, water trucks, buses, and delivery trucks, forming a dynamic monitoring network covering the entire city. For vehicles with fixed routes, such as buses and sanitation trucks, the system obtains the latitude and longitude sequence of a preset route through the onboard terminal, discretizing it into path nodes at second-level intervals. Each node includes information such as the stop name, estimated arrival time, and geographic coordinates. For vehicles with real-time trajectories, such as delivery trucks, location data is transmitted back in real time via 4G / 5G communication modules. After noise removal using a Gaussian filtering algorithm, a continuous trajectory is generated, and then sampled at 1-second intervals to generate dynamic path nodes. All nodes are geocoded using the UTM coordinate system to ensure the spatial reference consistency of trajectory data from different vehicles. The node sequence also includes vehicle type labels, such as "sanitation truck - cleaning route" and "delivery truck - delivery area," which facilitates the system's identification of the driving characteristics of different vehicles during data fusion.

[0062] A four-dimensional spatiotemporal index matrix is ​​established through the data fusion layer. Its row vectors correspond to the path node numbers in the dynamic path layer, while the column vectors integrate multi-dimensional environmental parameters such as pollutant concentration, temperature, humidity, wind speed, and wind direction. Each element in the matrix represents a multi-parameter monitoring value at a specific time and spatial location. For example, node number "W001" at 14:00:05 on May 10, 2024 corresponds to a carbon dioxide concentration of 850 ppm, a wind speed of 2.3 m / s, and a wind direction of 180°. To achieve rapid association between buildings and monitoring data, the system generates a unique spatial hash code for each building. This code is generated based on the geometric center coordinates of the building and the spatial relationship between surrounding path nodes: a circular area with a radius of 50 meters is defined centered on the building, and all path node numbers within this area are mapped to fixed-length strings using a hash function, forming a bidirectional association between the building and the monitoring nodes. When querying historical monitoring data around a building, the associated path nodes and corresponding data can be quickly retrieved simply through the spatial hash code, significantly improving data query efficiency.

[0063] In a preferred embodiment, the data collection frequency of each path node in the dynamic path layer over the past 24 hours is statistically analyzed, with frequency statistics measured in hours. The difference between the actual number of collections for each node and the preset minimum requirement is calculated. For building facade grids whose collection frequency is lower than the minimum requirement for 12 consecutive hours, they are marked as target grids. These grids are usually located in deep alleys of urban villages, on the shady side of buildings, or in the gaps between dense building clusters, making them difficult to cover by conventional vehicle routes and easily forming monitoring blind spots.

[0064] Once the target grid is determined, the system uses a heuristic algorithm to generate supplementary monitoring routes based on the real-time locations of nearby public vehicles and their intended routes. The algorithm optimizes for "minimum mileage increment," inserting the shortest path segment passing through the target grid onto the vehicle's current route to ensure the total mileage increment does not exceed a reasonable proportion of the original route. During route generation, the system prioritizes vehicles traveling within 500 meters of the target grid and sends route update commands via the vehicle navigation system's API interface. To avoid interfering with the driver's normal operation, the update command is disguised as a regular traffic congestion avoidance prompt, such as "Congested road ahead, we suggest detouring via Xiangqian Road." This prompt is displayed synchronously with the real-time traffic data from the map navigation system, ensuring the driver naturally accepts and executes the route adjustment.

[0065] When a target public vehicle enters a newly added route node, the onboard sensors automatically trigger a high-precision detection mode. In this mode, the sensor sampling frequency is increased to twice that of the standard mode, and simultaneous detection across all channels is enabled. An anti-interference filtering algorithm is also used to reduce noise in real time, minimizing the impact of environmental noise on the detection results. For example, when a delivery vehicle receives a supplementary route instruction and detours through a deep alley in an urban village, the sensors collect data at 0.5-second intervals, accurately capturing characteristic pollutants such as formaldehyde emitted by small workshops within the alley. After this data is transmitted back in real time via narrowband IoT, the system automatically updates the concentration distribution map of the corresponding building facade, providing more accurate boundary conditions for subsequent indoor pollution simulations.

[0066] In another preferred embodiment of the present invention, the specific process of generating the building facade concentration distribution map is as follows:

[0067] When a public vehicle enters the target road segment, the system delineates a surrounding area based on the geographic coordinates of the current path nodes and extracts a set of grids representing the exterior facades of all buildings within that area. The grid division precision is determined according to the building size: high-rise buildings are divided into 5m x 5m grids, and low-rise buildings into 3m x 3m grids, ensuring that the spatial resolution matches the monitoring requirements.

[0068] The system constructs a gas diffusion ellipse model based on real-time wind speed and direction. The ellipse is centered on the vehicle's location, with its major axis aligned with the wind direction and its minor axis inversely proportional to the wind speed, visually reflecting the impact of the wind field on pollutant diffusion. To determine the concentration weights for each grid cell, the system calculates the vertical distance from the grid center point to the edge of the ellipse; the smaller the distance, the greater the weight. An inverse distance weighting method is used to ensure the physical law of concentration decay with distance.

[0069] Before allocating measured concentration values, atmospheric stability correction is required. Correction parameters include temperature gradient and solar radiation intensity. Under inversion conditions, atmospheric stability is high, and pollutant diffusion is weak, necessitating upward correction of concentration values. Under strong solar radiation, turbulence is vigorous, and diffusion capacity is strong, requiring downward correction. The correction process is based on atmospheric boundary layer theory, establishing a mapping relationship between stability parameters and correction coefficients through empirical formulas.

[0070] After the corrected concentration values ​​are weighted and distributed to each grid, the system performs median aggregation every set interval to suppress outlier interference. Finally, a continuous and smooth concentration contour map is generated through cubic spline interpolation. This algorithm avoids the jagged effect of linear interpolation, clearly presents the gradient changes of pollutants on the building facade, and provides accurate boundary conditions for indoor pollution infiltration analysis.

[0071] In another preferred embodiment of the present invention, the specific process of calculating the gas permeation flux based on the window opening size and opening state is as follows:

[0072] The calculation of gas infiltration flux closely integrates the physical characteristics and real-time status of windows, achieving accurate ventilation efficiency assessment through multi-dimensional parameter analysis. First, the width and height data of window units are obtained from the static building floor plan to determine the standard window area, i.e., the theoretical maximum ventilation area. For different window types, differentiated methods are used to analyze actual ventilation capacity.

[0073] For sliding windows, miniature displacement sensors installed in the window frame track monitor the movement distance of the sash blades in real time. The distance is then normalized based on the total track length to generate an equivalent ventilation area reduction factor. For example, if the sash blade movement distance accounts for 40% of the total track length, the current ventilation area is 40% of the standard area. This method avoids treating a partially open window as a fully open state, ensuring that the calculated ventilation area accurately reflects the actual degree of opening.

[0074] The opening status of the casement window is captured by images obtained from a window corner vision sensor, and the opening angle is analyzed using a geometric projection transformation algorithm. This algorithm first identifies the edge features of the window frame and sash, then detects straight lines and calculates the included angle using Hough transform. When the opening angle is less than 30 degrees, the airflow around the window is predominantly turbulent. The system uses a reduction factor based on aerodynamic experimental fitting to reflect the impact of airflow turbulence on ventilation efficiency at small opening angles. When the angle is greater than 30 degrees, the airflow becomes laminar, and another set of reduction factors is used to reflect the smooth airflow characteristics at large opening angles, making the ventilation efficiency assessment more consistent with actual flow field characteristics.

[0075] For windows with structural damage such as broken glass or deformed frames, the system uses image edge detection algorithms to identify damage features and calculates an effective ventilation area compensation value based on the location and area of ​​the damaged area. For example, if there is a 10 cm × 15 cm uncovered damaged area in the lower left corner of the window, this area is directly used as the compensation value, representing an additional ventilation path and avoiding an underestimation of infiltration flux due to structural defects.

[0076] Multiplying the standard window area by the corresponding reduction factor and then adding the structural damage compensation value yields the effective ventilation area reflecting the current ventilation capacity. The system incorporates the pollutant concentration gradient at the window center point on the facade concentration distribution map—the concentration change per unit distance—to characterize the driving force of pollutant diffusion into the room. It also includes the real-time wind speed component along the window normal, obtained through wind speed vector decomposition to reflect the transport effect of external airflow on pollutants. Based on the principles of gas molecule diffusion dynamics, and integrating window ventilation capacity, external concentration gradient, and airflow driving force, the system achieves accurate calculation of gas infiltration per unit time.

[0077] In a preferred embodiment, the specific process of simulating the change in indoor pollutant gas concentration in the target building is as follows:

[0078] First, a three-dimensional Cartesian coordinate grid is established within the building's interior space. The grid division precision is set according to the building's function and monitoring needs. Residential buildings use a relatively loose grid, such as a basic grid of 1.5m × 1.5m × 1.5m, to ensure effective capture of concentration changes in key activity areas such as bedrooms and living rooms. Commercial and office buildings use a high-density grid of 1m × 1m × 1m to accommodate the complex airflow organization and pollutant diffusion characteristics in open spaces. After grid division, the indoor pollutant background concentration field is initialized, with the background value selected as the statistical median of historical monitoring datasets for similar buildings. For example, the carbon dioxide background concentration for residential buildings is defaulted to the median of daily activity periods in urban households. This value is derived by analyzing long-term monitoring data from a large number of similar buildings in the area, ensuring that the initial conditions closely resemble real-life scenarios.

[0079] A derivation system comprising three core equations is constructed. The mass conservation equation, using grid cells as the basic calculation unit, describes the mass balance of pollutants in indoor space, ensuring the dynamic conservation of inflow and outflow rates and source-sink terms. The turbulent mixing equation employs a k-ε two-equation model to characterize indoor airflow, capturing the enhancing effect of turbulence caused by ventilation systems and human activities on pollutant diffusion. The adsorption-sedimentation equation quantifies the adsorption and release processes of pollutants on indoor surfaces such as walls, floors, and furniture. Parameter settings are combined with material characteristics, such as the TVOC adsorption coefficient of latex-painted walls and the formaldehyde slow-release rate of wooden furniture.

[0080] The system synchronously loads basic pollution release parameters from a building type feature library, which is associated with building usage and typical activity scenarios. For residential buildings, the system maps carbon dioxide release rates to human basal metabolic rates. The basal metabolic rate of an adult male at rest affects the carbon dioxide release rate, and the system dynamically adjusts the source strength based on the number of people in the room. For scenarios involving cooking activities, the system matches total volatile organic compound (TVOC) release curves to the kitchen space volume. The shape of the release curves is obtained by fitting measured data of household cooking fume emissions, which can reflect the release patterns of TVOCs during cooking in kitchens of different volumes.

[0081] The complex derivation equations are decomposed into three independent computational units using an operator splitting method. The convection term employs an upwind scheme to calculate concentration transfer along the gas flow direction. This scheme, by identifying the direction of airflow velocity, prioritizes interpolation using concentration values ​​from the upstream grid, effectively suppressing numerical oscillations under strong convection conditions and ensuring a clear characterization of the concentration front. The diffusion term uses the central difference method to solve the concentration gradient-driven molecular diffusion process, combining the physical properties of different gases to achieve accurate simulation of the Brownian motion-dominated micro-diffusion process. The dynamic source term relies on the collaborative sensing of virtual monitoring points on the exterior facade and an indoor mobile sensing network to capture abnormal concentration fluctuations in real time. When a monitoring point detects a sudden increase in concentration amplitude within a short period, the system automatically matches a preset pollution source release model. If the pulse characteristics match the instantaneous peak characteristics of nicotine concentration (i.e., a steep rise lasting approximately 5 to 10 minutes), the system activates the smoking behavior release model, superimposing additional source strength into the corresponding room grid unit. If it matches the characteristics of cleaning agent volatilization (i.e., a gradual rise in concentration lasting more than 30 minutes), the system calls the chemical cleaning model, dynamically calculating the release rate based on the amount of cleaning agent used.

[0082] The concentration field is updated across the entire grid in 30-second time steps, and the discretized equations are solved using the Gauss-Seidel iterative method. When the rate of change of the Euclidean distance between concentration fields in adjacent time steps falls below a preset convergence threshold (determined by the building's usage; different thresholds are set for residential and commercial buildings based on their respective accuracy requirements), the indoor pollution distribution is considered to have reached a steady state, and the calculation terminates. The final output 3D concentration distribution matrix contains the pollutant concentration value corresponding to each grid cell, visually representing various pollution distribution phenomena. Examples include pollutant accumulation in poorly ventilated areas such as wardrobe corners or near air conditioning return vents, changes in indoor concentration gradients caused by external pollution seeping through windows when windows are open, and localized abnormal concentration increases caused by sudden activities such as cooking and smoking.

[0083] In another preferred embodiment of the invention, for each building, the percentage coefficient of the total effective ventilation area of ​​windows to the building's external surface area is first calculated. This coefficient reflects the building's natural ventilation potential. When the coefficient is lower than a preset threshold, it indicates that the total window area is insufficient or the opening state is restricted, which may lead to insufficient boundary conditions to support accurate indoor airflow simulation. In this case, a boundary condition insufficiency marker is generated.

[0084] Further analysis of the concentration gradient vector distribution characteristics of monitoring points on adjacent building facades was conducted, and the consistency of monitoring data was assessed by calculating the maximum angle between the vectors. If the maximum angle between the vectors exceeded 90 degrees, it indicated a significant conflict in the concentration change trends of adjacent monitoring points, which may be caused by monitoring equipment errors, sudden changes in meteorological conditions, or interference from local pollution sources. In this case, a data conflict marker would be generated.

[0085] By using gas path simulation technology, the volume ratio of indoor space covered by pollutant gas is calculated. This indicator reflects the model's ability to capture the range of pollutant diffusion. When the volume ratio does not reach the preset standard value, it indicates that there may be areas in the model that have not been fully simulated, such as narrow passages or complex structures. In this case, a blind spot marker will be generated.

[0086] Once any marker is detected, the corresponding building's map interface dynamically renders a topological anomaly warning map. The warning map uses heat-coding to annotate the spatial coordinates of anomalies, and different types of anomalies are distinguished by specific identifiers: insufficient boundary conditions are marked with blue circles, data conflicts with red triangles, and inference blind spots with yellow rectangles. This visualization method allows operators to intuitively identify the location and type of anomalies, providing clear guidance for subsequent remediation.

[0087] In a preferred embodiment, the repair operation after generating the marker is as follows:

[0088] When insufficient boundary conditions are generated, the average historical window permeability parameters of similar buildings are automatically used in the calculation. This parameter library is built based on long-term monitoring data of similar buildings in the region and includes the average permeability coefficients of windows with different orientations and structures. By introducing historical parameters, the deficiencies of the current boundary conditions can be compensated for, ensuring that the simulation model reasonably simulates the impact of natural ventilation on indoor air quality.

[0089] When data conflict markers are generated, multiple public vehicles are dispatched to perform collaborative monitoring in the target area. The vehicles travel along a pre-defined grid path, increasing the density of monitoring points and acquiring more comprehensive pollution distribution data. Subsequently, a spatial interpolation algorithm is used to reconstruct the concentration gradient field. This algorithm combines Kriging interpolation and inverse distance weighting to generate a smooth and continuous concentration distribution surface based on the monitoring point data, effectively eliminating data conflicts and improving the accuracy of boundary conditions.

[0090] When blind zone markers are generated during simulation, interior partition structure constraints are added to the computational grid. By analyzing building floor plans and site photographs, non-standard wall locations such as temporary partitions and furniture arrangements are identified, and equivalent infiltration channels are created in the model. These channels are configured with different resistance coefficients based on the permeability of the actual structure to ensure that pollutants can reasonably diffuse to areas that were originally insufficiently simulated, thereby expanding the coverage of gas movement paths.

[0091] All repair operations generate versioned correction records, including repair time, repair type, parameter adjustment details, etc., forming a complete operation audit trail. The corrected simulation results need to be re-verified for gas movement path coverage. If the verification fails, an adaptive grid resolution enhancement mechanism is triggered. This mechanism automatically adjusts the computational grid precision of the corresponding region based on the characteristics of the abnormal area, for example, refining the base grid to a higher density grid to improve the model's ability to capture complex airflows. This process is continuously iterated until all markers are eliminated and coverage meets the target, ensuring that the simulation results accurately reflect the true distribution of indoor pollutants.

[0092] Please refer to Figure 2 As shown, the present invention also includes a knowledge graph-based multi-source cross-domain behavioral data fusion and analysis system for implementing the above-described knowledge graph-based multi-source cross-domain behavioral data fusion and analysis method, comprising:

[0093] Gas sensors are used to periodically detect the concentration and location coordinates of pollutants along the driving path, generating a time-series gas concentration dataset.

[0094] The knowledge graph construction module is used to construct a spatiotemporal knowledge graph structure, which includes a static building layer, a dynamic path layer and a fusion data layer. The static building layer loads the three-dimensional outline data of buildings and identifies the window positions. The dynamic path layer maps the driving trajectory of public vehicles. The fusion data layer stores the collected time-series gas concentration dataset.

[0095] The concentration distribution generation module is used to match the current road segment in the dynamic path layer based on the real-time location of public vehicles, calculate the spatial relative relationship between public vehicles and the exterior facades of buildings on both sides, and map the gas concentration value to virtual monitoring points on the exterior surface of buildings by combining real-time wind direction and wind speed parameters, thereby generating a concentration distribution map of the exterior facade of buildings.

[0096] The pollutant gas simulation module is used to obtain the concentration distribution map of the exterior facade of the target building, calculate the gas infiltration flux based on window type, opening size and opening status, and simulate the indoor pollutant gas concentration change process of the target building through dynamic pressure balance equations.

[0097] The spatiotemporal evolution integration module is used to integrate the indoor pollutant gas concentration projection results along the time axis to form a spatiotemporal evolution map of air quality in the target area with a single building as the unit.

[0098] The foregoing has provided a detailed description of one embodiment of the present invention, but this description is merely a preferred embodiment and should not be construed as limiting the scope of the invention. All equivalent variations and modifications made within the scope of the claims of this invention should still fall within the patent coverage of this invention.

Claims

1. A method for fusion and analysis of multi-source cross-domain behavioral data based on knowledge graphs, characterized in that, Includes the following steps: Gas sensors are deployed on public vehicles in the target area to periodically detect the concentration values ​​and location coordinates of pollutants along the driving path, generating a time-series gas concentration dataset. A spatiotemporal knowledge graph structure is constructed, comprising a static building layer, a dynamic path layer, and a fusion data layer. The static building layer loads the three-dimensional contour data of buildings and identifies the window positions. The dynamic path layer maps the driving trajectories of public vehicles. The fusion data layer stores the collected time-series gas concentration dataset. Based on the real-time location of public vehicles, the current road segment in the dynamic path layer is matched, the spatial relative relationship between public vehicles and the exterior facades of buildings on both sides is calculated, and the gas concentration value is mapped to virtual monitoring points on the exterior facades of buildings in combination with real-time wind direction and speed parameters to generate a concentration distribution map of the exterior facades of buildings. Obtain the concentration distribution map of the exterior facade of the target building, calculate the gas infiltration flux based on window type, opening size and opening status, and deduce the indoor pollutant gas concentration change process of the target building through dynamic pressure balance equation; The indoor pollutant concentration projection results are integrated along a time axis to form a spatiotemporal evolution map of air quality in the target area, with each building as a unit.

2. The method for multi-source cross-domain behavioral data fusion and analysis based on knowledge graphs according to claim 1, characterized in that, The specific process by which the gas sensor detects polluting gases is as follows: Parallel independent detection channels are set up to capture carbon dioxide concentration, total volatile organic compound concentration and fine particulate matter concentration respectively. Each independent detection channel shares an air inlet and uses grating wave division technology to isolate spectral interference. When the vehicle is in a narrow alley, the low-speed detection mode is activated, and the sampling period of each gas component is extended to a multiple of the standard value. When the vehicle enters the main road, the high-speed detection mode is switched to collect the carbon dioxide concentration value first. All collected data is appended with millisecond-level timestamps, compressed using Huffman coding, and transmitted to the regional gateway via narrowband IoT.

3. The method for multi-source cross-domain behavioral data fusion and analysis based on knowledge graphs according to claim 1, characterized in that, The specific process for constructing the spatiotemporal knowledge graph structure is as follows: Digital surface models of urban village buildings are obtained from urban planning databases, the geometric topology of building facades is extracted, and the spatial coordinates of windows are marked; the dynamic path layer integrates the fixed routes, stops, and real-time trajectories of public vehicles, discretizing the continuous trajectory into a sequence of path nodes in seconds; the fusion data layer establishes a four-dimensional spatiotemporal index matrix, with the matrix row vectors corresponding to the path node numbers and the column vectors associated with pollutant gas concentration values, temperature values, humidity values, wind speed values, and wind direction angle values; A unique spatial hash code is generated for each building, and the spatial hash code is associated with all path nodes within a defined range around the building.

4. The method for multi-source cross-domain behavioral data fusion and analysis based on knowledge graphs according to claim 3, characterized in that, The data collection frequency of each path node in the dynamic path layer within the past set time period is counted. Building facade grids with a coverage frequency lower than the daily minimum requirement are marked as target grids. The real-time location and route to be traveled of nearby public vehicles are obtained. Supplementary monitoring paths covering the target grids are generated. The total mileage increment of the supplementary monitoring paths does not exceed the percentage limit of the original route to be traveled. The system sends route update instructions to nearby public vehicles. These instructions are embedded in the vehicle navigation system and disguised as regular traffic congestion avoidance prompts. When the target public vehicle enters the newly added route node, a high-precision detection mode is automatically triggered.

5. The method for multi-source cross-domain behavioral data fusion and analysis based on knowledge graphs according to claim 1, characterized in that, The specific process for generating the building facade concentration distribution map is as follows: When a public vehicle enters the target road segment, the set of building facade meshes within a set range around the current path node is obtained; a gas diffusion ellipse model is constructed based on real-time wind speed and direction, with the major axis of the ellipse aligned with the wind direction and the minor axis inversely proportional to the wind speed. Calculate the vertical distance from the center point of each facade grid to the edge of the diffusion ellipse, and determine the concentration allocation weighting coefficient based on the distance value; Atmospheric stability correction is applied to the measured pollutant concentration values ​​of vehicles. The correction parameters include temperature gradient value and solar radiation intensity value. The corrected pollutant concentration values ​​are then distributed to each facade grid according to weight coefficients. The median concentration of the grid is aggregated once every set time interval. The concentration contour map of the building facade is generated by cubic spline interpolation.

6. The method for multi-source cross-domain behavioral data fusion and analysis based on knowledge graphs according to claim 1, characterized in that, The specific process for calculating gas permeation flux based on window type, opening size, and opening status is as follows: Obtain physical dimension data of window units on the building facade, including the width and height of the window openings; identify the structural type of the window unit; for sliding windows, measure the proportion of the window sash movement distance to the total track length and convert it into an equivalent ventilation area reduction factor. For casement windows, real-time images of the window's open state are acquired, and the window opening angle is calculated using a geometric projection transformation algorithm. When the opening angle is less than 30 degrees, a ventilation efficiency reduction factor dominated by turbulence effect is used, and when the opening angle is greater than 30 degrees, a ventilation efficiency reduction factor dominated by laminar flow effect is used. In cases where there is damage to the window structure, the effective ventilation area compensation value is calculated based on the image edge damage characteristics. The final effective ventilation area is obtained by multiplying the standard window area value by the reduction factor and then adding the compensation value. Based on the pollutant gas concentration gradient value at the center point of the window in the building facade concentration distribution map, combined with the component data of the real-time wind speed value in the window normal direction, the gas permeation flux value per unit time is calculated according to the principle of gas molecule diffusion dynamics.

7. The method for multi-source cross-domain behavioral data fusion and analysis based on knowledge graphs according to claim 6, characterized in that, The specific process of the indoor pollutant gas concentration change process of the target building is as follows: Using the gas permeation flux as the boundary input condition, a three-dimensional Cartesian coordinate grid space is established in the interior space of the building; the indoor pollutant gas background concentration field is initialized, and derivation equations including mass conservation equation, turbulent mixing equation and adsorption sedimentation equation are constructed. The derivation equation is decomposed into three calculation units: convection term, diffusion term, and source term using the operator splitting method. The convection term uses an upwind scheme to calculate the concentration transfer process in the direction of gas flow. The diffusion term uses the finite difference method to solve the concentration gradient propagation process. The source term adjusts the pollution release intensity based on the abnormal gas concentration fluctuations collected from virtual monitoring points on the facade. The concentration field of the entire grid space is synchronously updated every 30 seconds. The iterative calculation is terminated when the rate of change of the Euclidean distance of the concentration field in adjacent time steps is lower than the preset convergence threshold. The pollutant gas concentration distribution matrix of each grid cell under steady-state conditions is output as the inversion result.

8. The method for multi-source cross-domain behavioral data fusion and analysis based on knowledge graphs according to claim 1, characterized in that, For the indoor pollutant concentration projection results of each building, calculate the percentage coefficient of the total effective ventilation area of ​​windows to the building's exterior area. When the coefficient is lower than a set threshold, generate a boundary condition deficiency marker. Analyze the distribution characteristics of the concentration gradient vector direction of monitoring points on adjacent facades of the building, calculate the maximum angle between vectors, and generate a data conflict marker if the maximum angle between vectors exceeds 90 degrees. Calculate the indoor space volume ratio covered by pollutant gas using gas movement path simulation technology. When the volume ratio does not reach a preset standard value, generate a projection blind zone marker. When any marker is detected, a topological anomaly warning map is dynamically rendered on the map interface of the corresponding building, and the spatial coordinates and type identifier of the anomaly are marked using a heat-coding method.

9. The method for multi-source cross-domain behavioral data fusion and analysis based on knowledge graphs according to claim 8, characterized in that, The repair operation after generating the marker is as follows: When the boundary conditions are insufficient, the average value of the historical window permeability parameters of similar buildings is automatically used in the calculation. When the data conflict is generated, multiple public vehicles are dispatched to perform collaborative monitoring in the target area, and the concentration gradient field is reconstructed using a spatial interpolation algorithm. When the simulation blind zone is generated, indoor partition structure constraints are added to the calculation grid, and equivalent permeability channels are created at non-standard wall locations. All repair operations generate versioned correction records. The corrected inference results are re-verified for gas motion path coverage. If the verification fails, the mesh resolution adaptive improvement mechanism is triggered until all markers are eliminated and the coverage meets the target.

10. A knowledge graph-based multi-source cross-domain behavioral data fusion and analysis system, used to implement the knowledge graph-based multi-source cross-domain behavioral data fusion and analysis method according to any one of claims 1-9, characterized in that, include: Gas sensors are used to periodically detect the concentration and location coordinates of pollutants along the driving path, generating a time-series gas concentration dataset. The knowledge graph construction module is used to construct a spatiotemporal knowledge graph structure, which includes a static building layer, a dynamic path layer and a fusion data layer. The static building layer loads the three-dimensional outline data of buildings and identifies the window positions. The dynamic path layer maps the driving trajectory of public vehicles. The fusion data layer stores the collected time-series gas concentration dataset. The concentration distribution generation module is used to match the current road segment in the dynamic path layer based on the real-time location of public vehicles, calculate the spatial relative relationship between public vehicles and the exterior facades of buildings on both sides, and map the gas concentration value to virtual monitoring points on the exterior surface of buildings by combining real-time wind direction and wind speed parameters, thereby generating a concentration distribution map of the exterior facade of buildings. The pollutant gas simulation module is used to obtain the concentration distribution map of the exterior facade of the target building, calculate the gas infiltration flux based on window type, opening size and opening status, and simulate the indoor pollutant gas concentration change process of the target building through dynamic pressure balance equations. The spatiotemporal evolution integration module is used to integrate the indoor pollutant gas concentration projection results along the time axis to form a spatiotemporal evolution map of air quality in the target area with a single building as the unit.

Images