Intelligent fire safety assessment method and system for building

Abstract

The invention discloses an intelligent fire safety assessment method and system for a building, and relates to the technical field of fire safety, and the method comprises the steps: carrying out the spatial registration and interpolation calculation of a risk thermodynamic diagram and a grading early warning grade, generating an evacuation guidance diagram, carrying out the dynamic path planning, and obtaining an optimal evacuation route; based on the optimal evacuation route and the evacuation guide map, carrying out space coverage comparison to obtain a route coverage range, and generating a route validity verification result through coverage efficiency matching; and performing risk association analysis according to the route validity verification result and the preprocessed multi-source data, generating a risk correction factor, and obtaining a safety assessment report through multi-criterion decision analysis. According to the invention, through the risk tensor integral model and the multi-objective optimization algorithm, real-time and accurate evaluation of the building fire safety risk is realized, and scientific guidance is provided for emergency evacuation.

Description

Technical Field

[0001] This invention relates to the field of fire safety technology, and in particular to an intelligent fire safety assessment method and system for buildings. Background Technology

[0002] As urban buildings become more high-rise, multifunctional, and integrated, fire safety assessment technology has gradually shifted from relying on manual experience to a comprehensive analysis system supported by digital and intelligent methods. Early fire safety assessments were primarily conducted through manual inspections of fire protection facilities, review of building drawings, and historical fire records. The assessment results were limited by the professional level of the assessors and the availability of data, making it difficult to cover potential risks in complex scenarios. In recent years, the widespread adoption of Building Information Modeling (BIM) technology has provided high-precision three-dimensional spatial data support for fire safety assessments, enabling parametric modeling of building structures and fire protection facilities, and visualization of spatial relationships.

[0003] The main shortcomings of existing technologies are as follows: First, they lack dynamic fault prediction capabilities. The core causes of building fire safety risks include dynamic factors such as the performance degradation of fire protection facilities, environmental anomalies, and personnel behavior. However, existing technologies mostly focus on verifying static risk indicators and lack means to predict potential faults such as the trend of facility performance degradation and abnormal fluctuations in environmental parameters. Second, the depth of multi-source data collaborative analysis is insufficient. Building fire safety involves multi-dimensional information such as building structure, equipment status, and personnel distribution. However, existing technologies mostly conduct assessments based on a single data source. The spatial correlation and temporal coupling between different data are not fully explored, resulting in a disconnect between risk assessment results and decision-making processes such as emergency evacuation route planning and rescue resource allocation, making it difficult to form a closed-loop management of "assessment-early warning-response". Summary of the Invention

[0004] In view of the aforementioned existing problems, the present invention is proposed.

[0005] Therefore, this invention provides an intelligent fire safety assessment method for buildings to address the problems of lack of dynamic fault prediction capabilities and insufficient multi-source data collaborative analysis.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution:

[0007] In a first aspect, the present invention provides an intelligent fire safety assessment method for buildings, comprising,

[0008] Collect multi-source data and preprocess it, then extract multi-dimensional features based on the preprocessed multi-source data;

[0009] Multidimensional features are input into the risk tensor integral model, the environmental dynamic fusion layer performs spatiotemporal feature fusion, and the risk evolution layer performs continuous time integration to generate a risk heat map and graded early warning levels.

[0010] Spatial registration and interpolation calculations are performed on the risk heat map and the graded early warning level to generate an evacuation guidance map and perform dynamic route planning to obtain the optimal evacuation route.

[0011] Based on the optimal evacuation route and evacuation guidance map, spatial coverage comparison is performed to obtain the route coverage range, and route effectiveness verification results are generated through coverage efficiency matching.

[0012] Based on the route validity verification results and preprocessed multi-source data, risk correlation analysis is performed to generate risk correction factors, and a safety assessment report is obtained through multi-criteria decision analysis.

[0013] As a preferred embodiment of the intelligent fire safety assessment method for buildings described in this invention, the multi-source data includes building BIM data, IoT sensor data, and personnel positioning data;

[0014] The preprocessing includes data cleaning, format conversion, deduplication, normalization, and outlier handling.

[0015] As a preferred embodiment of the intelligent fire safety assessment method for buildings described in this invention, the specific steps for extracting multidimensional features are as follows:

[0016] Based on building BIM data, structural risk characteristics of buildings are extracted through geometric and attribute information analysis.

[0017] For IoT sensor data, analyze the difference between adjacent time points to extract dynamic environmental risk characteristics;

[0018] Based on personnel location data, grid division and density clustering are performed to extract population distribution risk characteristics.

[0019] As a preferred embodiment of the intelligent fire safety assessment method for buildings described in this invention, the specific steps for generating the risk heat map and graded early warning levels are as follows:

[0020] The spatiotemporal tensor decomposition method is used to perform multi-dimensional feature alignment and cross-layer attention connection on the environmental dynamic fusion layer and the risk evolution layer to construct a risk tensor integral model;

[0021] Multidimensional features are input into the risk tensor integral model, and the environmental dynamic fusion layer fuses spatiotemporal features through spatiotemporal convolution to form dynamic risk field features.

[0022] The risk evolution layer uses the Euler-Lagrange explicit integration method to perform continuous-time integration and generate the risk intensity evolution trajectory;

[0023] The dynamic risk field characteristics and risk intensity evolution trajectory are spatiotemporally aligned and integrated through a feature fusion channel to generate a risk heat map and graded early warning levels.

[0024] As a preferred embodiment of the intelligent fire safety assessment method for buildings described in this invention, the specific steps for generating an evacuation guidance map and obtaining the optimal evacuation route are as follows:

[0025] Spatially register the risk heat map with the graded early warning level to generate risk early warning overlay data, and perform interpolation calculations to obtain the evacuation guidance map;

[0026] Based on the evacuation guidance map and personnel location data, a risk-weighted guidance map is generated, and an iterative search is performed to output the optimal evacuation route.

[0027] As a preferred embodiment of the intelligent fire safety assessment method for buildings described in this invention, the specific steps for generating route validity verification results are as follows:

[0028] The optimal evacuation route is compared with the evacuation guidance map to generate the coverage area boundary, and the coverage area is calculated to obtain the route coverage range.

[0029] Based on building BIM data and evacuation guidance maps, the coverage efficiency threshold is calculated using a parameter derivation method.

[0030] Based on the route coverage area, obtain the coverage efficiency index and match it with the coverage efficiency threshold to generate route validity verification results.

[0031] As a preferred embodiment of the intelligent fire safety assessment method for buildings described in this invention, the specific steps for obtaining the safety assessment report are as follows:

[0032] Regional overlay analysis was performed on the route validity verification results and preprocessed multi-source data to extract risk correlation characteristics of each region.

[0033] Perform local spatial correlation strength analysis on the risk correlation characteristics of each region to generate risk correction factors;

[0034] Based on the coverage efficiency index, graded early warning levels, and population distribution risk characteristics, the indicator weights are obtained, and the indicator weights are weighted and summed to construct a multi-criteria evaluation indicator set.

[0035] Based on a multi-criteria assessment indicator set and risk correction factors, a comprehensive safety score is obtained, and safety levels are classified according to the comprehensive safety score, and a safety assessment report is output.

[0036] Secondly, the present invention provides an intelligent fire safety assessment system for buildings, including a data acquisition module, a risk assessment module, a path planning module, an effective verification module, and a report generation module;

[0037] The data acquisition module is used to collect multi-source data and preprocess it, and extract multi-dimensional features based on the preprocessed multi-source data;

[0038] The risk assessment module is used to input multidimensional features into the risk tensor integral model, the environmental dynamic fusion layer performs spatiotemporal feature fusion, and the risk evolution layer performs continuous time integration to generate a risk heat map and graded early warning levels.

[0039] The route planning module is used to spatially register and interpolate the risk heat map with the graded warning level, generate the evacuation guidance map, and perform dynamic route planning to obtain the optimal evacuation route.

[0040] The effective verification module is used to perform spatial coverage comparison based on the optimal evacuation route and the evacuation guidance map, obtain the route coverage range, and generate route effectiveness verification results through coverage efficiency matching.

[0041] The report generation module is used to perform risk correlation analysis based on the route validity verification results and preprocessed multi-source data, generate risk correction factors, and obtain a safety assessment report through multi-criteria decision analysis.

[0042] Thirdly, the present invention provides a computer device including a memory and a processor, wherein the memory stores a computer program, wherein when the computer program is executed by the processor, it implements any step of the intelligent fire safety assessment method for buildings as described in the first aspect of the present invention.

[0043] Fourthly, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, it implements any step of the intelligent fire safety assessment method for buildings as described in the first aspect of the present invention.

[0044] The beneficial effects of this invention are as follows: by using the risk tensor integral model to perform spatiotemporal feature fusion and continuous time integration, the real-time performance of risk assessment is improved, the accuracy of risk prediction is enhanced, and a multi-objective optimization algorithm is used to iteratively search for the optimal evacuation route, making the evacuation plan more in line with the actual scenario, improving the accuracy of emergency response, and realizing real-time adaptation of evacuation routes. Attached Figure Description

[0045] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0046] Figure 1 A flowchart for an intelligent fire safety assessment method for buildings;

[0047] Figure 2 A schematic diagram of an intelligent fire safety assessment system for a building;

[0048] Figure 3 A flowchart for evacuation guidance and route planning;

[0049] Figure 4 This is a flowchart for a closed-loop safety assessment. Detailed Implementation

[0050] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0051] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0052] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0053] Reference Figures 1-4 As one embodiment of the present invention, this embodiment provides an intelligent fire safety assessment method for buildings, comprising the following steps:

[0054] S1. Collect multi-source data and preprocess it, then extract multi-dimensional features based on the preprocessed multi-source data;

[0055] S1.1 Multi-source data includes building BIM data, IoT sensor data, and personnel positioning data;

[0056] It should be noted that by reading the building's structured spatial information files, basic structural information of the building, including geometric dimensions, spatial layout, and component attributes, is collected. Simultaneously, millimeter-level building point cloud data is collected through laser scanning equipment. Three-dimensional images are generated by taking oblique photos with drones. The basic structural information of the building, the building point cloud data, and the three-dimensional images are compared, corrected, and fused to form complete building BIM data. Temperature, humidity, and energy consumption sensors are installed in public areas such as building equipment rooms and connected to IoT gateways to collect IoT sensor data. By deploying UWB tags and receivers, real-time three-dimensional coordinates of personnel are collected using the precise ranging of ultra-wideband signals to generate personnel positioning data.

[0057] S1.2 The preprocessing includes data cleaning, format conversion, deduplication, normalization and outlier handling.

[0058] It should be noted that, firstly, the collected multi-source data, including building BIM data, IoT sensor data, and personnel positioning data, undergoes data cleaning to remove incomplete, erroneous, or irrelevant data records. Next, format conversion is performed to ensure that all data sources adhere to a unified format standard for subsequent processing. Then, deduplication is performed to eliminate duplicate data entries and ensure the uniqueness of the dataset. The normalization step is used to map data values ​​from different sources to the same scale to avoid certain features from having an undue impact on the results due to excessive differences in magnitude. Finally, outlier handling is used to identify and correct data points that deviate from the normal range, ensuring the accuracy and reliability of the dataset.

[0059] S1.3 Based on building BIM data, geometric and attribute information are analyzed and extracted to extract building structural risk characteristics using spatial analysis method;

[0060] It should be noted that the geometric dimensions, topological relationships, and attribute parameters of components are extracted from the structural information of the building foundation in the building BIM data. Simultaneously, millimeter-level point cloud data acquired by laser scanning and 3D images generated by UAV oblique photography are used, and aligned to the same spatial reference. Then, a spatial analysis algorithm is used to compare the geometric dimensions of the components in the building BIM data with the millimeter-level point cloud data to obtain the geometric deviation between the actual length and the design length. Simultaneously, the attribute parameters in the building BIM data are compared with the actual measured values, marking components whose actual measured values ​​do not meet the attribute parameters. Combining the topological relationships in the building BIM data, the observed connection states are compared, marking components whose observed connection states do not meet the topological relationships. Finally, analysis results are generated, and risk characteristics such as geometric anomalies, attribute anomalies, and connection defects in the building structure are extracted based on the analysis results. The characteristic types, spatial locations, and degrees of deviation are identified, forming the building structure risk characteristics.

[0061] S1.4. Based on IoT sensor data, the time series difference method is used to analyze the difference between adjacent time points and extract dynamic environmental risk characteristics.

[0062] It should be noted that time-stamped environmental parameter data are collected through a sensor network to form multi-dimensional time series such as temperature and humidity time series, CO2 concentration time series, and equipment power time series. Then, based on the difference between the environmental parameter data at the current time point and the environmental parameter data at the previous time point, the first-order difference value of the multi-dimensional time series is obtained to generate a multi-dimensional difference series. Finally, the dynamic environmental risk characteristics are extracted through Z-score test, and the feature type, change magnitude, duration, and associated parameters are recorded to ultimately form the dynamic environmental risk characteristics.

[0063] S1.5 Based on personnel location data, spatial clustering is used to perform grid division and density clustering to extract population distribution risk characteristics;

[0064] It should be noted that, based on the building's floor plan layout, regular grids are divided, and the timestamps and spatial coordinates of personnel location data are mapped to the corresponding regular grids. The number of people gathered in each regular grid within a unit of time is counted to generate a three-dimensional dataset of "grid-time-number of people". Subsequently, the DBSCAN equal-density clustering algorithm is used to identify personnel gathering sub-regions within the regular grids, extract parameters such as the center coordinates, coverage area, and number of personnel location data points contained in the personnel gathering sub-regions, and obtain the crowd flow rate through a sliding time window. Combined with the functional area attributes of the building (such as evacuation routes and equipment rooms), the risk level is determined, and finally, the crowd distribution risk characteristics are output.

[0065] S2. Input multidimensional features into the risk tensor integral model, perform spatiotemporal feature fusion in the environmental dynamic fusion layer, and perform continuous time integration in the risk evolution layer to generate a risk heat map and graded early warning levels.

[0066] S2.1. Using the spatiotemporal tensor decomposition method, multi-dimensional feature alignment and cross-layer attention connection are performed on the environmental dynamic fusion layer and the risk evolution layer to construct a risk tensor integral model;

[0067] It should be noted that, based on environmental parameters collected by IoT sensors, the time series is divided into fixed time windows to generate a two-dimensional dataset of "time window-spatial grid". The spatial grid is divided into regular grid units for the building area. The mean, variance and other statistical characteristics of environmental parameters within the time window are statistically analyzed in each regular grid unit to form a dynamic environmental fusion layer. Historical risk characteristics and real-time evolution signals are integrated synchronously, aligned with the same time window and spatial grid, and risk indicators within each time window-spatial grid are statistically analyzed to form a risk evolution layer.

[0068] Using the spatiotemporal tensor decomposition method with time windows as the time dimension and spatial grids as the spatial dimension, the environmental dynamics fusion layer and the risk evolution layer are mapped one-to-one according to the "time window-spatial grid" index to extract time series features and spatial distribution features. A multi-head self-attention mechanism is adopted, using the statistical features of the environmental dynamics layer as queries and the corresponding risk indicators of the risk evolution layer as key-value pairs to construct an attention weight matrix to obtain cross-layer correlation features. The cross-layer correlation features are concatenated with the statistical features of the environmental dynamics fusion layer and input into a 3D convolutional layer with a kernel size of 3×3×3 for local feature aggregation. Then, tensor shrinkage is performed along the time and spatial dimensions to achieve high-order feature fusion across time and space, and finally output the constructed risk tensor integral model.

[0069] Next, the constructed risk tensor integral model is trained. First, the environment-risk joint dataset, which integrates the statistical features of the environment dynamic fusion layer and the risk indicators of the risk evolution layer, is divided into a sample set, a training set, and a validation set in a 7:2:1 ratio. The training set is Z-score normalized, and the data distribution is balanced by K-fold cross-validation to generate standardized training samples. The AdamW optimizer is used to backpropagate the risk tensor integral model to minimize the smoothing L1 loss function. After each round of training, the MAE (mean absolute error) is obtained through the validation set. Training is terminated when the MAE decreases to less than 0.01 for 5 consecutive rounds. Finally, the parameters of the trained risk tensor integral model are solidified using the torch.jit.trace parameter to output a deployable risk tensor integral model.

[0070] It should also be noted that historical risk characteristics mainly include the time, location, duration, severity, loss statistics, and corresponding fire response data of risk events such as past fires and equipment failures in the building. These are obtained through historical alarm records, accident investigation files, playback of surveillance videos, and manual record compilation.

[0071] The real-time evolution signal reflects the dynamic changes in the current risk status, covering real-time monitoring of smoke concentration, abnormal temperature values, fire lane blockage status, personnel location distribution, and fire protection facility operation status. It is collected in real time through IoT sensors, personnel positioning base stations, and fire protection facility status monitoring equipment deployed in the building.

[0072] Risk indicators include building structural risk, fire protection facility status, environmental dynamic risk, personnel safety risk, and historical risk. Among them, building structural risk and fire protection facility status are extracted from fire protection acceptance reports, environmental dynamic risk is obtained through real-time monitoring by IoT sensors, personnel safety risk is statistically analyzed after obtaining real-time location data using personnel positioning facilities such as Bluetooth beacons, and historical risk is obtained from fire accident databases or manually compiled accident files.

[0073] Spatiotemporal tensor decomposition is a mathematical method for analyzing multidimensional data with temporal and spatial correlations. Its core is to decompose the three-dimensional spatiotemporal tensor composed of "time-space-feature" into a combination of several low-rank subtensors. By mining the implicit correlations of each subtensor in the time, space and feature dimensions, it can efficiently extract and represent complex patterns in spatiotemporal data.

[0074] S2.2 Input multidimensional features into the risk tensor integral model, and the environment dynamic fusion layer fuses spatiotemporal features through spatiotemporal convolution to form dynamic risk field features;

[0075] It should be noted that after inputting the multidimensional features into the risk tensor integral model, the environment dynamic fusion layer uses a 3D convolution kernel to perform sliding window convolution on the input multidimensional features. The time dimension kernel captures the risk evolution trend over continuous time steps, and the spatial dimension kernel extracts the risk association pattern of adjacent grid cells. Then, a nonlinear transformation is performed through the ReLU activation function, and finally, dynamic risk field features are formed through feature aggregation.

[0076] S2.3 The risk evolution layer uses the Euler-Lagrange explicit integration method to perform continuous-time integration and generate the risk intensity evolution trajectory;

[0077] It should be noted that, firstly, the risk indicators of the risk evolution layer are used as inputs and aligned with the same time window and spatial grid to form a continuous risk input sequence; then, the average risk indicator of the most recent time window is extracted as the risk intensity value; subsequently, the risk intensity value of the next moment is adjusted and obtained through the Euler-Lagrange explicit integration method; finally, the risk intensity values ​​obtained from each time step based on the risk indicators are concatenated in chronological order to form a risk intensity evolution trajectory containing "timestamp-risk intensity".

[0078] It should also be noted that the steps for adjusting and obtaining the risk intensity value at the next moment using the Euler-Lagrange explicit integration method are as follows: First, extract the historical lag effect from the historical risk indicator records, and at the same time determine the current change based on the difference between the risk indicator and the previous moment's value; then integrate the risk intensity value with the historical lag effect and the current change, and obtain the risk intensity value at the next moment through integration.

[0079] S2.4. The dynamic risk field characteristics and risk intensity evolution trajectory are spatiotemporally aligned and integrated through the feature fusion channel to generate a risk heat map and graded early warning levels.

[0080] It should be noted that, based on the timestamp, the risk intensity evolution trajectory is spatiotemporally aligned with the dynamic risk field characteristics. For missing time steps in the risk intensity evolution trajectory, linear interpolation is used to supplement the corresponding risk intensity values ​​at those time points. For time steps in the dynamic risk field characteristics without corresponding trajectories, the nearest neighbor matching method is used to select risk field characteristics from adjacent time steps for alignment, ensuring a one-to-one correspondence between the risk intensity value and the spatial grid position at the same timestamp. Subsequently, based on the dynamic risk field characteristics, the local risk statistics of each spatial grid at each time step are obtained to generate spatial features. Based on the risk intensity evolution trajectory, the changes in risk intensity values ​​at adjacent time points are compared to obtain temporal features. Next, the extracted spatial and temporal features are merged in the data structure to form a three-dimensional feature tensor containing "timestamp-spatial grid-comprehensive risk" information. Min-Max normalization is then performed to generate a comprehensive risk value, which is mapped to a color gradient and filled with color according to the spatial grid position to generate a risk heatmap that intuitively displays the risk distribution. Finally, the comprehensive risk value is divided into warning levels according to a preset three-level risk threshold. For example, the first-level threshold range is set to [0.8, 1.0], the second-level threshold range is [0.5, 0.8], and the third-level threshold range is [0, 0.5]. When the comprehensive risk value is within the first-level threshold range, it is marked as a high-risk warning level; when it is within the second-level threshold range, it is marked as a medium-risk warning level; and when it is within the third-level threshold range, it is marked as a low-risk warning level. The divided risk warning levels are weighted and integrated to output a graded warning level.

[0081] It should also be noted that the preset three-level risk thresholds are defined based on the frequency of occurrence and severity of consequences of historical risk events.

[0082] S3. Spatial registration and interpolation calculations are used to generate evacuation guidance maps by matching the risk heat map with the graded early warning levels, and dynamic path planning is performed to obtain the optimal evacuation routes.

[0083] S3.1. Spatial registration of the risk heat map and the graded early warning level is performed by affine transformation to generate risk early warning overlay data. Kriging interpolation is used for interpolation calculation to fill the spatial gaps and generate an evacuation guidance map.

[0084] It should be noted that the spatial positions of the risk heat map and the graded warning level are adjusted by affine transformation to align them in the same spatial coordinate system. Then, for the spatial gaps that still exist after alignment, the Kriging interpolation method is used to estimate the comprehensive risk value at the gap by constructing a variogram function to fill the missing data areas. Finally, the interpolated complete risk heat map is superimposed with the graded warning level data to generate an evacuation guidance map that includes risk level labels and evacuation direction indicators.

[0085] S3.2 Based on the evacuation guidance map and personnel location data, a risk-weighted guidance map is generated using the dynamic weight assignment method, and the optimal evacuation route is output through iterative search using a multi-objective optimization algorithm.

[0086] It should be noted that the personnel location data is aligned with the spatial grid of the evacuation guidance map to determine the current spatial grid position of each person; then, weights are dynamically assigned based on the risk level of the evacuation guidance map to obtain the total weight of each potential evacuation route and generate a risk-weighted guidance map; then, a multi-objective optimization algorithm is used, with the total weight of the potential evacuation route, route length, and travel time as optimization objectives, to iteratively search in the risk-weighted guidance map, and balance the conflict between risk and efficiency by adjusting the nodes of the potential evacuation route, and finally output the optimal evacuation route.

[0087] S4. Based on the optimal evacuation route and evacuation guidance map, perform spatial coverage comparison to obtain the route coverage range, and generate route validity verification results through coverage efficiency matching.

[0088] S4.1 Based on the optimal evacuation route and evacuation guidance map, spatial overlay analysis is used to generate the coverage area boundary through spatial coverage comparison, and geometric calculation is used to calculate the coverage area to obtain the route coverage range.

[0089] It should be noted that the optimal evacuation route is spatially overlaid with the evacuation guidance map to ensure a one-to-one correspondence between the nodes of the optimal evacuation route and the grid positions of the evacuation guidance map. Then, a spatial coverage comparison is performed, traversing all nodes and line segments of the optimal evacuation route, and marking the grids passed through and the grids where the nodes are located on the evacuation guidance map to form a preliminary coverage area. Next, the convex hull algorithm in geometric measurement is used to generate a closed polygon boundary based on the grid boundary coordinates of the preliminary coverage area to determine the actual shape of the coverage area. Finally, the route coverage range is obtained based on the grid boundary coordinates using the grid counting method in geometric measurement.

[0090] S4.2. Based on building BIM data and evacuation guidance maps, the coverage efficiency threshold is calculated using a parameter derivation method. The expression is:

[0091]

[0092] Where η represents the coverage efficiency threshold, v represents the walking speed of personnel during emergency evacuation, t represents the safe allowable evacuation time, w represents the effective width of the evacuation passage, and S represents the total area of ​​the target area.

[0093] S4.3. Based on the route coverage area, the coverage ratio calculation method is used to obtain the coverage efficiency index, and the coverage efficiency is matched with the coverage efficiency threshold to generate the route validity verification result.

[0094] It should be noted that the coverage efficiency index is obtained based on the ratio of the route coverage area to the total evacuable area in the evacuation guidance map. Then, the coverage efficiency index is matched with the coverage efficiency threshold. If the coverage efficiency index is greater than or equal to the coverage efficiency threshold, it is determined to be "valid"; otherwise, it is "invalid". A route validity verification result containing "Route ID - Coverage Efficiency Index Value - Validity Conclusion" is generated.

[0095] S5. Based on the route validity verification results and the preprocessed multi-source data, risk correction factors are generated through risk correlation analysis, and a safety assessment report is obtained by using multi-criteria decision analysis.

[0096] S5.1. By performing spatial correlation analysis, the route validity verification results and preprocessed multi-source data are overlaid in a regional manner to extract the risk correlation characteristics of each region.

[0097] It should be noted that the route validity verification results are spatially overlaid with the preprocessed multi-source data, aligned based on a unified geographic coordinate system or regular grid index. Subsequently, spatial correlation analysis is performed to obtain the spatial covariance between the route coverage efficiency index and building structure risk, environmental dynamic risk, and population distribution risk within the region. The degree of dispersion is quantified based on the spatial covariance to obtain the spatial autocorrelation coefficient. Based on the value and magnitude of the spatial autocorrelation coefficient, the spatial correlation direction and clustering pattern between the route coverage efficiency index and building structure risk, environmental dynamic risk, and population distribution risk within the region are obtained, forming specific risk correlation characteristics such as "coverage efficiency-structural risk" and "population distribution-environmental risk".

[0098] S5.2. Spatial hotspot analysis is used to perform local spatial correlation strength analysis on the risk correlation characteristics of each region, and risk correction factors are generated.

[0099] It should be noted that spatial hotspot analysis is adopted. Based on the risk correlation characteristics of each region, the weighted sum of each region and its neighboring regions in terms of route coverage efficiency index and risk correlation characteristics is obtained to obtain the local spatial correlation strength of each region. The weights are then adjusted according to the local spatial correlation strength of each region to generate a risk correction factor.

[0100] S5.3 Based on the coverage efficiency index, graded early warning level, and population distribution risk characteristics, the weights of the indicators are obtained through the entropy method, and a multi-criteria evaluation indicator set is constructed by weighted integration.

[0101] It should be noted that entropy values ​​are obtained by assessing the dispersion of safety assessment indicators such as coverage efficiency index, graded early warning level, and population distribution risk characteristics. The difference coefficients of each safety assessment indicator are then obtained based on the entropy values. Subsequently, the difference coefficients of each safety assessment indicator are normalized to obtain the objective weights of coverage efficiency index, graded early warning level, and population distribution risk characteristics. Finally, a multi-criteria assessment indicator set containing coverage efficiency index, graded early warning level, and population distribution risk characteristics is generated by weighted summation.

[0102] It should also be noted that obtaining entropy values ​​by assessing the dispersion of safety assessment indicators such as coverage efficiency index, graded early warning level, and population distribution risk characteristics requires dividing the evacuation guidance map into assessment grid units using a regular grid. Then, the differences (dispersion) in the values ​​of each safety assessment indicator in different assessment grid units are statistically analyzed. Finally, the entropy value is calculated based on the magnitude of the differences: the greater the difference in the values ​​of the safety assessment indicators in each assessment grid unit, the smaller the entropy value; conversely, the smaller the difference, the larger the entropy value.

[0103] The difference coefficients of each safety assessment indicator are obtained based on the entropy value. The difference coefficients of each safety assessment indicator are calculated based on the entropy value of the coverage efficiency index, the graded early warning level, and the risk characteristics of population distribution. The smaller the entropy value, the larger the difference coefficient, and vice versa. The weaker the ability to distinguish is.

[0104] S5.4. Based on the multi-criteria assessment index set and risk correction factors, obtain a comprehensive safety score through multi-criteria decision analysis, classify the safety level based on the comprehensive safety score, and output a safety assessment report.

[0105] It should be noted that the indicator weights are calibrated by a risk correction factor to obtain the adjusted indicator weights; then, multi-criteria decision analysis is used to sum the indicator weights of each assessment unit with the adjusted indicator weights to obtain a comprehensive safety score. Next, the safety level is divided according to the comprehensive score. Finally, the comprehensive safety scores of each assessment unit are compiled, the causes of risks are analyzed, optimization suggestions are proposed, and a complete safety assessment report is formed.

[0106] This embodiment also provides an intelligent fire safety assessment system for buildings, including: a data acquisition module, a risk assessment module, a path planning module, an effective verification module, and a report generation module;

[0107] The data acquisition module is used to collect multi-source data and preprocess it, and extract multi-dimensional features based on the preprocessed multi-source data;

[0108] The risk assessment module is used to input multidimensional features into the risk tensor integral model, the environmental dynamic fusion layer performs spatiotemporal feature fusion, and the risk evolution layer performs continuous time integration to generate a risk heat map and graded early warning levels.

[0109] The route planning module is used to spatially register and interpolate the risk heat map with the graded warning level, generate the evacuation guidance map, and perform dynamic route planning to obtain the optimal evacuation route.

[0110] The effective verification module is used to perform spatial coverage comparison based on the optimal evacuation route and the evacuation guidance map, obtain the route coverage range, and generate route effectiveness verification results through coverage efficiency matching.

[0111] The report generation module is used to perform risk correlation analysis based on the route validity verification results and preprocessed multi-source data, generate risk correction factors, and obtain a safety assessment report through multi-criteria decision analysis.

[0112] This embodiment also provides a computer device applicable to the intelligent fire safety assessment method for buildings, including: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to realize the intelligent fire safety assessment method for buildings as proposed in the above embodiment.

[0113] The computer device can be a terminal, comprising a processor, memory, communication interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, carrier networks, NFC (Near Field Communication), or other technologies. The display screen can be an LCD screen or an e-ink screen. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad on the computer device's casing, or an external keyboard, touchpad, or mouse.

[0114] This embodiment also provides a storage medium storing a computer program, which, when executed by a processor, implements the intelligent fire safety assessment method for buildings as proposed in the above embodiments. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Red-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0115] In summary, this invention improves the real-time performance of risk assessment and enhances the accuracy of risk prediction by using a risk tensor integral model to fuse spatiotemporal features and perform continuous-time integration. Simultaneously, it employs a multi-objective optimization algorithm to iteratively search for the optimal evacuation route, making the evacuation plan more aligned with the actual scenario, improving the precision of emergency response, and achieving real-time adaptation of evacuation routes.

[0116] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for intelligent fire safety assessment of buildings, characterized in that: include, Collect multi-source data and preprocess it, then extract multi-dimensional features based on the preprocessed multi-source data; Multidimensional features are input into the risk tensor integral model, the environmental dynamic fusion layer performs spatiotemporal feature fusion, and the risk evolution layer performs continuous time integration to generate a risk heat map and graded early warning levels. Spatial registration and interpolation calculations are performed on the risk heat map and the graded early warning level to generate an evacuation guidance map and perform dynamic route planning to obtain the optimal evacuation route. Based on the optimal evacuation route and evacuation guidance map, spatial coverage comparison is performed to obtain the route coverage range, and route effectiveness verification results are generated through coverage efficiency matching. Based on the route validity verification results and preprocessed multi-source data, risk correlation analysis is performed to generate risk correction factors, and a safety assessment report is obtained through multi-criteria decision analysis.

2. The intelligent fire safety assessment method for buildings according to claim 1, characterized in that: The multi-source data includes building BIM data, IoT sensor data, and personnel positioning data; The preprocessing includes data cleaning, format conversion, deduplication, normalization, and outlier handling.

3. The intelligent fire safety assessment method for buildings according to claim 2, characterized in that: The specific steps for extracting multidimensional features are as follows: Based on building BIM data, structural risk characteristics of buildings are extracted through geometric and attribute information analysis. For IoT sensor data, analyze the difference between adjacent time points to extract dynamic environmental risk characteristics; Based on personnel location data, grid division and density clustering are performed to extract population distribution risk characteristics.

4. The intelligent fire safety assessment method for buildings according to claim 1, characterized in that: The specific steps for generating the risk heatmap and graded early warning levels are as follows. The spatiotemporal tensor decomposition method is used to perform multi-dimensional feature alignment and cross-layer attention connection on the environmental dynamic fusion layer and the risk evolution layer to construct a risk tensor integral model; Multidimensional features are input into the risk tensor integral model, and the environmental dynamic fusion layer fuses spatiotemporal features through spatiotemporal convolution to form dynamic risk field features. The risk evolution layer uses the Euler-Lagrange explicit integration method to perform continuous-time integration and generate the risk intensity evolution trajectory; The dynamic risk field characteristics and risk intensity evolution trajectory are spatiotemporally aligned and integrated through a feature fusion channel to generate a risk heat map and graded early warning levels.

5. The intelligent fire safety assessment method for buildings according to claim 1, characterized in that: The specific steps for generating the evacuation guidance map and obtaining the optimal evacuation route are as follows. Spatially register the risk heat map with the graded early warning level to generate risk early warning overlay data, and perform interpolation calculations to obtain the evacuation guidance map; Based on the evacuation guidance map and personnel location data, a risk-weighted guidance map is generated, and an iterative search is performed to output the optimal evacuation route.

6. The intelligent fire safety assessment method for buildings according to claim 1, characterized in that: The specific steps for generating the route validity verification result are as follows. The optimal evacuation route is compared with the evacuation guidance map to generate the coverage area boundary, and the coverage area is calculated to obtain the route coverage range. Based on building BIM data and evacuation guidance maps, the coverage efficiency threshold is calculated using a parameter derivation method. Based on the route coverage area, obtain the coverage efficiency index and match it with the coverage efficiency threshold to generate route validity verification results.

7. The intelligent fire safety assessment method for buildings according to claim 1, characterized in that: The specific steps for obtaining the security assessment report are as follows: Regional overlay analysis was performed on the route validity verification results and preprocessed multi-source data to extract risk correlation characteristics of each region. Perform local spatial correlation strength analysis on the risk correlation characteristics of each region to generate risk correction factors; Based on the coverage efficiency index, graded early warning levels, and population distribution risk characteristics, the indicator weights are obtained, and the indicator weights are weighted and summed to construct a multi-criteria evaluation indicator set. Based on a multi-criteria assessment indicator set and risk correction factors, a comprehensive safety score is obtained, and safety levels are classified according to the comprehensive safety score, and a safety assessment report is output.

8. An intelligent fire safety assessment system for buildings, based on the intelligent fire safety assessment method for buildings according to any one of claims 1 to 7, characterized in that: It includes a data acquisition module, a risk assessment module, a path planning module, an effective verification module, and a report generation module; The data acquisition module is used to collect multi-source data and preprocess it, and extract multi-dimensional features based on the preprocessed multi-source data; The risk assessment module is used to input multidimensional features into the risk tensor integral model, the environmental dynamic fusion layer performs spatiotemporal feature fusion, and the risk evolution layer performs continuous time integration to generate a risk heat map and graded early warning levels. The route planning module is used to spatially register and interpolate the risk heat map with the graded warning level, generate the evacuation guidance map, and perform dynamic route planning to obtain the optimal evacuation route. The effective verification module is used to perform spatial coverage comparison based on the optimal evacuation route and the evacuation guidance map, obtain the route coverage range, and generate route effectiveness verification results through coverage efficiency matching. The report generation module is used to perform risk correlation analysis based on the route validity verification results and preprocessed multi-source data, generate risk correction factors, and obtain a safety assessment report through multi-criteria decision analysis.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, it implements the steps of the intelligent fire safety assessment method for buildings according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by the processor, it implements the steps of the intelligent fire safety assessment method for buildings according to any one of claims 1 to 7.

Images