A method and system for detecting leaks in a building roof

Abstract

The application discloses a building roof leakage detection method and system, and relates to the technical field of monitoring and analysis, which comprises the following steps: guiding a drilling device to perform a punching operation based on a micropore path topology graph, collecting real-time deformation parameters of a surrounding area of a hole site and performing internal stress gradient analysis on a waterproof layer material to generate a hole site stress diffusion vector graph; performing space registration and fusion on a three-dimensional density distribution model of the waterproof layer and the hole site stress diffusion vector graph, calculating potential leakage probabilities corresponding to the waterproof layer around each hole site, performing time-space evolution simulation on the potential leakage probabilities, and outputting a dynamic leakage risk heat map; performing reverse tracking modeling of a leakage path in combination with the dynamic leakage risk heat map and the three-dimensional density distribution model of the waterproof layer to generate a roof leakage detection optimization model, and transmitting the roof leakage detection optimization model to a monitoring terminal to implement real-time leakage early warning. The application has the effect of improving leakage detection efficiency.

Description

Technical Field

[0001] This application relates to the field of monitoring and analysis technology, and in particular to a method and system for detecting roof leaks in buildings. Background Technology

[0002] As a crucial building envelope, the roof is constantly exposed to the natural environment and is prone to leaks due to factors such as aging waterproofing materials, construction defects, and external loads. This not only affects the building's functionality but can also lead to a chain of problems, including steel reinforcement corrosion and damage to interior finishes. Furthermore, the installation of equipment on the roof of new buildings (such as photovoltaic panels and air conditioning units) requires drilling into the waterproofing layer to secure the equipment. This drilling work can damage the integrity of the original waterproofing layer, exacerbating the risk of leaks. With the development of intelligent buildings, the need for proactive detection, dynamic risk assessment, and accurate early warning systems for roof leaks is becoming increasingly urgent.

[0003] In related technologies, traditional roof leak detection methods mainly include traditional visual inspection, infrared thermal imaging detection, and humidity sensor array monitoring. These methods lack proactive risk control during the construction phase (such as drilling holes in the waterproof layer), leading to increased leakage losses. Furthermore, the hole placement during drilling is often haphazard, failing to consider the structural characteristics of the waterproof layer. This can easily result in drilling in areas of stress concentration, exacerbating damage to the waterproof layer and causing low response efficiency. Consequently, the efficiency of roof leak detection is reduced, and there are areas for improvement. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this application provides a method and system for detecting roof leaks in buildings.

[0005] Firstly, this application provides a method for detecting roof leaks, comprising the following steps:

[0006] Step S1: Before drilling holes in the waterproof layer in the equipment installation area on the building roof, a multispectral sensor array is used to plan the pre-placement of holes, generate a micropore path topology map, and then the structure of the waterproof layer on the building roof is reconstructed by three-dimensional scanning based on the micropore path topology map to obtain a three-dimensional density distribution model of the waterproof layer.

[0007] Step S2: Guide the drilling equipment to perform drilling operations based on the micropore path topology map, and simultaneously collect real-time deformation parameters of the area around the hole. Analyze the internal stress gradient of the waterproof layer material based on the real-time deformation parameters to generate a stress diffusion vector map of the hole location.

[0008] Step S3: Spatial registration and fusion of the three-dimensional density distribution model of the waterproof layer and the stress diffusion vector map of the pore locations are performed to calculate the potential leakage probability of the waterproof layer around each pore location, and the spatiotemporal evolution simulation of the potential leakage probability is performed to output a dynamic leakage risk heat map.

[0009] Step S4: Leak path reverse tracking modeling is performed by combining the dynamic leakage risk thermal map and the waterproof layer three-dimensional density distribution model, a roof leak detection optimization model is generated, and the leak detection optimization model is transmitted to a monitoring terminal for real-time leak warning.

[0010] Preferably, the multispectral sensor array comprises an infrared thermal imaging unit, a millimeter wave radar unit, and a laser ranging unit.

[0011] The multispectral sensor array is arranged in a hexagonal honeycomb structure to form a redundant detection network.

[0012] The waterproof layer three-dimensional density distribution model comprises a dielectric constant distribution matrix of waterproof material and a thermal conductivity coefficient spatial gradient field.

[0013] Preferably, the step S1 specifically comprises the following steps:

[0014] The building roof corresponding bearing capacity parameters, waterproof layer thickness parameters and material aging parameters are extracted from the cloud database to construct a waterproof layer performance benchmark vector.

[0015] The equipment installation area is scanned by the multispectral sensor array in multiple modalities, and temperature gradient data, dielectric constant distribution data, and micro-deformation vibration data are synchronously collected.

[0016] A temperature conduction correlation model of waterproof layer aging state and temperature conduction rate is established based on the temperature gradient data, and a dielectric characteristic spatial distribution matrix is constructed in combination with the dielectric constant distribution data.

[0017] A waterproof layer structure fatigue damage degree surface model is constructed according to the micro-deformation vibration data, a density distribution feature tensor is generated by fusing the temperature conduction correlation model and the dielectric characteristic spatial distribution matrix.

[0018] The spatial resolution parameters in the density distribution feature tensor are optimized using an adaptive genetic algorithm to generate a micro-hole path topology graph.

[0019] The micro-hole path topology graph is calibrated in three-dimensional coordinates by the laser ranging unit, and a waterproof layer three-dimensional density distribution model is output based on the results of the three-dimensional coordinate calibration.

[0020] Preferably, the step S2 specifically comprises the following steps:

[0021] According to the micro-hole path topology graph, the drilling equipment is driven to perform drilling work, the real-time deformation parameters of the hole site surrounding area are synchronously collected by the pressure sensor and the micro-strain gauge, and the circumferential stress distribution data of the hole site are further confirmed.

[0022] The water content rate change inside the waterproof layer during drilling is measured to construct a water content rate gradient surface, the circumferential stress distribution data are fitted in a stress gradient field, and a two-dimensional stress diffusion map is generated.

[0023] mapping the two-dimensional stress diffusion map to the surface of the three-dimensional density distribution model of the waterproof layer, and fusing the water content gradient surface to construct a hole site stress diffusion vector diagram;

[0024] performing abnormal diffusion mode recognition on the hole site stress diffusion vector diagram to mark a dangerous diffusion area exceeding a preset stress threshold.

[0025] Preferably, the step S3 specifically comprises the following steps:

[0026] establishing a spatial coordinate conversion matrix of the three-dimensional density distribution model of the waterproof layer and the hole site stress diffusion vector diagram, and then performing spatial registration and fusion processing on the three-dimensional density distribution model of the waterproof layer and the hole site stress diffusion vector diagram based on the spatial coordinate conversion matrix;

[0027] extracting a corresponding dielectric constant distribution matrix and a thermal conductivity coefficient spatial gradient field of the waterproof material in the three-dimensional density distribution model of the waterproof layer to construct a multi-physical field coupling parameter space;

[0028] calculating a cosine similarity in the multi-physical field coupling parameter space to generate an initial leakage probability distribution matrix, performing feature importance analysis on the initial leakage probability distribution matrix, screening out key influencing factors to construct a leakage risk prediction model;

[0029] inputting historical leakage event data stored in the cloud database to train the leakage risk prediction model, performing spatiotemporal evolution iterative calculation of the leakage probability through a sliding time window method, performing stability correction on the result of the iterative calculation, and outputting a dynamic leakage risk thermodynamic map.

[0030] Preferably, the step S4 specifically comprises the following steps:

[0031] labeling a high-risk leakage area in the dynamic leakage risk thermodynamic map, extracting a density distribution feature tensor corresponding to the high-risk leakage area, and constructing a three-dimensional seepage diffusion model;

[0032] calculating a water flow velocity gradient field of different leakage paths according to the three-dimensional seepage diffusion model through a back propagation algorithm to generate a candidate water leakage path set;

[0033] simulating a leakage development process corresponding to each candidate water leakage path in the candidate water leakage path set to evaluate a leakage rate and an influence range of each candidate water leakage path;

[0034] establishing a water leakage propagation tree diagram based on the leakage rate and the influence range of each candidate water leakage path, constructing a roof water leakage detection optimization model according to the water leakage propagation tree diagram, setting multi-level early warning threshold parameters, and transmitting the water leakage detection optimization model to a monitoring terminal to implement real-time water leakage early warning.

[0035] In a second aspect, the application provides a building roof leakage detection system, comprising:

[0036] A dot planning module is configured to perform hole site pre-dot planning by using a multispectral sensor array before performing a waterproof layer punching operation in a device installation area of the building roof, to generate a micropore path topology map, and to perform three-dimensional scanning reconstruction on a waterproof layer structure of the building roof according to the micropore path topology map, so as to obtain a waterproof layer three-dimensional density distribution model;

[0037] A processing module is configured to guide a drilling device to perform a punching operation based on the micropore path topology map, to synchronously collect real-time deformation parameters of a hole site surrounding area, to perform waterproof layer material internal stress gradient analysis on the real-time deformation parameters, and to generate a hole site stress diffusion vector map;

[0038] An analysis module is configured to perform spatial registration and fusion on the waterproof layer three-dimensional density distribution model and the hole site stress diffusion vector map, to calculate a potential leakage probability of a waterproof layer corresponding to each hole site surrounding area, to perform space-time evolution simulation on the potential leakage probability, and to output a dynamic leakage risk heat map;

[0039] An optimization module is configured to perform leakage path reverse tracking modeling in combination with the dynamic leakage risk heat map and the waterproof layer three-dimensional density distribution model, to generate a roof leakage detection optimization model, and to transmit the roof leakage detection optimization model to a monitoring terminal to implement real-time leakage early warning.

[0040] In a third aspect, the application provides a computer readable storage medium, which stores instructions, when the instructions are run on a computer, the computer performs the building roof leakage detection method of any one of the above aspects.

[0041] In summary, the application has the following beneficial technical effects:

[0042] The application provides a building roof leakage detection method, which comprises the following steps: planning hole positions by a multi-spectral sensor array, generating a micro-hole path topology graph, reconstructing a waterproof layer structure of a building roof in three dimensions according to the micro-hole path topology graph, obtaining a waterproof layer three-dimensional density distribution model, guiding a drilling device to perform a drilling operation based on the micro-hole path topology graph, synchronously collecting real-time deformation parameters of a surrounding area of the hole position and performing internal stress gradient analysis on the waterproof layer material, generating a hole position stress diffusion vector graph, calculating a potential leakage probability of the waterproof layer corresponding to each hole position, simulating the potential leakage probability in space-time evolution, and outputting a dynamic leakage risk heat map; combining the dynamic leakage risk heat map and the waterproof layer three-dimensional density distribution model to perform reverse tracking modeling of a leakage path, generating a roof leakage detection optimization model, transmitting the roof leakage detection optimization model to a monitoring terminal to implement real-time leakage early warning, thereby improving the prospective risk prevention and control in the construction stage, reducing the situation of low response efficiency caused by blind hole position planning during the drilling operation and without considering the waterproof layer structure characteristics, and effectively improving the efficiency of building roof leakage detection. BRIEF DESCRIPTION OF DRAWINGS

[0043] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following will briefly introduce the drawings needed to be used in the description of the embodiments. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can also be obtained by those skilled in the art without creative labor.

[0044] Figure 1 It is a method flow chart of the building roof leakage detection of the embodiments of the present application.

[0045] Figure 2 It is a system schematic diagram of the building roof leakage detection of the embodiments of the present application. DETAILED DESCRIPTION

[0046] The following will be described in combination with the drawings. Figures 1-2 The present application will be further described in detail.

[0047] Embodiment 1

[0048] The embodiments of the present application disclose a building roof leakage detection method.

[0049] Referring to Figure 1 , a building roof leakage detection method comprises the following steps:

[0050] Step S1: before performing a waterproof layer drilling operation in a device installation area corresponding to a building roof, planning hole positions by a multi-spectral sensor array, generating a micro-hole path topology graph, and reconstructing a waterproof layer structure of a building roof in three dimensions according to the micro-hole path topology graph, obtaining a waterproof layer three-dimensional density distribution model;

[0051] Step S2: Guide the drilling equipment to perform drilling operations based on the micropore path topology map, and simultaneously collect real-time deformation parameters of the area around the hole. Analyze the internal stress gradient of the waterproof layer material based on the real-time deformation parameters to generate a stress diffusion vector map of the hole location.

[0052] Step S3: Spatial registration and fusion of the three-dimensional density distribution model of the waterproof layer and the stress diffusion vector map of the pore locations are performed to calculate the potential leakage probability of the waterproof layer around each pore location, and the spatiotemporal evolution simulation of the potential leakage probability is performed to output a dynamic leakage risk heat map.

[0053] Step S4: Combine the dynamic leakage risk heat map and the three-dimensional density distribution model of the waterproof layer to perform reverse tracing modeling of the leakage path, generate the roof leakage detection optimization model, and transmit the leakage detection optimization model to the monitoring terminal to implement real-time leakage early warning.

[0054] By adopting the above technical solution, a closed-loop management system for the entire process, from hole location planning to risk warning, is realized. Pre-planning reduces the damage to the waterproof layer caused by blind drilling, real-time stress analysis can promptly detect potential risks during the drilling process, dynamic risk heat map can intuitively present the spatiotemporal changes of leakage risk, and reverse tracking modeling improves the accuracy of leakage detection and the timeliness of warning. Compared with traditional post-detection methods, it can significantly reduce the losses caused by leakage. From data acquisition to model fusion to risk warning, a complete technical chain is formed, ensuring the systematic nature and accuracy of detection.

[0055] Furthermore, the multispectral sensor array includes an infrared thermal imaging unit, a millimeter-wave radar unit, and a laser ranging unit;

[0056] The multispectral sensor array is arranged in a hexagonal honeycomb structure to form a redundant detection network;

[0057] The three-dimensional density distribution model of the waterproof layer includes the dielectric constant distribution matrix of the waterproof material and the spatial gradient field of the thermal conductivity coefficient.

[0058] Specifically, the multispectral sensor array is composed of an infrared thermal imaging unit, a millimeter wave radar unit and a laser ranging unit, and each unit in the multispectral sensor array is arranged in a hexagonal honeycomb structure to form a redundant detection network. The infrared thermal imaging unit is used to collect temperature gradient data of the waterproof layer, the millimeter wave radar unit is used to obtain dielectric constant distribution data, and the laser ranging unit is used to realize three-dimensional coordinate calibration. The construction of the three-dimensional density distribution model of the waterproof layer is based on the data collected by the multispectral sensor array, the thermal conductivity coefficient spatial gradient field is obtained by processing the temperature data of the infrared thermal imaging unit, the dielectric constant distribution matrix of the waterproof material is constructed by using the detection results of the millimeter wave radar unit, and the three-dimensional coordinate data of the laser ranging unit is combined, the above parameters are integrated into the three-dimensional space to form the three-dimensional density distribution model of the waterproof layer containing the dielectric constant distribution matrix and the thermal conductivity coefficient spatial gradient field.

[0059] The multispectral sensor array in the above steps cooperates with multiple units, which can obtain waterproof layer information from multiple dimensions such as temperature, dielectric properties and spatial position. Compared with a single sensor, the detection dimension is more comprehensive and the data is more abundant. The redundant detection network of the hexagonal honeycomb structure can reduce the detection blind area, improve the reliability and redundancy of data acquisition, and even if some units fail, the continuity of overall detection can still be ensured. The dielectric constant and thermal conductivity coefficient in the three-dimensional density distribution model of the waterproof layer are key parameters reflecting the performance of the waterproof material. The dielectric constant can reflect the insulation and water content of the material, and the thermal conductivity coefficient can reflect the heat conduction performance and aging degree of the material. The combination of the two can more accurately depict the physical properties of the waterproof layer and provide reliable basic data for subsequent leakage risk assessment.

[0060] In the process of constructing the three-dimensional density distribution model of the waterproof layer, first, the size and arrangement density of the multispectral sensor array are determined according to the size and precision requirements of the equipment installation area, to ensure that the hexagonal honeycomb structure can achieve full coverage; then, each sensor unit is calibrated to ensure that the temperature measurement accuracy of the infrared thermal imaging unit, the dielectric constant detection accuracy of the millimeter wave radar unit, and the distance measurement error of the laser ranging unit are within the allowable range; next, each unit is started synchronously to collect data, the infrared thermal imaging unit generates a temperature gradient map by capturing the infrared radiation intensity of different areas, the millimeter wave radar unit calculates the spatial distribution of the dielectric constant by emitting millimeter waves and receiving reflected signals, and the laser ranging unit calculates the distance from the surface of the waterproof layer through the round-trip time of the laser pulse to obtain three-dimensional coordinate data; then, the collected temperature gradient data is processed, the spatial gradient field of the thermal conductivity coefficient is calculated by establishing a thermal conduction model, and the dielectric constant distribution matrix is constructed by interpolating and correcting the dielectric constant data; finally, the spatial gradient field of the thermal conductivity coefficient and the dielectric constant distribution matrix are spatially matched based on the three-dimensional coordinates obtained by the laser ranging, and integrated into a three-dimensional density distribution model of the waterproof layer. The three-dimensional density distribution model of the waterproof layer can comprehensively reflect the material properties and structural characteristics of the waterproof layer, providing accurate basic data support for subsequent hole planning and leakage detection.

[0061] It should be noted that the step S1 specifically comprises the following steps:

[0062] Extract the bearing capacity parameters, waterproof layer thickness parameters and material aging parameters corresponding to the building roof from the cloud database to construct a waterproof layer performance benchmark vector;

[0063] Multi-modal scanning of the equipment installation area is performed by a multispectral sensor array, and temperature gradient data, dielectric constant distribution data and micro-deformation vibration data are collected synchronously;

[0064] A temperature conduction correlation model of the aging state of the waterproof layer and the temperature conduction rate is established based on the temperature gradient data, and a dielectric characteristic spatial distribution matrix is constructed in combination with the dielectric constant distribution data;

[0065] A waterproof layer structure fatigue damage degree surface model is constructed according to the micro-deformation vibration data, a density distribution feature tensor is generated by fusing the temperature conduction correlation model and the dielectric characteristic spatial distribution matrix;

[0066] An adaptive genetic algorithm is used to optimize the spatial resolution parameters in the density distribution feature tensor to generate a micro-hole path topology graph;

[0067] The micro-hole path topology graph is calibrated by a laser ranging unit in three-dimensional coordinates, and a three-dimensional density distribution model of the waterproof layer is output based on the results of the three-dimensional coordinate calibration.

[0068] Specifically, first, the bearing capacity parameters, waterproof layer thickness parameters and material aging parameters of the building roof are extracted from the cloud database, and the above parameters are quantified to construct a waterproof layer performance benchmark vector as a reference benchmark for subsequent analysis; then the equipment installation area is scanned by a multi-spectral sensor array, and temperature gradient data, dielectric constant distribution data and micro-deformation vibration data are synchronously collected, wherein the temperature gradient data reflects the heat conduction difference of the waterproof layer, the dielectric constant distribution data reflects the insulation and water content characteristics of the material, and the micro-deformation vibration data reflects the stability of the structure; based on the temperature gradient data, the temperature conduction correlation model of the waterproof layer aging state and the temperature conduction rate is established by analyzing the correlation between the temperature conduction rate and the material aging degree in different regions, and the dielectric characteristic spatial distribution matrix is constructed by spatial interpolation and data fitting combined with the dielectric constant distribution data; according to the micro-deformation vibration data, the finite element analysis method is used to construct a waterproof layer structure fatigue damage degree surface model, which can reflect the fatigue degree of the structure at different positions; the temperature conduction correlation model, the dielectric characteristic spatial distribution matrix and the fatigue damage degree surface model are data fused to generate a density distribution feature tensor containing multi-dimensional features; the adaptive genetic algorithm is used to optimize the spatial resolution parameters in the density distribution feature tensor, and the optimal resolution is found through selection, crossover, mutation and other operations to generate a micro-hole path topology graph that can accurately reflect the hole distribution; finally, the three-dimensional coordinate calibration of the micro-hole path topology graph is performed by the laser ranging unit, and the waterproof layer three-dimensional density distribution model containing spatial position information is output based on the calibration result.

[0069] By combining the historical data with the real-time detection data through the above steps, the reference parameters extracted by the cloud database provide a historical reference for the detection, avoiding the deviation that may be caused by relying only on real-time data; the multi-dimensional data collected by multi-modal scanning can comprehensively reflect the performance of the waterproof layer, and the cooperative analysis of temperature, dielectric and deformation data can more accurately depict the state of the waterproof layer than single data; the fusion of the temperature conduction correlation model, the dielectric characteristic matrix and the fatigue damage surface model forms a more comprehensive density distribution feature tensor, providing multi-factor support for micro-hole path planning; the application of the adaptive genetic algorithm improves the efficiency and accuracy of spatial resolution optimization, making the micro-hole path topology graph more in line with actual needs and reducing unnecessary damage to the waterproof layer caused by drilling; the three-dimensional coordinate calibration of the laser ranging ensures the spatial accuracy of the three-dimensional density distribution model, providing a precise spatial reference for subsequent drilling operations and risk assessment.

[0070] In constructing the waterproof layer performance benchmark vector, the parameters in the cloud database need to be screened and standardized, abnormal values are removed and parameters of different dimensions are converted to a uniform order of magnitude. For example, the bearing capacity parameter is converted to the bearing value per unit area, the thickness parameter is converted to the millimeter level value, and the aging parameter is quantified by aging period and performance decay rate. Then the above parameters are integrated into the benchmark vector by vector synthesis method. In establishing the temperature conduction correlation model, the temperature gradient data are first smoothed to remove noise, and then the temperature conduction rate of different aging degree areas is compared, and the functional relationship between aging state and conduction rate is fitted by regression analysis method to form the temperature conduction correlation model. In constructing the dielectric property spatial distribution matrix, the dielectric constant data collected by the millimeter wave radar are gridded, the missing data are supplemented by Kriging interpolation method, and then the data are organized into the dielectric property spatial distribution matrix by matrix operation. In constructing the waterproof layer structure fatigue damage degree surface model, the micro-deformation vibration data are converted to strain values, combined with the fatigue limit of the material, and simulated by finite element software to generate fatigue damage degrees at different positions, and then the fatigue damage degree surface model of the waterproof layer structure is formed by surface fitting. In generating the density distribution feature tensor by fusing the above three models, the data of each model need to be converted to a unified data format and spatial grid, and the data fusion is realized by tensor product operation to retain the feature information of each dimension. In optimizing the spatial resolution parameter, the fitness function of the adaptive genetic algorithm is set as the comprehensive index of resolution and data error, the crossover probability and mutation probability are dynamically adjusted to improve the convergence speed and optimization accuracy of the algorithm, and it is ensured that the generated micro-hole path topology map can accurately reflect the hole distribution and reduce data redundancy. Finally, in the three-dimensional coordinate calibration, the three-dimensional coordinate data collected by the laser ranging unit are matched with the two-dimensional coordinates of the micro-hole path topology map, the two-dimensional topology map is mapped to the three-dimensional space through the coordinate conversion matrix, and then the spatial parameters of the three-dimensional density distribution model of the waterproof layer are corrected to ensure that the spatial position of the model is consistent with the actual roof structure, and accurate path guidance is provided for subsequent drilling operations.

[0071] It should be noted that the step S2 specifically comprises the following steps:

[0072] According to the micro-hole path topology map, the drilling equipment is driven to perform drilling operation, the real-time deformation parameters of the hole position surrounding area are collected by synchronously activating the pressure sensor and the micro-strain gauge, and then the hole position circumferential stress distribution data are confirmed;

[0073] The water content rate change in the waterproof layer corresponding to the drilling process is measured, the water content rate gradient surface is constructed, the stress gradient field fitting of the circumferential stress distribution data is performed, and the two-dimensional stress diffusion map is generated;

[0074] The two-dimensional stress diffusion map is mapped to the surface of the three-dimensional density distribution model of the waterproof layer, and the water content gradient surface is fused to construct a hole site stress diffusion vector diagram.

[0075] An abnormal diffusion mode recognition is performed on the hole site stress diffusion vector diagram, and a dangerous diffusion area exceeding a preset stress threshold is marked.

[0076] Specifically, first, according to the hole site coordinates and path planning in the micro-hole path topological graph, the drilling equipment is driven to perform the drilling operation according to the preset trajectory. In the drilling process, the pressure sensor mounted on the drilling equipment and the micro-strain gauge around the hole site are activated synchronously, and the real-time deformation parameters of the surrounding area of the hole site are collected in real time, including displacement, strain value, etc. Through analysis and calculation of the above parameters, the hole site circumferential stress distribution data are confirmed, which reflect the stress change of the surrounding structure in the drilling process. At the same time, the water content change in the waterproof layer corresponding to the drilling process is measured, the water content data at different depths are collected through the humidity sensor pre-buried near the drilling, and the water content gradient surface is constructed by using the spatial interpolation method. The water content gradient surface reflects the distribution difference of the water content in space. Based on the hole site circumferential stress distribution data, a stress gradient field fitting is performed by using a stress field fitting algorithm to generate a two-dimensional stress diffusion map reflecting the stress diffusion direction and intensity. The two-dimensional stress diffusion map is mapped to the surface of the three-dimensional density distribution model of the waterproof layer through coordinate conversion, so that the stress data are associated with the three-dimensional spatial position, and then the information of the water content gradient surface is fused, and a hole site stress diffusion vector diagram containing the stress diffusion direction, intensity and water content distribution is constructed through data superposition and vector synthesis. Finally, an abnormal diffusion mode recognition is performed on the hole site stress diffusion vector diagram by using an abnormal detection algorithm, a preset stress threshold is set, and the area exceeding the preset stress threshold is marked as a dangerous diffusion area, prompting the possible structural damage risk.

[0077] Through the above steps, the real-time monitoring of the drilling process is realized. Through the real-time data collection of the pressure sensor and the micro-strain gauge, the influence of drilling on the surrounding structure can be grasped in time, and the structural damage caused by blind drilling is avoided. The measurement of the water content change and the construction of the gradient surface combine stress analysis with the water content state of the material. Because the high water content may exacerbate the damage of stress to the waterproof layer, the fusion of the two makes the risk assessment more comprehensive. The conversion of the two-dimensional stress diffusion map to the three-dimensional vector diagram realizes the spatialization of the stress data, which is more in line with the three-dimensional characteristics of the actual structure. The abnormal diffusion mode recognition can quickly locate the dangerous area, provide timely warning for the operator, and facilitate the adjustment of drilling parameters or the suspension of operation to reduce the potential leakage risk.

[0078] In collecting real-time deformation parameters, the pressure sensor and the micro-strain gauge need to be calibrated to ensure the accuracy of the measurement data. The pressure sensor collects the pressure changes during drilling, and the micro-strain gauge collects the strain response of the surrounding material. Through the data acquisition system, the above parameters are transmitted to the processing terminal in real time. After filtering and amplification processing, the circumferential stress distribution data of the hole site is calculated. When constructing the moisture content gradient surface, the moisture content data collected by the humidity sensor need to be temperature compensated to eliminate the influence of temperature on the measurement results. Then, the inverse distance weighted interpolation method is used to perform spatial interpolation on the discrete moisture content data to generate a continuous moisture content gradient surface. The slope of the surface reflects the rate of change of moisture content. When fitting the stress gradient field, based on the circumferential stress distribution data of the hole site, the least squares method is used to fit the stress distribution function in the radial and circumferential directions to generate a two-dimensional stress diffusion map. The color or grayscale in the map represents the stress magnitude, and the arrow represents the diffusion direction. When constructing the hole site stress diffusion vector map, first, establish the coordinate correspondence between the two-dimensional stress diffusion map and the three-dimensional density distribution model of the waterproof layer. Through affine transformation, the two-dimensional map is mapped to the surface of the three-dimensional model. Then, the information of the moisture content gradient surface is used as a weight factor to correct the stress vectors in the three-dimensional space, so that the size and direction of the stress vectors reflect the influence of stress and moisture content at the same time, forming a vector map containing multiple factors. When identifying abnormal diffusion patterns, the stress threshold is preset according to the compressive strength and safety factor of the waterproof layer material. The clustering algorithm is used to divide the data in the stress diffusion vector map into normal and abnormal regions. The dangerous diffusion area exceeding the threshold is highlighted to provide a clear basis for subsequent drilling adjustment and ensure the safety of the waterproof layer structure during drilling.

[0079] It should be noted that the step S3 specifically comprises the following steps:

[0080] A spatial coordinate conversion matrix is established between the three-dimensional density distribution model of the waterproof layer and the hole site stress diffusion vector map, and then the three-dimensional density distribution model of the waterproof layer and the hole site stress diffusion vector map are spatially registered and fused based on the spatial coordinate conversion matrix;

[0081] The corresponding dielectric constant distribution matrix and thermal conductivity coefficient spatial gradient field of the waterproof material in the three-dimensional density distribution model of the waterproof layer are extracted to construct a multi-physical field coupling parameter space.

[0082] The cosine similarity is calculated in the multi-physical field coupling parameter space to generate an initial leakage probability distribution matrix. The initial leakage probability distribution matrix is analyzed for feature importance, and the key influencing factors are selected to construct a leakage risk prediction model.

[0083] The historical leakage event data stored in the cloud database is input into the leakage risk prediction model, the spatio-temporal evolution iterative calculation of the leakage probability is performed through the sliding time window method, the stability of the result of the iterative calculation is corrected, and the dynamic leakage risk heat map is output.

[0084] Specifically, first, a spatial coordinate conversion matrix of the waterproof layer three-dimensional density distribution model and the hole site stress diffusion vector diagram is established, the coordinates of a plurality of corresponding feature points in the waterproof layer three-dimensional density distribution model and the hole site stress diffusion vector diagram are collected, the conversion parameters are calculated by using the least square method, and then the waterproof layer three-dimensional density distribution model and the hole site stress diffusion vector diagram are spatially registered and fused based on the matrix, so that the data is aligned in the same spatial coordinate system; the dielectric constant distribution matrix and the thermal conductivity coefficient spatial gradient field of the waterproof material corresponding to the waterproof layer three-dimensional density distribution model are extracted from the fused model, and the above parameters are taken as the core variables of the multi-physical field to construct a multi-physical field coupling parameter space, which contains the electrical and thermal characteristics of the material; in the multi-physical field coupling parameter space, the cosine similarity of the dielectric constant and the thermal conductivity coefficient at different positions with the historical leakage point parameters is calculated, and an initial leakage probability distribution matrix is generated by the similarity size, each element in the initial leakage probability distribution matrix represents the initial leakage probability of the corresponding position; the initial leakage probability distribution matrix is analyzed for feature importance, the feature importance score of each parameter is calculated by using the random forest algorithm, and the key influencing factors with greater influence on the leakage probability, such as high water cut area and low dielectric constant area, are screened out, and a leakage risk prediction model is constructed based on the above factors; the historical leakage event data, including the leakage position, occurrence time, and environmental conditions, are extracted from the cloud database, and the above data is input into the leakage risk prediction model as a training sample, the model parameters are adjusted for training, and the prediction accuracy of the model is improved; the spatio-temporal evolution iterative calculation of the leakage probability is performed by using the sliding time window method, a fixed time window size is set, the real-time monitoring data is updated in each window, the leakage probability is recalculated, and the probability distribution is more in line with the actual change through multiple iterations; the result of the iterative calculation is corrected for stability, the abnormal fluctuation value is removed, the probability data is smoothed by using the exponential smoothing method, and finally the dynamic leakage risk heat map that can dynamically reflect the change of the leakage risk is output.

[0085] The consistency of different model data is ensured by the spatial registration fusion in the above steps, and the analysis error caused by coordinate difference is avoided; the key physical properties of the material are integrated in the multi-physical field coupling parameter space, compared with single physical field analysis, which can more comprehensively reflect the potential conditions of leakage occurrence; the cosine similarity calculation and feature importance analysis improve the accuracy of the initial leakage probability calculation, and the model focuses more on the key influencing factors; the generalization ability of the leakage risk prediction model is enhanced by the historical data training, which can better adapt to different roof environments; the sliding time window method realizes the dynamic update of the leakage probability, making the risk assessment time-effective; the stability correction ensures the reliability of the dynamic leakage risk thermal map, providing an intuitive and accurate risk reference for subsequent water leakage detection and early warning.

[0086] In the establishment of the space coordinate conversion matrix, first, at least three non-collinear feature points are selected in the waterproof layer three-dimensional density distribution model and the hole site stress diffusion vector diagram, such as the hole site center, the waterproof layer edge point and the like, the coordinate values of the above-mentioned points in the waterproof layer three-dimensional density distribution model and the hole site stress diffusion vector diagram are obtained, then the least square method is used to solve the parameters of the conversion matrix, and the deviation of the converted model data in the space position is ensured to be less than a preset threshold; when constructing the multi-physical field coupling parameter space, the dielectric constant distribution matrix and the thermal conductivity coefficient space gradient field are converted into three-dimensional grid data, each grid element contains two parameters of dielectric constant and thermal conductivity coefficient, the two parameters are integrated into a coupling parameter by defining the weight coefficient of the parameter, and the multi-physical field coupling parameter space is formed; when calculating the initial leakage probability distribution matrix, first, the dielectric constant and thermal conductivity coefficient data of the historical leakage points are collected as standard samples, then the cosine similarity of each grid element in the multi-physical field coupling parameter space and the standard sample is calculated, the higher the similarity, the greater the initial leakage probability, and the similarity value is normalized to form an initial probability matrix; when constructing the leakage risk prediction model, the initial leakage probability is taken as the output variable, the dielectric constant, the thermal conductivity coefficient, the water content and the like are taken as the input variables, the neural network model is taken as the basic framework, the input layer receives the characteristic factors, the hidden layer carries out nonlinear mapping, and the output layer outputs the predicted leakage probability; when training the model, the historical leakage event data is divided into a training set and a test set according to a ratio of 7:3, the weight and bias of the model are adjusted through the training set, the prediction accuracy of the model is verified by using the test set, and the cross-validation method is used to avoid overfitting; in the sliding time window method, the size of the time window is determined according to the leakage development speed of the waterproof layer, and is generally set to 1-2 hours, in each window, the newly collected temperature, humidity, stress and the like are input into the model, the leakage probability is recalculated, and dynamic updating is realized; when correcting the stability, the 3σ criterion is used to identify and eliminate abnormal values, then the exponential smoothing method is used to process the probability data, the smoothing coefficient is determined according to the volatility of the data, the thermal map can reflect the risk change trend and avoid the interference brought by short-term fluctuations, and finally the generated dynamic leakage risk thermal map can clearly show the leakage risk level at different times and different positions, thereby providing accurate decision support for roof leakage detection.

[0087] It should be noted that the step S4 specifically comprises the following steps:

[0088] In the dynamic leakage risk thermal map, the high-risk leakage area is marked, the density distribution feature tensor corresponding to the high-risk leakage area is extracted, and a three-dimensional seepage diffusion model is constructed;

[0089] According to the three-dimensional seepage diffusion model, the water flow velocity gradient field of different leakage paths is calculated through the back propagation algorithm, and a candidate leakage path set is generated;

[0090] simulate a leakage development process corresponding to each candidate leakage path in the candidate leakage path set, and evaluate a leakage rate and an influence range of each candidate leakage path;

[0091] Based on the leakage rate and the influence range of each candidate leakage path, a leakage propagation tree diagram is established, a roof leakage detection optimization model is constructed according to the leakage propagation tree diagram, multi-level early warning threshold parameters are set, and the roof leakage detection optimization model is transmitted to a monitoring terminal to implement real-time leakage early warning.

[0092] Specifically, first, in the dynamic leakage risk thermal map, a high-risk leakage area is marked according to a preset high-risk threshold, the high-risk leakage area is a key focus of potential water leakage, a density distribution feature tensor corresponding to the area is extracted, including dielectric constant, thermal conductivity coefficient, structure density and other parameters, a three-dimensional seepage diffusion model is constructed based on the above parameters and combined with the principle of fluid mechanics, which can simulate the penetration process of water flow in the waterproof layer; according to the three-dimensional seepage diffusion model, the possible water leakage source is deduced from the high-risk leakage area by using the back propagation algorithm, the water flow velocity gradient field on different leakage paths is calculated, the size of the velocity gradient reflects the speed of water flow on different paths, and a candidate leakage path set containing multiple possible paths is generated based on this; the leakage development process of each candidate leakage path in the candidate leakage path set is simulated, different initial conditions (such as initial water quantity, water pressure, etc.) are set, and the leakage state of each path at different time points is calculated by using finite element simulation software, the leakage rate (water leakage quantity per unit time) and the influence range (size of the area affected by leakage) of each candidate leakage path are evaluated; based on the leakage rate and the influence range of each candidate leakage path, a tree structure modeling method is used to establish a leakage propagation tree diagram, the root of the tree is the possible water leakage source, the branches are different propagation paths, and the leaves are the influence range, the propagation relationship of water leakage is clearly displayed through the tree diagram; according to the leakage propagation tree diagram, a roof leakage detection optimization model is constructed in combination with the evaluation results of the leakage rate and the influence range, multi-level early warning threshold parameters are set in the roof leakage detection optimization model, such as a first-level early warning corresponding to a lower leakage rate and a smaller influence range, a second-level early warning corresponding to a medium risk, and a third-level early warning corresponding to a high risk; the roof leakage detection optimization model is transmitted to a monitoring terminal, the terminal receives the risk evaluation results output by the model in real time, when the risk of a certain area reaches the corresponding early warning threshold, the corresponding early warning mechanism is triggered immediately, such as sound and light alarm, short message notification, etc., to implement real-time leakage early warning.

[0093] The precise labeling of high-risk areas by the above steps makes the water leakage detection more targeted, avoiding the inefficiency of comprehensive investigation; the application of the three-dimensional seepage diffusion model and the back propagation algorithm improves the accuracy of water leakage path tracing, which can quickly locate the potential water leakage source; the simulation and evaluation of the leakage development process provide a quantitative basis for risk level classification, making the early warning more scientific; the water leakage propagation tree diagram intuitively displays the propagation relationship of water leakage, which is convenient for operators to understand the development trend of water leakage; the setting of multi-level early warning thresholds realizes differentiated early warning, which can take corresponding measures according to the risk level, improving the practicality and response efficiency of early warning; the real-time early warning of the monitoring terminal can timely discover problems in the early stage of water leakage, reducing property loss and structural damage caused by water leakage.

[0094] In constructing the three-dimensional seepage diffusion model, first, the boundary conditions (such as the upper and lower surfaces of the waterproof layer) and the initial conditions (such as the initial water content) of the model are determined according to the density distribution tensor of the high-risk leakage area, Darcy's law is used as the basic equation of seepage motion, combined with the permeability parameters of the waterproof layer material, the three-dimensional area is divided into grids through finite element discretization, and the numerical model of the seepage control equation is established; in calculating the water flow velocity gradient field, the back propagation algorithm takes the high-risk area as the target output, adjusts the permeability coefficients and other parameters of each path, so that the water flow velocity distribution output by the model matches the actual monitoring speed data, and then the water flow velocity gradient field of each path is obtained; in generating the candidate water leakage path set, according to the direction and size of the water flow velocity gradient field, the paths with larger water flow velocity gradient are selected as candidate paths, and a clustering algorithm is used to remove repeated or similar paths, ensuring the diversity and representativeness of the set; in simulating the leakage development process, the computational fluid dynamics software is used to numerically simulate each candidate water leakage path, the time step is set, and the change of water leakage amount and the expansion of the affected area are calculated in each time step, and the leakage state at different times is obtained through iterative calculation; in establishing the water leakage propagation tree diagram, the potential water leakage source is taken as the root node, the branches are constructed according to the connection relationship and propagation order of the paths, each branch is labeled with the corresponding leakage rate and influence range parameters, and the propagation priority of water leakage is reflected through the hierarchical structure of the tree; in constructing the roof water leakage detection optimization model, the parameters of the water leakage propagation tree diagram are integrated into the input variables of the model, a fuzzy control algorithm is used to set multi-level early warning thresholds, the thresholds are determined based on the leakage rate and influence range corresponding to different risk levels in historical data, and the risk assessment results are mapped to the corresponding early warning levels through fuzzy rules; after the roof water leakage detection optimization model is transmitted to the monitoring terminal, the terminal receives the data of the field sensors in real time through the interface, inputs the roof water leakage detection optimization model for calculation, and when the result exceeds a certain level of early warning threshold, the preset early warning program is started immediately to ensure that relevant personnel can take maintenance measures in time and minimize water leakage loss.

[0095] Example 2

[0096] The embodiment of the application also discloses a building roof leakage detection system.

[0097] With reference to Figure 2 The building roof leakage detection system comprises the following modules.

[0098] A dot planning module is configured to, before performing a waterproof layer punching operation in a device installation area of a building roof, perform hole site pre-dot planning by using a multispectral sensor array, generate a micropore path topology graph, and perform three-dimensional scanning reconstruction on a waterproof layer structure of the building roof according to the micropore path topology graph, so as to obtain a waterproof layer three-dimensional density distribution model.

[0099] A processing module is configured to guide a drilling device to perform a punching operation based on the micropore path topology graph, synchronously collect real-time deformation parameters of a hole site surrounding area, perform waterproof layer material internal stress gradient analysis on the real-time deformation parameters, and generate a hole site stress diffusion vector graph.

[0100] An analysis module is configured to perform spatial registration and fusion on the waterproof layer three-dimensional density distribution model and the hole site stress diffusion vector graph, calculate potential leakage probabilities of a waterproof layer corresponding to each hole site surrounding area, perform time-space evolution simulation on the potential leakage probabilities, and output a dynamic leakage risk heat map.

[0101] An optimization module is configured to perform leakage path reverse tracking modeling in combination with the dynamic leakage risk heat map and the waterproof layer three-dimensional density distribution model, generate a roof leakage detection optimization model, and transmit the roof leakage detection optimization model to a monitoring terminal to implement real-time leakage early warning.

[0102] The above content is merely an example and description of the concept of the application, and those skilled in the art can make various modifications or supplements or use similar ways to replace the described specific embodiments, as long as the concept of the application is not deviated, which should belong to the protection scope of the application.

[0103] In the description of the present specification, the description of the terms "one embodiment", "example", "specific example" and the like means that the specific features, structures, materials or characteristics described in combination with the embodiment or example are contained in at least one embodiment or example of the present application. In the present specification, the illustrative description of the above terms does not necessarily refer to the same embodiment or example. Moreover, the described specific features, structures, materials or characteristics can be combined in any one or more embodiments or examples in a suitable manner.

[0104] The preferred embodiments of the application disclosed above are only to facilitate the elucidation of the application. The preferred embodiments do not describe all the details of the application and limit the application to the specific embodiments. Obviously, many modifications and variations can be made in light of the teachings above. The description is chosen and described in order to provide the best illustration of the application principles and their practical application, so that those skilled in the art can well understand and utilize the application.

Claims

1. A method of detecting leaks in a building roof, characterized by, The method comprises the following steps: Step S1: Before the waterproof layer punching operation is performed on the equipment installation area corresponding to the building roof, the hole position pre-dispensing plan is performed by the multispectral sensor array, the micro-hole path topology graph is generated, and the waterproof layer structure of the building roof is reconstructed by three-dimensional scanning according to the micro-hole path topology graph, so as to obtain the three-dimensional density distribution model of the waterproof layer; Step S2: Based on the micro-hole path topology graph, the drilling equipment is guided to perform the punching operation, the real-time deformation parameters of the surrounding area of the hole position are synchronously collected, the real-time deformation parameters are analyzed for the internal stress gradient of the waterproof layer material, and the hole position stress diffusion vector graph is generated; Step S3: The three-dimensional density distribution model of the waterproof layer is fused with the hole position stress diffusion vector graph in space, the potential leakage probability of the surrounding waterproof layer corresponding to each hole position is calculated, the potential leakage probability is simulated for time-space evolution, and a dynamic leakage risk heat map is output; Step S4: The dynamic leakage risk heat map and the three-dimensional density distribution model of the waterproof layer are combined to perform reverse tracking modeling of the leakage path, a roof leakage detection optimization model is generated, and the roof leakage detection optimization model is transmitted to a monitoring terminal to implement real-time leakage early warning; The multispectral sensor array comprises an infrared thermal imaging unit, a millimeter wave radar unit and a laser ranging unit; The multispectral sensor array is arranged in a hexagonal honeycomb structure to form a redundant detection network; The three-dimensional density distribution model of the waterproof layer comprises a waterproof material dielectric constant distribution matrix and a thermal conductivity coefficient spatial gradient field; Step S4 specifically comprises the following steps: In the dynamic leakage risk heat map, a high-risk leakage area is marked, a density distribution feature tensor corresponding to the high-risk leakage area is extracted, and a three-dimensional seepage diffusion model is constructed; According to the three-dimensional seepage diffusion model, the water flow velocity gradient field of different leakage paths is calculated by a back propagation algorithm to generate a candidate leakage path set; The development process of each candidate leakage path in the candidate leakage path set is simulated, and the leakage rate and influence range of each candidate leakage path are evaluated; Based on the leakage rate and influence range of each candidate leakage path, a leakage propagation tree diagram is established, a roof leakage detection optimization model is constructed according to the leakage propagation tree diagram, multi-level early warning threshold parameters are set, and the leakage detection optimization model is transmitted to a monitoring terminal to implement real-time leakage early warning.

2. The method of claim 1, wherein, Step S1 specifically comprises the following steps: The bearing capacity parameters, waterproof layer thickness parameters and material aging parameters corresponding to the building roof are extracted from a cloud database to construct a waterproof layer performance benchmark vector; The equipment installation area is scanned by the multispectral sensor array in multiple modes, and temperature gradient data, dielectric constant distribution data and micro-deformation vibration data are synchronously collected; A temperature conduction correlation model of the waterproof layer aging state and temperature conduction rate is established based on the temperature gradient data, and a dielectric characteristic spatial distribution matrix is constructed in combination with the dielectric constant distribution data; A waterproof layer structure fatigue damage degree surface model is constructed according to the micro-deformation vibration data, a density distribution feature tensor is generated by fusing the temperature conduction correlation model and the dielectric characteristic spatial distribution matrix; The spatial resolution parameters in the density distribution feature tensor are optimized by using an adaptive genetic algorithm to generate a micro-hole path topology graph. The micro-hole path topology is calibrated by a laser ranging unit, and a waterproof layer three-dimensional density distribution model is output based on the calibration result.

3. The method of claim 1, wherein, The step S2 specifically includes the following steps: According to the micro-hole path topology, a drilling device is driven to perform a drilling operation, and a pressure sensor and a micro-strain gauge are synchronously activated to collect real-time deformation parameters of a hole site surrounding area, and then hole site circumferential stress distribution data is confirmed; A waterproof layer internal corresponding water content rate change in a drilling process is measured, a water content rate gradient surface is constructed, a stress gradient field fitting is performed on the circumferential stress distribution data, and a two-dimensional stress diffusion map is generated; The two-dimensional stress diffusion map is mapped to a surface of the waterproof layer three-dimensional density distribution model, and a hole site stress diffusion vector diagram is constructed by fusing the water content rate gradient surface; An abnormal diffusion mode recognition is performed on the hole site stress diffusion vector diagram, and a dangerous diffusion area exceeding a preset stress threshold is marked.

4. The method of claim 1, wherein, The step S3 specifically includes the following steps: A space coordinate conversion matrix of the waterproof layer three-dimensional density distribution model and the hole site stress diffusion vector diagram is established, and then the space coordinate conversion matrix is used for space registration fusion processing of the waterproof layer three-dimensional density distribution model and the hole site stress diffusion vector diagram; A waterproof material dielectric constant distribution matrix and a thermal conductivity coefficient space gradient field corresponding to the waterproof layer three-dimensional density distribution model are extracted, and a multi-physical field coupling parameter space is constructed; Cosine similarities of dielectric constants and thermal conductivity coefficients at different positions and historical leakage point parameters are calculated in the multi-physical field coupling parameter space, an initial leakage probability distribution matrix is generated, a feature importance analysis is performed on the initial leakage probability distribution matrix, key influence factors are screened, and a leakage risk prediction model is constructed; The historical leakage event data stored in the cloud database are input to train the leakage risk prediction model, a time-space evolution iterative calculation of leakage probability is performed by using a sliding time window method, a stability correction is performed on a result of the iterative calculation, and a dynamic leakage risk thermodynamic map is output.

5. A building roof leakage detection system applied to the building roof leakage detection method of any one of claims 1-4, characterized in that, It comprises: A dot planning module is used to perform hole site pre-dot planning by using a multi-spectral sensor array before a waterproof layer drilling operation is performed on a corresponding equipment installation area of a building roof, to generate a micro-hole path topology, and to reconstruct a waterproof layer three-dimensional density distribution model according to a three-dimensional scanning of the building roof waterproof layer structure; A processing module is used to guide a drilling device to perform a drilling operation based on the micro-hole path topology, to synchronously collect real-time deformation parameters of a hole site surrounding area, and to perform a waterproof layer material internal stress gradient analysis on the real-time deformation parameters, to generate a hole site stress diffusion vector diagram; An analysis module is used to perform space registration fusion of the waterproof layer three-dimensional density distribution model and the hole site stress diffusion vector diagram, to calculate potential leakage probabilities of each hole site surrounding waterproof layer, to perform a time-space evolution simulation on the potential leakage probabilities, and to output a dynamic leakage risk thermodynamic map; An optimization module is used to perform a leakage path reverse tracking modeling in combination with the dynamic leakage risk thermodynamic map and the waterproof layer three-dimensional density distribution model, to generate a roof leakage detection optimization model, and to transmit the roof leakage detection optimization model to a monitoring terminal to implement a real-time leakage early warning.

6. A computer-readable storage medium, characterized in that: The computer is caused to execute the building roof leakage detection method according to any one of claims 1-4 when the instructions are run on the computer.

Images