A building risk real-time evaluation method and system based on multi-modal remote sensing images

Abstract

The application provides a building risk real-time evaluation method and system based on multi-modal remote sensing images, relates to the technical field of building safety monitoring, and obtains high-precision radar and infrared images, identifies a building metal structure area, and extracts a preliminary and temperature anomaly area; signals are collected through an internal vibration monitoring network, potential damage areas are determined in combination with intensity distribution and anomaly area information, and vibration characteristics are extracted; a physical correlation model of a metal microstructure and a radar signal is established, a response map is generated through space-time matching, a radar signal attenuation amount caused by vibration is calculated, and comparison and fusion are performed with a fatigue threshold value, temperature anomalies are corrected in combination, a risk index of weld cracking is generated, a sampling frequency is adjusted according to a risk level, an evaluation result containing the risk index and coordinates is output, multi-modal data fusion and dynamic monitoring adjustment can be achieved, and real-time and accurate evaluation of the weld cracking risk of a building metal structure can be achieved.

Description

Technical Field

[0001] This application relates to the field of building safety monitoring technology, and in particular to a method and system for real-time building risk assessment based on multimodal remote sensing images. Background Technology

[0002] In industrial buildings, the intense vibrations generated by the continuous operation of heavy equipment constantly impact the building's metal structure. Over time, this can easily lead to fatigue damage in the metal structure, such as weld cracking and component deformation, seriously threatening the structural safety of the building and the normal operation of production. Therefore, there is an urgent need for a method that can monitor the fatigue damage status of metal structures in real time and accurately, so as to detect potential risks in a timely manner, take maintenance measures in advance, and avoid safety accidents.

[0003] Currently, a common approach to monitoring fatigue damage to metal structures in industrial buildings caused by vibrations from heavy equipment is to utilize distributed sensor networks and data analysis systems. This involves deploying various types of sensors at key locations in the building's metal structure, such as beam-column joints and welds. These sensors include accelerometers to monitor vibration acceleration and strain gauges to measure structural strain. The sensors collect data in real time and transmit it to the data analysis system, which processes and analyzes the data to identify signs of fatigue damage in the metal structure.

[0004] However, this existing approach has significant drawbacks. Firstly, the number and distribution of sensors cannot fully cover the entire metal structure, creating monitoring blind spots, and early fatigue damage in some critical areas may not be detected in time. Secondly, relying solely on sensor-collected data provides a limited range of information. Industrial building environments are complex, and limited data such as vibration acceleration and structural strain are insufficient to accurately determine the true extent and trend of fatigue damage in metal structures, easily leading to misjudgments or omissions, and failing to provide a reliable basis for building safety maintenance. Summary of the Invention

[0005] The purpose of this application is to provide a method and system for real-time building risk assessment based on multimodal remote sensing imagery, in order to solve the problems of blind spots and single information dimension in the existing distributed sensor network monitoring, which lead to misjudgment and missed judgment.

[0006] To address the aforementioned technical problems, in a first aspect, this application provides a method for real-time building risk assessment based on multimodal remote sensing imagery, comprising:

[0007] Acquire high-precision radar and infrared images, identify building metal structure areas from the high-precision radar images and extract preliminary abnormal areas reflecting structural anomalies, and identify temperature anomaly areas within the metal structure areas from the infrared images.

[0008] By deploying a vibration monitoring network within a building to collect specific vibration signals during operation, and combining the intensity distribution of the specific vibration signals, the spatial information of the preliminary abnormal area and the temperature abnormal area, potential damage areas are identified and vibration characteristics associated with the potential damage areas are extracted.

[0009] Under vibration stress, a physical correlation model describing the relationship between changes in the microstructure of metal and the attenuation of radar signal reflection is established. Spatiotemporal matching technology is used to match the vibration characteristics with the scattering characteristics of high-precision radar images to generate a response relationship diagram.

[0010] Based on the physical correlation model, the radar signal attenuation caused by vibration energy in the response relationship diagram is calculated. The radar signal attenuation is dynamically compared and fused with a preset metal fatigue threshold. At the same time, the fusion result is corrected according to the spatial distribution characteristics of the temperature anomaly zone to generate a risk index for weld cracking in the potential damage zone.

[0011] Based on the risk level of the risk index, the sampling frequency configuration of the vibration monitoring network for the metal components in the potential damage zone is switched, and an evaluation result containing the risk index and the corresponding location coordinates of the potential damage zone is output.

[0012] Optionally, the step of establishing a physical correlation model describing the changes in the metal's microstructure and the attenuation of radar signal reflection under vibration stress, and using spatiotemporal matching technology to match the vibration characteristics with the scattering characteristics of high-precision radar images to generate a response relationship diagram, includes:

[0013] Based on the physical mechanism by which changes in the microstructure of metal under vibration stress affect radar signal reflection capability, a mapping relationship is established between the dominant vibration frequency and vibration energy attenuation rate in the vibration characteristics and the radar signal reflection attenuation coefficient, which serves as a physical correlation model.

[0014] The dominant vibration frequency and vibration energy attenuation rate in the vibration characteristics are input into the physical correlation model to generate the predicted reflection attenuation distribution of the radar signal;

[0015] High-precision radar images are acquired synchronously with the vibration feature acquisition time. The actual radar echo signal intensity values ​​of each spatial location point in the building metal structure area are extracted from the high-precision radar images to form the actual scattering intensity distribution of the radar signal.

[0016] The predicted reflection attenuation distribution and the actual scattering intensity distribution are spatiotemporally registered and numerically correlated to generate a response relationship diagram that reflects the quantitative relationship between the predicted attenuation and the actual scattering intensity, with spatial location as the coordinate.

[0017] Optionally, based on the physical correlation model, the radar signal attenuation caused by vibration energy in the response relationship diagram is calculated. The radar signal attenuation is then dynamically compared and fused with a preset metal fatigue threshold. Simultaneously, the fusion result is corrected according to the spatial distribution characteristics of the temperature anomaly zone to generate a risk index for weld cracking in the potential damage zone, including:

[0018] Based on the physical correlation model, the radar signal attenuation directly caused by vibration energy is analyzed from the response relationship diagram;

[0019] Calculate the spatial average value of the radar signal attenuation within each potential damage zone to obtain the characteristic attenuation.

[0020] A dynamic comparison mechanism is established between the characteristic attenuation amount and the preset metal fatigue threshold to calculate the deviation between the characteristic attenuation amount and the metal fatigue threshold in real time and obtain a risk assessment value.

[0021] Acquire spatial distribution data of temperature anomaly zones that spatially overlap with the potential damage zone, calculate temperature influence correction coefficient based on the spatial distribution data, and correct the risk assessment value according to the temperature influence correction coefficient;

[0022] The revised risk assessment value is converted into a risk index characterizing weld cracking in the potential damage zone.

[0023] Optionally, the step of acquiring spatial distribution data of temperature anomaly areas that spatially overlap with the potential damage area, calculating a temperature influence correction coefficient based on the spatial distribution data, and correcting the risk assessment value according to the temperature influence correction coefficient includes:

[0024] Obtain a temperature anomaly area that overlaps with the spatial range of the potential damage area, and extract the temperature data of all pixels within the temperature anomaly area as spatial distribution data;

[0025] Based on the spatial distribution data, calculate the highest temperature value, the average temperature value, and the ratio of the area of ​​the temperature anomaly region to the total area of ​​the potential damage region in the temperature anomaly region.

[0026] The highest temperature value, average temperature value, and proportional value are input into a preset temperature stress relationship function to calculate the temperature influence correction coefficient.

[0027] The preliminary risk assessment value is multiplied by the temperature influence correction coefficient to obtain the corrected risk assessment value.

[0028] Optionally, the specific vibration signals collected by the vibration monitoring network deployed within the building, combined with the intensity distribution of the specific vibration signals, the spatial information of the preliminary anomaly zone and the temperature anomaly zone, are used to determine potential damage zones and extract vibration features associated with the potential damage zones, including:

[0029] A vibration monitoring network consisting of multiple vibration sensors is deployed within the building, and the vibration monitoring network is operated to collect specific vibration signals generated during the operation of the equipment;

[0030] Based on the specific vibration signal, an intensity distribution reflecting the magnitude of vibration energy is generated;

[0031] The intensity distribution, the spatial information of the preliminary abnormal area, and the spatial information of the temperature abnormal area are superimposed, and the area in the intensity distribution where the vibration energy is higher than the fourth set threshold and spatially overlaps with the preliminary abnormal area and the temperature abnormal area is selected as the potential damage area.

[0032] The vibration sensor is located within the potential damage area, and the vibration dominant frequency and vibration energy decay rate are extracted from the specific vibration signal collected by the vibration sensor as vibration characteristics.

[0033] Optionally, the step of switching the sampling frequency configuration of the vibration monitoring network for the metal components in the potential damage zone based on the risk level of the risk index, and outputting an evaluation result including the risk index and the corresponding coordinates of the potential damage zone, includes:

[0034] Multiple consecutively divided risk level intervals are pre-defined, and a specific target sampling frequency is configured for each risk level interval;

[0035] The risk index is matched with the risk level range to determine the risk level corresponding to the risk index;

[0036] The target sampling frequency corresponding to the risk level is queried, and a control command is sent to the vibration sensor deployed within the potential damage area to switch the sampling frequency of the vibration sensor to the target sampling frequency.

[0037] Generate and output structured assessment results, which include the risk index and corresponding risk level, the location coordinates of the potential damage area, and the target sampling frequency.

[0038] Optionally, the steps of acquiring high-precision radar and infrared images, identifying the building's metal structure area from the high-precision radar images and extracting preliminary anomaly areas reflecting structural anomalies, and identifying temperature anomaly areas within the metal structure area from the infrared images, include:

[0039] High-precision radar images of the target building are acquired using radar sensors, while infrared images of the same target building are acquired using infrared thermal imagers.

[0040] The area with a reflected signal intensity higher than a first set threshold is identified from the high-precision radar image as the building metal structure area;

[0041] Within the building's metal structure area, regions where the intensity variation of reflected signals exceeds a second preset threshold are identified as preliminary anomaly zones.

[0042] The areas in the building's metal structure zone identified from the infrared image whose temperature values ​​are higher than a third preset threshold are designated as temperature anomaly zones.

[0043] Secondly, this application provides a real-time building risk assessment system based on multimodal remote sensing imagery, comprising:

[0044] The acquisition module is used to acquire high-precision radar images and infrared images, identify the building metal structure area from the high-precision radar images and extract the preliminary abnormal area reflecting structural anomalies, and identify the temperature abnormal area within the metal structure area from the infrared images.

[0045] The extraction module is used to collect specific vibration signals when the vibration monitoring network is deployed in the building, and to determine potential damage areas and extract vibration features associated with the potential damage areas by combining the intensity distribution of the specific vibration signals, the spatial information of the preliminary abnormal area and the temperature abnormal area.

[0046] The matching module is used to establish a physical correlation model describing the changes in the microstructure of metal and the attenuation of radar signal reflection under vibration stress, and to use spatiotemporal matching technology to match the vibration characteristics with the scattering characteristics of high-precision radar images to generate a response relationship diagram.

[0047] The generation module is used to calculate the radar signal attenuation caused by vibration energy in the response relationship diagram based on the physical correlation model, dynamically compare and fuse the radar signal attenuation with a preset metal fatigue threshold, and correct the fusion result according to the spatial distribution characteristics of the temperature anomaly zone to generate a risk index for weld cracking in the potential damage zone.

[0048] The output module is used to switch the sampling frequency configuration of the vibration monitoring network for the metal components in the potential damage zone based on the risk level of the risk index, and output the evaluation result including the risk index and the corresponding location coordinates of the potential damage zone.

[0049] Thirdly, this application provides an electronic device, comprising:

[0050] Memory, used to store computer programs;

[0051] A processor is configured to execute the computer program to implement the steps of the real-time building risk assessment method based on multimodal remote sensing imagery as described in the first aspect above.

[0052] Fourthly, this application provides a computer-readable storage medium storing a computer program that, when executed by a processor, can implement the steps of the real-time building risk assessment method based on multimodal remote sensing imagery as described in the first aspect above.

[0053] The method for real-time building risk assessment based on multimodal remote sensing imagery provided in this application acquires high-precision radar and infrared images and identifies relevant anomaly areas. It combines these with signals and spatial information collected by a vibration monitoring network within the building to determine potential damage areas and extract vibration characteristics. This comprehensive approach captures structural anomaly information, laying the foundation for subsequent assessments. A physical correlation model is established and a response relationship diagram is generated through spatiotemporal matching, effectively linking vibration characteristics with radar scattering properties and improving the accuracy of data fusion. The attenuation is calculated based on the model and fused with a threshold. A risk index is generated by correcting for temperature anomaly area characteristics, accurately quantifying weld cracking risk. The sampling frequency is switched according to the risk level, and results containing the risk index and location coordinates are output. This ensures both real-time and targeted monitoring while providing accurate data for risk assessment. Overall, this method achieves real-time, accurate, and comprehensive assessment of weld cracking risk in building metal structures, effectively improving the efficiency and reliability of building safety monitoring.

[0054] Furthermore, based on the physical mechanism by which changes in the microstructure of metal under vibration stress affect radar signal reflection capability, a physical correlation model is established, mapping the vibration dominant frequency, vibration energy attenuation rate, and radar signal reflection attenuation coefficient in the vibration characteristics. Then, the vibration dominant frequency and vibration energy attenuation rate are input into the model to generate the predicted radar signal reflection attenuation distribution. Next, synchronous high-precision radar images are acquired, and the actual radar echo signal intensity values ​​at each spatial location point in the metal structure area are extracted to form the actual scattering intensity distribution. Finally, spatiotemporal registration and numerical correlation analysis are performed on the predicted and actual distributions to generate a response relationship diagram reflecting the quantitative relationship between the two, using spatial location as coordinates. This scheme, by establishing a precise physical correlation model and combining it with synchronously acquired image data for spatiotemporal registration and correlation analysis, generates a response relationship diagram that accurately reflects the quantitative correlation between vibration characteristics and radar scattering characteristics. This provides a reliable and accurate data foundation for subsequent calculations of radar signal attenuation and assessment of weld cracking risk, improving the scientific rigor and accuracy of the entire risk assessment process. Attached Figure Description

[0055] To more clearly illustrate the technical solutions of the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0056] Figure 1 A flowchart illustrating a method for real-time building risk assessment based on multimodal remote sensing imagery, provided in an embodiment of this application;

[0057] Figure 2 A schematic diagram illustrating the specific implementation process of a real-time building risk assessment method based on multimodal remote sensing imagery provided in this application embodiment;

[0058] Figure 3 A schematic diagram illustrating a specific implementation of a real-time building risk assessment method based on multimodal remote sensing imagery provided in this application embodiment;

[0059] Figure 4 This is a schematic diagram of the structure of a real-time building risk assessment system based on multimodal remote sensing imagery, provided as an embodiment of this application. Detailed Implementation

[0060] In fatigue damage monitoring of metal structures in industrial buildings, existing solutions utilizing distributed sensor networks and data analysis systems have significant shortcomings. The sheer number and distribution of sensors cannot fully cover the metal structure, creating monitoring blind spots and making it difficult to detect early fatigue damage in some critical areas in a timely manner. Furthermore, relying solely on limited data such as vibration acceleration and structural strain provides a single dimension of information, making it difficult to accurately determine the true extent and development trend of damage in complex environments. This can easily lead to misjudgments or omissions, failing to provide a reliable basis for safety maintenance.

[0061] To address the aforementioned issues, this application proposes a real-time risk assessment method for buildings based on multimodal remote sensing imagery. This method acquires high-precision radar and infrared images to identify anomalous areas, combines signals and spatial information collected by an in-building vibration monitoring network to determine potential damage areas and extract vibration characteristics, establishes a physical correlation model, and generates a response relationship diagram through spatiotemporal matching. It then calculates attenuation and generates a risk index, finally switching the sampling frequency according to the risk level and outputting the results. This scheme, through multimodal data fusion, provides broader coverage, eliminates monitoring blind spots, and integrates multi-dimensional information such as radar, infrared, and vibration data, enabling more accurate assessment of damage conditions. It effectively solves the problems of incomplete monitoring and limited information in existing schemes, leading to misjudgments and omissions, thus improving the accuracy and reliability of the assessment and providing strong support for the safety maintenance of industrial buildings.

[0062] To enable those skilled in the art to better understand the present application, the present application will be further described in detail below with reference to the accompanying drawings and specific embodiments. Obviously, the described embodiments are merely some embodiments of the present application, and not all embodiments. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0063] The core of this application is to provide a method for real-time building risk assessment based on multimodal remote sensing imagery, and a flowchart of one specific implementation is shown below. Figure 1 As shown, the method includes:

[0064] S101. Acquire high-precision radar images and infrared images, identify the building metal structure area from the high-precision radar images and extract the preliminary abnormal area reflecting structural anomalies, and identify the temperature anomaly area within the metal structure area from the infrared images.

[0065] Optionally, step S101 may specifically include the following steps:

[0066] S1011. Acquire high-precision radar images of the target building through radar sensors, and simultaneously acquire infrared images of the same target building through an infrared thermal imager.

[0067] S1012. Identify areas with reflected signal strength higher than a first set threshold from the high-precision radar image as building metal structure areas;

[0068] S1013. Within the building's metal structure area, extract the region where the intensity of the reflected signal changes more than a second set threshold as a preliminary abnormal region.

[0069] S1014. Identify areas in the building's metal structure area whose temperature values ​​are higher than a third preset threshold from the infrared image as temperature anomaly areas.

[0070] In the above scheme, high-precision radar imagery refers to images acquired by radar sensors that clearly present detailed information about buildings, including data such as the reflection of radar signals from different areas of the building. Infrared imagery is captured by an infrared thermal imager, reflecting the temperature distribution of different parts of the building. The building's metal structure area refers to the structural area of ​​the building made of metal materials identified from the radar imagery. The preliminary anomaly area is a region within the building's metal structure area where the intensity of the radar reflection signal varies significantly, potentially indicating structural anomalies. The temperature anomaly area is a region within the building's metal structure area where the temperature, as identified from the infrared imagery, is significantly higher than the surrounding area. The first threshold is a standard value used to determine whether the intensity of the reflected signal in the radar imagery is sufficient to identify a building's metal structure area; the second threshold is a standard value used to determine whether the variation in the intensity of the reflected signal within the building's metal structure area reaches an abnormal level; and the third threshold is a standard value used to determine whether the temperature in the infrared imagery is abnormally high.

[0071] In this embodiment, step S1011 first involves simultaneously acquiring image data of the target building using a radar sensor and an infrared thermal imager. The radar sensor employs synthetic aperture radar technology, emitting radar waves of a specific frequency towards the building. These radar waves are reflected upon encountering the building surface. The sensor receives the reflected waves and records their intensity, phase, and other information, generating a high-precision radar image through signal processing algorithms. Simultaneously, the infrared thermal imager uses infrared detectors to capture the infrared radiation energy emitted from various parts of the building, converting it into electrical signals through photoelectric conversion. These signals are then processed by image processing algorithms to generate an infrared image, visually reflecting the temperature distribution on the building surface. For example, when monitoring an industrial plant, the radar sensor emits radar waves from multiple angles and receives reflected signals. After processing, a radar image containing structural reflection information from the plant's metal frame, walls, etc., is obtained. Simultaneously, the infrared thermal imager captures images, obtaining an infrared image showing temperature differences in different areas of the plant.

[0072] Secondly, in step S1012, the metal structure area of ​​the building is identified based on high-precision radar imagery. Since metal materials have a strong ability to reflect radar waves, their reflected signal intensity is significantly higher than that of other building materials such as concrete and wood. Therefore, a threshold segmentation algorithm is used. A first set threshold is determined based on experimental data of the radar reflection characteristics of common building materials. Then, each pixel in the radar image is traversed, and pixels with reflected signal intensity greater than the first set threshold are marked. Noise interference is then removed using a morphological processing algorithm, and adjacent marked points are connected to form a continuous region, which is the metal structure area of ​​the building. For example, in the acquired radar image of an industrial plant, the first set threshold is set to 50. After threshold segmentation and morphological processing, areas in the image with reflected signal intensity exceeding 50 are identified as the metal structure area of ​​the building, consisting of metal columns, beams, etc.

[0073] Next, step S1013 analyzes the radar image sequence corresponding to the building's metal structure area. A time-series analysis algorithm is used to calculate the variation amplitude of the reflected signal intensity of each pixel at different time points. Then, a second threshold is set based on the fluctuation range of the reflected signal intensity under normal metal structure conditions. Areas containing pixels with variation amplitudes exceeding the second threshold are designated as preliminary anomaly zones. These areas may have significantly altered radar reflection characteristics due to structural deformation, cracks, etc. For example, in the metal beam structure area of ​​an industrial plant, by analyzing radar images over 30 consecutive days, the variation amplitude of the reflected signal intensity at each point is calculated. The second threshold is set to 10. If the variation amplitude of a certain section of the beam reaches 15, exceeding the threshold, this area is extracted as a preliminary anomaly zone.

[0074] Finally, in step S1014, the infrared image and radar image are spatially registered to accurately locate the building's metal structure area in the infrared image. Then, the temperature values ​​of each pixel within this area are extracted. Subsequently, a third set threshold is determined based on the normal operating temperature range of the metal structure. Areas with temperatures exceeding the third set threshold are identified as temperature anomaly zones. These areas may exhibit abnormal temperature increases due to friction, excessive current, or other factors, indicating potential structural problems. For example, after spatial registration of the metal column structure area in an industrial plant within the infrared image, the temperature values ​​of each point are extracted. Based on an ambient temperature of 30℃, a third set threshold of 50℃ is set. If the temperature at a connection point on the column reaches 65℃, exceeding the threshold, this area is identified as a temperature anomaly zone.

[0075] In a practical application, to monitor the condition of the metal structure of an industrial building, staff used radar sensors to scan the building, acquiring high-precision radar images. Simultaneously, infrared thermal imagers captured infrared images. Based on experimental data of the radar reflection characteristics of common building materials, a first threshold of 60 was set in the high-precision radar images. Analysis revealed areas with reflection signal intensities higher than 60, identified as metal structural areas, such as the building's metal load-bearing frame. Within these load-bearing frame areas, the variation in reflection signal intensity was calculated for each part. Based on the normal fluctuation range of reflection signal intensity in a metal structure, a second threshold of 15 was set. A section of the metal frame with a variation greater than 15 was extracted as a preliminary anomaly area. Further examination of the infrared images corresponding to the metal load-bearing frame revealed an ambient temperature of 32℃. A third threshold of 52℃ was set. A connection point of this section of the metal frame, with a temperature exceeding 52℃, was identified as having a temperature of 59℃ in the infrared image, thus being identified as a temperature anomaly area.

[0076] The overall solution described in S101, by acquiring two different types of images and performing step-by-step analysis and identification, can accurately determine the metal structure areas within a building, as well as the preliminary anomaly areas and temperature anomaly areas that may contain structural abnormalities. This provides fundamental information for further assessment of potential damage to the building's metal structure, helps to comprehensively understand the anomalies in the building's metal structure, and lays a solid foundation for subsequent risk assessment work.

[0077] S102. When a vibration monitoring network is deployed inside a building to collect specific vibration signals during operation, and combined with the intensity distribution of the specific vibration signals, the spatial information of the preliminary abnormal area and the temperature abnormal area, the potential damage area is determined and the vibration characteristics associated with the potential damage area are extracted.

[0078] Optionally, step S102 may specifically include the following steps:

[0079] S1021. Deploy a vibration monitoring network consisting of multiple vibration sensors within the building, and operate the vibration monitoring network to collect specific vibration signals generated during the operation of the equipment;

[0080] S1022. Based on the specific vibration signal, generate an intensity distribution reflecting the magnitude of the vibration energy;

[0081] S1023. The intensity distribution, the spatial information of the preliminary abnormal area, and the spatial information of the temperature abnormal area are superimposed, and the area in the intensity distribution where the vibration energy is higher than the fourth set threshold and spatially overlaps with the preliminary abnormal area and the temperature abnormal area is selected as the potential damage area.

[0082] S1024. Locate the vibration sensor within the potential damage area, and extract the vibration dominant frequency and vibration energy decay rate from the specific vibration signal collected by the vibration sensor as vibration characteristics.

[0083] In the above scheme, the vibration monitoring network is a system composed of multiple vibration sensors used to collect vibration signals from equipment operating within the building, including vibration data collected by each sensor. Specific vibration signals refer to vibration data generated during equipment operation that may be related to damage to the building's metal structure. Intensity distribution reflects the distribution of vibration energy at different locations within the building. The potential damage zone is an area identified as potentially susceptible to damage after considering the spatial information of vibration signal intensity, preliminary anomaly zones, and temperature anomaly zones. Vibration characteristics are parameters extracted from the signals collected by vibration sensors within the potential damage zone, reflecting the vibration characteristics; specifically, these refer to the dominant vibration frequency and vibration energy decay rate. The fourth threshold is a standard value used to determine whether the vibration energy has reached a level that may cause damage.

[0084] In this embodiment, firstly, in step S1021, based on the distribution characteristics of the building's metal structure, multiple vibration sensors are uniformly and comprehensively deployed at key locations such as the metal load-bearing frame and metal components connecting equipment to the building. After the sensors are deployed, they are connected to the terminal monitoring equipment via data transmission lines to form a vibration monitoring network. When heavy equipment in the building starts operating, the vibrations generated by the equipment are transmitted to various parts through the building structure. After sensing these vibrations, the vibration sensors convert them into electrical signals and transmit them to the terminal in real time, thereby collecting specific vibration signals from the equipment during operation. For example, on the metal load-bearing frame of a building, a vibration sensor is arranged every 3 meters, for a total of 8 sensors. When a large stamping machine in the factory operates, the sensors collect the vibration signals transmitted by the machine to the metal load-bearing frame. These signals include information such as the intensity and frequency of the vibration.

[0085] Secondly, in step S1022, the electrical signals transmitted by each vibration sensor are preprocessed. A filtering algorithm removes noise interference from the signals, retaining only the valid vibration signals. Then, a Fourier transform algorithm is used to convert the preprocessed time-domain vibration signals into frequency-domain signals, and the energy value of each signal is calculated. Next, based on the installation coordinates of each vibration sensor, the corresponding vibration energy value is associated with the location information. An interpolation algorithm is then used to fill in the energy values ​​for the entire building's metal structure area, ultimately generating an intensity distribution reflecting the magnitude of vibration energy with the building's spatial location as the coordinate. For example, after preprocessing and calculating the energy of vibration signals collected by eight sensors in an industrial building, the vibration energy values ​​at each sensor location are obtained as 50, 45, 35, 48, 37, 54, 42, and 37, respectively. Interpolation is then performed using their location coordinates to generate the intensity distribution of vibration energy at each point on the metal load-bearing frame, clearly showing which sections have higher vibration energy.

[0086] Next, in step S1023, the spatial coordinates of the intensity distribution, the initial anomaly zone, and the temperature anomaly zone are transformed to the same coordinate system for spatial overlay. Using the overlay analysis function in GIS spatial analysis technology, the spatial information of these three data points is overlaid to obtain their spatial overlap. Simultaneously, based on the material properties of the building's metal structure and the long-term impact data of equipment vibration, a fourth threshold is set to determine whether the vibration energy has reached a level that may cause damage. In the overlay results, areas with vibration energy higher than the fourth threshold and simultaneously located within both the initial anomaly zone and the temperature anomaly zone are selected as potential damage zones. For example, in an industrial building, the fourth threshold is set to 40. Overlaying areas with energy higher than 40 in the intensity distribution with the initial anomaly zone (a 5-meter-long section of metal frame) and the temperature anomaly zone (a 2-square-meter connection point on this section of metal frame) reveals that a 3-meter-long area within this section of metal frame simultaneously meets the criteria of vibration energy higher than 40, being within the initial anomaly zone, and being within the temperature anomaly zone. Therefore, this 3-meter-long area is identified as a potential damage zone.

[0087] Finally, in step S1024, based on the spatial coordinates of the potential damage area, vibration sensors located within that area are identified in the vibration monitoring network. The signals collected by these sensors accurately reflect the vibration conditions of the potential damage area. Vibration characteristics are then extracted from the specific vibration signals collected by these sensors. For the dominant vibration frequency, spectral analysis is performed to find the frequency corresponding to the peak value in the power spectrum; this frequency is the dominant vibration frequency. For the vibration energy decay rate, vibration signals over a period of time are selected, and the vibration energy values ​​at different times are calculated. Linear fitting is used to obtain the slope of the straight line showing the energy change over time; the absolute value of this slope is the vibration energy decay rate. For example, in a 3-meter-long potential damage area of ​​a building, there are two vibration sensors. Spectral analysis of the signals collected by these two sensors reveals that the frequency corresponding to the peak value of the power spectrum is 50 Hz, i.e., the dominant vibration frequency is 50 Hz. Selecting vibration energy data over 10 seconds, the energy decays from 60 to 20, and linear fitting yields a decay rate of 4. These two parameters together constitute the vibration characteristics of the potential damage area.

[0088] In practical applications, a monitoring network of eight piezoelectric vibration sensors, spaced 3 meters apart, is constructed on the metal load-bearing frame of an industrial building. When the large stamping equipment in the factory is running, the sensors collect vibration signals and transmit them to the terminal. After signal preprocessing, Fourier transform is used to calculate the energy values ​​at each sensor location: 35, 42, 50, 48, 39, 55, 46, and 37. These values ​​are then combined with coordinate interpolation to generate an intensity distribution. This distribution is then transformed into the same coordinate system as the initial anomaly area (a 5-meter-long metal frame) and the temperature anomaly area (a 2-square-meter connection area on the frame). Using GIS overlay analysis, a fourth threshold of 40 is set, and a 3-meter-long overlapping area is selected as a potential damage zone. Within the location zone, the power spectrum of two sensors is analyzed, and the peak power frequency corresponding to 50 Hz is identified as the dominant vibration frequency. Ten seconds of energy data are collected, and linear fitting yields a vibration energy attenuation rate of 4, which decreases from 60 Hz to 20 Hz. These two values ​​are used as the vibration characteristics of this area.

[0089] The overall solution described in S102 involves deploying a vibration monitoring network to collect signals, processing them to obtain the intensity distribution, and then combining this with spatial information from preliminary and temperature anomaly areas to determine potential damage zones. Vibration characteristics are then extracted, enabling precise identification of areas potentially susceptible to damage and acquisition of vibration characteristic parameters for those areas. This provides crucial information for subsequent analysis of the relationship between vibration and structural damage, facilitating a more in-depth assessment of the risks associated with building metal structures.

[0090] S103. Under the action of vibration stress, establish a physical correlation model describing the changes in the metal microstructure and the attenuation of radar signal reflection, and use spatiotemporal matching technology to match the vibration characteristics with the scattering characteristics of high-precision radar images to generate a response relationship diagram.

[0091] Optionally, step S103 may specifically include the following steps:

[0092] S1031. Based on the physical mechanism by which changes in the microstructure of metal under vibration stress affect radar signal reflection capability, establish the mapping relationship between the dominant vibration frequency and vibration energy attenuation rate in the vibration characteristics and the radar signal reflection attenuation coefficient, as a physical correlation model.

[0093] S1032. Input the vibration dominant frequency and vibration energy attenuation rate in the vibration characteristics into the physical correlation model to generate the predicted reflection attenuation distribution of the radar signal;

[0094] S1033. Synchronously acquire high-precision radar images that are synchronized with the vibration feature acquisition time, and extract the actual radar echo signal intensity values ​​of each spatial location point in the building metal structure area from the high-precision radar images to form the actual scattering intensity distribution of the radar signal.

[0095] S1034. Perform spatiotemporal registration and numerical correlation analysis on the predicted reflection attenuation distribution and the actual scattering intensity distribution to generate a response relationship diagram that reflects the quantitative relationship between the predicted attenuation and the actual scattering intensity, with spatial location as coordinates.

[0096] In the above scheme, vibration stress refers to the force exerted on the building's metal structure by equipment vibration. Metal microstructure change refers to the alteration of the fine structures such as grains and grain boundaries within the metal under vibration stress. Radar signal reflection attenuation refers to the weakening of radar signal reflection intensity after encountering a metal structure. The physical correlation model describes the correspondence between vibration characteristics and radar signal reflection attenuation coefficients, reflecting the impact of metal microstructure changes on radar reflection. The predicted reflection attenuation distribution is the distribution of radar signal attenuation at different spatial locations obtained after inputting vibration characteristics into the physical correlation model. Synchronized high-precision radar imagery refers to radar imagery acquired at the same time as the vibration characteristics, reflecting the radar reflection of the metal structure at that moment. The actual scattering intensity distribution is the actual radar echo signal intensity distribution at each location of the building's metal structure, extracted from the synchronized radar imagery. Spatiotemporal matching technology is used to align and correlate data acquired from different sources and at different times in time and space. The response relationship diagram is a graph using spatial location as coordinates, illustrating the quantitative relationship between predicted reflection attenuation and actual scattering intensity.

[0097] In the embodiments of this application, such as Figure 2 As shown, in step S1031, a metal material mechanics experiment is conducted to observe microscopic changes such as grain dislocation, grain boundary slip, and microcrack formation within the metal under different vibration stresses. Simultaneously, corresponding radar signal reflection data is recorded. Analysis reveals that these microscopic changes alter the surface roughness and electromagnetic properties of the metal, leading to variations in the radar signal reflection attenuation. Based on this, a large amount of sample data on the dominant vibration frequency, vibration energy attenuation rate, and radar signal reflection attenuation coefficient are collected. A multiple linear regression algorithm is used to fit the data, determining the mathematical expression between the three: the dominant vibration frequency f and the vibration energy attenuation rate v as independent variables, and the radar signal reflection attenuation coefficient k as the dependent variable. The calculation formula can be expressed as follows: Where a and b are regression coefficients, and c is a constant term, calculated from sample data to construct a physical correlation model. For example, in an experiment on a metal sample, the reflection attenuation coefficient was measured to be 0.2 at a dominant vibration frequency of 30 Hz and an energy attenuation rate of 2; at a dominant frequency of 60 Hz and an attenuation rate of 5, the coefficient was 0.5. Substituting these data into the formula, regression analysis yields a = 0.005, b = 0.05, and c = 0.05, resulting in the model formula. .

[0098] Secondly, the vibration characteristic parameters extracted from the potential damage area in step S1032, namely the dominant vibration frequency and the vibration energy attenuation rate, are organized according to the format required by the model. These parameters are then input into the physical correlation model, which uses the built-in mathematical formulas to calculate the radar signal reflection attenuation at each spatial location within the potential damage area. Specifically, assuming a location has coordinates (x, y), the dominant vibration frequency f and energy attenuation rate v of the potential damage area are input into the model to calculate the basic attenuation coefficient for that location. For example, substituting the dominant vibration frequency of 50 Hz and the energy attenuation rate v of the potential damage area into the formula yields... Then, by combining the position correction coefficients of each grid cell, the attenuation amount of different grid cells is obtained. For example, the attenuation amount of one grid cell after correction is 0.35, while that of another cell is 0.42. Finally, by combining the spatial coordinate information of the potential damage area, a gridding process is used to divide the entire potential damage area into several grid cells, and each cell is assigned a corresponding attenuation value, ultimately forming the predicted reflection attenuation distribution.

[0099] Next, the acquisition time of the vibration characteristics is determined in step S1033. Then, the radar sensor is controlled to scan the target building at the same time to ensure that the acquired high-precision radar image and the vibration characteristics are completely synchronized in time, thus guaranteeing the timeliness and matching of the data. The synchronously acquired radar image is preprocessed, and the metal structure area of ​​the building is separated using an image segmentation algorithm. Then, pixel intensity extraction technology is used to extract the radar echo signal intensity value I(x,y) of each spatial location point (x,y) in this area. The calculation is based on the amplitude A of the reflected signal received by the radar sensor, and the formula is as follows: ,in These are constants related to the radar system. These intensity values ​​are then correlated with their corresponding spatial coordinates, and distributed to each grid cell using the same grid division method as the predicted reflection attenuation distribution, forming the actual scattering intensity distribution. For example, at 10:00 AM, when characteristics such as a dominant vibration frequency of 50 Hz are acquired, radar images of industrial buildings are obtained. After processing, the amplitude of the reflected signal in a certain grid cell is measured to be 0.28. Then the strength value of this element is The other unit has an amplitude of 0.27 and an intensity value of This forms the actual scattering intensity distribution.

[0100] Finally, in step S1034, the predicted reflection attenuation distribution and the actual scattering intensity distribution are spatiotemporally registered. Spatially, a feature point matching algorithm is used to find the corresponding building structure feature points in the two distributions, and coordinate transformation is used to align their spatial positions completely. Temporally, the acquisition times of both are checked to ensure they are at the same moment, eliminating the influence of time differences. Then, numerical correlation analysis is performed on the two registered distributions to calculate the correlation value R(x,y) between the predicted reflection attenuation k(x,y) and the actual scattering intensity I(x,y) of each grid cell. The calculation formula is in the form of a ratio. Finally, using spatial coordinates (x, y) as the horizontal and vertical axes, and the correlation value R(x, y) as the data basis, a response relationship diagram is generated using visualization plotting technology. Different colors or values ​​in the diagram represent different degrees of correlation. For example, after spatiotemporal registration of the predicted reflection attenuation distribution and the actual scattering intensity distribution of a building, for a certain grid cell k(x, y) = 0.35 and I(x, y) = 80, the correlation value is calculated using the formula. The other unit has k(x,y)=0.42 and I(x,y)=75, with associated values... Based on this, a response relationship diagram is generated, which clearly shows the quantitative relationship between the two at various locations within the potential damage area.

[0101] In practical applications, a physical correlation model is established experimentally for monitoring potential damage zones in industrial buildings. The formula is as follows: Where k is the radar signal reflection attenuation coefficient, f is the dominant vibration frequency, and v is the vibration energy attenuation rate. Substituting the dominant vibration frequency of 50 Hz and the energy attenuation rate of 4 into the model, we get k = 0.005 × 50 + 0.05 × 4 + 0.05 = 0.5, generating the predicted reflection attenuation distribution. Simultaneously, synchronous radar images are acquired, and the actual scattering intensity of one grid cell is calculated to be 80, while another is 75, forming the actual scattering intensity distribution. After spatiotemporal registration of the two, [f is used]. Calculate the correlation values ​​respectively Generate a response relationship graph.

[0102] The overall scheme of S103 described above establishes a physical correlation model to link vibration characteristics with radar signal reflection attenuation. This is then combined with synchronized radar imagery to obtain the actual scattering intensity distribution, and a response relationship diagram is generated through spatiotemporal matching. This process effectively combines vibration characteristics with radar scattering properties, providing an important bridge for accurately calculating radar signal attenuation and assessing weld cracking risk, thus improving the scientific rigor and accuracy of risk assessment.

[0103] S104. Based on the physical correlation model, calculate the radar signal attenuation caused by vibration energy in the response relationship diagram, dynamically compare and fuse the radar signal attenuation with the preset metal fatigue threshold, and correct the fusion result according to the spatial distribution characteristics of the temperature anomaly zone to generate the risk index of weld cracking in the potential damage zone.

[0104] Optionally, step S104 may specifically include the following steps:

[0105] S1041. Based on the physical correlation model, the radar signal attenuation directly caused by vibration energy is analyzed from the response relationship diagram;

[0106] S1042. Calculate the spatial average value of the radar signal attenuation in each potential damage zone to obtain the characteristic attenuation.

[0107] S1043. Establish a dynamic comparison mechanism between the characteristic attenuation amount and the preset metal fatigue threshold, calculate the deviation between the characteristic attenuation amount and the metal fatigue threshold in real time, and obtain a risk assessment value.

[0108] S1044. Obtain spatial distribution data of temperature anomaly areas that spatially overlap with the potential damage area, calculate temperature influence correction coefficient based on the spatial distribution data, and correct the risk assessment value according to the temperature influence correction coefficient.

[0109] Specifically, step S1044 includes the following processes: acquiring a temperature anomaly area that overlaps with the spatial range of the potential damage area; extracting the temperature data of all pixels within the temperature anomaly area as spatial distribution data; based on the spatial distribution data, calculating the highest temperature value, average temperature value, and the ratio of the temperature anomaly area to the total area of ​​the potential damage area; inputting the highest temperature value, average temperature value, and ratio into a preset temperature stress relationship function to calculate the temperature influence correction coefficient; and multiplying the preliminary risk assessment value by the temperature influence correction coefficient to obtain the corrected risk assessment value.

[0110] S1045. Convert the corrected risk assessment value into a risk index that characterizes weld cracking in the potential damage zone.

[0111] In the above scheme, radar signal attenuation refers to the specific numerical value of the reduction in reflection intensity of a radar signal after encountering a metal structure under the action of vibration energy. Characteristic attenuation is the spatial average value of radar signal attenuation within the potential damage zone, representing the overall attenuation level of that area. The metal fatigue threshold is a pre-set critical value used to determine whether the metal structure has experienced fatigue damage. The dynamic comparison mechanism is a method of comparing the characteristic attenuation with the metal fatigue threshold in real time and calculating the degree of deviation between the two. The deviation degree is the degree of difference between the characteristic attenuation and the metal fatigue threshold, used to reflect the magnitude of the risk. The risk assessment value is obtained based on the deviation degree and is a preliminary numerical measure of the risk of the potential damage zone. The spatial distribution data of the temperature anomaly zone refers to information such as the temperature and area of ​​the temperature anomaly zone overlapping with the potential damage zone. The temperature influence correction coefficient is calculated based on the spatial distribution data of the temperature anomaly zone and is used to correct the risk assessment value. The temperature-stress relationship function is a function used to calculate the temperature influence correction coefficient based on temperature-related parameters. The risk index is obtained by converting the corrected risk assessment value and is an indicator that intuitively represents the risk of weld cracking in the potential damage zone.

[0112] In this embodiment, firstly, based on the established physical correlation model, step S1041 uses data analysis technology to filter out signal changes caused solely by vibration energy from the response relationship diagram, excluding attenuation interference caused by other factors such as temperature and environment. Specifically, using the correspondence between vibration characteristics and attenuation coefficients in the model, each data point in the response relationship diagram is reverse-engineered to determine whether it is caused by vibration energy, and finally the radar signal attenuation at each location is extracted. For example, if the correlation value of a point in a response relationship diagram is 0.005, combined with the corresponding parameters of vibration energy and attenuation in the physical correlation model, the radar signal attenuation caused by vibration energy at that point is analyzed to be 0.35.

[0113] Secondly, in step S1042, radar signal attenuation data for all locations within the potential damage zone are collected to ensure coverage of every grid cell within the coverage area. Then, an arithmetic mean algorithm is used to sum all attenuation data and divide by the number of data points; the resulting average is the characteristic attenuation. This process eliminates the influence of local anomalies and comprehensively reflects the attenuation level of the entire potential damage zone. For example, if a potential damage zone contains 6 grid cells with attenuations of 0.3, 0.35, 0.4, 0.45, 0.38, and 0.42, summing these values ​​gives a total of 2.3, which, when divided by 6, yields a characteristic attenuation of approximately 0.38.

[0114] Next, in step S1043, based on the fatigue characteristics of metallic materials and engineering experience, a metal fatigue threshold is preset. This threshold represents the critical value at which radar signal attenuation begins to appear when the metallic structure begins to show fatigue damage. Then, through real-time data interaction technology, the characteristic attenuation amount is dynamically compared with the metal fatigue threshold using the formula: Calculate the degree of deviation between the two. Then, based on the magnitude of the deviation, convert it into a risk assessment value using a mapping algorithm. The greater the deviation, the higher the risk assessment value, thus intuitively reflecting the potential risk. For example, if the metal fatigue threshold is set to 0.3 and the characteristic decay is 0.38, the calculated deviation is... The risk assessment value of 27 was obtained by transformation through a mapping algorithm.

[0115] Then, in step S1044, using spatial overlay analysis, a temperature anomaly area overlapping with the potential damage area is identified, and the temperature data of all pixels within this area is extracted as spatial distribution data. Next, the highest temperature, average temperature, and ratio of the temperature anomaly area to the total area of ​​the potential damage area are calculated. These three parameters are input into a preset temperature stress relationship function to calculate the temperature influence correction coefficient. Finally, the risk assessment value is multiplied by the correction coefficient to obtain the corrected risk assessment value. For example, if a temperature anomaly area has a highest temperature of 58℃, an average temperature of 54℃, and a ratio of 0.2, the corresponding correction coefficient formula is substituted into this value. The calculated correction factor is 0.66, and the risk assessment value of 27 multiplied by 0.66 yields the corrected risk assessment value of 17.82.

[0116] Finally, through step S1045, a clear conversion rule is set to divide the corrected risk assessment value into several intervals. Each interval corresponds to a risk index; the larger the interval value, the higher the corresponding risk index, thus intuitively reflecting the degree of risk. Then, an interval matching algorithm is used to determine the interval in which the corrected risk assessment value falls, thereby determining the corresponding risk index. For example, if the conversion rule is set as 0-20 corresponding to risk index 1, 21-40 corresponding to risk index 2, and 41 and above corresponding to risk index 3, the corrected risk assessment value of 17.82 falls in the 0-20 interval, corresponding to a risk index of 1. This index represents the degree of risk of weld cracking in the potential damage zone.

[0117] In practical applications, during the monitoring of potential damage zones in an industrial building, the radar signal attenuation caused by vibration energy was analyzed from the response diagram based on a physical correlation model. The attenuation values ​​at various locations within this area were 0.35, 0.4, 0.45, 0.4, and 0.35, respectively. The spatial average of these attenuation values ​​was calculated to be 0.39, yielding a characteristic attenuation of 0.39. A preset metal fatigue threshold of 0.3 was used, and the deviation was calculated through a dynamic comparison mechanism. The risk assessment value is 30. The temperature anomaly area overlapping with the potential damage area is obtained; its spatial distribution data shows a maximum temperature of 59℃, an average temperature of 56℃, and the ratio of the temperature anomaly area to the total area of ​​the potential damage area is 0.25. This data is then input into the temperature stress relationship function: The correction factor is 1.4. Multiplying the risk assessment value of 30 by 1.4 yields the corrected risk assessment value of 42. According to the conversion rule, 42 corresponds to a risk index of 2, meaning that the risk index for weld cracking in this potential damage zone is 2.

[0118] The overall scheme of S104 above calculates the radar signal attenuation caused by vibration energy, combines it with the metal fatigue threshold to obtain a risk assessment value, and then converts it into a risk index after being corrected for the characteristics of the temperature anomaly zone. It comprehensively considers the influence of vibration and temperature on the metal structure, so that the obtained risk index can more accurately reflect the actual risk of weld cracking in the potential damage zone, and provides a reliable basis for subsequent risk assessment.

[0119] S105. Based on the risk level of the risk index, switch the sampling frequency configuration of the vibration monitoring network for the metal components in the potential damage zone, and output the evaluation result including the risk index and the corresponding location coordinates of the potential damage zone.

[0120] Optionally, step S105 may specifically include the following steps:

[0121] S1051. Pre-define multiple consecutively divided risk level intervals and configure a specific target sampling frequency for each risk level interval;

[0122] S1052. Match the risk index with the risk level range to determine the risk level corresponding to the risk index;

[0123] S1053. Query the target sampling frequency corresponding to the risk level, and send a control command to the vibration sensor deployed within the potential damage area to switch the sampling frequency of the vibration sensor to the target sampling frequency.

[0124] S1054. Generate and output structured evaluation results, wherein the evaluation results include the risk index and the corresponding risk level, the location coordinates of the potential damage area and the target sampling frequency.

[0125] In the above scheme, the risk level intervals are pre-defined continuous intervals used to define the levels of risk index, with each interval corresponding to a risk level. The target sampling frequency is the frequency at which the vibration sensor collects signals for each risk level interval; the higher the risk level, the higher the frequency. The risk level is determined based on the risk level interval in which the risk index falls, and is used to indicate the degree of risk of the potential damage area. The control command is a command signal used to adjust the sampling frequency of the vibration sensor. The structured assessment result is a standardized report containing information such as the risk index, risk level, location coordinates of the potential damage area, and target sampling frequency. The location coordinates are numerical values ​​describing the specific location of the potential damage area within the building, used for precise positioning.

[0126] In this embodiment, firstly, step S1051 determines the number of risk levels based on the safety standards for building metal structures, past damage cases, and the performance of the monitoring system. Generally, these are divided into three levels: low, medium, and high. Then, an interval division algorithm is used to divide the possible range of risk index values ​​into multiple continuous and non-overlapping intervals, each corresponding to a risk level. Next, based on the monitoring accuracy requirements of each risk level, a target sampling frequency is configured for each level, adopting a positive correlation between level and frequency. That is, the higher the risk level, the higher the target sampling frequency, ensuring that more dense data can be collected in high-risk areas. For example, risk indices 1-2 are classified as low-risk (Level 1), corresponding to a target sampling frequency of 10 times / second; 3-4 are classified as medium-risk (Level 2), corresponding to 20 times / second; and 5 and above are classified as high-risk (Level 3), corresponding to 30 times / second. These configurations are stored in the system's parameter database.

[0127] Secondly, in step S1052, the risk index of the potential damage area is extracted from the previous calculation results. Then, an interval matching algorithm is called to compare the risk index with the risk level intervals preset in step S1051 one by one to determine which interval the risk index falls into. The algorithm first monitors whether the risk index is greater than or equal to the lower limit of the interval and less than or equal to the upper limit of the interval. If it falls within the range of a certain interval, the risk level corresponding to that interval is the current risk level. For example, if the risk index of the potential damage area is 2, and the comparison shows that it falls within the interval of 1-2, the corresponding risk level is 1.

[0128] Next, in step S1053, based on the risk level determined in step S1052, the system queries the system parameter database for the target sampling frequency corresponding to that risk level. Then, using wireless communication technology, a control command is sent to all vibration sensors deployed within the potential damage area. This command includes the specific value of the target sampling frequency and its effective time. After receiving the command, the sensor's internal frequency adjustment module adjusts the operating frequency of its sampling circuit according to the command, thereby switching the sampling frequency to the target sampling frequency. For example, if the target sampling frequency for risk level 1 is 10 times / second, after the system sends the command to the sensors in that area, the sensors will adjust their sampling frequency from 5 times / second to 10 times / second.

[0129] Finally, step S1054 collects information such as the risk index, risk level, location coordinates of the potential damage area, and the target sampling frequency set by the vibration sensor. Then, using data formatting technology, this information is integrated into structured data according to a preset template, with the location coordinates specified to a precise range. Finally, the structured assessment results are sent to the monitoring terminal's display screen for display through the system's output module, and simultaneously stored in the database for later retrieval, allowing staff to intuitively understand the risk situation and monitoring configuration of the potential damage area. For example, the generated assessment results include a risk index of 2, a risk level of 1, location coordinates (X: 10-15 meters, Y: 8-12 meters, Z: 3-5 meters), and a target sampling frequency of 10 times / second, all of which will be clearly displayed on the monitoring screen.

[0130] In a practical application, the monitoring system for an industrial building pre-sets risk level ranges and corresponding target sampling frequencies. Risk indices 1-2 correspond to low risk level 1 (10 samplings / second), 3-4 to medium risk level 2 (20 samplings / second), and 5 and above to high risk level 3 (30 samplings / second). These configurations are stored in the system parameter database. A potential damage area in the building has a risk index of 3. After comparing it using an interval matching algorithm, it is determined to be in the 3-4 range, corresponding to a risk level of 2. Based on risk level 2, the database retrieves a target sampling frequency of 20 samplings / second. The system then sends a control command via Wi-Fi to the vibration sensors in the potential damage area, adjusting the sensor's original sampling frequency from 10 samplings / second to 20 samplings / second. Finally, relevant information is collected, and data formatting technology is used to generate a structured assessment result containing risk index 3, risk level 2, location coordinates (X: 12-18 meters, Y: 6-10 meters, Z: 2-4 meters), and the target sampling frequency of 20 samplings / second. This result is then sent to the monitoring terminal display for display and stored in the database.

[0131] The overall solution described in S105 determines the risk level based on the risk index and then adjusts the sampling frequency of the vibration monitoring network. This enables focused monitoring of high-risk areas while outputting assessment results containing key information. This not only allows for more efficient use of monitoring resources and timely understanding of risk changes in potential damage areas, but also provides staff with clear and comprehensive assessment information, helping them make targeted maintenance decisions and ensuring the safe operation of the building.

[0132] The following is a complete example for steps 101-105, such as Figure 3 As shown, during the monitoring of industrial building E, staff used radar scanners and infrared thermometers to acquire high-precision radar and infrared images. In the radar images, a first threshold of 55 was set, and areas such as metal pipe supports with reflected signal intensity exceeding this value were identified as metal structure areas. Within these areas, the variation in reflected signal intensity for each part was calculated, and a second threshold of 12 was set; pipe supports with a variation greater than this value were extracted as preliminary anomaly areas. Simultaneously, temperature anomaly areas were identified from the infrared images corresponding to the metal pipe supports. With an ambient temperature of 28°C and a third threshold of 48°C, the interface portion of the pipe support with a temperature of 55°C was identified as a temperature anomaly area.

[0133] On the metal pipe support of building E, a monitoring network of six piezoelectric vibration sensors were installed at 2.5-meter intervals. When the ventilation equipment in the plant was running, the sensors collected vibration signals and transmitted them to the terminal. After signal preprocessing, the energy values ​​of each sensor location were calculated using Fourier transform: 28, 35, 42, 39, 33, and 40. The intensity distribution was generated by coordinate interpolation. This distribution was then transformed into the same coordinate system as the initial anomaly area (a 4-meter-long pipe support) and the temperature anomaly area (a 1.5-square-meter interface area on the support). Using GIS overlay analysis and setting a fourth threshold of 35, a 2.5-meter-long overlapping area was selected as the potential damage area. Two sensors within this area were located, and their signals were subjected to spectral analysis to obtain a power spectrum peak frequency of 45 Hz. Eight seconds of energy data were collected, and linear fitting yielded a decay rate of 4 from 50 Hz to 18 Hz. These two parameters were used as the vibration characteristics of this area.

[0134] In monitoring potential damage zones, a physical correlation model is established experimentally, with the following formula: Let k be the radar signal reflection attenuation coefficient, f be the dominant vibration frequency, and v be the vibration energy attenuation rate. Substituting the dominant vibration frequency of 45 Hz and the energy attenuation rate of 4 Hz into the model, the radar signal reflection attenuation coefficient k is obtained as 0.49, generating the predicted reflection attenuation distribution. Simultaneously, synchronous radar images are acquired, and the actual scattering intensity of one grid cell is calculated to be 75, while that of another is 70, forming the actual scattering intensity distribution. After spatiotemporal registration of the two, the values ​​are 0.0065 and 0.007 respectively, generating the response relationship diagram.

[0135] Based on the physical correlation model, the radar signal attenuation caused by vibration energy in the potential damage zone of industrial building E was analyzed from the response relationship diagram. The attenuation values ​​at various locations within this zone were 0.32, 0.38, 0.43, 0.39, and 0.34, respectively. The spatial average value was calculated to be 0.372, yielding a characteristic attenuation of 0.37. A preset metal fatigue threshold of 0.28 was used, and the deviation was calculated through a dynamic comparison mechanism. The risk assessment value is 32. The temperature anomaly zone overlapping with the potential damage area is obtained; its spatial distribution data shows a maximum temperature of 55℃ and an average temperature of 52℃. The ratio of the area of ​​the temperature anomaly zone to the total area of ​​the potential damage area is 0.2. This data is input into the temperature stress relationship function: correction coefficient = 0.012 × maximum temperature + 0.012 × average temperature + 1.2 × ratio, resulting in a correction coefficient of 1.524. Multiplying the risk assessment value 32 by 1.524 yields a corrected risk assessment value of approximately 48.77. According to the conversion rules, 48.77 corresponds to a risk index of 3, meaning the risk index for weld cracking in this potential damage area is 3.

[0136] In the monitoring system, risk level ranges and corresponding target sampling frequencies are pre-defined. Risk indices 1-2 correspond to low risk level 1 (8 times / second), 3-4 to medium risk level 2 (18 times / second), and 5 and above to high risk level 3 (28 times / second). These configurations are stored in the system parameter database. The risk index of the potential damage area of ​​the building is 3. After comparing with the interval matching algorithm, it is determined to be in the 3-4 range, corresponding to a risk level of 2. Based on risk level 2, the target sampling frequency of 18 times / second is found in the database. The system sends a control command via Bluetooth to the vibration sensor in the potential damage area to adjust the sensor's original sampling frequency of 8 times / second to 18 times / second. Finally, relevant information is collected, and data formatting technology is used to generate a structured evaluation result containing risk index 3, risk level 2, location coordinates (X: 8-14 meters, Y: 5-9 meters, Z: 1-3 meters), and target sampling frequency of 18 times / second. This result is sent to the monitoring terminal display screen for display and stored in the database.

[0137] Figure 4This application provides a schematic diagram of a specific implementation of a real-time building risk assessment system based on multimodal remote sensing imagery, as illustrated in the following embodiment. Figure 4 The system may include:

[0138] The acquisition module 41 is used to acquire high-precision radar images and infrared images, identify the building metal structure area from the high-precision radar images and extract the preliminary abnormal area reflecting structural anomalies, and identify the temperature abnormal area within the metal structure area from the infrared images.

[0139] Extraction module 42 is used to collect specific vibration signals when the vibration monitoring network equipment is deployed in the building, and to determine potential damage areas and extract vibration features associated with the potential damage areas by combining the intensity distribution of the specific vibration signals, the spatial information of the preliminary abnormal area and the temperature abnormal area.

[0140] The matching module 43 is used to establish a physical correlation model describing the changes in the metal microstructure and the attenuation of radar signal reflection under vibration stress, and to use spatiotemporal matching technology to match the vibration characteristics with the scattering characteristics of high-precision radar images to generate a response relationship diagram.

[0141] The generation module 44 is used to calculate the radar signal attenuation caused by vibration energy in the response relationship diagram based on the physical correlation model, dynamically compare and fuse the radar signal attenuation with a preset metal fatigue threshold, and correct the fusion result according to the spatial distribution characteristics of the temperature anomaly zone to generate a risk index for weld cracking in the potential damage zone.

[0142] The output module 45 is used to switch the sampling frequency configuration of the vibration monitoring network for the metal components in the potential damage zone based on the risk level of the risk index, and output the evaluation result including the risk index and the corresponding location coordinates of the potential damage zone.

[0143] The real-time building risk assessment system based on multimodal remote sensing imagery in this application is used to implement the aforementioned real-time building risk assessment method based on multimodal remote sensing imagery. Therefore, the specific implementation of the real-time building risk assessment system based on multimodal remote sensing imagery can be found in the embodiment section of the real-time building risk assessment method based on multimodal remote sensing imagery above. The specific implementation can be referred to the description of the corresponding embodiments, which will not be repeated here.

[0144] This application also provides an electronic device, comprising: a memory for storing a computer program; and a processor for executing the computer program to implement the steps of any of the above-described methods for real-time building risk assessment based on multimodal remote sensing imagery.

[0145] This application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of any of the above-described methods for real-time building risk assessment based on multimodal remote sensing imagery.

[0146] In one exemplary embodiment, the aforementioned computer-readable storage medium may include, but is not limited to, various media capable of storing computer programs, such as USB flash drives, read-only memory, random access memory, portable hard drives, magnetic disks, or optical disks.

[0147] Embodiments of the present invention also provide a computer program product, which includes a computer program that, when executed by a processor, implements the steps in any of the embodiments of the real-time building risk assessment method based on multimodal remote sensing imagery.

[0148] Those skilled in the art will further recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0149] The foregoing has provided a detailed description of a method and system for real-time building risk assessment based on multimodal remote sensing imagery, as provided in this application. Specific examples have been used to illustrate the principles and implementation methods of this application. The descriptions of the embodiments above are merely for the purpose of helping to understand the method and its core ideas. It should be noted that those skilled in the art can make various improvements and modifications to this application without departing from its principles, and these improvements and modifications also fall within the protection scope of this application.

Claims

1. A method for real-time building risk assessment based on multimodal remote sensing imagery, characterized in that, include: Acquire high-precision radar and infrared images, identify building metal structure areas from the high-precision radar images and extract preliminary abnormal areas reflecting structural anomalies, and identify temperature anomaly areas within the metal structure areas from the infrared images. By deploying a vibration monitoring network within a building to collect specific vibration signals during operation, and combining the intensity distribution of the specific vibration signals, the spatial information of the preliminary abnormal area and the temperature abnormal area, potential damage areas are identified and vibration characteristics associated with the potential damage areas are extracted. Under vibration stress, a physical correlation model describing the relationship between changes in the microstructure of metal and the attenuation of radar signal reflection is established. Spatiotemporal matching technology is used to match the vibration characteristics with the scattering characteristics of high-precision radar images to generate a response relationship diagram. Based on the physical correlation model, the radar signal attenuation caused by vibration energy in the response relationship diagram is calculated. The radar signal attenuation is dynamically compared and fused with a preset metal fatigue threshold. At the same time, the fusion result is corrected according to the spatial distribution characteristics of the temperature anomaly zone to generate a risk index for weld cracking in the potential damage zone. Based on the risk level of the risk index, the sampling frequency configuration of the vibration monitoring network for the metal components in the potential damage zone is switched, and an evaluation result containing the risk index and the corresponding location coordinates of the potential damage zone is output. The generation of the response relationship diagram includes: Based on the physical mechanism by which changes in the microstructure of metal under vibration stress affect radar signal reflection capability, a mapping relationship is established between the dominant vibration frequency and vibration energy attenuation rate in the vibration characteristics and the radar signal reflection attenuation coefficient, serving as a physical correlation model. The dominant vibration frequency and vibration energy attenuation rate in the vibration characteristics are input into the physical correlation model to generate a predicted reflection attenuation distribution of the radar signal. High-precision radar images acquired synchronously with the vibration characteristics are obtained, and the actual radar echo signal intensity values ​​at each spatial location within the building's metal structure area are extracted from the high-precision radar images to form the actual scattering intensity distribution of the radar signal. The predicted reflection attenuation distribution and the actual scattering intensity distribution are spatiotemporally registered and numerically correlated to generate a response relationship diagram reflecting the quantitative relationship between the predicted attenuation and the actual scattering intensity, using spatial location as coordinates. The risk index for weld cracking in the potential damage zone includes: Based on the physical correlation model, the radar signal attenuation directly caused by vibration energy is extracted from the response relationship diagram; the spatial average value of the radar signal attenuation in each potential damage zone is calculated to obtain the characteristic attenuation; a dynamic comparison mechanism between the characteristic attenuation and a preset metal fatigue threshold is established, and the deviation between the characteristic attenuation and the metal fatigue threshold is calculated in real time to obtain a risk assessment value; spatial distribution data of temperature anomaly zones that spatially overlap with the potential damage zones are acquired, a temperature influence correction coefficient is calculated based on the spatial distribution data, and the risk assessment value is corrected according to the temperature influence correction coefficient; the corrected risk assessment value is converted into a risk index characterizing weld cracking in the potential damage zone.

2. The method according to claim 1, characterized in that, The step of acquiring spatial distribution data of temperature anomaly areas that spatially overlap with the potential damage area, calculating a temperature influence correction coefficient based on the spatial distribution data, and correcting the risk assessment value according to the temperature influence correction coefficient includes: Obtain a temperature anomaly area that overlaps with the spatial range of the potential damage area, and extract the temperature data of all pixels within the temperature anomaly area as spatial distribution data; Based on the spatial distribution data, calculate the highest temperature value, the average temperature value, and the ratio of the area of ​​the temperature anomaly region to the total area of ​​the potential damage region in the temperature anomaly region. The highest temperature value, average temperature value, and proportional value are input into a preset temperature stress relationship function to calculate the temperature influence correction coefficient. The preliminary risk assessment value is multiplied by the temperature influence correction factor to obtain the corrected risk assessment value.

3. The method according to claim 1, characterized in that, The specific vibration signals acquired by the vibration monitoring network deployed within the building, combined with the intensity distribution of the specific vibration signals, the spatial information of the preliminary abnormal area and the temperature abnormal area, determine the potential damage area and extract vibration features associated with the potential damage area, including: A vibration monitoring network consisting of multiple vibration sensors is deployed within the building, and the vibration monitoring network is operated to collect specific vibration signals generated during the operation of the equipment; Based on the specific vibration signal, an intensity distribution reflecting the magnitude of vibration energy is generated; The intensity distribution, the spatial information of the preliminary abnormal area, and the spatial information of the temperature abnormal area are superimposed, and the area in the intensity distribution where the vibration energy is higher than the fourth set threshold and spatially overlaps with the preliminary abnormal area and the temperature abnormal area is selected as the potential damage area. The vibration sensor is located within the potential damage area, and the vibration dominant frequency and vibration energy decay rate are extracted from the specific vibration signal collected by the vibration sensor as vibration characteristics.

4. The method according to claim 1, characterized in that, Based on the risk level of the risk index, the sampling frequency configuration of the vibration monitoring network for metal components within the potential damage zone is switched, and an evaluation result including the risk index and the corresponding coordinates of the potential damage zone is output, including: Multiple consecutively divided risk level intervals are pre-defined, and a specific target sampling frequency is configured for each risk level interval; The risk index is matched with the risk level range to determine the risk level corresponding to the risk index; The target sampling frequency corresponding to the risk level is queried, and a control command is sent to the vibration sensor deployed within the potential damage area to switch the sampling frequency of the vibration sensor to the target sampling frequency. Generate and output structured assessment results, which include the risk index and corresponding risk level, the location coordinates of the potential damage area, and the target sampling frequency.

5. The method according to claim 1, characterized in that, The process of acquiring high-precision radar and infrared images, identifying building metal structure areas from the high-precision radar images and extracting preliminary anomaly areas reflecting structural abnormalities, and identifying temperature anomaly areas within the metal structure areas from the infrared images includes: High-precision radar images of the target building are acquired using radar sensors, while infrared images of the same target building are acquired using infrared thermal imagers. The area with a reflected signal intensity higher than a first set threshold is identified from the high-precision radar image as the building metal structure area; Within the building's metal structure area, regions where the intensity variation of reflected signals exceeds a second preset threshold are identified as preliminary anomaly zones. The areas in the building's metal structure zone identified from the infrared image whose temperature values ​​are higher than a third preset threshold are designated as temperature anomaly zones.

6. A real-time building risk assessment system based on multimodal remote sensing imagery, characterized in that, include: The acquisition module is used to acquire high-precision radar images and infrared images, identify the building metal structure area from the high-precision radar images and extract the preliminary abnormal area reflecting structural anomalies, and identify the temperature abnormal area within the metal structure area from the infrared images. The extraction module is used to collect specific vibration signals when the vibration monitoring network is deployed in the building, and to determine potential damage areas and extract vibration features associated with the potential damage areas by combining the intensity distribution of the specific vibration signals, the spatial information of the preliminary abnormal area and the temperature abnormal area. The matching module is used to establish a physical correlation model describing the changes in the microstructure of metal and the attenuation of radar signal reflection under vibration stress, and to use spatiotemporal matching technology to match the vibration characteristics with the scattering characteristics of high-precision radar images to generate a response relationship diagram. The generation module is used to calculate the radar signal attenuation caused by vibration energy in the response relationship diagram based on the physical correlation model, dynamically compare and fuse the radar signal attenuation with a preset metal fatigue threshold, and correct the fusion result according to the spatial distribution characteristics of the temperature anomaly zone to generate a risk index for weld cracking in the potential damage zone. The output module is used to switch the sampling frequency configuration of the vibration monitoring network for the metal components in the potential damage zone based on the risk level of the risk index, and output the evaluation result including the risk index and the corresponding location coordinates of the potential damage zone. The generation of the response relationship diagram includes: Based on the physical mechanism by which changes in the microstructure of metal under vibration stress affect radar signal reflection capability, a mapping relationship is established between the dominant vibration frequency and vibration energy attenuation rate in the vibration characteristics and the radar signal reflection attenuation coefficient, serving as a physical correlation model. The dominant vibration frequency and vibration energy attenuation rate in the vibration characteristics are input into the physical correlation model to generate a predicted reflection attenuation distribution of the radar signal. High-precision radar images acquired synchronously with the vibration characteristics are obtained, and the actual radar echo signal intensity values ​​at each spatial location within the building's metal structure area are extracted from the high-precision radar images to form the actual scattering intensity distribution of the radar signal. The predicted reflection attenuation distribution and the actual scattering intensity distribution are spatiotemporally registered and numerically correlated to generate a response relationship diagram reflecting the quantitative relationship between the predicted attenuation and the actual scattering intensity, using spatial location as coordinates. The risk index for weld cracking in the potential damage zone includes: Based on the physical correlation model, the radar signal attenuation directly caused by vibration energy is extracted from the response relationship diagram; the spatial average value of the radar signal attenuation in each potential damage zone is calculated to obtain the characteristic attenuation; a dynamic comparison mechanism between the characteristic attenuation and a preset metal fatigue threshold is established, and the deviation between the characteristic attenuation and the metal fatigue threshold is calculated in real time to obtain a risk assessment value; spatial distribution data of temperature anomaly zones that spatially overlap with the potential damage zones are acquired, a temperature influence correction coefficient is calculated based on the spatial distribution data, and the risk assessment value is corrected according to the temperature influence correction coefficient; the corrected risk assessment value is converted into a risk index characterizing weld cracking in the potential damage zone.

7. An electronic device, characterized in that, include: Memory, used to store computer programs; A processor, configured to execute the computer program to implement the steps of the method for real-time building risk assessment based on multimodal remote sensing imagery as described in any one of claims 1 to 5.

8. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, enables the implementation of the real-time building risk assessment method based on multimodal remote sensing imagery as described in any one of claims 1 to 5.

Images