Building deformation monitoring method and system based on space-borne InSAR data

Abstract

The present application relates to the technical field of building monitoring, and in particular to a building deformation monitoring method and system based on space-borne InSAR data. The method comprises: periodically collecting space-borne InSAR data of a target building; performing time series difference processing on the collected data to obtain a plurality of sets of tilt monitoring data, crack monitoring data and settlement monitoring data of the target building, the collection of all the monitoring data constituting a deformation monitoring data set, and coupling each set of data to obtain a building deformation value; constructing a deformation state classification model using a clustering algorithm, pre-training the model based on the building deformation value, and the output of the model being a deformation state classification result; inputting the building deformation value corresponding to the current period into the pre-trained model to output a current deformation state classification result, and issuing a deformation warning signal, which greatly simplifies the deformation monitoring processing flow, ensures the timeliness of building deformation monitoring and warning, and effectively improves the accuracy of structure state evaluation.

Description

Technical Field

[0001] This invention relates to the field of building monitoring technology, and specifically to a method and system for monitoring building deformation based on spaceborne InSAR data. Background Technology

[0002] Spaceborne InSAR, or interferometric synthetic aperture radar, is a radar technology used for geodesy and remote sensing. InSAR technology uses two or more synthetic aperture radar (SAR) images, utilizing the phase differences of the waves returning to the satellite to calculate the topography, landforms, and minute surface changes of a target area. This technology can potentially measure millimeter-level deformations spanning from days to years. Unlike visible or infrared light, radar waves can penetrate most clouds, fog, and smoke to observe surface objects, and are equally effective in darkness. Therefore, due to its wide-area coverage and excellent monitoring capabilities even in adverse weather conditions at night, InSAR has also found widespread application in the construction industry.

[0003] For example, Chinese patent CN118330627A provides a slope aperture radar displacement monitoring and early warning system based on the sparrow search algorithm, achieving intelligent monitoring and early warning over all times, all weather conditions, and a large area. However, this existing technology still has the following technical problems:

[0004] 1. Existing technologies use differential interferometry (D-InSAR) to process the phase difference of SAR images. However, it is assumed that the atmospheric delay phase and noise phase are equal in two measurements. However, in practical applications, atmospheric conditions may change over time (such as humidity and temperature fluctuations), and noise may not be completely canceled due to environmental interference (such as dust in mining areas and equipment vibration), resulting in phase difference calculation errors and affecting the accuracy of displacement monitoring.

[0005] 2. Existing technologies require multiple steps to process SAR images, such as image registration, phase unwrapping, and atmospheric phase compensation. In particular, they rely on the least norm method to unwrap the phase and the least squares method to calculate the deformation. Furthermore, the sparrow search algorithm requires complex parameter adjustments, resulting in high computational complexity and a large amount of computation. Consequently, the building deformation monitoring system experiences response delays, making it difficult to meet the requirements for timely early warning.

[0006] 3. The early warning logic of the existing technology is based on whether the stability coefficient output by the sparrow search algorithm exceeds the reference range. However, relying solely on the analysis of a single stability coefficient and the early warning mechanism with a fixed parameter range to judge the risk will undoubtedly ignore the complex working conditions of other deformation dimensions, resulting in low accuracy of structural state assessment. Summary of the Invention

[0007] To address the aforementioned technical issues, this invention proposes a building deformation monitoring method and system based on spaceborne InSAR data, aiming to significantly simplify the monitoring and processing procedures and effectively improve the accuracy of structural condition assessment while ensuring the timeliness of building deformation monitoring and early warning.

[0008] In one aspect, this application provides a method for monitoring building deformation based on spaceborne InSAR data, including the following steps:

[0009] Periodically collect spaceborne InSAR data of the target building;

[0010] Temporal differential processing was performed on the spaceborne InSAR data to obtain several sets of tilt monitoring data, crack monitoring data and settlement monitoring data of the target building. The set of all tilt monitoring data, crack monitoring data and settlement monitoring data constitutes the deformation monitoring dataset of the target building.

[0011] Couple each set of tilt monitoring data, crack monitoring data and settlement monitoring data in the deformation monitoring dataset, and record the coupling result as the building deformation value;

[0012] A clustering algorithm is used to construct a deformation state classification model, and the deformation state classification model is pre-trained based on the building deformation value. The output of the model is the deformation state classification result, and the deformation state categories include qualified, abnormal and severe.

[0013] Input the building deformation value corresponding to the current period into the pre-trained deformation state classification model, and output the current deformation state classification result.

[0014] A deformation warning signal is issued based on the current deformation classification results.

[0015] In some embodiments, time-series differential processing is performed on spaceborne InSAR data to obtain several sets of tilt monitoring data, crack monitoring data, and settlement monitoring data of the target building, including:

[0016] The differences in tilt, crack volume, and settlement between adjacent periods of spaceborne InSAR data are calculated. Each period is considered as a set, resulting in several sets of tilt monitoring data, crack monitoring data, and settlement monitoring data for the target building.

[0017] In some embodiments, a clustering algorithm is used to construct a deformation state classification model, and the model is pre-trained based on the building deformation values. The output of the model is the deformation state classification result, including:

[0018] Step 201: Construct a deformation state classification model and pre-determine three primary cluster centers K1, K2, and K3, which represent qualified deformation state, abnormal deformation state, and severe deformation state, respectively.

[0019] Step 202: Randomly select three data points from the building deformation values ​​as the primary cluster centers K1, K2, and K3, respectively;

[0020] Step 203: Calculate the distance from each data point in the building deformation value to the center of each primary cluster, and assign the data points to the nearest primary cluster to form three intermediate clusters;

[0021] Step 204: Calculate the mean of all data points in each intermediate cluster, and use it as the new primary cluster center;

[0022] Step 205: Repeat steps 203-204 until the preset convergence condition is met, and obtain the pre-trained deformation state classification model.

[0023] In some embodiments, step 204, calculating the mean of all data points within each intermediate cluster as the new primary cluster center, includes:

[0024] When there are no data points in the intermediate cluster during the iteration process, the data point farthest from the current primary cluster center is selected as the new primary cluster center.

[0025] In some embodiments, pre-training a deformation state classification model based on building deformation values ​​includes:

[0026] Store the building deformation values ​​for the most recent Q periods. When a new building deformation value is added for R periods, the deformation state classification model is re-pre-trained based on the latest Q+R building deformation values.

[0027] In some embodiments, issuing a deformation warning signal based on the current deformation classification result further includes:

[0028] The building stability value was calculated based on the deformation monitoring dataset;

[0029] Based on the current deformation classification results and building stability values, a deformation warning signal is issued.

[0030] In some embodiments, the deformation monitoring dataset includes the settlement and crack volume of data points. Based on the deformation monitoring dataset, the building stability value is calculated, including:

[0031] Based on the settlement amount of the data points, the settlement rate is calculated, and the tilt reference value is calculated using the two-point difference method.

[0032] The settlement rate, tilt reference value, and crack volume were normalized by using the maximum allowable value constraint method, and then weighted summation was performed to obtain the building stability value.

[0033] In some embodiments, based on the current deformation classification result and the building stability value, a deformation warning signal is issued, including:

[0034] The preset threshold range for building stability values ​​is (E1, E2).

[0035] If the current deformation classification result is qualified and the building stability value is greater than or equal to E2, a good deformation warning signal will be issued;

[0036] If the current deformation classification result is qualified and the building stability value is within the threshold range (E1, E2), a relatively stable deformation warning signal will be issued.

[0037] If the current deformation classification result is abnormal and the building stability value is within the threshold range (E1, E2), a poor deformation warning signal will be issued.

[0038] If the current deformation classification result is abnormal and the building stability value is less than or equal to E1, an unstable deformation warning signal will be issued.

[0039] If the current deformation classification result is severe, an extremely poor deformation warning signal will be issued under any building stability value.

[0040] Secondly, this application provides a building deformation monitoring system based on spaceborne InSAR data, including:

[0041] The data acquisition module is used to periodically acquire spaceborne InSAR data of the target building;

[0042] The data preprocessing module is used to perform time-series differential processing on the spaceborne InSAR data to obtain several sets of tilt monitoring data, crack monitoring data and settlement monitoring data of the target building. The collection of all tilt monitoring data, crack monitoring data and settlement monitoring data constitutes the deformation monitoring dataset of the target building.

[0043] The deformation value calculation module is used to couple each set of tilt monitoring data, crack monitoring data and settlement monitoring data in the deformation monitoring dataset, and record the coupling result as the building deformation value.

[0044] The deformation state classification module is used to construct a deformation state classification model and pre-train the deformation state classification model based on the building deformation value. The output of the model is the deformation state classification result. The categories of deformation state include qualified, abnormal and severe. The building deformation value corresponding to the current period is input into the pre-trained deformation state classification model, and the current deformation state classification result is output.

[0045] The early warning module is used to issue deformation early warning signals based on the current deformation classification results.

[0046] In some embodiments, the system further includes:

[0047] The stability value calculation module is used to calculate the building stability value based on the deformation monitoring dataset;

[0048] The early warning module is used to issue a deformation early warning signal based on the current deformation classification result and the building stability value.

[0049] The beneficial technical effects of the present invention include at least the following:

[0050] 1. A building deformation monitoring method and system based on spaceborne InSAR data is adopted. Through the collaborative innovation of "temporal differential processing - multidimensional data coupling - data-driven dynamic clustering and classification", a closed-loop optimized monitoring mechanism is formed. Specifically: First, a dynamic baseline is constructed through a dual differential processing mechanism (spatial phase difference + temporal difference). The adjacent period data is used as a moving reference frame, which avoids the cumulative error of traditional fixed time reference point monitoring. At the same time, it solves the error problem caused by assuming that the atmospheric state remains unchanged in the existing technology. Cleaner deformation monitoring data can be obtained directly without complex atmospheric phase compensation processing. Secondly, by weighted coupling of three key deformation monitoring data—tilt, cracks, and settlement—building deformation values ​​are generated, forming a multi-dimensional integrated deformation index. Compared to the single stability coefficient in existing technologies, this can reduce the false alarm rate of early warnings to a certain extent. Finally, the dynamic deformation state classification model built based on clustering algorithms avoids the complex computational load of sparrow search algorithms in existing technologies, significantly improving the timeliness of early warnings. These three elements work together to form a technical chain of "error suppression → comprehensive quantification → dynamic classification," achieving a synergistic breakthrough in the accuracy of building deformation monitoring and data processing efficiency, resulting in a beneficial effect of "1+1>2."

[0051] 2. Existing early warning logic is based on whether the stability coefficient output by the sparrow search algorithm exceeds the reference range. However, relying solely on the analysis of a single stability coefficient and a fixed parameter range early warning mechanism to judge risk will undoubtedly ignore the complex working conditions of other deformation dimensions, resulting in low accuracy of structural state assessment. Therefore, this application proposes a "qualitative-quantitative dual-modal collaborative decision-making" monitoring and early warning method to achieve full-chain optimization of building deformation monitoring from data acquisition to risk decision-making. Specifically: First, atmospheric delay and noise interference are suppressed through time-series differential processing to generate high signal-to-noise ratio data. Then, a deformation state classification model constructed using a clustering algorithm provides a qualitative judgment of the deformation state. The system provides a quantitative assessment by combining the calculation of building stability values ​​with the classification of building deformation into qualified, abnormal, and severe conditions. The two work together to achieve dual verification of qualitative and quantitative results. Finally, it triggers a five-level early warning signal (good, relatively stable, poor, unstable, and extremely poor). This effectively solves the problems of atmospheric interference, high computational complexity, single early warning, and rigidity that rely solely on fixed thresholds in existing technologies. Furthermore, it adapts to different building types and environmental changes through a dynamic threshold adjustment mechanism (adaptively updating the stability value threshold range based on the cluster mean and standard deviation). While ensuring the timeliness of building deformation monitoring and early warning, it effectively improves the accuracy and reliability of the overall building deformation status assessment and effectively reduces the false alarm rate.

[0052] Other features and advantages of the present invention will be disclosed in detail in the following detailed description and accompanying drawings. Attached Figure Description

[0053] The invention will be further described below with reference to the accompanying drawings:

[0054] Figure 1 This is a flowchart of a building deformation monitoring method based on spaceborne InSAR data, according to an embodiment of the present invention.

[0055] Figure 2 This is a flowchart of a building deformation monitoring method based on spaceborne InSAR data, according to an embodiment of the present invention.

[0056] Figure 3 This is a schematic diagram of the building deformation monitoring system based on spaceborne InSAR data, according to an embodiment of the present invention. Detailed Implementation

[0057] The technical solutions of the embodiments of the present invention will be explained and described below with reference to the accompanying drawings. However, the following embodiments are only preferred embodiments of the present invention and not all of them. Other embodiments obtained by those skilled in the art based on the embodiments in the implementation methods without creative effort are all within the protection scope of the present invention.

[0058] In the following description, terms such as “inner,” “outer,” “upper,” “lower,” “left,” and “right” are used only to indicate orientation or positional relationship for the convenience of describing the embodiments and simplifying the description, and are not intended to indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the present invention.

[0059] Please see the appendix Figure 1 , Figure 1 A flowchart illustrating a building deformation monitoring method based on spaceborne InSAR data, provided in one embodiment of this specification, is shown.

[0060] like Figure 1 As shown, this method for monitoring building deformation based on spaceborne InSAR data may include at least the following steps:

[0061] Step 101: Periodically collect spaceborne InSAR data of the target building.

[0062] The revisit period of a single SAR satellite is typically 12-24 days (e.g., Sentinel-1 is 12 days), which makes it difficult to capture sudden building deformations. Therefore, multiple SAR satellites (e.g., the COSMO-SkyMed constellation) can be deployed to shorten the revisit period and provide a data foundation for building safety monitoring with hourly to dayly responses. This embodiment will not elaborate on this aspect.

[0063] Step 102: Perform time-series differential processing on the spaceborne InSAR data to obtain several sets of tilt monitoring data, crack monitoring data and settlement monitoring data of the target building. The set of all tilt monitoring data, crack monitoring data and settlement monitoring data constitutes the deformation monitoring dataset of the target building.

[0064] Specifically, in this embodiment, time-series differential processing is performed on the spaceborne InSAR data to obtain several sets of tilt monitoring data, crack monitoring data, and settlement monitoring data of the target building, including:

[0065] The differences in tilt, crack volume, and settlement between adjacent periods of spaceborne InSAR data are calculated. Each period is considered as a set, resulting in several sets of tilt monitoring data, crack monitoring data, and settlement monitoring data for the target building.

[0066] For example, the settlement amount identified by the satellite-borne InSAR data of the target building in the m-th cycle is subtracted from the settlement amount of the target building in the (m-1)-th cycle to obtain settlement monitoring data; the tilt angle identified by the satellite-borne InSAR data of the target building in the m-th cycle is subtracted from the tilt angle of the target building in the (m-1)-th cycle to obtain tilt monitoring data; and the crack volume identified by the satellite-borne InSAR data of the target building in the m-th cycle is subtracted from the crack volume of the target building in the (m-1)-th cycle to obtain crack monitoring data.

[0067] The methods for identifying settlement, tilt, and crack volume based on spaceborne InSAR data can all refer to existing technologies. For example, settlement data is the most direct identification result of spaceborne InSAR data, and the settlement data preprocessing method in Chinese Patent CN117723016A "A method for monitoring building settlement based on PS-InSAR technology and level instrument measurement" can be referenced. Similarly, the identification of tilt data can refer to the tilt data preprocessing method in Chinese Patent CN106772377A "A method for monitoring building deformation based on InSAR", etc. This embodiment will not elaborate further.

[0068] In this embodiment, the deformation monitoring dataset of the target building can include both change monitoring data obtained by time-series differential processing of spaceborne InSAR data and raw monitoring data (such as settlement, tilt, and crack volume obtained by data point identification) obtained from spaceborne InSAR data in each cycle.

[0069] Existing technologies employ differential interferometry (D-InSAR) to process SAR image phase differences, but they assume that atmospheric delay phase and noise phase are equal in two measurements. However, in practical applications, atmospheric conditions may change over time (e.g., humidity and temperature fluctuations), and noise may not be completely canceled out due to environmental interference (e.g., dust in mining areas, equipment vibration), leading to phase difference calculation errors and affecting displacement monitoring accuracy. To address this, this embodiment proposes a "dual-difference" processing mechanism, which includes both traditional InSAR spatial phase difference and superimposed temporal series difference. It uses adjacent periods of spaceborne InSAR data as a dynamic baseline to establish a moving reference frame for building conditions, avoiding the accumulated errors of traditional fixed-time reference point monitoring. Simultaneously, temporal difference processing cancels out common errors (e.g., atmospheric delay and orbital errors). Furthermore, because atmospheric conditions are similar in adjacent periods, the error problem caused by the assumption of constant atmospheric conditions in existing technologies is actively resolved. Cleaner deformation monitoring data can be obtained directly without complex atmospheric phase compensation processing, improving displacement monitoring accuracy and providing a reliable data foundation for subsequent analysis.

[0070] Step 103: Couple each set of tilt monitoring data, crack monitoring data and settlement monitoring data in the deformation monitoring dataset, and record the coupling result as the building deformation value;

[0071] Specifically, in this embodiment, the coupling of each set of tilt monitoring data, crack monitoring data, and settlement monitoring data is achieved by weighted summation of each set of tilt monitoring data, crack monitoring data, and settlement monitoring data. The weight factors of the tilt monitoring data, crack monitoring data, and settlement monitoring data can be preset based on experience, or adjusted and optimized according to specific building types, service life, geological conditions, and other factors. As deformation monitoring data accumulates, the weight factors can be further optimized and verified through data analysis and machine learning methods. This embodiment does not limit this process.

[0072] Understandably, compared to existing technologies that rely solely on the Sparrow Search algorithm to predict slope displacement instability and determine risk using stability coefficients, the building deformation values ​​in this embodiment comprehensively quantify risk by considering three key deformation factors (tilt, cracks, and settlement), covering a more comprehensive range of deformation dimensions. This significantly reduces computational complexity while effectively reflecting the overall deformation of the target building, avoiding misjudgments based on a single parameter, and improving the accuracy of subsequent analysis and early warning.

[0073] Step 104: A deformation state classification model is constructed using a clustering algorithm, and the deformation state classification model is pre-trained based on the building deformation values. The output of the model is the deformation state classification result, and the deformation state categories include qualified, abnormal, and severe.

[0074] Furthermore, in this embodiment, a clustering algorithm is used to construct a deformation state classification model, and the model is pre-trained based on the building deformation values. The output of the model is the deformation state classification result, including:

[0075] Step 201: Construct a deformation state classification model and pre-determine three primary cluster centers K1, K2, and K3, which represent qualified deformation state (K1), abnormal deformation state (K2), and severe deformation state (K3), respectively.

[0076] Step 202: Randomly select three data points {u1, u2, u3} from the building deformation values ​​as the primary cluster centers K1, K2, and K3, respectively;

[0077] It is understandable that the random selection of the initial cluster center in this embodiment depends on the distribution of the current building deformation value dataset, rather than a fixed preset, which helps the model adapt to the data characteristics of different monitoring periods (such as the difference in deformation patterns caused by seasonal changes).

[0078] Step 203, calculate x for each data point in the building deformation value.i The distance to the centers of each primary cluster {u1, u2, u3} is given, and the data point x is... i The building is assigned to the nearest primary cluster, forming three intermediate clusters U1, U2, and U3, thereby classifying the building deformation values.

[0079] In this embodiment, the method for calculating the distance from each data point in the building deformation value to the center of each primary cluster can be Euclidean distance, Mahalanobis distance, etc., and this embodiment does not limit it.

[0080] Step 204: Calculate the mean of all data points in each intermediate cluster, and use it as the new primary cluster center.

[0081] Understandably, for each intermediate cluster, the mean of all data points within it is calculated as the new primary cluster center.

[0082] On the other hand, in this embodiment, step 204, which calculates the mean of all data points within each intermediate cluster as the new primary cluster center, further includes:

[0083] When there are no data points in the intermediate cluster during the iteration process, the data point farthest from the current primary cluster center is selected as the new primary cluster center.

[0084] Understandably, this implementation also provides an anomaly absorption mechanism for the deformation state classification model. When an empty cluster occurs during an iteration, the data point furthest from the current primary cluster center is automatically selected as the new primary cluster center to prevent the model from failing due to sudden changes in data distribution, thereby effectively improving the stability of the model.

[0085] Step 205: Repeat steps 203-204 until the preset convergence condition is met, and obtain the pre-trained deformation state classification model.

[0086] The preset convergence condition can be a preset number of iterations or the change in the primary cluster center being less than a threshold. This embodiment does not limit this.

[0087] It is understandable that in each iteration of this embodiment, the primary cluster center is readjusted according to the actual allocation of the current data points. By continuously recalculating the mean of all data points in the intermediate cluster, the deformation state classification model can capture changes in the distribution of building deformation value data, such as the intensification or mitigation of building deformation trends. This allows for dynamic adjustment of the classification criteria, enabling the cluster center to gradually approach the true data distribution in order to reflect the inherent structure of building deformation values.

[0088] Furthermore, in this embodiment, the pre-training of the deformation state classification model based on building deformation values ​​also includes:

[0089] Store the building deformation values ​​of the most recent Q periods (Q can be 20-30). When a new R period of building deformation values ​​is added, the deformation state classification model is re-pre-trained based on the latest Q+R building deformation values.

[0090] Understandably, in this embodiment, a sliding window mechanism is used to periodically retrain the deformation state classification model with newly monitored building deformation data. When new data arrives, the deformation state classification model re-executes the clustering process, thereby achieving dynamic updates to adapt to long-term changes in building deformation (such as the cumulative effect of foundation settlement) and improving the sensitivity of the deformation state classification model to monitoring gradual deformation.

[0091] This embodiment constructs a data-driven dynamic clustering system by combining a clustering model with temporal difference processing and multidimensional data coupling. It simplifies the complex computational process of sparrow search algorithm modeling and least squares calculation in existing technologies into a lightweight monitoring chain of "data cleaning → rapid classification". After suppressing noise and coupling multidimensional deformation factors to obtain building deformation values ​​through temporal difference processing, the model automatically classifies data categories through unsupervised learning. The building deformation value data is directly input into the deformation state classification model constructed by the clustering algorithm to output the deformation state classification results. While ensuring data reliability, it significantly shortens the data processing time and greatly improves the timeliness of early warning, providing real-time guarantee for the monitoring of sudden building deformation.

[0092] Step 105: Input the building deformation value corresponding to the current period into the pre-trained deformation state classification model, and output the current deformation state classification result.

[0093] Step 106: Issue a deformation warning signal based on the current deformation classification results.

[0094] In this embodiment, the deformation warning signal includes, but is not limited to, sending a warning text message to a preset responsible person, activating the on-site sound and light alarm device, pushing a red pop-up window and vibration reminder to the mobile monitoring APP, generating an electronic work order containing a QR code of deformation parameters and dispatching it to the maintenance department, etc. This embodiment does not limit these actions.

[0095] Specifically, the method for issuing a deformation early warning signal based on the current deformation classification results is as follows:

[0096] If the current deformation classification result is qualified, a good deformation warning signal will be issued;

[0097] If the current deformation classification result is abnormal, a poor deformation warning signal will be issued.

[0098] If the current deformation classification result is severe, a very poor deformation warning signal will be issued.

[0099] In summary, this embodiment establishes a closed-loop optimized monitoring mechanism through the collaborative innovation of "temporal differential processing - multidimensional data coupling - data-driven dynamic clustering and classification." Specifically: First, a dynamic baseline is constructed using a dual-difference processing mechanism (spatial phase difference + temporal difference), with adjacent period data serving as a moving reference frame. This avoids the accumulated errors of traditional fixed-time reference point monitoring and solves the error problem caused by the assumption of constant atmospheric conditions in existing technologies. Cleaner deformation monitoring data can be obtained directly without complex atmospheric phase compensation processing. Second, through tilting... The three key deformation factors of tilt, cracks, and settlement are weighted and coupled to generate building deformation values, forming a multi-dimensional integrated deformation index. Compared with the single stability coefficient of existing technologies, this can effectively reduce the false alarm rate of early warning. Finally, the dynamic deformation state classification model based on clustering algorithm avoids the complex computation of sparrow search algorithm in existing technologies, greatly improving the timeliness of early warning. The three factors work together to form a technical chain of "error suppression → comprehensive quantification → dynamic classification", achieving a synergistic breakthrough in the accuracy of building deformation monitoring and data processing efficiency, achieving a beneficial effect of "1+1>2".

[0100] Example 2:

[0101] Please see the appendix Figure 2 , Figure 2 A flowchart illustrating a building deformation monitoring method based on spaceborne InSAR data, provided in yet another embodiment of this specification, is shown.

[0102] like Figure 2 As shown, this method for monitoring building deformation based on spaceborne InSAR data includes the following steps:

[0103] Step 301: Periodically collect spaceborne InSAR data of the target building;

[0104] Step 302: Perform time-series differential processing on the spaceborne InSAR data to obtain several sets of tilt monitoring data, crack monitoring data and settlement monitoring data of the target building. The set of all tilt monitoring data, crack monitoring data and settlement monitoring data constitutes the deformation monitoring dataset of the target building.

[0105] Step 303: Couple each set of tilt monitoring data, crack monitoring data and settlement monitoring data in the deformation monitoring dataset, and record the coupling result as the building deformation value;

[0106] Step 304: A deformation state classification model is constructed using a clustering algorithm, and the deformation state classification model is pre-trained based on the building deformation values. The output of the model is the deformation state classification result, and the categories of deformation state include qualified, abnormal, and severe.

[0107] Step 305: Input the building deformation value corresponding to the current period into the pre-trained deformation state classification model, and output the current deformation state classification result.

[0108] Step 306: Calculate the building stability value based on the deformation monitoring dataset.

[0109] Specifically, in this embodiment, the deformation monitoring dataset includes the settlement and crack volume of data points. Based on the deformation monitoring dataset, the building stability value is calculated, including:

[0110] Step 401: Based on the settlement amount of the data points, the settlement rate is calculated, and the tilt reference value is calculated using the two-point difference method, which can be specifically expressed as:

[0111]

[0112] Among them, C a Let B represent the settlement rate, ΔD represent the difference in settlement between two consecutive observations of the same data point, and ΔT represent the time interval between two consecutive observations. a Indicates the tilt reference value, B b and B c These represent the settlement amounts at two data points along the tilt direction of the target building, where L represents B. b and B c The horizontal distance between two corresponding data points.

[0113] Step 402: The settlement rate, tilt reference value, and crack volume are normalized using the maximum allowable value constraint method, and then weighted and summed to obtain the building stability value. Building stability value S a It can be represented as:

[0114]

[0115] Where vmax represents the maximum allowable settlement rate, θmax represents the maximum allowable tilt, A represents the crack volume, and Amax represents the maximum allowable crack volume. The maximum allowable values ​​are all set according to building codes. H1, H2, and H3 represent the corresponding preset weights. In this embodiment, the weight allocation in the calculation process of building stability value is determined empirically to adapt to the building stability status assessment of different building structure types (for example, H2 = 0.6 is suitable for high-rise buildings).

[0116] Understandably, this embodiment first assesses the settlement dimension by measuring the settlement rate per unit time, reflecting the dynamic process of settlement and helping to identify sudden settlement events. Second, it calculates the tilt reference value using the two-point difference method, and assesses the tilt dimension by measuring the ratio of the settlement difference between the two points to the horizontal distance, reflecting the rate of change of the tilt angle, which is used to quantify the building tilt trend and helps to avoid tilt monitoring errors caused by the limitations of radar line of sight (LOS) direction in existing technologies. Next, it normalizes the three dimensions of tilt, settlement, and crack volume by using maximum allowable value constraints to achieve comparability of multi-source heterogeneous data. Finally, it adapts to different building structure types by weighting empirical coefficients and obtains the building stability value by weighted summation to comprehensively assess the deformation stability state of the target building.

[0117] Step 307: Based on the current deformation classification results and building stability values, issue a deformation warning signal.

[0118] Specifically, in this embodiment, based on the current deformation classification result and the building stability value, a deformation warning signal is issued, including:

[0119] The preset threshold range for building stability values ​​is (E1, E2).

[0120] If the current deformation classification result is qualified and the building stability value is greater than or equal to E2, a good deformation warning signal will be issued;

[0121] If the current deformation classification result is qualified and the building stability value is within the threshold range (E1, E2), a relatively stable deformation warning signal will be issued.

[0122] If the current deformation classification result is abnormal and the building stability value is within the threshold range (E1, E2), a poor deformation warning signal will be issued.

[0123] If the current deformation classification result is abnormal and the building stability value is less than or equal to E1, an unstable deformation warning signal will be issued.

[0124] If the current deformation classification result is severe, an extremely poor deformation warning signal will be issued under any building stability value.

[0125] It is understandable that within the threshold range (E1, E2) of building stability value, E1 is the lower critical limit of stability, and E2 is the upper critical limit of stability.

[0126] Steps 301-305 above can be referred to steps 101-105 in Embodiment 1, and will not be repeated here.

[0127] The existing early warning logic is based on whether the stability coefficient output by the sparrow search algorithm exceeds the reference range. However, relying solely on the analysis of a single stability coefficient and the early warning mechanism with a fixed parameter range to judge risk will undoubtedly ignore the complex working conditions of other deformation dimensions, resulting in low accuracy of structural state assessment. To address this, this embodiment proposes a "qualitative-quantitative dual-modal collaborative decision-making" monitoring and early warning method. This method optimizes the entire chain of building deformation monitoring, from data acquisition to risk decision-making. Specifically, firstly, it suppresses atmospheric delay and noise interference through time-series differential processing to generate high signal-to-noise ratio data. Then, it provides a qualitative judgment of the deformation state (qualified / abnormal / severe) by constructing a deformation state classification model using a clustering algorithm. Simultaneously, it provides a quantitative assessment by combining building stability values. The two work together to achieve dual verification of "qualitative + quantitative" results. Finally, it collaboratively triggers five-level early warning signals (good, relatively stable, poor, unstable, and extremely poor). This effectively solves the problems of atmospheric interference, high computational complexity, single early warning, and rigidity that rely solely on fixed threshold judgments in existing technologies. Furthermore, it adapts to different building types and environmental changes through a dynamic threshold adjustment mechanism (adaptively updating the stable value threshold range based on the cluster mean and standard deviation) to ensure the accuracy and reliability of overall building deformation monitoring and effectively reduce the false alarm rate of early warnings.

[0128] On the other hand, this embodiment also includes a dynamic threshold adjustment mechanism, specifically:

[0129] Use the preset threshold range of building stability values ​​as the initial threshold range;

[0130] When the building deformation values ​​are added for every R cycles and the deformation state classification model is re-pre-trained, the deformation state classification model recalculates the primary cluster centers. The threshold interval is then updated based on the data distribution of the primary cluster centers, and the updated threshold interval can be expressed as:

[0131] E2=μ k2 -1.5σ k2

[0132] E1=μ k3 -1.5σ k3

[0133] Where, μ k2 and σ k2 μ represents the mean and standard deviation of the cluster centers of anomalous deformation states, respectively. k3 and σ k3 represents the mean and standard deviation of the cluster center in the severely deformed state, respectively.

[0134] Understandably, the threshold dynamic adjustment mechanism proposed in this embodiment achieves dynamic calibration of the threshold parameter as the building deformation data distribution is adjusted by binding the threshold interval with the statistical characteristics of the cluster centers. Specifically, each additional R-cycle of data triggers model retraining → cluster center update → threshold interval reconstruction → the new threshold guides the warning for the next cycle. Compared to a preset fixed threshold, this dual closed-loop system of "model iteration + threshold iteration" ensures that the threshold parameter is always synchronized with the latest deformation pattern, thereby automatically adapting to the data characteristics of different building types and different monitoring stages (such as changes in deformation characteristics between the construction and operation phases), effectively reducing the false alarm rate. For example, when a building enters the aging stage, causing the baseline deformation value to shift overall, the threshold interval will be adjusted synchronously with the drift of the cluster centers, avoiding the risk of missed alarms caused by a preset fixed threshold.

[0135] Example 3:

[0136] Please see the appendix Figure 3 , Figure 3 This is a schematic diagram of a building deformation monitoring system based on spaceborne InSAR data, provided as another embodiment of this specification.

[0137] like Figure 3 As shown, the building deformation monitoring system based on spaceborne InSAR data can include at least a data acquisition module 1, a data preprocessing module 2, a deformation value calculation module 3, a deformation state classification module 4, and an early warning module 5, wherein:

[0138] Data acquisition module 1 is used to periodically acquire spaceborne InSAR data of the target building;

[0139] Data preprocessing module 2 is used to perform time-series differential processing on spaceborne InSAR data to obtain several sets of tilt monitoring data, crack monitoring data and settlement monitoring data of the target building. The collection of all tilt monitoring data, crack monitoring data and settlement monitoring data constitutes the deformation monitoring dataset of the target building.

[0140] The deformation value calculation module 3 is used to couple each set of tilt monitoring data, crack monitoring data and settlement monitoring data in the deformation monitoring dataset, and record the coupling result as the building deformation value.

[0141] Deformation state classification module 4 is used to construct a deformation state classification model and pre-train the deformation state classification model based on the building deformation value. The output of the model is the deformation state classification result. The categories of deformation state include qualified, abnormal and severe. Input the building deformation value corresponding to the current period into the pre-trained deformation state classification model and get the current deformation state classification result.

[0142] Early warning module 5 is used to issue deformation early warning signals based on the current deformation classification results.

[0143] Furthermore, in this embodiment, the building deformation monitoring system based on spaceborne InSAR data may also include:

[0144] The stability value calculation module is used to calculate the building stability value based on the deformation monitoring dataset;

[0145] The early warning module is also used to issue deformation early warning signals based on the current deformation classification results and building stability values.

[0146] It is understood that the technical concept of the building deformation monitoring system based on spaceborne InSAR data provided in this embodiment is similar to the technical concept of the building deformation monitoring method based on spaceborne InSAR data provided in the previous embodiment, and will not be repeated here.

[0147] The above description is merely a preferred embodiment disclosed in this application and an explanation of the technical principles employed. Those skilled in the art should understand that the scope of protection involved in this disclosure is not limited to technical solutions formed by specific combinations of the above-described technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the above-disclosed concept. For example, technical solutions formed by substituting the above features with (but not limited to) technical features with similar functions disclosed in this disclosure.

[0148] Furthermore, while the operations are described in a specific order, this should not be construed as requiring these operations to be performed in the specific order shown or in a sequential order. In certain environments, multitasking and parallel processing may be advantageous. Similarly, while several specific implementation details are included in the above discussion, these should not be construed as limiting the scope of this disclosure. Certain features described in the context of individual embodiments may also be implemented in combination in a single embodiment. Conversely, various features described in the context of a single embodiment may also be implemented individually or in any suitable sub-combination in multiple embodiments.

Claims

1. A building deformation monitoring method based on spaceborne InSAR data, characterized in that, Includes the following steps: Periodically collect spaceborne InSAR data of the target building; Temporal differential processing was performed on the spaceborne InSAR data to obtain several sets of tilt monitoring data, crack monitoring data and settlement monitoring data of the target building. The set of all tilt monitoring data, crack monitoring data and settlement monitoring data constitutes the deformation monitoring dataset of the target building. Couple each set of tilt monitoring data, crack monitoring data and settlement monitoring data in the deformation monitoring dataset, and record the coupling result as the building deformation value; A clustering algorithm is used to construct a deformation state classification model, and the deformation state classification model is pre-trained based on the building deformation value. The output of the model is the deformation state classification result, and the deformation state categories include qualified, abnormal and severe. Input the building deformation value corresponding to the current period into the pre-trained deformation state classification model, and output the current deformation state classification result. A deformation warning signal is issued based on the current deformation classification results; The deformation monitoring dataset includes the settlement and crack volume of data points, and issues deformation early warning signals based on the current deformation classification results. It also includes: Based on the settlement amount of the data points, the settlement rate is calculated, and the tilt reference value is calculated using the two-point difference method. The settlement rate, tilt reference value and crack volume are normalized by the maximum allowable value constraint method, and then weighted summation is performed to obtain the building stability value. Based on the current deformation classification results and building stability values, a deformation warning signal is issued.

2. The building deformation monitoring method based on spaceborne InSAR data as described in claim 1, characterized in that, Temporal differential processing was performed on the spaceborne InSAR data to obtain several sets of tilt monitoring data, crack monitoring data, and settlement monitoring data of the target building, including: The differences in tilt, crack volume, and settlement between adjacent periods of spaceborne InSAR data are calculated. Each period is considered as a set, resulting in several sets of tilt monitoring data, crack monitoring data, and settlement monitoring data for the target building.

3. The building deformation monitoring method based on spaceborne InSAR data as described in claim 1, characterized in that, A clustering algorithm is used to construct a deformation state classification model, and the model is pre-trained based on the building deformation values. The output of the model is the deformation state classification result, including: Step 201: Construct a deformation state classification model and pre-determine three primary cluster centers K1, K2, and K3, which represent qualified deformation state, abnormal deformation state, and severe deformation state, respectively. Step 202: Randomly select three data points from the building deformation values ​​as the primary cluster centers K1, K2, and K3, respectively; Step 203: Calculate the distance from each data point in the building deformation value to the center of each primary cluster, and assign the data points to the nearest primary cluster to form three intermediate clusters; Step 204: Calculate the mean of all data points in each intermediate cluster, and use it as the new primary cluster center; Step 205: Repeat steps 203-204 until the preset convergence condition is met, and obtain the pre-trained deformation state classification model.

4. The building deformation monitoring method based on spaceborne InSAR data as described in claim 3, characterized in that, In step 204, the mean of all data points within each intermediate cluster is calculated and used as the new primary cluster center, including: When there are no data points in the intermediate cluster during the iteration process, the data point farthest from the current primary cluster center is selected as the new primary cluster center.

5. The building deformation monitoring method based on spaceborne InSAR data as described in claim 1, characterized in that, Pre-training of a deformation state classification model based on building deformation values ​​includes: Store the building deformation values ​​for the most recent Q periods. When a new building deformation value is added for R periods, the deformation state classification model is re-pre-trained based on the latest Q+R building deformation values.

6. The building deformation monitoring method based on spaceborne InSAR data as described in claim 1, characterized in that, Based on the current deformation classification results and building stability values, a deformation warning signal is issued, including: Preset the threshold range (E1, E2) for building stability values. If the current deformation classification result is qualified and the building stability value is greater than or equal to E2, a good deformation warning signal will be issued; If the current deformation classification result is qualified and the building stability value is within the threshold range (E1, E2), a relatively stable deformation warning signal will be issued. If the current deformation classification result is abnormal and the building stability value is within the threshold range (E1, E2), a poor deformation warning signal will be issued. If the current deformation classification result is abnormal and the building stability value is less than or equal to E1, an unstable deformation warning signal will be issued. If the current deformation classification result is severe, an extremely poor deformation warning signal will be issued under any building stability value.

7. A building deformation monitoring system based on spaceborne InSAR data, characterized in that, include: The data acquisition module is used to periodically acquire spaceborne InSAR data of the target building; The data preprocessing module is used to perform time-series differential processing on the spaceborne InSAR data to obtain several sets of tilt monitoring data, crack monitoring data and settlement monitoring data of the target building. The collection of all tilt monitoring data, crack monitoring data and settlement monitoring data constitutes the deformation monitoring dataset of the target building. The deformation value calculation module is used to couple each set of tilt monitoring data, crack monitoring data and settlement monitoring data in the deformation monitoring dataset, and record the coupling result as the building deformation value. The deformation state classification module is used to construct a deformation state classification model and pre-train the deformation state classification model based on the building deformation value. The output of the model is the deformation state classification result. The categories of deformation state include qualified, abnormal and severe. The building deformation value corresponding to the current period is input into the pre-trained deformation state classification model, and the current deformation state classification result is output. The early warning module is used to issue deformation early warning signals based on the current deformation classification results; The deformation monitoring dataset includes the settlement and crack volume of data points, and the early warning module also includes: The stability value calculation module is used to calculate the settlement rate based on the settlement amount of data points, and to calculate the tilt reference value using the two-point difference method. The settlement rate, tilt reference value and crack volume are normalized using the maximum allowable value constraint method, and then weighted summation is performed to obtain the building stability value. The early warning module is used to issue a deformation early warning signal based on the current deformation classification result and the building stability value.

Images