A method for monitoring the health state of a building based on vibration signals

Abstract

The application relates to the technical field of building health monitoring, and discloses a building health state monitoring method based on vibration signals. The method collects vibration data of different monitoring points of a building, and obtains external influence data such as meteorological data, traffic data, underground water level data, construction activity data and earthquake activity data; after uniform time calibration and preprocessing of the above data, natural frequency, vibration energy, cepstrum, three-dimensional vibration trajectory, ground reflection wave, elastic wave propagation and external influence characteristics are extracted; the extracted characteristics are input into a building evaluation model to obtain health state evaluation results including a health index, an abnormal type and a risk level; further combined with evaluation results and spatial position relationships of buildings in a target region, a group abnormal linkage relationship is identified, regional synchronous abnormality, single-body abnormality or conduction abnormality is determined, and regional safety evaluation results and risk early warning information are generated. The method improves the accuracy and early warning timeliness of building and building group health monitoring.

Description

Technical Field

[0001] This application relates to the field of building health monitoring technology, and in particular to a method for monitoring the health status of buildings based on vibration signals. Background Technology

[0002] As urban construction expands, the number of high-rise buildings, large-span buildings, and dense building clusters continues to increase. During long-term service, these buildings are susceptible to structural aging, material fatigue, foundation settlement, changes in groundwater levels, surrounding construction, traffic vibrations, extreme weather, and seismic activity, posing structural safety risks.

[0003] Existing building health monitoring methods mainly include manual inspection, local sensor monitoring, active excitation modal testing, and passive monitoring based on environmental vibration. Manual inspection is highly subjective and time-consuming, making it difficult to detect hidden damage in a timely manner; local sensor monitoring has a limited range and cannot reflect the overall building condition and foundation response changes; active excitation testing is costly to implement, limited by site conditions, and unsuitable for long-term continuous monitoring; existing environmental vibration monitoring mostly relies on single indicators such as natural frequency and vibration amplitude, making it difficult to comprehensively identify overall stiffness changes, local damage, changes in internal reflection paths, foundation anomalies, and three-dimensional abnormal vibration modes.

[0004] Meanwhile, building vibration responses are easily affected by external factors such as weather, traffic, groundwater levels, construction activities, and seismic activity. Existing methods lack comprehensive analysis of the relationship between external influence data and vibration anomalies, which can easily lead to false alarms or missed alarms. In urban building clusters, multiple buildings may experience linked anomalies due to common external disturbances, the influence of adjacent buildings, or changes in the foundation environment. Existing individual monitoring methods are insufficient to identify the relationships between group anomalies and generate regional safety assessment results. Summary of the Invention

[0005] To overcome, to some extent, the problems in related technologies regarding building health monitoring that rely on a single vibration index, have difficulty distinguishing between external disturbances and structural anomalies, have difficulty identifying abnormal linkages among building groups, and have insufficient accuracy in regional structural safety assessments, this application provides a method for monitoring building health status based on vibration signals.

[0006] The proposed solution is as follows:

[0007] A method for monitoring the health status of buildings based on vibration signals, comprising: Vibration data of the building is collected by vibration sensors deployed at different monitoring points in the building; Obtain external impact data related to the building; the external impact data includes at least one of meteorological data, traffic data, groundwater level data, construction activity data, and seismic activity data; The vibration data and external influence data are uniformly time-calibrated, and then subjected to noise reduction, filtering, outlier removal and standardization to obtain the data to be analyzed. Building vibration characteristics and external influence characteristics are extracted based on the data to be analyzed; the building vibration characteristics include at least one of the following: natural frequency characteristics, vibration energy characteristics, cepstral characteristics, three-dimensional vibration trajectory characteristics, foundation reflected wave characteristics, and elastic wave propagation characteristics; The building vibration characteristics and external influence characteristics are input into a pre-trained building assessment model, which outputs a building health status assessment result; the building health status assessment result includes a health index, anomaly type, and risk level; Based on the building health status assessment results and the spatial relationship of buildings within the target area, generate building cluster status association data for the target area; Based on the building group status correlation data, correlation analysis is performed on the changes in the time of occurrence of anomalies, anomaly type and risk level among the buildings with anomalies in the target area to identify the linkage relationship of group anomalies; Based on the aforementioned group anomaly linkage relationship, the cause of the anomaly in the target area is determined and the regional security assessment result of the target area is generated; the cause of the anomaly in the target area includes regional synchronous anomaly caused by common external influences, individual anomaly caused by changes in local structural state, and transmission anomaly caused by changes in adjacent buildings or foundation environment; Based on the health status assessment results of each building and the regional safety assessment results of the target area, risk warning information is generated; Based on the building health status assessment results for each building and the regional safety assessment results for the target area, the degree of abnormal contribution of each monitoring point is determined. Based on the degree of abnormal contribution, key monitoring points are identified from multiple monitoring points, and the building structure area where the key monitoring points are located is identified as the key monitoring area; When the health index corresponding to the key monitoring point decreases or the risk level increases, the sampling frequency of the vibration sensor corresponding to the key monitoring point is increased. When the anomaly type corresponding to the key monitoring point persists for multiple consecutive evaluation periods, the data collection time window corresponding to the key monitoring point shall be extended. When multiple monitoring points within the key monitoring area show abnormal correlations, the scope of the key monitoring area is expanded, or adjacent structural areas are adjusted to become new key monitoring areas.

[0008] Preferably, the method further includes: Obtain structural information of the building; the structural information includes at least one of the following: building height, floor distribution, load-bearing structure layout, foundation location, support node location, and span direction; The key structural regions of the building are determined based on the structural information; the key structural regions include at least one of the following: overall vibration response region, local structural response region, foundation response region, and environmental vibration background region; The monitoring point locations of the vibration sensor are determined based on the key areas of the structure; including: When the building is a layered structure, the locations of the top, middle structural layer and bottom foundation of the building are determined as monitoring points. When the building is a high-rise building, multiple vertical monitoring levels are determined based on the floor height and floor distribution, and the top of the building, the multiple vertical monitoring levels, and the foundation position at the bottom of the building are determined as monitoring point positions; When the building is a large-span building, multiple lateral monitoring areas are determined according to the span direction and the location of the support nodes, and the lateral monitoring areas and the location of the support nodes are determined as monitoring point locations.

[0009] Preferably, extracting building vibration characteristics based on the data to be analyzed includes: The vibration data in the data to be analyzed is segmented according to a preset sliding time window to obtain multiple continuous vibration analysis segments; Spectral analysis was performed on each vibration analysis segment to extract the principal mode frequency corresponding to the spectral peak. Based on the variation amplitude and trend of the main modal frequency within a continuous sliding time window, natural frequency characteristics are generated. The natural frequency characteristics are compared with the historical reference frequency characteristics of the corresponding building to determine the overall stiffness variation state of the building.

[0010] Preferably, extracting building vibration characteristics based on the data to be analyzed further includes: The combination of monitoring points to be analyzed is determined from the vibration data corresponding to different monitoring points of the building; Cross-correlation processing is performed on vibration data from different monitoring points in the same combination of monitoring points to be analyzed, and the structural response function between different monitoring points is reconstructed. Autocorrelation processing was performed on vibration data from the same monitoring point to extract internal building reflection response information. Based on the structural response function and internal reflection response information, the changes in the propagation path and reflection path of vibration within the building structure are determined, and vibration interference characteristics are generated.

[0011] Preferably, extracting building vibration characteristics based on the data to be analyzed further includes: The vibration propagation delay between different monitoring points is determined based on the structural response function. Based on the spatial relationship between different monitoring points and the vibration propagation time delay, the elastic wave velocity change information is determined; The elastic wave velocity variation information is divided according to the frequency band range to obtain the elastic wave velocity characteristics of low frequency band, mid frequency band, and high frequency band. The overall structural state of a building is characterized by the low-frequency elastic wave velocity characteristics, the local structural state of a building is characterized by the mid-frequency elastic wave velocity characteristics, and the surface or micro-damage state of a building is characterized by the high-frequency elastic wave velocity characteristics.

[0012] Preferably, extracting building vibration characteristics based on the data to be analyzed further includes: The vibration data in the data to be analyzed is subjected to frequency domain transformation to obtain vibration spectrum data; The vibration spectrum data is subjected to logarithmic processing and inverse transform processing to obtain cepstral data; Based on the cepstral data, extract the cepstral peak position, cepstral peak intensity, and cepstral peak variation trend; The reflection period variation corresponding to the internal reflection path of the building is determined based on the cepstral peak position; Based on the cepstral peak intensity and the cepstral peak variation trend, cepstral features are generated to characterize the layering, cracks, or local loosening of the internal structure of a building.

[0013] Preferably, extracting building vibration characteristics based on the data to be analyzed further includes: Low-frequency vibration response data are extracted from vibration data corresponding to building foundation monitoring points and surrounding ground surface monitoring points. Based on the low-frequency vibration response data, determine the changes in ground reflected waves, the state of low-frequency vibration enhancement, and the state of changes in ground elastic wave velocity. The changes in ground reflected waves, the enhanced state of low-frequency vibration, and the changes in ground elastic wave velocity are compared with the historical ground reference characteristics to obtain the ground response characteristics; Based on the aforementioned foundation response characteristics, at least one of the following foundation anomalies can be identified: uneven foundation settlement, soil softening, and groundwater level changes.

[0014] Preferably, extracting building vibration characteristics based on the data to be analyzed further includes: Obtain vibration components of the same monitoring point in different directions; Construct a three-dimensional vibration trajectory corresponding to the monitoring point based on the vibration components in different directions; The main vibration direction, major axis, minor axis, and deflection state of the trajectory are determined based on the three-dimensional vibration trajectory. The main vibration direction, major axis of the trajectory, minor axis of the trajectory, and trajectory deflection state are compared with the historical reference trajectory to determine the vibration state in the abnormal direction; Based on the abnormal directional vibration state, a three-dimensional vibration trajectory feature is generated to characterize the abnormal vibration mode of the building.

[0015] Preferably, the method further includes: Record the building health status assessment results for each building in chronological order of assessment time to generate a historical sequence of building health status. Based on the historical sequence of the building's health status, extract the trends of health index changes, abnormality type evolution trends, and risk level changes; Based on the trends in health index changes, abnormality type evolution trends, and risk level changes, the degradation trend of building structural performance is determined. By comparing the degradation trend of the building's structural performance with the historical baseline degradation trend, the risk development status of the building can be predicted; Based on the risk development status, a lifespan prediction result and maintenance reminder information are generated; the lifespan prediction result is used to characterize the remaining safe use status of the building under the current degradation trend.

[0016] The technical solution provided in this application may include the following beneficial effects: This technical solution collects vibration data from different monitoring points on a building and combines it with external influence data such as meteorology, traffic, groundwater level, construction activities, and seismic activity for unified time calibration and preprocessing. This allows for correlation analysis between the building's own vibration response and external disturbance factors under the same time reference, improving the completeness of monitoring data and the reliability of assessment. By extracting multi-dimensional building vibration characteristics such as natural frequency, vibration energy, cepstrum, three-dimensional vibration trajectory, foundation reflected waves, and elastic wave propagation, and inputting them together with external influence characteristics into the building assessment model, it is possible to analyze multiple aspects such as overall stiffness changes, local anomalies, changes in internal reflection paths, changes in foundation response, and changes in spatial vibration modes. This approach comprehensively assesses the health status of buildings from multiple perspectives, reducing false alarms and missed alarms caused by monitoring a single indicator. Furthermore, by correlating the health status assessment results and spatial relationships of multiple buildings within the target area, it can identify the group linkages between changes in the timing, type, and risk level of anomalies. This allows for the differentiation between synchronous anomalies caused by shared external influences, individual anomalies caused by changes in local structural conditions, and transmission anomalies caused by changes in adjacent buildings or the foundation environment. Consequently, regional safety assessment results and risk warning information can be generated based on individual building health assessments, improving the accuracy and continuity of building health monitoring and the timeliness of regional structural safety warnings.

[0017] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and do not limit this application. Attached Figure Description

[0018] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0019] Figure 1 This is a schematic flowchart of a method for monitoring the health status of a building based on vibration signals, provided in one embodiment of this application. Figure 2 This is a schematic diagram of the layout of vibration sensors at different monitoring points of a building, provided in one embodiment of this application. Detailed Implementation

[0020] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numbers in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this application as detailed in the appended claims.

[0021] Example 1 Figure 1 This is a flowchart illustrating a method for monitoring the health status of a building based on vibration signals, provided in one embodiment of this application. (Refer to...) Figure 1 A method for monitoring the health status of buildings based on vibration signals, comprising: S11. Vibration data of the building is collected by vibration sensors installed at different monitoring points of the building; S12. Obtain external impact data related to the building; external impact data includes at least one of meteorological data, traffic data, groundwater level data, construction activity data, and seismic activity data; S13. Perform unified time calibration on the vibration data and external influence data, and perform noise reduction, filtering, outlier removal and standardization to obtain the data to be analyzed. S14. Extract building vibration characteristics and external influence characteristics based on the data to be analyzed; building vibration characteristics include at least one of the following: natural frequency characteristics, vibration energy characteristics, cepstral characteristics, three-dimensional vibration trajectory characteristics, foundation reflected wave characteristics, and elastic wave propagation characteristics; S15. Input the building vibration characteristics and external influence characteristics into the pre-trained building assessment model, and output the building health status assessment results; the building health status assessment results include health index, anomaly type and risk level; S16. Generate building group status association data for the target area based on the building health status assessment results and the spatial location relationship of each building in the target area. S17. Based on the building group status correlation data, conduct correlation analysis on the changes in the time of occurrence of anomalies, anomaly type and risk level among the buildings with anomalies in the target area, and identify the linkage relationship of group anomalies. S18. Based on the linkage relationship of group anomalies, determine the cause of anomalies in the target area and generate the regional security assessment result of the target area; the cause of anomalies in the target area includes regional synchronous anomalies caused by common external influences, individual anomalies caused by changes in local structural state, and transmission anomalies caused by changes in adjacent buildings or foundation environment; S19. Based on the health status assessment results of each building and the regional safety assessment results of the target area, generate risk warning information.

[0022] For ease of understanding, the following explains some key terms in this embodiment: A vibration sensor is a device used to sense and measure the vibration response of a building. It is typically deployed at key monitoring points in a building to collect the dynamic response of the structure under environmental excitations in real time, such as acceleration, velocity, or displacement signals.

[0023] External impact data refers to data on non-structural factors that may affect the structural health of a building, including but not limited to meteorological data, traffic data, groundwater level data, surrounding construction activity data, and seismic activity data. This data is used to interpret or correct for changes in structural condition reflected in vibration data.

[0024] The data to be analyzed refers to vibration data and external influence data that have undergone uniform time calibration, denoising, filtering, outlier removal, and standardization. This data provides a high-quality and consistent foundation for subsequent feature extraction and model input.

[0025] Building vibration characteristics are quantitative indicators extracted from the data to be analyzed, which characterize the structural dynamics of a building. These characteristics include natural frequency characteristics, vibration energy characteristics, cepstral characteristics, three-dimensional vibration trajectory characteristics, ground reflection wave characteristics, and elastic wave propagation characteristics. These characteristics are used to reflect information such as the overall stiffness of the building, local damage, material properties, and foundation response.

[0026] A building assessment model is a pre-trained intelligent model, such as one built using machine learning or deep learning algorithms. This model takes building vibration characteristics and external influence characteristics as input, learns patterns from historical data, and outputs a building health status assessment result.

[0027] Building health status assessment results are the output of a building assessment model, comprehensively reflecting the current health condition of the building. These results typically include health indices, anomaly types, and risk levels.

[0028] Building cluster status association data refers to data generated by integrating the health status assessment results of each building within a target area with their spatial location relationships. This data is used to analyze the mutual influence between buildings and the overall status within the area.

[0029] The linkage relationship of anomalies in a group refers to the interrelationship patterns among multiple buildings within a target area identified by analyzing the correlation data of the status of a building group, showing changes in the timing, type, or risk level of anomalies. This relationship helps to discover regional, transmissive, or synchronous anomalies.

[0030] The causes of anomalies in the target area are the root causes identified through source analysis of the interconnected relationships of identified group anomalies. These can be categorized into regional synchronous anomalies caused by common external influences, individual anomalies caused by changes in local structural states, and transmission anomalies caused by changes in adjacent buildings or the foundation environment.

[0031] This embodiment provides a method for monitoring the health status of buildings based on vibration signals, specifically including the following steps: First, vibration data of the building is collected by vibration sensors deployed at different monitoring points. The vibration sensors can be accelerometers, velocities, or displacement sensors, which are installed at different locations on the building structure. Vibration data can be collected continuously or periodically at preset time intervals.

[0032] Secondly, data on the external impacts of the building are obtained. This data can be acquired from various sources. For example, meteorological data can be obtained from local weather stations or public meteorological databases; traffic data can be obtained from urban traffic management departments or road monitoring systems; groundwater level data can be provided by groundwater monitoring wells or hydrological stations; construction activity data can be provided by surrounding construction units or obtained through on-site surveys; and seismic activity data can be obtained from earthquake monitoring agencies. This data is used to assist in the analysis and interpretation of the building's vibration response.

[0033] Furthermore, the vibration data and external influence data are uniformly time-calibrated, and then denoised, filtered, outlier removed, and standardized to obtain the data to be analyzed. Uniform time calibration ensures the consistency of the vibration data and external influence data in the time dimension, for example, by aligning the timestamps of all data to the same reference clock. Denoising can be performed using methods such as wavelet transform or empirical mode decomposition to eliminate random noise in the signal. Filtering can apply digital filters, such as Butterworth filters, to remove interference components within a specific frequency range. Outlier removal can be based on statistical methods, such as the three-standard-deviation criterion, to identify and remove anomalous data points caused by sensor malfunctions or instantaneous impacts. Standardization can transform data with different dimensions to a uniform numerical range, for example, through min-max normalization or Z-score standardization.

[0034] Subsequently, building vibration characteristics and external influence characteristics are extracted from the data to be analyzed. Building vibration characteristics may include at least one of the following: natural frequency characteristics, vibration energy characteristics, cepstral characteristics, three-dimensional vibration trajectory characteristics, ground reflection wave characteristics, and elastic wave propagation characteristics. For example, natural frequency characteristics can be obtained by performing a Fourier transform on the vibration data to identify the dominant frequency in the spectrum. Vibration energy characteristics can be characterized by calculating the root mean square value or energy integral of the vibration signal. External influence characteristics can be extracted from external influence data, such as average wind speed and maximum wind pressure from meteorological data, peak traffic flow from traffic data, and the amplitude of changes in groundwater level data.

[0035] Next, the building vibration characteristics and external impact characteristics are input into a pre-trained building assessment model, which outputs the building health status assessment results. The building assessment model can be a classification or regression model built based on machine learning algorithms (such as support vector machines, random forests, or neural networks). During the training phase, the model learns using a large amount of historical monitoring data and corresponding building health status labels, thereby establishing a mapping relationship between features and health status. The output building health status assessment results include a health index (e.g., a score of 0-100), anomaly type (e.g., "normal," "minor damage," "foundation settlement"), and risk level (e.g., "low risk," "medium risk," "high risk").

[0036] Furthermore, based on the building health status assessment results and spatial relationships of each building within the target area, cluster status association data for the target area is generated. For example, the health index, anomaly type, risk level, and geographical coordinates or relative location information of each building can be integrated into a database or geographic information system to form an association dataset containing status information of all buildings within the area.

[0037] Therefore, based on the correlation data of the building complex's status, a correlation analysis is performed on the changes in the timing, type, and risk level of anomalies among buildings exhibiting abnormalities within the target area to identify the linkage relationships of group anomalies. For example, when multiple adjacent buildings all experience anomalies of the "foundation settlement" type within a short period of time, or when the risk level of one building increases and the risk level of its surrounding buildings also increases, this can be identified as a linkage relationship of group anomalies. This correlation analysis can employ time series analysis, spatial cluster analysis, or graph theory methods.

[0038] Based on this, and according to the interconnected relationships of the anomalies within the group, the causes of the anomalies in the target area are determined, and a regional safety assessment result for the target area is generated. Specifically, if multiple buildings exhibit anomalies simultaneously and external impact data shows significant changes (e.g., regional heavy rainfall leading to a general rise in groundwater levels), the cause of the anomaly can be identified as a synchronous anomaly caused by a common external impact. If the anomaly is limited to a single building and its internal vibration characteristics indicate a specific structural problem, it is identified as an individual anomaly caused by a change in the local structural state. If an anomaly in one building leads to anomalies in adjacent buildings, it is identified as a transmission anomaly caused by changes in adjacent buildings or the foundation environment. The regional safety assessment result is a comprehensive safety judgment of the entire target area, such as "the overall area is safe, but some areas require attention" or "the area has a moderate risk, and a comprehensive investigation is recommended."

[0039] Finally, based on the health status assessment results of each building and the regional safety assessment results of the target area, risk warning information is generated. This risk warning information can include the specific location of the abnormal building, the type of abnormality, the risk level, the cause of the abnormality, and recommended measures. For example, when the health index of a building falls below a threshold or the risk level reaches "high risk," the system can automatically generate a warning message and notify relevant management personnel through a visual interface, email, or SMS.

[0040] The method also includes: Based on the building health status assessment results for each building and the regional safety assessment results for the target area, the degree of abnormal contribution of each monitoring point is determined. Based on the degree of abnormal contribution, key monitoring points were identified from multiple monitoring points, and the building structure areas where the key monitoring points were located were identified as key monitoring areas; When the health index corresponding to a key monitoring point decreases or the risk level increases, the sampling frequency of the vibration sensor corresponding to the key monitoring point is increased. When the anomaly type corresponding to a key monitoring point persists for multiple consecutive assessment periods, the data collection time window for the key monitoring point shall be extended. When multiple monitoring points within a key monitoring area show abnormal correlations, the scope of the key monitoring area should be expanded, or adjacent structural areas should be adjusted to become new key monitoring areas.

[0041] Specifically, the degree of anomaly contribution refers to the role and impact of a specific monitoring point in the overall anomaly of a building or region. Its determination can be based on various indicators, such as the magnitude of the decline in the monitoring point's health index, the degree of increase in its risk level, the correlation between the anomaly type and the overall anomaly, and the importance of the monitoring point's structural location. For example, if the health index of a monitoring point drops sharply and its anomaly type shows a strong correlation with other anomaly buildings in the area, then the anomaly contribution of that monitoring point is high. Based on this, key monitoring points refer to those monitoring points that have a significant impact on the building's health status or the regional safety assessment results and have a high degree of anomaly contribution. This can be determined by setting a threshold to filter monitoring points whose anomaly contribution exceeds a preset threshold, or by ranking and selecting the monitoring points with the highest anomaly contribution. Key monitoring areas refer to the building structural areas containing these key monitoring points, such as a specific floor, structural unit, or foundation. Identifying these areas allows for precise location of anomalies.

[0042] When the health index corresponding to a key monitoring point declines or the risk level increases, it indicates that there may be a risk of structural damage or performance degradation in that area. In this case, increasing the sampling frequency of the vibration sensor—that is, collecting more data points per unit time—can capture more detailed changes in the vibration signal, such as the appearance of high-frequency components or changes in subtle vibration patterns. This provides richer and more accurate raw data for subsequent anomaly diagnosis, helping to detect early signs of structural damage in a timely manner.

[0043] Furthermore, when the anomaly type corresponding to a key monitoring point persists across multiple consecutive assessment periods, this indicates that the anomaly may be persistent or evolving. In such cases, extending the data acquisition time window—that is, continuously acquiring data over a period of time rather than just momentary acquisition within a short period—allows for the acquisition of vibration data over a longer time series. This enables a better analysis of the long-term trend, periodic changes, or evolutionary patterns of the anomaly, and helps assess its stability, development speed, and cumulative impact on the structural performance of the building.

[0044] Furthermore, when multiple monitoring points within a key monitoring area show correlated anomalies, this may indicate that the anomaly has spread from a localized area to adjacent areas, or that the root cause of the anomaly has a wide-ranging impact. In this case, expanding the scope of the key monitoring area, such as including adjacent floors, structural units, or foundation areas, can comprehensively cover the areas potentially affected by the anomaly. Alternatively, if analysis suggests that the source of the anomaly may be located in an adjacent structural area not currently covered by the key monitoring area, this adjacent structural area can be designated as a new key monitoring area to ensure effective allocation of monitoring resources and prevent further spread or omission of the anomaly.

[0045] This technical solution collects vibration data from different monitoring points on a building and combines it with external influence data such as meteorology, traffic, groundwater level, construction activities, and seismic activity for unified time calibration and preprocessing. This allows for correlation analysis between the building's own vibration response and external disturbance factors under the same time reference, improving the completeness of monitoring data and the reliability of assessment. By extracting multi-dimensional building vibration characteristics such as natural frequency, vibration energy, cepstrum, three-dimensional vibration trajectory, foundation reflected waves, and elastic wave propagation, and inputting them together with external influence characteristics into the building assessment model, it is possible to analyze multiple aspects such as overall stiffness changes, local anomalies, changes in internal reflection paths, changes in foundation response, and changes in spatial vibration modes. This approach comprehensively assesses the health status of buildings from multiple perspectives, reducing false alarms and missed alarms caused by monitoring a single indicator. Furthermore, by correlating the health status assessment results and spatial relationships of multiple buildings within the target area, it can identify the group linkages between changes in the timing, type, and risk level of anomalies. This allows for the differentiation between synchronous anomalies caused by shared external influences, individual anomalies caused by changes in local structural conditions, and transmission anomalies caused by changes in adjacent buildings or the foundation environment. Consequently, regional safety assessment results and risk warning information can be generated based on individual building health assessments, improving the accuracy and continuity of building health monitoring and the timeliness of regional structural safety warnings.

[0046] Example 2 In some embodiments, the method further includes: Obtain structural information about the building; structural information includes at least one of the following: building height, floor distribution, load-bearing structure layout, foundation location, support node location, and span direction; The key structural regions of a building are determined based on structural information; the key structural regions include at least one of the following: overall vibration response region, local structural response region, foundation response region, and environmental vibration background region. The location of vibration sensor monitoring points is determined based on key structural areas; including: When the building is a layered structure, the top of the building, the middle structural layer of the building, and the foundation of the building are determined as the monitoring point locations; When the building is a high-rise building, multiple vertical monitoring levels are determined based on the floor height and floor distribution, and the top of the building, multiple vertical monitoring levels, and the foundation position at the bottom of the building are determined as monitoring point locations; When the building is a large-span building, multiple lateral monitoring areas are determined based on the span direction and the location of the support nodes, and the lateral monitoring areas and the location of the support nodes are determined as the monitoring point locations.

[0047] Specifically, the structural information may include at least one of the following: building height, floor distribution, load-bearing structure layout, foundation location, support node location, and span direction. Among these, building height provides overall dimensional information and influences its overall vibration mode; floor distribution reveals the vertical distribution of mass and stiffness, significantly impacting inter-story displacement and shear response; the layout of load-bearing structures (such as shear walls, frame columns, beams, etc.) determines the load transfer path and structural stiffness distribution; foundation location is the critical interface connecting the building and the foundation, affecting foundation-structure interaction; support node location is particularly important for large-span structures, serving as stress concentration and deformation-sensitive areas; and the span direction indicates the main force direction and deformation characteristics of the structure. By obtaining this information, a comprehensive understanding of the building's structural characteristics can be achieved.

[0048] Based on this, the structural critical areas of the building are determined according to the structural information. Based on the acquired structural information, the areas within the building that are most sensitive to external excitations, most prone to damage, or have the greatest impact on overall structural safety can be identified—these are the structural critical areas. These areas are the focus of vibration monitoring, providing data that best reflects changes in the building's health status. The structural critical areas may include at least one of the following: overall vibration response area, local structural response area, foundation response area, and environmental vibration background area. Specifically, the overall vibration response area typically refers to the top of the building or areas sensitive to changes in overall stiffness; its vibration data reflects the overall modal and stiffness changes of the building. Local structural response areas refer to specific components (such as beams, columns, and floor slabs) or connection nodes; damage in these areas may lead to a decrease in local stiffness or stress concentration. The foundation response area focuses on the connection between the building and the foundation, used to monitor foundation settlement, liquefaction, or foundation damage. The environmental vibration background area is used to collect environmental noise and ground vibration data for background noise subtraction and environmental impact assessment during data processing.

[0049] Subsequently, the monitoring point locations of the vibration sensors are determined based on the critical structural areas. After identifying the critical structural areas, the monitoring point locations of the vibration sensors will be selectively chosen within or near these areas. This structural characteristic-based deployment strategy aims to ensure that the sensors can capture vibration signals that best reflect changes in the building's health condition, avoiding inefficient monitoring or omission of crucial information due to indiscriminate deployment.

[0050] Figure 2 This is a schematic diagram of the vibration sensor layout at different monitoring points of a building according to one embodiment of this application, with reference to... Figure 2 When the building is a layered structure, the monitoring points are located at the top, middle structural layers, and foundation. Preferably, additional monitoring points are set up around the perimeter of the building. For layered structures, the vibration response typically exhibits significant inter-layer shear or bending deformation. Placing sensors at the top of the building allows for the capture of the maximum displacement or acceleration of the overall vibration response; placing them in the middle structural layers helps monitor inter-layer deformation and localized damage; and placing them at the foundation effectively monitors the interaction between the foundation and the structure, as well as the stability of the foundation. This layered deployment method comprehensively reflects the vertical vibration characteristics of layered structures.

[0051] When the building is a high-rise, multiple vertical monitoring levels are determined based on the floor height and distribution, and the top of the building, the multiple vertical monitoring levels, and the foundation location are identified as monitoring point locations. Due to their great height and high flexibility, high-rise buildings exhibit more complex vibration modes, potentially including higher-order modes and torsional vibrations. Therefore, in addition to the top and foundation locations, monitoring points need to be set according to the floor height and distribution, for example, at regular intervals or at points of abrupt changes in structural stiffness, forming multiple vertical monitoring levels. This allows for more precise capture of the vibration response of high-rise buildings at different heights, identifying potential localized damage or overall instability risks.

[0052] When the building is a long-span structure, multiple lateral monitoring areas are determined based on the span direction and the location of support nodes, and these lateral monitoring areas and support node locations are designated as monitoring point locations. A characteristic of long-span buildings is that their main stress direction and deformation are concentrated in the span direction. Therefore, monitoring points should be strategically placed at key locations along the span direction, such as the mid-span and quarter-span points, to capture bending or torsional deformation. Simultaneously, support node locations are critical points for load transfer and stress concentration; placing sensors at these locations can effectively monitor the health of the support system. This combined lateral and node-based deployment allows for a comprehensive reflection of the structural behavior of long-span buildings.

[0053] Example 3 In some embodiments, building vibration characteristics are extracted based on the data to be analyzed, including: The vibration data in the data to be analyzed is segmented according to a preset sliding time window to obtain multiple continuous vibration analysis segments; Perform spectral analysis on each vibration analysis segment and extract the principal mode frequencies corresponding to the spectral peaks; Natural frequency characteristics are generated based on the variation amplitude and trend of the main modal frequencies within a continuous sliding time window; By comparing the natural frequency characteristics with the historical reference frequency characteristics of the corresponding building, the overall stiffness variation state of the building can be determined.

[0054] Specifically, when extracting building vibration characteristics, the vibration data in the analysis data, which has undergone uniform time calibration, denoising, filtering, outlier removal, and standardization, must first be processed. This processing is achieved by setting a preset sliding time window, which slides across the vibration data stream at a fixed length. Each slide can overlap to some extent, thus dividing the continuous vibration data into multiple continuous vibration analysis segments. For example, the length of the preset sliding time window can be set from several seconds to tens of seconds, depending on the building's structural characteristics and monitoring requirements. The sliding step size can be set to 50% or less of the time window length to ensure sufficient continuity between adjacent segments while capturing dynamic changes in vibration characteristics.

[0055] Subsequently, spectral analysis is performed on each obtained vibration analysis segment. Spectral analysis aims to transform the time-domain vibration signal into the frequency domain to reveal its frequency components. Commonly used spectral analysis methods include Fast Fourier Transform (FFT) or Power Spectral Density (PSD) estimation. Through spectral analysis, peaks with significant energy in the spectrum can be identified; the frequencies corresponding to these peaks are the principal modal frequencies. The principal modal frequencies are the natural frequencies of a building under specific vibration modes, and their values ​​are closely related to the building's mass and stiffness.

[0056] After obtaining the principal modal frequencies of continuous vibration analysis segments, it is necessary to further analyze the amplitude and trend of these principal modal frequencies. For example, the percentage change of principal modal frequencies within adjacent time windows can be calculated, or the evolution curve of principal modal frequencies over a period of time can be fitted using methods such as regression analysis. This comprehensive information on the amplitude and trend of these changes constitutes the natural frequency characteristics. Natural frequency characteristics not only reflect the frequency value at a specific moment, but more importantly, reveal the dynamic process of the building structure's stiffness changing over time.

[0057] Finally, the generated natural frequency characteristics are compared with the historical reference frequency characteristics of the corresponding building. Historical reference frequency characteristics are typically established within the normal fluctuation range when the building is in good health or through long-term monitoring. This comparison allows us to determine whether the current building's natural frequency characteristics deviate from the normal range, and the degree and direction of this deviation. For example, if the natural frequency characteristics show a continuous decrease in the principal modal frequency or a significant drop below the historical reference, it can be determined that the overall stiffness of the building has decreased, thus indicating the state of change in the building's overall stiffness.

[0058] In some embodiments, extracting building vibration characteristics based on the data to be analyzed further includes: The combination of monitoring points to be analyzed is determined from the vibration data corresponding to different monitoring points of the building; Cross-correlation processing is performed on vibration data from different monitoring points in the same combination of monitoring points to be analyzed, and the structural response function between different monitoring points is reconstructed. Autocorrelation processing was performed on vibration data from the same monitoring point to extract internal building reflection response information. Based on the structural response function and internal reflection response information, the changes in the propagation path and reflection path of vibration within the building structure are determined, and vibration interference characteristics are generated.

[0059] Specifically, the step of determining the combination of monitoring points to be analyzed from vibration data corresponding to different monitoring points within the building aims to select vibration sensors on specific areas or components inside the building to form pairs or combinations of monitoring points for subsequent analysis. The selection of these combinations of monitoring points can be based on the building's structural characteristics, potential weaknesses, or areas requiring focused attention. For example, sensors located at both ends of the same beam segment, at different locations on the same floor slab, or on adjacent columns can be selected to form a combination. In this way, the dynamic response of specific structural units can be focused, laying the foundation for in-depth analysis of local structural behavior.

[0060] Subsequently, cross-correlation processing is performed on the vibration data from different monitoring points within the same set of monitoring points to reconstruct the structural response function between the different monitoring points. Cross-correlation processing is a signal processing technique used to measure the similarity between two signals at different time delays. In this step, by performing cross-correlation calculations on the vibration data from selected different monitoring points (e.g., monitoring point A and monitoring point B), a function can be obtained that characterizes the vibration propagation characteristics from monitoring point A to monitoring point B, i.e., the structural response function. This function includes information such as the time delay, attenuation, and phase change of the vibration signal propagating in the structure, reflecting the transmission path and medium characteristics of vibration energy within the structure. For example, when the structural stiffness changes, the peak position or shape in the cross-correlation function will change.

[0061] Simultaneously, autocorrelation processing is performed on vibration data from the same monitoring point to extract reflection response information within the building. Autocorrelation processing measures the similarity between a signal and itself at different time delays. For vibration data collected from a single monitoring point, autocorrelation processing can reveal the reflected wave information generated after the vibration wave propagates inside the building and encounters structural discontinuities (such as cracks, voids, material interfaces, etc.). The peak value in the autocorrelation function typically corresponds to the round-trip propagation time of the vibration wave within the structure. This reflection response information can indirectly reflect the integrity and uniformity of the building's internal structure. For example, when a new crack appears inside the building, it may introduce a new reflection peak or alter the characteristics of existing reflection peaks in the autocorrelation function.

[0062] Based on this, according to the structural response function and internal reflection response information, the changes in the propagation path and reflection path of vibration within the building structure are determined, generating vibration interference characteristics. This step comprehensively analyzes the structural response function (characterizing vibration propagation) obtained through cross-correlation processing and the internal reflection response information (characterizing vibration reflection) obtained through autocorrelation processing. By comparing these functions at the current moment with the baseline functions under the building's healthy state, subtle changes in the vibration propagation path (such as propagation velocity and attenuation) and reflection path (such as reflection intensity and reflection time) can be identified. These changes may be caused by factors such as structural damage, material degradation, or loose connections. After quantifying these changes, vibration interference characteristics are generated, which can serve as a sensitive indicator of changes in the internal structural state of the building, for example, indicating local stiffness loss, crack propagation, or delamination.

[0063] In some embodiments, extracting building vibration characteristics based on the data to be analyzed further includes: The vibration propagation delay between different monitoring points is determined based on the structural response function; Based on the spatial relationship between different monitoring points and the vibration propagation time delay, the elastic wave velocity change information is determined; The elastic wave velocity variation information is divided according to the frequency band range to obtain the elastic wave velocity characteristics of low frequency band, mid frequency band, and high frequency band. The overall structural state of a building is characterized by low-frequency elastic wave velocity characteristics, the local structural state of a building is characterized by mid-frequency elastic wave velocity characteristics, and the surface or micro-damage state of a building is characterized by high-frequency elastic wave velocity characteristics.

[0064] Specifically, determining the vibration propagation delay between different monitoring points based on the structural response function can be achieved by analyzing the peak position of the cross-correlation function. The peak value of the cross-correlation function on the time axis typically corresponds to the time required for the vibration signal to propagate from one monitoring point to another. For example, peak detection can be performed on the cross-correlation function, and the time when the peak occurs is determined as the vibration propagation delay.

[0065] After determining the vibration propagation time delay between different monitoring points, the velocity variation of the elastic wave can be calculated by combining the spatial relationships between these monitoring points. Elastic wave velocity is a direct reflection of the structural material properties and damage state, and it is calculated by dividing the distance between monitoring points by the vibration propagation time delay. The spatial relationships between monitoring points can be obtained through a pre-measured sensor layout map or a high-precision positioning system (such as GPS). By comparing the calculated elastic wave velocity with the historical baseline velocity of the building under healthy conditions, the variation information of the elastic wave velocity can be obtained; this variation information can be an absolute change or a relative rate of change.

[0066] To identify structural damage at different scales, this application divides the elastic wave velocity variation information according to frequency bands, thereby obtaining low-frequency, mid-frequency, and high-frequency elastic wave velocity characteristics. The sensitivity of elastic waves to structural damage varies at different frequencies. This can typically be achieved by bandpass filtering the original vibration signal and then calculating the propagation delay and elastic wave velocity for each frequency band. For example, the low-frequency band can be set to 0-10Hz, the mid-frequency band to 10-100Hz, and the high-frequency band to above 100Hz. The specific frequency band division can be optimized based on the building's structural type, material properties, and potential damage modes.

[0067] Among them, the low-frequency elastic wave velocity characteristics are mainly used to characterize the overall structural state of a building. Because low-frequency elastic waves have longer wavelengths and stronger penetrating power, their velocity changes can reflect the overall stiffness of the building, the integrity of connections between major load-bearing components, and other macroscopic structural conditions. For example, a significant decrease in low-frequency elastic wave velocity may indicate a reduction in the overall structural stiffness of the building or extensive damage to major load-bearing components.

[0068] Mid-frequency elastic wave velocity characteristics are used to characterize the local structural condition of a building. Mid-frequency elastic waves are more sensitive to mesoscale structural changes such as damage to local components (e.g., beams, columns, slabs) and crack propagation. Therefore, localized anomalous changes in mid-frequency elastic wave velocity may indicate problems such as cracks, material degradation, or loose connections in components within a specific area.

[0069] High-frequency elastic wave velocity characteristics are used to characterize the surface or micro-damage state of buildings. High-frequency elastic waves have short wavelengths and are highly sensitive to micro-damage such as minute defects, micro-cracks, and early-stage material degradation on material surfaces. Therefore, subtle changes in high-frequency elastic wave velocity may reveal early signs of damage such as coating peeling, concrete carbonation, and micro-cracks caused by steel corrosion on building surfaces.

[0070] In some embodiments, extracting building vibration characteristics based on the data to be analyzed further includes: The vibration data in the data to be analyzed is transformed in the frequency domain to obtain the vibration spectrum data; Logarithmic and inverse transform processes are performed on the vibration spectrum data to obtain cepstral data. Extract the cepstral peak position, cepstral peak intensity, and cepstral peak variation trend from the cepstral data; Determine the reflection period variation corresponding to the internal reflection path of the building based on the cepstral peak position; Based on the cepstral peak intensity and the cepstral peak variation trend, cepstral features are generated to characterize the layering, cracks, or local loosening of the internal structure of a building.

[0071] Specifically, firstly, the vibration data in the data to be analyzed undergoes frequency domain transformation to obtain vibration spectrum data. Frequency domain transformation is the process of converting a time-domain signal into a frequency-domain signal, aiming to reveal the frequency components and their intensities contained in the signal. Common frequency domain transformation methods include Fast Fourier Transform (FFT) or Short-Time Fourier Transform (STFT). Through frequency domain transformation, the original vibration time series data can be transformed into vibration spectrum data, which contains information on the distribution of vibration energy at different frequencies. This is helpful for subsequent analysis of the signal's periodicity, harmonic components, and resonant frequencies.

[0072] Secondly, the vibration spectrum data undergoes logarithmic processing and inverse transform processing to obtain cepstrum data. Logarithmic processing typically involves performing a logarithmic operation on the amplitude of the vibration spectrum data to compress the dynamic range and convert multiplicative components into additive components. Inverse transform processing involves performing an inverse Fourier transform (IFFT) or inverse short-time Fourier transform on the logarithmically processed spectrum data to obtain cepstrum data. Cepstrum is a signal processing technique that performs an inverse Fourier transform on the logarithm of a signal's spectrum, aiming to separate the excitation source (such as an impact) and the system response (such as multipath reflections) in the signal. In building vibration analysis, cepstrum data can effectively reveal periodic structures in the signal, especially periodic delays caused by internal reflection paths.

[0073] Based on this, cepstral peak positions, cepstral peak intensities, and cepstral peak variation trends are extracted from the cepstral data. Cepstral data typically exhibits peaks at specific "quefrencies." The cepstral peak positions correspond to the periodic repetition delays in the signal, such as the round-trip time of a vibration wave reflecting and propagating between different structural layers within a building. The cepstral peak intensities reflect the significance of these periodic reflections. The cepstral peak variation trend refers to the dynamic evolution of the cepstral peak positions and intensities over different time periods or under different monitoring conditions. These parameters can be extracted using peak detection algorithms, threshold settings, and time series analysis.

[0074] Furthermore, the reflection period variation corresponding to the reflection path within the building is determined based on the cepstral peak position. Structural delamination, cracks, or localized loosening within the building can cause reflection and scattering of vibration waves during propagation, resulting in a multipath effect. These reflected waves manifest as specific peaks in the cepstral domain. The cepstral peak position directly corresponds to the propagation delay time of these reflected waves, i.e., the reflection period. By monitoring changes in these reflection periods, changes in the building's internal structure can be indirectly inferred, such as crack expansion or exacerbated delamination, as these changes affect the propagation path and velocity of vibration waves.

[0075] Finally, based on the cepstral peak intensity and its variation trend, cepstral features are generated to characterize the delamination, cracks, or localized loosening of the building's internal structure. The cepstral peak intensity reflects the energy of the reflected signal. When delamination, cracks, or loosening occur in the building's internal structure, the characteristics of the reflecting interface change, thus affecting the intensity of the reflected wave. The cepstral peak variation trend provides dynamic information about the evolution of these internal structural states over time. For example, the propagation of cracks may lead to the emergence of new reflection paths or changes in the intensity of existing reflection paths. By comprehensively analyzing the cepstral peak intensity and its variation trend, diagnostically significant cepstral features can be generated, which can sensitively indicate whether damage such as delamination, cracks, or localized loosening exists in the building's internal structure.

[0076] In some embodiments, extracting building vibration characteristics based on the data to be analyzed further includes: Low-frequency vibration response data are extracted from vibration data corresponding to building foundation monitoring points and surrounding ground surface monitoring points. The changes in ground reflected waves, the state of low-frequency vibration enhancement, and the state of ground elastic wave velocity changes are determined based on low-frequency vibration response data. By comparing the changes in ground reflected waves, the enhanced state of low-frequency vibration, and the changes in ground elastic wave velocity with historical ground benchmark characteristics, the ground response characteristics are obtained. Identify at least one of the following foundation anomalies based on foundation response characteristics: uneven foundation settlement, soil softening, and groundwater level changes.

[0077] Specifically, in order to obtain vibration information closely related to the characteristics of the building foundation and surrounding soil medium, this application extracts low-frequency vibration response data from vibration data corresponding to monitoring points on the building foundation and the surrounding ground surface. The vibration sensor can be an accelerometer, velocity sensor, or displacement sensor, and its sampling frequency and range should cover the target low-frequency range, for example, 0.1 Hz to 20 Hz. After preprocessing (such as detrending and removing high-frequency noise) the collected raw vibration data, the low-frequency vibration response data is extracted through bandpass filtering (e.g., setting the cutoff frequency to 20 Hz). These data reflect the dynamic response of the foundation under external excitations (such as wind loads, traffic loads, and seismic micro-motions).

[0078] Based on this, the changes in ground reflected waves, the state of low-frequency vibration enhancement, and the state of ground elastic wave velocity variation are determined according to the low-frequency vibration response data. Changes in ground reflected waves can be identified by performing cross-correlation analysis or spectral analysis on vibration data between foundation monitoring points and surface monitoring points to determine the arrival time, amplitude, and phase changes of the reflected waves. The state of low-frequency vibration enhancement can be determined by calculating the root mean square value, peak value, or integral value of the energy spectral density of the low-frequency vibration response data in a specific low-frequency band and comparing it with historical benchmark values. The state of ground elastic wave velocity variation can be determined by using multi-point vibration data for wave velocity inversion. For example, by measuring the propagation time delay of vibration waves between different monitoring points and combining this with the spatial distance between the monitoring points, the propagation velocity of elastic waves (such as Rayleigh waves or shear waves) can be calculated.

[0079] Furthermore, the changes in foundation reflected waves, low-frequency vibration enhancement, and foundation elastic wave velocity are compared with historical foundation benchmark characteristics to obtain foundation response characteristics. Historical benchmark characteristics are collected and established during the initial construction phase of the building or under known foundation health conditions, representing the normal operating state of the foundation. Comparison can be performed using statistical methods such as calculating the Euclidean distance, correlation coefficient, or relative rate of change between the current state and the benchmark state. These quantified change indicators are then combined to form a multidimensional foundation response characteristic vector.

[0080] Finally, based on the foundation response characteristics, at least one of the following foundation anomalies is identified: uneven foundation settlement, soil softening, and groundwater level changes. For example, uneven foundation settlement may manifest as increased differences in low-frequency vibration responses between different foundation monitoring points, or an inconsistent decreasing trend in the elastic wave velocity across different regions, as well as local distortion of the reflected wave path; soil softening typically leads to a general decrease in the elastic wave velocity, an overall enhancement of the low-frequency vibration response, and attenuation of the reflected wave amplitude or a delay in arrival time; groundwater level changes may alter the effective stress of the foundation, thereby affecting the elastic wave velocity (e.g., rising water levels may lead to a decrease in the shear wave velocity of saturated soil, or changes in pore water pressure may affect the reflected wave characteristics), and may be accompanied by changes in low-frequency vibration characteristics. The identification process can employ rule-based expert systems, pattern recognition algorithms (such as support vector machines, neural networks), or cluster analysis to match the current foundation response characteristics with a pre-trained anomaly pattern library, thereby identifying the specific foundation anomaly.

[0081] In some embodiments, extracting building vibration characteristics based on the data to be analyzed further includes: Obtain vibration components of the same monitoring point in different directions; Construct three-dimensional vibration trajectories corresponding to monitoring points based on vibration components in different directions; The main vibration direction, major axis, minor axis, and deflection state of the trajectory are determined based on the three-dimensional vibration trajectory. By comparing the main vibration direction, the major axis of the trajectory, the minor axis of the trajectory, and the trajectory deflection state with the historical baseline trajectory, the vibration state in the abnormal direction is determined. Three-dimensional vibration trajectory features are generated based on the abnormal direction vibration state to characterize the abnormal vibration mode of the building.

[0082] Specifically, when acquiring vibration components in different directions at the same monitoring point, this is typically achieved by deploying multi-axial vibration sensors (such as triaxial accelerometers or triaxial displacement sensors) at the building monitoring point. These sensors can simultaneously collect time-series vibration data of the building in the three orthogonal directions of X, Y, and Z, thus providing basic data for subsequent three-dimensional trajectory construction.

[0083] When constructing the three-dimensional vibration trajectory corresponding to the monitoring point based on the vibration components in different directions, the time-series vibration components collected from the same monitoring point in different directions are integrated. For example, if vibration data in three directions X(t), Y(t), and Z(t) are collected, then at each time point t, the spatial position of the monitoring point can be represented as (X(t), Y(t), Z(t)). Connecting these consecutive spatial position points forms the three-dimensional motion trajectory of the monitoring point over a period of time. This trajectory intuitively reflects the actual vibration path of the building in space.

[0084] When determining the principal vibration direction, trajectory major axis, trajectory minor axis, and trajectory deflection state based on the three-dimensional vibration trajectory, multivariate statistical methods such as principal component analysis (PCA) can be used to analyze the constructed three-dimensional vibration trajectory. The principal vibration direction usually corresponds to the direction with the largest variance in the trajectory data, i.e., the direction of the first principal component, which indicates the most significant vibration trend of the building at that monitoring point. The trajectory major axis and trajectory minor axis represent the maximum and minimum extension range of the trajectory in the principal vibration direction and its orthogonal directions, respectively, and can be quantified by the eigenvalues ​​of the principal components. The trajectory deflection state describes the degree of rotation or tilt of the principal vibration direction or the overall spatial attitude of the trajectory relative to a certain reference direction or historical benchmark, and can be obtained by comparing the angular changes of the principal component vectors over different time periods.

[0085] When comparing the main vibration direction, major axis, minor axis, and deflection state with historical baseline trajectories to determine abnormal vibration states, it is first necessary to establish historical baseline trajectory parameters for the building under normal and healthy conditions. These baseline parameters can be statistical averages, confidence intervals, or specific thresholds. Then, the currently monitored main vibration direction, major axis, minor axis, and deflection state are compared with these historical baselines. If there is a significant deviation between the current parameters and the baseline parameters, such as exceeding a preset threshold range, an abnormal vibration state is determined to exist. This comparison helps identify abnormal responses in specific directions of the building, such as changes in vibration patterns due to localized damage.

[0086] When generating three-dimensional vibration trajectory features to characterize the abnormal vibration modes of a building based on the aforementioned abnormal vibration states, the abnormal vibration states identified in the comparison results are quantified and encoded to form a structured feature vector. This feature vector may include the type of anomaly (e.g., deflection of the principal vibration direction, significant shortening or lengthening of the major axis, abnormal increase in the minor axis, etc.), the degree of anomaly (e.g., deflection angle, numerical value of the rate of change of axis length), and the frequency or duration of the anomaly. This feature vector, as a three-dimensional vibration trajectory feature, can comprehensively and concisely describe the abnormalities in the building's spatial vibration modes, providing crucial input for subsequent building health status assessment models.

[0087] Example 4 In some embodiments, the method further includes: Record the building health status assessment results for each building in chronological order of assessment time to generate a historical sequence of building health status. Based on the historical sequence of building health status, extract the trends of health index changes, abnormality type evolution trends, and risk level changes; The trend of building structural performance degradation is determined based on the trends of health index changes, abnormality type evolution, and risk level changes. By comparing the degradation trend of building structural performance with the historical baseline degradation trend, the risk development status of the building can be predicted; Based on the risk development status, life prediction results and maintenance tips are generated; the life prediction results are used to characterize the remaining safe service status of the building under the current degradation trend.

[0088] Specifically, the method involves recording the building health status assessment results for each building in chronological order, generating a historical sequence of building health status. The building health status assessment results are health indices, anomaly types, and risk levels output by a pre-trained building assessment model based on vibration and external influence data. These assessment results, along with assessment timestamps, are stored and managed in chronological order of their generation, forming a continuous time-series data set, i.e., the historical sequence of building health status. This historical sequence data can be stored in a dedicated time-series database or archived through a structured file system to ensure data integrity, traceability, and efficient retrieval.

[0089] Based on this, and according to the historical sequence of the building's health status, the trends in health index changes, anomaly type evolution trends, and risk level changes are extracted. For extracting health index trends, time series analysis methods, such as moving averages, exponential smoothing, or regression analysis, can be used to smooth the data and identify long-term upward or downward trends. For example, the rate of increase or decrease can be quantified by calculating the slope of the health index over time. For extracting anomaly type evolution trends, the frequency and duration of different anomaly types in the historical sequence, as well as possible transition patterns between types, can be statistically analyzed, such as the evolution from minor damage to more serious structural problems. For extracting risk level change trends, the rise and fall patterns of risk levels over time can be analyzed, such as the gradual transition from low risk to medium risk. State transition models can be used to analyze the evolution path.

[0090] Furthermore, based on the trends in health index changes, anomaly type evolution, and risk level changes, the structural performance degradation trend of the building is determined. This is typically achieved by constructing a multi-factor fusion model, which can be based on machine learning algorithms (such as support vector machines, neural networks) or expert systems. Using the extracted trends in health index changes, anomaly type evolution, and risk level changes as input features, the model learns the intrinsic correlation between these trends in historical data and the actual structural performance degradation of the building, outputting a quantitative structural performance degradation trend index, such as "slight degradation," "moderate degradation," or "severe degradation," or a continuous degradation rate value.

[0091] Subsequently, the degradation trend of the building's structural performance is compared with the historical baseline degradation trend to predict the building's risk development status. The historical baseline degradation trend can be an average degradation curve obtained from long-term monitoring data of a large number of similar buildings, or an ideal degradation curve calculated based on building design codes, material aging models, and other theories. The comparison method can employ statistical methods, such as hypothesis testing or correlation analysis, to assess the similarity or difference between the current building's degradation rate and pattern and the baseline trend. Based on the comparison results, the probability and time of the building reaching a certain preset risk threshold in the future can be predicted. For example, if the current degradation rate is significantly faster than the baseline, its risk level is predicted to increase in a shorter period of time.

[0092] Finally, based on the risk development status, a lifespan prediction result and maintenance reminder information are generated. The lifespan prediction result is used to characterize the remaining safe service status of the building under the current degradation trend. The lifespan prediction result is based on the predicted risk development status, combined with a preset failure threshold (e.g., when the health index falls below a certain critical value or the risk level reaches "high risk"), and calculates the time required for the building to reach that threshold through extrapolation or a prediction model, presented in the form of years, months, etc. The maintenance reminder information is a targeted maintenance suggestion automatically generated based on the predicted risk development status, the evolution trend of anomaly types, and the degradation trend of structural performance. For example, if the degradation rate of a certain component is predicted to accelerate, the system may prompt for "regular inspection," "reinforcement," or "component replacement," and provide suggested maintenance time, maintenance content, and priority.

[0093] The following example will provide a more detailed explanation of the above technical solution: Within an urban area, there are multiple high-rise buildings, such as Building B1, Building B2, and Building B3. These buildings are subject to a variety of external factors over a long period, including vibrations from surrounding traffic, seasonal weather changes (such as strong winds and sudden temperature drops), nearby subway construction activities, and fluctuations in groundwater levels. Traditional monitoring methods often struggle to comprehensively assess the health status of these buildings, particularly in distinguishing between structural damage and environmental disturbances, and in identifying abnormal linkages between building clusters.

[0094] This scheme first involves continuously collecting vibration data from the top, middle structural layers, foundation locations, and surrounding areas of buildings B1, B2, and B3 by deploying vibration sensors. Simultaneously, the system acquires external impact data related to the target area A, including meteorological data from weather stations (such as wind speed, wind direction, and temperature), traffic data from traffic monitoring systems (such as traffic flow and frequency of heavy-duty vehicles), groundwater level data from groundwater monitoring wells, construction activity data from the construction site (such as piling and blasting time and intensity), and seismic activity data from seismic networks.

[0095] Subsequently, the collected vibration data and external influence data were uniformly time-calibrated to ensure that all data were aligned on the time axis. Next, these data were denoised and filtered to eliminate environmental interference and sensor noise, outliers were removed to eliminate occasional erroneous data, and standardization was performed to eliminate dimensional differences, thus obtaining the data to be analyzed for subsequent analysis.

[0096] Based on this data to be analyzed, the system extracts the building vibration characteristics and external influence characteristics of each building. In extracting building vibration characteristics, this scheme employs a multi-dimensional analysis method. For example, by performing sliding time window segmentation and spectral analysis on the vibration data, the main modal frequencies are extracted, and natural frequency characteristics are generated based on their variation amplitude and trend to determine the overall stiffness variation state of the building. Simultaneously, by performing cross-correlation processing on vibration data from different monitoring points, the structural response function is reconstructed, and by performing autocorrelation processing on data from the same monitoring point, internal reflection response information is extracted, thereby determining the propagation path and reflection path changes within the structure and generating vibration interference characteristics. Further, the vibration propagation time delay between different monitoring points is determined based on the structural response function, and elastic wave velocity variation information is calculated by combining spatial positional relationships. This information is then divided by frequency band to characterize the overall structure, local structure, and surface or micro-damage states of the building. In addition, cepstral data is obtained through frequency domain transformation, logarithmic processing, and inverse transformation. The cepstral peak positions, intensities, and variation trends are extracted to generate cepstral characteristics, which are used to characterize the internal structural delamination, cracks, or local loosening states of the building. For foundation conditions, variations in ground reflected waves, low-frequency vibration enhancement, and elastic wave velocity are extracted from low-frequency vibration response data at foundation monitoring points and surrounding surface monitoring points. These are then compared with historical baseline characteristics to identify anomalous conditions such as uneven foundation settlement, soil softening, or groundwater level changes. Simultaneously, vibration components at the same monitoring point in different directions are acquired to construct a three-dimensional vibration trajectory. The dominant vibration direction, major axis, minor axis, and deflection state are determined and compared with historical baseline trajectories to generate three-dimensional vibration trajectory features, used to characterize anomalous vibration modes. When extracting external influence features, wind load and temperature change features are extracted from meteorological data, traffic load features from traffic data, groundwater level change features from groundwater level data, construction vibration features from construction activity data, and seismic response features from seismic activity data.

[0097] Subsequently, these multi-dimensional building vibration characteristics and external influence characteristics are input into a pre-trained building assessment model. This model, trained on extensive historical data, can comprehensively analyze these characteristics and output health status assessment results for each building, including a health index (quantifying health level), anomaly type (such as overall stiffness reduction, local cracks, foundation settlement, etc.), and risk level (such as low, medium, and high risk). Compared to existing technologies that rely solely on a single vibration index or fail to adequately consider external influences, this approach significantly improves the accuracy and comprehensiveness of the assessment through multi-feature fusion and the combination of external influence data. It avoids misjudging environmental disturbances as structural damage or masking early structural anomalies within external disturbances.

[0098] After obtaining the health status assessment results of each building, the system generates building status association data for target area A based on these results and the spatial relationship between buildings B1, B2, and B3. For example, it records that building B1 developed localized cracks at a certain point in time, building B2 experienced a decrease in overall stiffness during the same period, while building B3 remained normal.

[0099] Next, based on the correlation data of building cluster status, the system performs correlation analysis on the changes in the occurrence time, anomaly type, and risk level of anomalies among buildings exhibiting abnormalities within target area A. For example, if buildings B1 and B2 simultaneously exhibit similar vibration pattern anomalies during the same strong winds, the system will identify this as a group anomaly linkage. If the foundation settlement anomaly of building B1 precedes the local structural stress anomaly of the adjacent building B2, a transmission anomaly may be identified. This correlation analysis capability is not available in existing single-building monitoring methods; it can reveal the complex interactions within a building cluster.

[0100] Based on the identified interconnected anomalies, the system further determines the causes of anomalies in target area A and generates a regional safety assessment result. For example, if multiple buildings simultaneously experience abnormal vibrations during strong winds, it is identified as a synchronous anomaly caused by a common external influence (strong winds). If only building B1 develops localized cracks, unrelated to external influences or adjacent buildings, it is identified as an individual anomaly caused by changes in local structural conditions. If uneven settlement of the foundation of building B1 leads to structural stress concentration in adjacent building B2, it is identified as a transmission anomaly caused by changes in adjacent buildings or the foundation environment. Finally, a regional safety assessment report containing these analytical results is generated.

[0101] Finally, based on the health status assessment results of each building and the regional safety assessment results of target area A, risk warning information is generated. For example, a warning is sent to relevant management departments, pointing out that "building B1 has a moderate risk of localized damage, building B2 has a slight risk of overall stiffness reduction, and target area A is experiencing regional synchronous anomalies caused by strong winds, recommending a comprehensive inspection of all high-rise buildings in the area." This comprehensive warning information not only focuses on individual buildings but also provides a comprehensive safety situation awareness at the regional level, providing strong support for urban public safety management.

[0102] It is understood that the same or similar parts in the above embodiments can be referred to each other, and the contents not described in detail in some embodiments can be referred to the same or similar contents in other embodiments.

[0103] Any process or method description in the flowchart or otherwise herein can be understood as representing a module, segment, or portion of code comprising one or more executable instructions for implementing a particular logical function or process, and the scope of the preferred embodiments of this application includes additional implementations in which functions may be performed not in the order shown or discussed, including substantially simultaneously or in reverse order according to the functions involved, as should be understood by those skilled in the art to which embodiments of this application pertain.

[0104] It should be understood that various parts of this application can be implemented using hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented using software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0105] Those skilled in the art will understand that all or part of the steps of the methods in the above embodiments can be implemented by a program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, the program includes one or a combination of the steps of the method embodiments.

[0106] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0107] Although embodiments of this application have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting this application. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of this application.

Claims

1. A method for monitoring the health status of buildings based on vibration signals, characterized in that, include: Vibration data of the building is collected by vibration sensors deployed at different monitoring points in the building; Obtain external impact data related to the building; the external impact data includes at least one of meteorological data, traffic data, groundwater level data, construction activity data, and seismic activity data; The vibration data and external influence data are uniformly time-calibrated, and then subjected to noise reduction, filtering, outlier removal and standardization to obtain the data to be analyzed. Building vibration characteristics and external influence characteristics are extracted based on the data to be analyzed; the building vibration characteristics include at least one of the following: natural frequency characteristics, vibration energy characteristics, cepstral characteristics, three-dimensional vibration trajectory characteristics, foundation reflected wave characteristics, and elastic wave propagation characteristics; The building vibration characteristics and external influence characteristics are input into a pre-trained building assessment model, which outputs a building health status assessment result; the building health status assessment result includes a health index, anomaly type, and risk level; Based on the building health status assessment results and the spatial location relationship of each building in the target area, generate the building group status association data of the target area; Based on the building group status correlation data, correlation analysis is performed on the changes in the time of occurrence of anomalies, anomaly type and risk level among the buildings with anomalies in the target area to identify the linkage relationship of group anomalies; Based on the aforementioned group anomaly linkage relationship, the cause of the anomaly in the target area is determined and the regional security assessment result of the target area is generated; the cause of the anomaly in the target area includes regional synchronous anomaly caused by common external influences, individual anomaly caused by changes in local structural state, and transmission anomaly caused by changes in adjacent buildings or foundation environment; Based on the health status assessment results of each building and the regional safety assessment results of the target area, risk warning information is generated; Based on the building health status assessment results for each building and the regional safety assessment results for the target area, the degree of abnormal contribution of each monitoring point is determined. Based on the degree of abnormal contribution, key monitoring points are identified from multiple monitoring points, and the building structure area where the key monitoring points are located is identified as the key monitoring area; When the health index corresponding to the key monitoring point decreases or the risk level increases, the sampling frequency of the vibration sensor corresponding to the key monitoring point is increased. When the anomaly type corresponding to the key monitoring point persists for multiple consecutive evaluation periods, the data collection time window corresponding to the key monitoring point shall be extended. When multiple monitoring points within the key monitoring area show abnormal correlations, the scope of the key monitoring area is expanded, or adjacent structural areas are adjusted to become new key monitoring areas.

2. The method according to claim 1, characterized in that, The method further includes: Obtain structural information of the building; the structural information includes at least one of the following: building height, floor distribution, load-bearing structure layout, foundation location, support node location, and span direction; The key structural regions of the building are determined based on the structural information; the key structural regions include at least one of the following: overall vibration response region, local structural response region, foundation response region, and environmental vibration background region; Determining the monitoring point location of the vibration sensor based on the key areas of the structure includes: When the building is a layered structure, the locations of the top, middle structural layer and bottom foundation of the building are determined as monitoring points. When the building is a high-rise building, multiple vertical monitoring levels are determined based on the floor height and floor distribution, and the top of the building, the multiple vertical monitoring levels, and the foundation position at the bottom of the building are determined as monitoring point positions; When the building is a large-span building, multiple lateral monitoring areas are determined according to the span direction and the location of the support nodes, and the lateral monitoring areas and the location of the support nodes are determined as monitoring point locations.

3. The method according to claim 1, characterized in that, The building vibration characteristics are extracted based on the data to be analyzed, including: The vibration data in the data to be analyzed is segmented according to a preset sliding time window to obtain multiple continuous vibration analysis segments; Spectral analysis was performed on each vibration analysis segment to extract the principal mode frequency corresponding to the spectral peak. Based on the variation amplitude and trend of the main modal frequency within a continuous sliding time window, natural frequency characteristics are generated. The natural frequency characteristics are compared with the historical reference frequency characteristics of the corresponding building to determine the overall stiffness variation state of the building.

4. The method according to claim 1, characterized in that, Extracting building vibration characteristics based on the data to be analyzed also includes: The combination of monitoring points to be analyzed is determined from the vibration data corresponding to different monitoring points of the building; Cross-correlation processing is performed on vibration data from different monitoring points in the same combination of monitoring points to be analyzed, and the structural response function between different monitoring points is reconstructed. Autocorrelation processing was performed on vibration data from the same monitoring point to extract internal building reflection response information. Based on the structural response function and internal reflection response information, the changes in the propagation path and reflection path of vibration within the building structure are determined, and vibration interference characteristics are generated.

5. The method according to claim 4, characterized in that, Extracting building vibration characteristics based on the data to be analyzed also includes: The vibration propagation delay between different monitoring points is determined based on the structural response function. Based on the spatial relationship between different monitoring points and the vibration propagation time delay, the elastic wave velocity change information is determined; The elastic wave velocity variation information is divided according to the frequency band range to obtain the elastic wave velocity characteristics of low frequency band, mid frequency band, and high frequency band. The overall structural state of a building is characterized by the low-frequency elastic wave velocity characteristics, the local structural state of a building is characterized by the mid-frequency elastic wave velocity characteristics, and the surface or micro-damage state of a building is characterized by the high-frequency elastic wave velocity characteristics.

6. The method according to claim 1, characterized in that, Extracting building vibration characteristics based on the data to be analyzed also includes: The vibration data in the data to be analyzed is subjected to frequency domain transformation to obtain vibration spectrum data; The vibration spectrum data is subjected to logarithmic processing and inverse transform processing to obtain cepstral data; Based on the cepstral data, extract the cepstral peak position, cepstral peak intensity, and cepstral peak variation trend; The reflection period variation corresponding to the internal reflection path of the building is determined based on the cepstral peak position; Based on the cepstral peak intensity and the cepstral peak variation trend, cepstral features are generated to characterize the layering, cracks, or local loosening of the internal structure of a building.

7. The method according to claim 1, characterized in that, Extracting building vibration characteristics based on the data to be analyzed also includes: Low-frequency vibration response data are extracted from vibration data corresponding to building foundation monitoring points and surrounding ground surface monitoring points. Based on the low-frequency vibration response data, determine the changes in ground reflected waves, the state of low-frequency vibration enhancement, and the state of changes in ground elastic wave velocity. The changes in ground reflected waves, the enhanced state of low-frequency vibration, and the changes in ground elastic wave velocity are compared with the historical ground reference characteristics to obtain the ground response characteristics; Based on the aforementioned foundation response characteristics, at least one of the following foundation anomalies can be identified: uneven foundation settlement, soil softening, and groundwater level changes.

8. The method according to claim 1, characterized in that, Extracting building vibration characteristics based on the data to be analyzed also includes: Obtain vibration components of the same monitoring point in different directions; Construct a three-dimensional vibration trajectory corresponding to the monitoring point based on the vibration components in different directions; The main vibration direction, major axis, minor axis, and deflection state of the trajectory are determined based on the three-dimensional vibration trajectory. The main vibration direction, major axis of the trajectory, minor axis of the trajectory, and trajectory deflection state are compared with the historical reference trajectory to determine the vibration state in the abnormal direction; Based on the abnormal directional vibration state, a three-dimensional vibration trajectory feature is generated to characterize the abnormal vibration mode of the building.

9. The method according to claim 1, characterized in that, The method further includes: Record the building health status assessment results for each building in chronological order of assessment time to generate a historical sequence of building health status. Based on the historical sequence of the building's health status, extract the trends of health index changes, abnormality type evolution trends, and risk level changes; Based on the trends in health index changes, abnormality type evolution trends, and risk level changes, the degradation trend of building structural performance is determined. By comparing the degradation trend of the building's structural performance with the historical baseline degradation trend, the risk development status of the building can be predicted; Based on the risk development status, a lifespan prediction result and maintenance reminder information are generated; the lifespan prediction result is used to characterize the remaining safe use status of the building under the current degradation trend.

Images