A high-rise building abnormal vibration rapid judgment method based on time-frequency analysis

Abstract

The application discloses a high-rise building abnormal vibration rapid judgment method based on time-frequency analysis, and through steps one to three, a power spectrum density time-frequency diagram of vibration acceleration data of the high-rise building is constructed, and representative frequency bands, frequency components and occurrence time rules thereof are identified; through step four, a dynamic account is constructed as a conventional dynamic reference to determine conventional frequency components of the high-rise building and occurrence rules thereof; thus, when the high-rise building has an abnormal vibration event, the vibration acceleration data to be judged can be compared with the dynamic account through step five, so as to quickly distinguish abnormal components from a conventional background, accurately capture an abnormal candidate window containing an abnormal signal, quickly lock an abnormal occurrence time, effectively avoid false positives and false negatives caused by only relying on amplitude alarm, and improve the speed and accuracy of high-rise building abnormal vibration identification; and the application has applicability to scenes with or without long-term monitoring systems.

Description

Technical Field

[0001] This invention relates to the fields of building structural health monitoring, high-rise building dynamic response analysis, and building safety management, specifically a method for rapid assessment of abnormal vibrations in high-rise buildings based on time-frequency analysis. Background Technology

[0002] High-rise buildings are characterized by high flexibility, low damping, and sensitivity to environmental and operational conditions. Their dynamic response is influenced by multiple factors, including the structure itself, environmental conditions, equipment operation, personnel activities, traffic, and surrounding construction. As a large number of high-rise buildings enter their long-term service phase, the need to identify the normal response patterns of their dynamic characteristics and to quickly interpret abnormal conditions is becoming increasingly prominent.

[0003] Various vibration analysis and condition assessment techniques have been developed in the field of structural health monitoring. Operational modal analysis and random subspace identification methods can extract modal parameters such as the natural frequencies, mode shapes, and damping ratios of a structure. Signal processing methods such as short-time spectrum analysis and power spectral density analysis can be used to identify the frequency components of non-stationary vibration signals; however, current applications are mostly limited to the analysis of single time windows, single signal segments, or single spectral indices.

[0004] In building operation and maintenance practice, on-site dynamic testing and long-term structural health monitoring can obtain continuous vibration data, but there is still a lack of a methodological framework to transform massive monitoring data into an accumulative and interpretable dynamic reference system. There is an urgent need for a set of structural dynamic monitoring data analysis methods that can be used for the scientific judgment of abnormal vibration events.

[0005] Currently, the existing technical solutions related to the above-mentioned structural dynamic monitoring data analysis methods include the following two:

[0006] One existing technical solution is a vibration amplitude monitoring and threshold alarm method based on a long-term structural health monitoring system. This method typically involves deploying long-term monitoring sensors in the building structure to continuously collect acceleration or other monitoring data, acquiring vibration signals in real time and calculating vibration amplitude (such as peak acceleration or root mean square value). When the amplitude exceeds a preset threshold, an alarm is triggered, indicating a possible abnormal vibration event.

[0007] The aforementioned existing technical solution 1 has the following shortcomings:

[0008] First, the vibration amplitude contains limited information related to the abnormal event. This method cannot reveal the time-frequency characteristics, spatial distribution, and potential sources of abnormal signals, making it difficult to support subsequent event tracing and scientific handling.

[0009] Secondly, it lacks long-term accumulation and reference of the laws governing the normal dynamic response of buildings. This method cannot transform long-term dynamic monitoring data into effective structural dynamic information, and cannot provide the normal frequency characteristics of buildings as a background reference for analysis when analyzing abnormal vibrations.

[0010] Third, amplitude threshold alone cannot effectively distinguish between abnormal vibration and amplification under normal operating conditions. For example, increased personnel activity during working hours, starting of air conditioning equipment, high-frequency operation of elevators, or short-term environmental disturbances may all lead to an increase in vibration amplitude, but they do not necessarily constitute an abnormal event. Relying solely on vibration amplitude for anomaly identification can easily lead to false alarms.

[0011] The second existing technical solution is a traditional method for dynamic testing and identification of building structures' dynamic parameters. This method typically involves deploying accelerometers, velocity sensors, or displacement sensors on several floors or at key locations within a building to collect vibration response data of the structure under environmental or artificial excitation over a certain period of time. Then, methods such as Fourier spectrum analysis, power spectral density analysis, operational modal analysis, or random subspace identification are used to identify the structure's natural frequencies, damping ratios, and mode shapes, and the structural state is assessed by tracking the changes of these parameters over time.

[0012] The aforementioned existing technical solution 1 has the following disadvantages:

[0013] First, this method focuses on the structure itself and struggles to identify the normal operating frequencies caused by large equipment. It primarily focuses on parameters such as the structure's natural frequencies, damping ratio, and mode shapes, lacking a systematic analysis of the overall composition, patterns, and sources of operating frequencies from equipment such as air conditioning systems, elevators, rooftop fans, window cleaning equipment, and rail transit systems.

[0014] Secondly, limited-time dynamic testing and parameter analysis cannot reflect the differences in vibration background of a building throughout the day, week, or long-term operation. The vibration response of high-rise buildings varies with office hours, nighttime, equipment start-up and shutdown, traffic operation, and environmental changes, making it difficult to establish the building's regular dominant frequency composition and occurrence patterns through short-term testing. Summary of the Invention

[0015] The technical problem to be solved by this invention is to provide a method for rapid assessment of abnormal vibrations in high-rise buildings based on time-frequency analysis, so as to solve the problem of how to quickly assess abnormal vibration events in emergency scenarios where existing routine monitoring data is lacking.

[0016] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is as follows:

[0017] A rapid assessment method for abnormal vibrations in high-rise buildings based on time-frequency analysis is proposed. The applicable high-rise buildings include high-rise buildings, super high-rise buildings, and other tall building structures that are highly sensitive to environmental excitations, equipment excitations, and changes in operating conditions. This includes:

[0018] Step 1: Obtaining Vibration Acceleration Data

[0019] Continuously acquire vibration acceleration data of high-rise buildings at multiple floors and multiple key structural locations within each floor to obtain the differences in vibration response of high-rise buildings at different heights and at different key structural locations; for example, when a high-rise building is a steel-concrete composite frame-core tube system, the key structural locations in the floors are generally near the core tube, near the outer frame, and at the mid-span of the floor slab.

[0020] The vibration acceleration data can come from either the existing long-term structural health monitoring system of the high-rise building or from the emergency monitoring system temporarily deployed after an abnormal event occurs.

[0021] Step 2: Construction of multi-scale power spectral density time-frequency plot:

[0022] The vibration acceleration data obtained in step one were processed using the sliding window short-time spectrum analysis method, and power spectral density time-frequency diagrams (i.e., PSD time-frequency diagrams) were constructed at various time scales.

[0023] In addition to constructing the power spectral density time-frequency plot, step two can also be replaced by constructing time-frequency plots of other spectral representations, as long as they can reflect the energy distribution of different frequency components, such as amplitude spectrum, energy spectrum or other frequency domain representations.

[0024] Step 3: Identification of representative frequency bands, their frequency components, and their timing patterns:

[0025] From each power spectral density time-frequency plot constructed in step two, identify representative frequency bands and extract the frequency components of the representative frequency bands and their occurrence time patterns throughout the day;

[0026] Step 4: Constructing the Power Ledger:

[0027] If the vibration acceleration data includes routine monitoring data of the high-rise building during normal operation, the data obtained from the routine monitoring data through steps two and three is constructed into a power ledger according to a preset data structuring method. In the routine monitoring scenario, the routine monitoring data preferably uses data covering a complete daily cycle, such as data from a continuous week, to reflect the differences between weekdays and non-weekdays, and between working and non-working hours. The data structuring method of the power ledger can adopt a structure of "frequency characteristics—temporal patterns—energy levels—spatial manifestations—source attribution," and can be organized according to vibration source type or frequency band range.

[0028] If the vibration acceleration data does not include the normal monitoring data, the monitoring data of the first continuous monitoring period is processed by steps two and three to construct a power ledger according to a preset data structuring method. In emergency scenarios, the time period of the first continuous monitoring period can be adjusted as needed, preferably 12 or 24 hours covering both day and night. For example, in emergency scenarios lacking long-term monitoring foundation, the first 24 hours of continuous monitoring data after the sensor system used to collect the vibration acceleration data enters the site is used as the initial power reference.

[0029] Step 5: Identification of Abnormal Candidate Windows

[0030] The vibration acceleration data obtained in step one is recorded as the vibration acceleration data to be analyzed. Based on the frequency components and occurrence time patterns of the power spectral density time-frequency diagram to be analyzed and its representative frequency bands obtained through steps two and three, and the difference between the power spectral density time-frequency diagram to be analyzed and the data recorded in the power ledger as a normal background reference during the same period, abnormal candidate windows that may contain abnormal vibration events of the high-rise building in the power spectral density time-frequency diagram to be analyzed are identified.

[0031] Therefore, this invention constructs a power spectral density time-frequency diagram of vibration acceleration data of high-rise buildings through steps one to three, and identifies representative frequency bands, their frequency components, and occurrence time patterns. Then, through step four, a power ledger is constructed as a reference for conventional power to clarify the conventional frequency components and occurrence patterns of high-rise buildings. Thus, when abnormal vibration events occur in high-rise buildings, step five compares the vibration acceleration data to be analyzed with the power ledger to quickly distinguish abnormal components from the conventional background. It can accurately capture abnormal candidate windows containing abnormal signals, quickly pinpoint the time of abnormal occurrence, and effectively avoid false alarms and missed alarms caused by relying solely on amplitude alarms, thereby improving the speed and accuracy of abnormal vibration identification in high-rise buildings.

[0032] Furthermore, step four of this invention can not only construct a dynamic ledger when the vibration acceleration data includes normal monitoring data, but also construct a dynamic ledger using the monitoring data of the first continuous monitoring period in emergency scenarios after an abnormal vibration event that lacks existing monitoring data, so as to quickly establish an initial dynamic reference. Therefore, this invention is applicable to scenarios with or without a long-term monitoring system.

[0033] The construction method of the power spectral density time-frequency map in step two is as follows: the vibration acceleration data is segmented according to the window length and step size set by the various time scales, and the spectrum estimation is performed on each segment to calculate the power spectral density value (PSD) of each frequency point of the segment, forming a power spectral density time-frequency map with time as the horizontal axis, frequency as the vertical axis, and power spectral density value as the color representation.

[0034] The window length refers to the signal duration covered by the analysis window corresponding to a single power spectral density time-frequency plot, and the step size refers to the time interval between the starting points of the analysis windows corresponding to two adjacent power spectral density time-frequency plots. It directly controls the overlap rate of the analysis windows and determines the time sampling density of the power spectral density time-frequency plot.

[0035] In such power spectral density time-frequency graphs, horizontal bright bars indicate that a certain frequency component persists or recurs over a long period of time, while vertical bright bars indicate that the frequency energy suddenly increases at a certain moment or within a short period of time.

[0036] Among them, the representative frequency bands mentioned in step three are divided into three categories according to their occurrence time patterns:

[0037] Type I frequency stripes: These exist continuously or recurringly throughout the day, appearing as horizontal, bright stripes in the power spectral density time-frequency plot that are continuously visible or recurringly stable throughout the day. For example... Figure 2a The 2.4Hz stripe; among them, the first type of frequency stripe can be stably identified in the power spectral density time-frequency diagrams at three time scales, which may correspond to the main mode of the structure, the mode of local components, or the frequency of long-term operating mechanical equipment.

[0038] The second type of frequency stripe: It appears clearly only during a specific time period of the day, and weakens or disappears during the rest of the time. In the power spectral density time-frequency plot, it appears as a horizontal bright stripe that is only located during the specific time period, for example... Figure 2b The 16.5Hz band is included; among them, the second type of frequency band may correspond to building equipment operating during fixed time periods; for example, a frequency that mainly appears during daytime working hours and disappears during non-working hours at night may be related to air conditioning systems, office equipment, human activities, or the operation of property equipment; a frequency that mainly appears at night may be related to nighttime cleaning equipment, nighttime fan operation, or other special operating conditions. For the second type of frequency band, the source should be attributed by combining equipment operation records, property work orders, human activity patterns, and on-site visits.

[0039] The third type of frequency stripe: These appear intermittently at certain time intervals, and in the power spectral density time-frequency graph, they appear as horizontal bright stripes exhibiting periodic changes in brightness. For example... Figure 2c The 38Hz band is included. The third type of frequency band may correspond to equipment and events that operate intermittently or periodically, such as elevators, crowds, and rail transit. For the third type of frequency band, its repetition period, duration, frequency of occurrence, and typical occurrence times should be carefully recorded.

[0040] The abnormal candidate windows identified in step five differ from the power ledger used as a normal background reference from any one or more of the following:

[0041] Type 1: In the power spectral density time-frequency graph to be analyzed, there are short-term vertical brightening bands with a frequency coverage exceeding a preset minimum frequency range threshold. This means that multiple frequency components simultaneously increase in intensity within a short period, which manifests as vertical bright bands in the power spectral density time-frequency graph. For example... Figure 3 Four distinctly abnormal vertical bright bands were observed in the middle.

[0042] Category 2: The power spectral density time-frequency plot to be analyzed contains new representative frequency bands or brightened areas that are not recorded in the power ledger.

[0043] The third type of difference is that the representative frequency bands of the power spectral density time-frequency diagram to be analyzed have different occurrence time patterns compared with the data recorded in the power ledger. For example, the representative frequency bands of frequency components in a specific time period suddenly increase outside the regular occurrence time period recorded in the power ledger.

[0044] Preferably, step two constructs power spectral density time-frequency maps at three time scales. These three time scales include a long window scale, a medium window scale, and a short window scale, with the window length and step size decreasing sequentially. The power spectral density time-frequency maps obtained from these three time scales are denoted as the long window power spectral density time-frequency map, the medium window power spectral density time-frequency map, and the short window power spectral density time-frequency map, respectively. The window length, step size, and analysis frequency range of the three time scales can be adjusted according to the dominant frequency range of high-rise buildings, the type of conventional vibration sources, the duration of abnormal events, and the monitoring sampling frequency. They can also be flexibly adjusted according to the height of high-rise buildings, structural systems, expected anomaly types, and sampling frequencies.

[0045] Step five employs a coarse-to-fine identification strategy, including: First, comparing the long-window power spectral density time-frequency map of the vibration acceleration data to be analyzed with the data recorded in the power ledger during the same period to identify long-window power spectral density time-frequency maps and their corresponding long-window time intervals that may contain abnormal vibration events of the high-rise building; then, comparing the medium-window power spectral density time-frequency map of the vibration acceleration data to be analyzed belonging to the long-window time interval with the data recorded in the power ledger during the same period to identify medium-window power spectral density time-frequency maps and their corresponding medium-window time intervals that may contain abnormal vibration events of the high-rise building; finally, comparing the short-window power spectral density time-frequency map of the vibration acceleration data to be analyzed belonging to the medium-window time interval with the data recorded in the power ledger during the same period to identify short-window power spectral density time-frequency maps and their corresponding short-window time intervals that may contain abnormal vibration events of the high-rise building, which are the abnormal candidate windows. This enables accurate and rapid locking of abnormal candidate windows, pinpointing the precise time, duration, and number of repetitions of abnormal vibration events in high-rise buildings.

[0046] Therefore, this invention constructs power spectral density time-frequency maps at three time scales—long window, medium window, and short window—through step two, and adopts a coarse-to-fine identification strategy in step five to take into account the identification of long-term stability, time-period regularity, and short-term abrupt change characteristics. This allows for the rapid and accurate identification of abnormal candidate windows containing abnormal signal segments from massive vibration acceleration data, thereby improving the efficiency of abnormal vibration identification.

[0047] Preferably, the three time scales are respectively set as follows:

[0048] Long window scale: The window length ranges from 1 to 2 hours, the step size from 10 to 20 minutes, and the analysis frequency range is 0 to 10 Hz. Therefore, the power spectral density time-frequency plot under the long window scale can be used to observe low-frequency stable bands and their long-term variation characteristics, for example... Figure 2a ;

[0049] Medium window scale: The window length is between 10 and 20 minutes, the step size is between 1 and 5 minutes, and the analysis frequency range is 0 to 20 Hz. Therefore, the power spectral density time-frequency plot at the medium window scale can balance overall viewing with time-frequency resolution, making it a primary scale for identifying regular frequency bands and establishing a power ledger. Figure 2b ;

[0050] Short window scale: The window length ranges from 1 to 5 minutes, the step size from 10 to 60 seconds, and the analysis frequency range is 0 to 50 Hz. Therefore, the power spectral density time-frequency plot under the short window scale can be used to capture minute-level short-term energy abrupt changes and locate anomalous vertical bright bands, for example... Figure 2c .

[0051] Preferably, in step one, the acquired vibration acceleration data covers at least the low, middle, and high zones of the high-rise building; and the vibration acceleration data is obtained by monitoring with a triaxial accelerometer to obtain the vibration response in the X, Y, and Z directions; the sampling frequency of the triaxial accelerometer is between 50Hz and 200Hz to take into account the analysis of conventional equipment frequencies, local component frequencies, and abnormal vibration frequencies in the range of 0Hz to 50Hz, thus meeting the requirements for general abnormal vibration events in high-rise buildings.

[0052] As a preferred embodiment of the present invention: the method for rapid assessment of abnormal vibrations in high-rise buildings further includes:

[0053] Step Six: Frequency and Spatial Response Analysis:

[0054] For the abnormal candidate windows identified in step five, frequency feature analysis and multi-point spatial response analysis are performed:

[0055] The frequency feature analysis is as follows: extract the power spectral density curve (i.e., PSD curve) of the abnormal signal within the abnormal candidate window and the power spectral density curve of the normal signal within a preset time period before and after the abnormal candidate window (i.e., the normal PSD curve of several time periods before and after the abnormal event), and compare them to determine the frequency range and frequency value of the power spectral density enhancement of the abnormal signal relative to the normal signal, so as to determine whether the frequency band selectivity of the abnormal signal is wideband indiscriminate amplification or specific frequency band selective enhancement.

[0056] The multi-point spatial response analysis involves comparing the vibration responses between different floors and between different key structural locations on the same floor within the same abnormal time period, for the same abnormal candidate window, to determine the spatial distribution pattern and propagation path characteristics of the abnormal vibration response during the abnormal time period. Specifically, this includes:

[0057] The power spectral density energy distribution of abnormal signals within the frequency band of the abnormal candidate window is compared on different floors to analyze the changing trend of abnormal vibration response with the floor and determine whether the abnormal vibration response shows the characteristic of increasing or decreasing with the height of the high-rise building.

[0058] Compare the power spectral density curves and time history response differences of the two key structural locations, the core tube area and the outer frame structure area, on the same floor to determine whether the abnormal vibration response first manifests in the core tube area or the outer frame structure area.

[0059] By comparing the order in which vibration acceleration data from different floors reached their peak values ​​during abnormal periods, the direction of propagation of abnormal signals can be determined.

[0060] Therefore, by performing frequency characteristic analysis and multi-point spatial response analysis in step six, this invention clarifies the frequency band selectivity, spatial distribution pattern, and propagation path characteristics of abnormal signals, providing a clear basis for identifying the abnormal vibration source region and equipment type. Thus, this invention can support rapid source tracing of abnormal vibration events and significantly shorten the judgment cycle from abnormal vibration event to abnormal source confirmation.

[0061] Step six, based on frequency characteristic analysis and multi-point spatial response analysis, assesses the abnormal vibration events occurring at the candidate windows of the high-rise building, including:

[0062] If the frequency selectivity of the abnormal signal is a wideband indiscriminate amplification with an overall increase across the entire frequency band, it indicates that the abnormal vibration event belongs to one or more of wideband disturbance, impact, and overall noise enhancement.

[0063] If the frequency selectivity of the abnormal signal is manifested as a specific frequency selective enhancement in which the power spectral density is significantly enhanced only in a specific frequency band, it indicates that the abnormal vibration event has frequency selectivity, which is usually more conducive to tracing the source of vibration, because different devices or components often have relatively stable operating frequencies or excitation frequencies.

[0064] Based on the spatial distribution pattern and propagation path characteristics, the possible types of abnormal vibration sources are gradually narrowed down, and the location of suspected vibration source equipment or suspected vibration source components is initially identified.

[0065] Preferably, the power log in step four also records the energy level of each representative frequency band, its spatial distribution characteristics on each floor, and the possible vibration sources corresponding to the representative frequency bands. This is obtained through on-site visits, comparison of equipment logs and operation records. Among them, for frequency components that can be attributed through on-site investigation, their corresponding vibration sources are clearly marked. For frequency components that cannot be accurately attributed temporarily, they are also directly included in the power log as a routine reference.

[0066] Step six also includes:

[0067] The abnormal enhancement frequency is compared with the data recorded in the power ledger, wherein the abnormal enhancement frequency is the frequency band range and frequency value of the power spectral density enhancement of the abnormal signal relative to the normal signal.

[0068] If the abnormally enhanced frequency is close to the frequency components of the representative frequency band recorded in the power log, it indicates that the abnormal excitation may have excited the existing structural frequency or local component frequency of the high-rise building.

[0069] If the abnormally enhanced frequency is close to the normal operating frequency of a certain device, but the occurrence time, energy level or duration of the abnormally enhanced frequency is significantly different from the normal characteristics of the device, it may indicate that the device is operating under abnormal conditions.

[0070] If the abnormally enhanced frequency is not recorded in the power log, it indicates that there may be a new vibration source, temporary working conditions, or non-routine excitation.

[0071] Preferably, the method for rapid assessment of abnormal vibrations in high-rise buildings further includes:

[0072] Step 7: Quick assessment of the event:

[0073] By combining the results of the anomaly candidate window identification in step five and the results of the frequency and spatial response analysis in step six, a rapid judgment conclusion on the abnormal vibration event is formed.

[0074] The analysis includes: the precise time window of the abnormal vibration, the characteristics and frequency values ​​of the abnormal enhanced frequency band, the comparative analysis of the abnormal frequency and the dynamic ledger, the spatial distribution pattern and propagation direction of the abnormal signal, and the preliminary identification of candidate vibration source areas or equipment types.

[0075] Based on the above analysis, we can further guide the on-site implementation of targeted verification work such as equipment reproduction testing and operational condition checks, ultimately confirming the abnormal vibration source and handling the incident.

[0076] Compared with the prior art, the present invention has the following beneficial effects:

[0077] First, this invention constructs a power spectral density time-frequency diagram of vibration acceleration data of high-rise buildings through steps one to three, and identifies representative frequency bands, their frequency components, and occurrence time patterns. Then, through step four, a power ledger is constructed as a reference for conventional power to clarify the conventional frequency components and occurrence patterns of high-rise buildings. Thus, when an abnormal vibration event occurs in a high-rise building, step five compares the vibration acceleration data to be analyzed with the power ledger to quickly distinguish abnormal components from the conventional background. This can accurately capture abnormal candidate windows containing abnormal signals, quickly pinpoint the time of abnormal occurrence, and effectively avoid false alarms and missed alarms caused by relying solely on amplitude alarms, thereby improving the speed and accuracy of abnormal vibration identification in high-rise buildings.

[0078] Furthermore, step four of this invention can not only construct a dynamic ledger when the vibration acceleration data includes normal monitoring data, but also construct a dynamic ledger using the monitoring data of the first continuous monitoring period in emergency scenarios after an abnormal vibration event that lacks existing monitoring data, so as to quickly establish an initial dynamic reference. Therefore, this invention is applicable to scenarios with or without a long-term monitoring system.

[0079] Second, the present invention constructs power spectral density time-frequency maps at three time scales—long window scale, medium window scale, and short window scale—through step two, and adopts a coarse-to-fine identification strategy in step five to take into account the identification of long-term stability, time-period regularity, and short-term abrupt change characteristics. This can quickly and accurately lock the abnormal candidate window containing abnormal signal segments from massive vibration acceleration data, thereby improving the efficiency of abnormal vibration identification.

[0080] Third, through step six, the present invention performs frequency characteristic analysis and multi-point spatial response analysis to clarify the frequency band selectivity, spatial distribution pattern and propagation path characteristics of abnormal signals, providing a clear basis for locking the abnormal vibration source area and equipment type. Therefore, the present invention can support the rapid tracing of abnormal vibration events and significantly shorten the judgment cycle from abnormal vibration event to abnormal vibration source confirmation.

[0081] Fourth, through steps one to seven, this invention forms a systematic and operable method for rapid assessment of abnormal vibration events in high-rise buildings, filling the gap in the existing technology in this complete methodological framework. It is particularly suitable for high-rise buildings and large public buildings that are already in operation, as well as building safety management scenarios that require long-term dynamic performance tracking, abnormal vibration identification, vibration source tracing, and operation and maintenance risk assessment. Attached Figure Description

[0082] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments:

[0083] Figure 1 This is a flowchart of the method for rapid assessment of abnormal vibrations in high-rise buildings according to the present invention;

[0084] Figure 2a This is an example of the power spectral density time-frequency plot constructed in step two of this invention at a long window scale;

[0085] Figure 2b This is an example of the power spectral density time-frequency plot constructed at the mid-window scale in step two of this invention;

[0086] Figure 2c This is an example of the power spectral density time-frequency plot constructed in step two of this invention at a short window scale;

[0087] Figure 3 This is a schematic diagram comparing the power spectral density time-frequency graph of the power spectrum density to be analyzed with the power spectral density time-frequency graph recorded in the power ledger as a normal background in step five of this invention.

[0088] Figure 4a This is an example of a power spectral density curve comparison diagram for the frequency characteristic analysis of the core tube in step six of the present invention;

[0089] Figure 4b This is an example of a power spectral density curve comparison diagram for frequency characteristic analysis of the external frame in step six of the present invention.

[0090] Figure 5a This is an example of the Z-axis acceleration time history curve for multi-point spatial response analysis of the outer frame on different floors in step six of the present invention;

[0091] Figure 5b This is an example of the Z-axis acceleration time history curve for multi-point spatial response analysis of the outer frame and core tube on the same floor in step six of this invention. Detailed Implementation

[0092] The present invention will now be described in detail with reference to the embodiments and accompanying drawings to help those skilled in the art better understand the inventive concept of the present invention. However, the scope of protection of the claims of the present invention is not limited to the following embodiments. For those skilled in the art, all other embodiments obtained without creative effort without departing from the inventive concept of the present invention are within the scope of protection of the present invention.

[0093] Example 1

[0094] like Figure 1 As shown, this invention discloses a rapid assessment method for abnormal vibrations in high-rise buildings based on time-frequency analysis. The applicable high-rise buildings include high-rise buildings, super high-rise buildings, and other tall building structures that are highly sensitive to environmental excitations, equipment excitations, and changes in operating conditions. This includes:

[0095] Step 1: Obtaining Vibration Acceleration Data

[0096] Continuously acquire vibration acceleration data of high-rise buildings at multiple floors and multiple key structural locations within each floor to obtain the differences in vibration response of high-rise buildings at different heights and at different key structural locations; for example, when a high-rise building is a steel-concrete composite frame-core tube system, the key structural locations in the floors are generally near the core tube, near the outer frame, and at the mid-span of the floor slab.

[0097] The vibration acceleration data can come from either the existing long-term structural health monitoring system of the high-rise building or from the emergency monitoring system temporarily deployed after an abnormal event occurs.

[0098] Step 2: Construction of multi-scale power spectral density time-frequency plot:

[0099] The vibration acceleration data obtained in step one were processed using the sliding window short-time spectrum analysis method, and power spectral density time-frequency diagrams (i.e., PSD time-frequency diagrams) were constructed at various time scales.

[0100] The construction method of the power spectral density time-frequency map in step two is as follows: the vibration acceleration data is segmented according to the window length and step size set by the various time scales, and the spectrum estimation is performed on each segment to calculate the power spectral density value (PSD) of each frequency point of the segment, forming a power spectral density time-frequency map with time as the horizontal axis, frequency as the vertical axis, and power spectral density value as the color representation.

[0101] The window length refers to the signal duration covered by the analysis window corresponding to a single power spectral density time-frequency plot, and the step size refers to the time interval between the starting points of the analysis windows corresponding to two adjacent power spectral density time-frequency plots. It directly controls the overlap rate of the analysis windows and determines the time sampling density of the power spectral density time-frequency plot.

[0102] In such power spectral density time-frequency graphs, horizontal bright bars indicate that a certain frequency component persists or recurs over a long period of time, while vertical bright bars indicate that the frequency energy suddenly increases at a certain moment or within a short period of time.

[0103] In addition to constructing the power spectral density time-frequency plot, step two can also be replaced by constructing time-frequency plots of other spectral representations, as long as they can reflect the energy distribution of different frequency components, such as amplitude spectrum, energy spectrum or other frequency domain representations.

[0104] Step 3: Identification of representative frequency bands, their frequency components, and their timing patterns:

[0105] From each power spectral density time-frequency plot constructed in step two, identify representative frequency bands and extract the frequency components of the representative frequency bands and their occurrence time patterns throughout the day;

[0106] Among them, the representative frequency bands mentioned in step three are divided into three categories according to their occurrence time patterns:

[0107] Type I frequency stripes: These exist continuously or recurringly throughout the day, appearing as horizontal, bright stripes in the power spectral density time-frequency plot that are continuously visible or recurringly stable throughout the day. For example... Figure 2a The 2.4Hz stripe; among them, the first type of frequency stripe can be stably identified in the power spectral density time-frequency diagrams at three time scales, which may correspond to the main mode of the structure, the mode of local components, or the frequency of long-term operating mechanical equipment.

[0108] The second type of frequency stripe: It appears clearly only during a specific time period of the day, and weakens or disappears during the rest of the time. In the power spectral density time-frequency plot, it appears as a horizontal bright stripe that is only located during the specific time period, for example... Figure 2b The 16.5Hz band is included; among them, the second type of frequency band may correspond to building equipment operating during fixed time periods; for example, a frequency that mainly appears during daytime working hours and disappears during non-working hours at night may be related to air conditioning systems, office equipment, human activities, or the operation of property equipment; a frequency that mainly appears at night may be related to nighttime cleaning equipment, nighttime fan operation, or other special operating conditions. For the second type of frequency band, the source should be attributed by combining equipment operation records, property work orders, human activity patterns, and on-site visits.

[0109] The third type of frequency stripe: These appear intermittently at certain time intervals, and in the power spectral density time-frequency graph, they appear as horizontal bright stripes exhibiting periodic changes in brightness. For example... Figure 2cThe 38Hz band is included. The third type of frequency band may correspond to equipment and events that operate intermittently or periodically, such as elevators, crowds, and rail transit. For the third type of frequency band, its repetition period, duration, frequency of occurrence, and typical occurrence times should be carefully recorded.

[0110] Step 4: Constructing the Power Ledger:

[0111] If the vibration acceleration data includes routine monitoring data of the high-rise building during normal operation, the data obtained from the routine monitoring data through steps two and three is constructed into a power ledger according to a preset data structuring method. In the routine monitoring scenario, the routine monitoring data preferably uses data covering a complete daily cycle, such as data from a continuous week, to reflect the differences between weekdays and non-weekdays, and between working and non-working hours. The data structuring method of the power ledger can adopt a structure of "frequency characteristics—temporal patterns—energy levels—spatial manifestations—source attribution," and can be organized according to vibration source type or frequency band range.

[0112] If the vibration acceleration data does not include the normal monitoring data, the monitoring data of the first continuous monitoring period is processed by steps two and three to construct a power ledger according to a preset data structuring method. In emergency scenarios, the time period of the first continuous monitoring period can be adjusted as needed, preferably 12 or 24 hours covering both day and night. For example, in emergency scenarios lacking long-term monitoring foundation, the first 24 hours of continuous monitoring data after the sensor system used to collect the vibration acceleration data enters the site is used as the initial power reference.

[0113] Step 5: Identification of Abnormal Candidate Windows

[0114] The vibration acceleration data obtained in step one is recorded as the vibration acceleration data to be analyzed. Based on the frequency components and occurrence time patterns of the power spectral density time-frequency diagram to be analyzed and its representative frequency bands obtained through steps two and three, and the difference between the power spectral density time-frequency diagram to be analyzed and the data recorded in the power ledger as a normal background reference during the same period, abnormal candidate windows that may contain abnormal vibration events of the high-rise building in the power spectral density time-frequency diagram to be analyzed are identified.

[0115] The abnormal candidate windows identified in step five differ from the power ledger used as a normal background reference from any one or more of the following:

[0116] Type 1: In the power spectral density time-frequency graph to be analyzed, there are short-term vertical brightening bands with a frequency coverage exceeding a preset minimum frequency range threshold. This means that multiple frequency components simultaneously increase in intensity within a short period, which manifests as vertical bright bands in the power spectral density time-frequency graph. For example... Figure 3Four distinctly abnormal vertical bright bands were observed in the middle.

[0117] Category 2: The power spectral density time-frequency plot to be analyzed contains new representative frequency bands or brightened areas that are not recorded in the power ledger.

[0118] The third type of difference is that the representative frequency bands of the power spectral density time-frequency diagram to be analyzed have different occurrence time patterns compared with the data recorded in the power ledger. For example, the representative frequency bands of frequency components in a specific time period suddenly increase outside the regular occurrence time period recorded in the power ledger.

[0119] Therefore, this invention constructs a power spectral density time-frequency diagram of vibration acceleration data of high-rise buildings through steps one to three, and identifies representative frequency bands, their frequency components, and occurrence time patterns. Then, through step four, a power ledger is constructed as a reference for conventional power to clarify the conventional frequency components and occurrence patterns of high-rise buildings. Thus, when abnormal vibration events occur in high-rise buildings, step five compares the vibration acceleration data to be analyzed with the power ledger to quickly distinguish abnormal components from the conventional background. It can accurately capture abnormal candidate windows containing abnormal signals, quickly pinpoint the time of abnormal occurrence, and effectively avoid false alarms and missed alarms caused by relying solely on amplitude alarms, thereby improving the speed and accuracy of abnormal vibration identification in high-rise buildings.

[0120] Furthermore, step four of this invention can not only construct a dynamic ledger when the vibration acceleration data includes normal monitoring data, but also construct a dynamic ledger using the monitoring data of the first continuous monitoring period in emergency scenarios after an abnormal vibration event that lacks existing monitoring data, so as to quickly establish an initial dynamic reference. Therefore, this invention is applicable to scenarios with or without a long-term monitoring system.

[0121] The above is the basic implementation method of this embodiment one, and further optimizations, improvements and limitations can be made based on this basic implementation method:

[0122] Preferably, step two constructs power spectral density time-frequency maps at three time scales. These three time scales include a long window scale, a medium window scale, and a short window scale, with the window length and step size decreasing sequentially. The power spectral density time-frequency maps obtained from these three time scales are denoted as the long window power spectral density time-frequency map, the medium window power spectral density time-frequency map, and the short window power spectral density time-frequency map, respectively. The window length, step size, and analysis frequency range of the three time scales can be adjusted according to the dominant frequency range of high-rise buildings, the type of conventional vibration sources, the duration of abnormal events, and the monitoring sampling frequency. They can also be flexibly adjusted according to the height of high-rise buildings, structural systems, expected anomaly types, and sampling frequencies.

[0123] Step five employs a coarse-to-fine identification strategy, including: First, comparing the long-window power spectral density time-frequency map of the vibration acceleration data to be analyzed with the data recorded in the power ledger during the same period to identify long-window power spectral density time-frequency maps and their corresponding long-window time intervals that may contain abnormal vibration events of the high-rise building; then, comparing the medium-window power spectral density time-frequency map of the vibration acceleration data to be analyzed belonging to the long-window time interval with the data recorded in the power ledger during the same period to identify medium-window power spectral density time-frequency maps and their corresponding medium-window time intervals that may contain abnormal vibration events of the high-rise building; finally, comparing the short-window power spectral density time-frequency map of the vibration acceleration data to be analyzed belonging to the medium-window time interval with the data recorded in the power ledger during the same period to identify short-window power spectral density time-frequency maps and their corresponding short-window time intervals that may contain abnormal vibration events of the high-rise building, which are the abnormal candidate windows. This enables accurate and rapid locking of abnormal candidate windows, pinpointing the precise time, duration, and number of repetitions of abnormal vibration events in high-rise buildings.

[0124] Therefore, this invention constructs power spectral density time-frequency maps at three time scales—long window, medium window, and short window—through step two, and adopts a coarse-to-fine identification strategy in step five to take into account the identification of long-term stability, time-period regularity, and short-term abrupt change characteristics. This allows for the rapid and accurate identification of abnormal candidate windows containing abnormal signal segments from massive vibration acceleration data, thereby improving the efficiency of abnormal vibration identification.

[0125] Preferably, the three time scales are respectively set as follows:

[0126] Long window scale: The window length ranges from 1 to 2 hours, the step size from 10 to 20 minutes, and the analysis frequency range is 0 to 10 Hz. Therefore, the power spectral density time-frequency plot under the long window scale can be used to observe low-frequency stable bands and their long-term variation characteristics, for example... Figure 2a ;

[0127] Medium window scale: The window length is between 10 and 20 minutes, the step size is between 1 and 5 minutes, and the analysis frequency range is 0 to 20 Hz. Therefore, the power spectral density time-frequency plot at the medium window scale can balance overall viewing with time-frequency resolution, making it a primary scale for identifying regular frequency bands and establishing a power ledger. Figure 2b ;

[0128] Short window scale: The window length ranges from 1 to 5 minutes, the step size from 10 to 60 seconds, and the analysis frequency range is 0 to 50 Hz. Therefore, the power spectral density time-frequency plot under the short window scale can be used to capture minute-level short-term energy abrupt changes and locate anomalous vertical bright bands, for example... Figure 2c .

[0129] Preferably, in step one, the acquired vibration acceleration data covers at least the low, middle, and high zones of the high-rise building; and the vibration acceleration data is obtained by monitoring with a triaxial accelerometer to obtain the vibration response in the X, Y, and Z directions; the sampling frequency of the triaxial accelerometer is between 50Hz and 200Hz to take into account the analysis of conventional equipment frequencies, local component frequencies, and abnormal vibration frequencies in the range of 0Hz to 50Hz, thus meeting the requirements for general abnormal vibration events in high-rise buildings.

[0130] Example 2

[0131] Based on the above embodiment one, this embodiment two also adopts the following preferred implementation method:

[0132] The method for rapid assessment of abnormal vibrations in high-rise buildings also includes:

[0133] Step Six: Frequency and Spatial Response Analysis:

[0134] For the abnormal candidate windows identified in step five, frequency feature analysis and multi-point spatial response analysis are performed:

[0135] The frequency feature analysis is as follows: extract the power spectral density curve (i.e., PSD curve) of the abnormal signal within the abnormal candidate window and the power spectral density curve of the normal signal within a preset time period before and after the abnormal candidate window (i.e., the normal PSD curve of several time periods before and after the abnormal event), and compare them to determine the frequency range and frequency value of the power spectral density enhancement of the abnormal signal relative to the normal signal, so as to determine whether the frequency band selectivity of the abnormal signal is wideband indiscriminate amplification or specific frequency band selective enhancement.

[0136] The multi-point spatial response analysis involves comparing the vibration responses between different floors and between different key structural locations on the same floor within the same abnormal time period, for the same abnormal candidate window, to determine the spatial distribution pattern and propagation path characteristics of the abnormal vibration response during the abnormal time period. Specifically, this includes:

[0137] The power spectral density energy distribution of abnormal signals within the frequency band of the abnormal candidate window is compared on different floors to analyze the changing trend of abnormal vibration response with the floor and determine whether the abnormal vibration response shows the characteristic of increasing or decreasing with the height of the high-rise building.

[0138] Compare the power spectral density curves and time history response differences of the two key structural locations, the core tube area and the outer frame structure area, on the same floor to determine whether the abnormal vibration response first manifests in the core tube area or the outer frame structure area.

[0139] By comparing the order in which vibration acceleration data from different floors reached their peak values ​​during abnormal periods, the direction of propagation of abnormal signals can be determined.

[0140] Step six, based on frequency characteristic analysis and multi-point spatial response analysis, assesses the abnormal vibration events occurring at the candidate windows of the high-rise building, including:

[0141] If the frequency selectivity of the abnormal signal is a wideband indiscriminate amplification with an overall increase across the entire frequency band, it indicates that the abnormal vibration event belongs to one or more of wideband disturbance, impact, and overall noise enhancement.

[0142] If the frequency selectivity of the abnormal signal is manifested as a specific frequency selective enhancement in which the power spectral density is significantly enhanced only in a specific frequency band, it indicates that the abnormal vibration event has frequency selectivity, which is usually more conducive to tracing the source of vibration, because different devices or components often have relatively stable operating frequencies or excitation frequencies.

[0143] Based on the spatial distribution pattern and propagation path characteristics, the possible types of abnormal vibration sources are gradually narrowed down, and the location of suspected vibration source equipment or suspected vibration source components is initially identified.

[0144] Therefore, by performing frequency characteristic analysis and multi-point spatial response analysis in step six, this invention clarifies the frequency band selectivity, spatial distribution pattern, and propagation path characteristics of abnormal signals, providing a clear basis for identifying the abnormal vibration source region and equipment type. Thus, this invention can support rapid source tracing of abnormal vibration events and significantly shorten the judgment cycle from abnormal vibration event to abnormal source confirmation.

[0145] The above is the basic implementation method of this embodiment two, and further optimizations, improvements and limitations can be made based on this basic implementation method:

[0146] Preferably, the power log in step four also records the energy level of each representative frequency band, its spatial distribution characteristics on each floor, and the possible vibration sources corresponding to the representative frequency bands. This is obtained through on-site visits, comparison of equipment logs and operation records. Among them, for frequency components that can be attributed through on-site investigation, their corresponding vibration sources are clearly marked. For frequency components that cannot be accurately attributed temporarily, they are also directly included in the power log as a routine reference.

[0147] Step six also includes:

[0148] The abnormal enhancement frequency is compared with the data recorded in the power ledger, wherein the abnormal enhancement frequency is the frequency band range and frequency value of the power spectral density enhancement of the abnormal signal relative to the normal signal.

[0149] If the abnormally enhanced frequency is close to the frequency components of the representative frequency band recorded in the power log, it indicates that the abnormal excitation may have excited the existing structural frequency or local component frequency of the high-rise building.

[0150] If the abnormally enhanced frequency is close to the normal operating frequency of a certain device, but the occurrence time, energy level or duration of the abnormally enhanced frequency is significantly different from the normal characteristics of the device, it may indicate that the device is operating under abnormal conditions.

[0151] If the abnormally enhanced frequency is not recorded in the power log, it indicates that there may be a new vibration source, temporary working conditions, or non-routine excitation.

[0152] Preferably, the method for rapid assessment of abnormal vibrations in high-rise buildings further includes:

[0153] Step 7: Quick assessment of the event:

[0154] By combining the results of the anomaly candidate window identification in step five and the results of the frequency and spatial response analysis in step six, a rapid judgment conclusion on the abnormal vibration event is formed.

[0155] The analysis includes: the precise time window of the abnormal vibration, the characteristics and frequency values ​​of the abnormal enhanced frequency band, the comparative analysis of the abnormal frequency and the dynamic ledger, the spatial distribution pattern and propagation direction of the abnormal signal, and the preliminary identification of candidate vibration source areas or equipment types.

[0156] Based on the above analysis, we can further guide the on-site implementation of targeted verification work such as equipment reproduction testing and operational condition checks, ultimately confirming the abnormal vibration source and handling the incident.

[0157] The following example, using the rapid assessment of an abnormal vibration event in a super high-rise building in Guangzhou, illustrates the implementation process and technical effects of the rapid assessment method for abnormal vibrations in high-rise buildings of this invention:

[0158] The case study involves a high-rise office building in Guangzhou, with a main structure employing a steel-concrete composite frame-core tube system. It has 5 underground floors and 68 floors above ground, with a total height of approximately 318 meters. During normal use, the building experienced abnormal vibrations, with users reporting vibrations across the middle and upper floors. This example demonstrates how the method described in this invention can be applied to quickly assess abnormal vibration events in emergency scenarios where existing routine monitoring data is lacking.

[0159] Step 1: Obtaining Vibration Acceleration Data

[0160] To obtain vibration data at different floors and structural locations of the building, three-dimensional accelerometers were deployed vertically along the building at floors -5, 1st, 5th, 21st, 37th, 53rd, 65th, and the roof. Measurement points on each floor were strategically located near the core tube, near the outer frame, and at the mid-span of the floor slab. The sampling frequency was set to 100Hz. The first 24 hours of continuous monitoring data after site entry was used as the initial dynamic reference.

[0161] Step 2: Construction of multi-scale PSD time-frequency plots:

[0162] Taking the measuring point at the 65th floor core tube as an example, the vibration data in the first 24 hours were used to construct the PSD time-frequency diagram using three time scales:

[0163] Long window dimensions: window length 1 hour, step size 10 minutes, frequency range 0-10Hz, such as... Figure 2a As shown, the PSD time-frequency plot of the Z-axis vibration acceleration data of the 65th core tube measuring point is displayed on a long window scale. The plot shows a stable frequency band that exists continuously throughout the day, taking the 2.4 Hz band as an example.

[0164] Medium window scale: window length 10 minutes, step size 2 minutes, frequency range 0-20Hz, such as... Figure 2b As shown, the PSD time-frequency plot of the Z-axis vibration acceleration data of the 65-story core tube measuring point is shown at the medium window scale. The plot shows a frequency band that appears only clearly at specific times, taking the 16.5 Hz band as an example. This frequency mainly appears during the daytime working hours and basically disappears at night.

[0165] Short window scale: window length 2 minutes, step size 30 seconds, frequency range 0-50Hz, such as... Figure 2c As shown, this is the PSD time-frequency plot of the Z-axis vibration acceleration data of the 65-story core tube measuring point under the short window scale. The plot shows a periodic and intermittent frequency band, taking the 38 Hz band as an example. This frequency appears more frequently during the daytime working hours, decreases significantly at night, and weakens accordingly during lunch breaks and evening rest periods.

[0166] Step 3: Extraction of the composition and occurrence patterns of the main frequency:

[0167] Extract three representative frequency bands from the time-frequency plot:

[0168] Category 1: Stable frequency bands that persist throughout the day. Represented by the 2.4Hz band, these are continuously visible throughout the day in the long-window time-frequency diagram and can be stably identified at the measurement points in the core tube on both the 53rd and 65th floors.

[0169] The second category consists of frequency bands that appear clearly only during specific time periods. Represented by the 16.5Hz band, these bands primarily appear during daytime working hours and largely disappear at night.

[0170] The third category consists of frequency bands that appear periodically and intermittently. Represented by bands around 38Hz, these bands appear more frequently during the daytime working hours and decrease significantly at night, exhibiting an intermittent pattern related to elevator operation.

[0171] Step 4: Constructing the Power Ledger:

[0172] Referring to Table 1 below, organize the above identification results into a power ledger according to "frequency characteristics - temporal patterns - energy levels - spatial manifestations - source attribution":

[0173] Table 1

[0174]

[0175] Step 5: Identification of Abnormal Candidate Windows

[0176] like Figure 3 As shown, the PSD time-frequency plot of the Z-axis vibration acceleration data from the measuring points of the 65th floor outer frame during the period from 14:00 to 16:00 on a certain day is compared with the conventional time-frequency background during the same period. The analysis scale is a 2-minute window length and a 30-second step. The comparison shows that four obvious abnormal vertical bright bars appeared during this period, and their common feature is a significant enhancement of energy in the 4-8 Hz frequency band within a short period of time. Similar enhancements were observed at multiple measuring points during the same period, and the time corresponding to the vertical bars coincided with the time of subjective vibration feedback from users, and the abnormal candidate window was locked.

[0177] Step Six: Frequency and Spatial Response Analysis:

[0178] Frequency characteristic analysis: Figure 4a and Figure 4b The PSD curves of the fourth abnormal vibration period (15:44:30–15:46:30) are compared with the normal periods before and after it. The results show that the energy increase during the abnormal period was not uniform across the entire frequency band, but rather that the energy in several frequency bands within the 4–8 Hz range was significantly enhanced, with some frequencies approaching the stable dominant frequency in the initial dynamic log. This indicates that the abnormal event has frequency-selective excitation characteristics.

[0179] Spatial response analysis: Based on the comparison of PSD curves of the core tube and outer frame measuring points on different floors during the abnormal period, different floors all showed similar frequency band enhancement characteristics, and the overall response of the outer frame was stronger than that of the core tube. Figure 5a and Figure 5b For time history comparison at different measuring points, the high-region abnormal response reached its peak value before the low-region abnormal response, and the response at the outer frame measuring points on the same floor was earlier and stronger than that at the core tube measuring points. Combining the spatial response and propagation path characteristics, it is indicated that the abnormal excitation source is located in the upper region of the building, and the energy first acts on the outer structure, converging the vibration source range to the area of ​​related equipment outside the roof layer.

[0180] Step 7: Quick assessment of the event:

[0181] Based on the above analysis, a rapid assessment conclusion was reached: the abnormal vibration target frequency band is 4–8 Hz, the abnormal characteristics are selective enhancement of a specific frequency band rather than broadband amplification, the spatial distribution shows preferential response in high-rise areas and the outer frame responding before the core tube, and the propagation direction is from the outer perimeter of the roof to the lower part of the building. The primary suspect was identified as the window cleaning equipment on the outer side of the roof. Subsequent targeted reproduction tests verified this conclusion, proving the effectiveness of the assessment results obtained using this method.

[0182] This invention is not limited to the specific embodiments described above. Based on the above content and in accordance with common technical knowledge and conventional methods in the field, without departing from the basic technical concept of this invention, this invention can also make other equivalent modifications, substitutions or alterations, all of which fall within the protection scope of this invention.

Claims

1. A method for rapid assessment of abnormal vibrations in high-rise buildings based on time-frequency analysis, characterized in that, include: Step 1: Obtaining Vibration Acceleration Data Continuously acquire vibration acceleration data of high-rise buildings on multiple floors and at multiple key structural locations within each floor; Step 2: Construction of multi-scale power spectral density time-frequency plot: The vibration acceleration data obtained in step one were processed using the sliding window short-time spectrum analysis method, and power spectral density time-frequency diagrams were constructed at multiple time scales. Step 3: Identification of representative frequency bands, their frequency components, and their timing patterns: From each power spectral density time-frequency plot constructed in step two, identify representative frequency bands and extract the frequency components of the representative frequency bands and their occurrence time patterns throughout the day; Step 4: Constructing the Power Ledger: If the vibration acceleration data includes the normal monitoring data of the high-rise building during normal operation, the data obtained by processing the normal monitoring data through steps two and three shall be constructed into a power ledger according to a preset data structuring method. If the vibration acceleration data does not include the normal monitoring data, the monitoring data of the first continuous monitoring period is processed by steps two and three to construct a power ledger according to a preset data structuring method. Step 5: Identification of Abnormal Candidate Windows The vibration acceleration data obtained in step one is recorded as the vibration acceleration data to be analyzed. Based on the difference between the power spectral density time-frequency diagram to be analyzed and the frequency components and occurrence time patterns of the representative frequency bands obtained by processing in steps two and three, and the data recorded in the power ledger in the same period, abnormal candidate windows that may contain abnormal vibration events of the high-rise building in the power spectral density time-frequency diagram to be analyzed are identified.

2. The method for rapid assessment of abnormal vibrations in high-rise buildings based on time-frequency analysis according to claim 1, characterized in that: The construction method of the power spectral density time-frequency diagram in step two is as follows: the vibration acceleration data is segmented according to the window length and step size set by the various time scales, and the spectrum estimation is performed on each segment to calculate the power spectral density value of each frequency point of the segment, forming a power spectral density time-frequency diagram with time as the horizontal axis, frequency as the vertical axis, and power spectral density value as the color representation.

3. The method for rapid assessment of abnormal vibrations in high-rise buildings based on time-frequency analysis according to claim 1, characterized in that: The representative frequency bands described in step three are divided into three categories according to their occurrence time patterns: First type of frequency stripe: It exists continuously or appears repeatedly and stably throughout the day; The second type of frequency band: It appears clearly only during a specific time period of the day, and weakens or disappears during the rest of the time outside the specific time period; The third type of frequency stripe: it appears intermittently at certain time intervals.

4. The method for rapid assessment of abnormal vibrations in high-rise buildings based on time-frequency analysis according to claim 1, characterized in that: The abnormal candidate windows identified in step five differ from the power ledger in any one or more of the following ways: Type 1: In the power spectral density time-frequency graph to be analyzed, there are short-term vertical brightening stripes with a frequency coverage range exceeding the preset minimum frequency range threshold; Category 2: The power spectral density time-frequency plot to be analyzed contains new representative frequency bands that are not recorded in the power ledger. The third type of difference is that the representative frequency bands of the power spectral density time-frequency diagram to be analyzed have different occurrence time patterns compared to the data recorded in the power ledger.

5. The method for rapid assessment of abnormal vibrations in high-rise buildings based on time-frequency analysis according to claim 1, characterized in that: Step two constructs power spectral density time-frequency maps at three time scales. The three time scales include a long window scale, a medium window scale, and a short window scale, where the window length and step size decrease sequentially. The power spectral density time-frequency maps obtained are denoted as the long window power spectral density time-frequency map, the medium window power spectral density time-frequency map, and the short window power spectral density time-frequency map, respectively. Step five employs a coarse-to-fine identification strategy, including: First, comparing the long-window power spectral density time-frequency map of the vibration acceleration data to be analyzed with the data recorded in the power ledger during the same period to identify long-window power spectral density time-frequency maps and their corresponding long-window time intervals that may contain abnormal vibration events of the high-rise building; then, comparing the medium-window power spectral density time-frequency map of the vibration acceleration data to be analyzed belonging to the long-window time interval with the data recorded in the power ledger during the same period to identify medium-window power spectral density time-frequency maps and their corresponding medium-window time intervals that may contain abnormal vibration events of the high-rise building; finally, comparing the short-window power spectral density time-frequency map of the vibration acceleration data to be analyzed belonging to the medium-window time interval with the data recorded in the power ledger during the same period to identify short-window power spectral density time-frequency maps and their corresponding short-window time intervals that may contain abnormal vibration events of the high-rise building, which are the abnormal candidate windows.

6. The method for rapid assessment of abnormal vibrations in high-rise buildings based on time-frequency analysis according to claim 5, characterized in that: The three time scales are respectively set as follows: Long window scale: The window length is between 1 and 2 hours, the step size is between 10 and 20 minutes, and the analysis frequency range is 0 to 10 Hz; Medium window scale: The window length is between 10 and 20 minutes, the step size is between 1 and 5 minutes, and the analysis frequency range is 0 to 20 Hz; Short window scale: The window length is between 1 and 5 minutes, the step size is between 10 and 60 seconds, and the analysis frequency range is 0 to 50 Hz.

7. The method for rapid assessment of abnormal vibrations in high-rise buildings based on time-frequency analysis according to claim 1, characterized in that: In step one, the acquired vibration acceleration data covers at least the low, middle and high zones of the high-rise building; and the vibration acceleration data is obtained by monitoring a triaxial accelerometer; the sampling frequency of the triaxial accelerometer is between 50Hz and 200Hz.

8. The method for rapid assessment of abnormal vibrations in high-rise buildings based on time-frequency analysis according to claim 1, characterized in that: The method for rapid assessment of abnormal vibrations in high-rise buildings also includes: Step Six: Frequency and Spatial Response Analysis: For the abnormal candidate windows identified in step five, frequency feature analysis and multi-point spatial response analysis are performed: The frequency feature analysis is as follows: extract the power spectral density curve of the abnormal signal within the abnormal candidate window and the power spectral density curve of the normal signal within a preset time period before and after the abnormal candidate window, and compare them to determine the frequency band range and frequency value of the power spectral density enhancement of the abnormal signal relative to the normal signal, so as to determine whether the frequency band selectivity of the abnormal signal is wideband indiscriminate amplification or specific frequency band selective enhancement. The multi-point spatial response analysis involves comparing the vibration responses between different floors and between different key structural locations on the same floor within the same abnormal time period, for the same abnormal candidate window, to determine the spatial distribution pattern and propagation path characteristics of the abnormal vibration response during the abnormal time period. Specifically, this includes: The power spectral density energy distribution of abnormal signals within the frequency band of the abnormal candidate window is compared on different floors to analyze the changing trend of abnormal vibration response with the floor and determine whether the abnormal vibration response shows the characteristic of increasing or decreasing with the height of the high-rise building. Compare the power spectral density curves and time history responses of the core tube region and the outer frame structure region on the same floor to determine whether the abnormal vibration response first manifests in the core tube region or the outer frame structure region. By comparing the order in which vibration acceleration data from different floors reached their peak values ​​during abnormal periods, the direction of propagation of abnormal signals can be determined.

9. The method for rapid assessment of abnormal vibrations in high-rise buildings based on time-frequency analysis according to claim 8, characterized in that: Step six, based on the frequency characteristic analysis conclusions and the multi-point spatial response analysis conclusions, assesses the abnormal vibration events occurring at the abnormal candidate windows of the high-rise building, including: If the frequency selectivity of the abnormal signal is a wideband indiscriminate amplification with an overall increase across the entire frequency band, it indicates that the abnormal vibration event belongs to one or more of wideband disturbance, impact, and overall noise enhancement. If the frequency selectivity of the abnormal signal is manifested as a specific frequency selective enhancement in which the power spectral density is significantly enhanced only within a certain specific frequency band, it indicates that the abnormal vibration event has frequency selectivity. Based on the spatial distribution pattern and propagation path characteristics, the possible types of abnormal vibration sources are gradually narrowed down, and the location of suspected vibration source equipment or suspected vibration source components is initially identified.

10. The method for rapid assessment of abnormal vibrations in high-rise buildings based on time-frequency analysis according to claim 8, characterized in that: The power ledger in step four also records the energy level of each representative frequency strip, its spatial distribution characteristics on each floor, and the possible vibration sources corresponding to the representative frequency strip. Step six also includes: The abnormal enhancement frequency is compared with the data recorded in the power ledger, wherein the abnormal enhancement frequency is the frequency band range and frequency value of the power spectral density enhancement of the abnormal signal relative to the normal signal. If the abnormally enhanced frequency is close to the frequency components of the representative frequency band recorded in the power log, it indicates that the abnormal excitation may have excited the existing structural frequency or local component frequency of the high-rise building. If the abnormally enhanced frequency is close to the normal operating frequency of a certain device, but the occurrence time, energy level or duration of the abnormally enhanced frequency is significantly different from the normal characteristics of the device, it may indicate that the device is operating under abnormal conditions. If the abnormally enhanced frequency is not recorded in the power log, it indicates that there may be a new vibration source, temporary working conditions, or non-routine excitation.

Images