A bridge scour dynamics identification method based on health monitoring data

By collecting vibration acceleration data of the bridge infrastructure, using Fourier transform and kernel density estimation methods, time-frequency abnormal segments are identified, and the problem of difficult monitoring of the bridge foundation erosion depth is solved, and the rapid diagnosis and safety warning of the bridge structure is achieved, reducing the The cost of testing is available and the ability to monitor dynamically for a long time is available.

CN114036974BActive Publication Date: 2024-11-26SOUTHEAST UNIV

Patent Information

Application Number
CN202111203755.8
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2021-10-15
Publication Date
2024-11-26
Estimated Expiration
2041-10-15

AI Technical Summary

Technical Problem

The existing technology is difficult to effectively monitor and early warning the depth of the bridge foundation, especially in underwater environments, which makes it difficult to ensure the safety of the bridge structure, and the traditional detection methods are costly and do not have long-term dynamic monitoring capabilities.

Method used

By collecting vibration acceleration data of the bridge infrastructure in the erosion state, combining Fourier transform and kernel density estimation methods, establishing a probability distribution model, identifying time-frequency abnormal segments, setting up and down control thresholds, and realizing dynamics of the bridge erosion depth Identification and early warning.

Benefits of technology

It realizes rapid detection and diagnosis of bridge infrastructure, provides early warning of bridge structure safety, reduces detection costs, has long-term dynamic monitoring capabilities, and avoids the need for direct underwater observation.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN114036974B_ABST
    Figure CN114036974B_ABST
Patent Text Reader

Abstract

The present invention discloses a bridge scour dynamic identification method based on health monitoring data, comprising: collecting the acceleration-time curve of the bridge foundation structure when it vibrates: using the health monitoring system to collect the acceleration-time curve of each bridge foundation structure when it vibrates in the scour state; calculating the warning control threshold of the time-frequency change abnormal warning of the benchmark mode of the first scour bridge assessment; identifying the abnormal segment in the frequency segment of the scour bridge to be identified; identifying the abnormal time-frequency sequence in the abnormal time-frequency segment: after the abnormal sequence completes the scour warning, updating the warning control threshold of the random fluctuation of the time-frequency feature of the bridge scour benchmark mode itself, so as to prepare for the next abnormal change identification and scour warning; the present invention dynamically identifies the foundation scour depth by analyzing the dynamic characteristics of the structural system, and the identification method can also realize the technical characteristics of long-term dynamic underwater foundation scour monitoring and early warning.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of bridge health monitoring, and in particular to a method for dynamically identifying foundation scour depth by analyzing structural system dynamic characteristics based on bridge monitoring vibration acceleration time history data. Background Art

[0002] Bridge foundation scour is one of the main reasons for the failure of bridge structure function and loss of its safety performance. Taking the United States as an example, from 1966 to 2005, 58% of the collapsed bridges (1,502 bridges) in the United States were related to bridge foundation structure scour. The U.S. Department of Transportation has regarded bridge foundation scour as one of the most common reasons for the failure of highway bridge structure function and safety performance. In my country, especially in the eastern region, the pile length of highway bridges over water is generally designed to be 10-50m. According to the regular inspection data of old and old bridges on site, it is found that more than 5m is the common depth of bridge foundation scour, and the scour depth can even exceed 10m under certain hydrological conditions. For large bridges across rivers, lakes and seas, the balanced scour of the foundation structure can reach more than 20m. In addition, since scour occurs below the water surface, there is usually no sign of scour damage to the bridge foundation, which seriously endangers the safety performance of the bridge structure and the smooth operation of the traffic highway network.

[0003] In order to predict the development of local scour patterns in advance, reasonably conduct safety assessment and reinforcement decisions of bridge foundations during the operation stage, and prevent catastrophic structural collapse caused by bridge foundation scour, it is obviously necessary to conduct long-term and regular inspection and diagnosis of the scour status of bridge foundations. For a long time, the scour status of bridge foundations has been mainly based on subjective experience judgments of daily inspections, with low accuracy. Although scour model experiments can be carried out for individual extra-large bridges, the model similarity ratio is difficult to determine, and the human and material costs of the experiment are high, and it cannot be widely used in general bridge design; in recent years, the newly emerged underwater detection equipment, such as sonar, TDR technology, multi-beam detection systems, and underwater robots, are still limited by terrain factors to detect the depth of foundation scour, and the cost of a single detection is high, and long-term dynamic underwater monitoring cannot be achieved. Summary of the invention

[0004] The technical problem to be solved by the present invention is to address the deficiencies of the above-mentioned prior art and to propose a method for dynamically identifying the foundation scour depth by analyzing the dynamic characteristics of the structural system based on the bridge monitoring vibration acceleration time history data. The method has wide applicability, relatively low cost, and can realize long-term dynamic underwater foundation scour monitoring and early warning.

[0005] In order to solve the above technical problems, the technical solution adopted by the present invention is: a bridge scour dynamics identification method based on health monitoring data, comprising the following steps:

[0006] Step 1, collecting the acceleration-time curve when the bridge foundation structure vibrates: using the health monitoring system to collect the acceleration-time curve of each bridge foundation structure when each bridge foundation structure vibrates under the scouring state, and pre-processing the acceleration-time curve with anti-interference factors;

[0007] Step 2, obtaining a bridge scour reference modal frequency-time curve: the acceleration-time curve in step 1 is subjected to Fourier transformation to obtain a scour reference modal frequency-time curve;

[0008] Step 3, determine the value of the significance level α;

[0009] Step 3.1, using the kernel density estimation method to establish the probability distribution model of the bridge scour assessment benchmark modal time-frequency, converting the scour benchmark modal frequency into a random variable that obeys the standard normal distribution;

[0010] Step 3.2: Based on the random variables that obey the standard normal distribution and in combination with the Shewhart mean control chart, preliminarily set the value of the significance level α, obtain the probability distribution function corresponding to the significance level α, and establish a normal distribution probability model;

[0011] Step 3.3, according to the value range of the significance level α that is initially set, perform recognition sensitivity calibration;

[0012] Step 4, bring α into the normal distribution probability model, and calculate the upper control threshold UCL and the lower control threshold LCL of the time-frequency change abnormal warning of the first scour bridge assessment benchmark mode;

[0013] Step 5: Identify abnormal segments in the frequency segments of the bridge to be identified:

[0014] Step 5.1, determining the frequency segment of the bridge to be identified that exceeds the upper control threshold UCL or the lower control threshold LCL as a time-frequency abnormal segment;

[0015] Step 6: Identify abnormal time-frequency sequences in abnormal time-frequency segments:

[0016] Step 6.1: The time-frequency anomaly segment includes multiple time-frequency sequences, and the time-frequency anomaly sequence among the multiple time-frequency sequences is identified:

[0017] Set the recognition parameters of the time-frequency anomaly sequence. The recognition parameters of the time-frequency anomaly sequence include the time length ratio parameter P of the abnormal reference frequency sequence. L / U ', the time interval parameter Ts' between two adjacent abnormal frequencies, the mean change difference parameter M of the scour reference frequency s ';

[0018] Step 6.2: Calculate the time length ratio parameter P of the abnormal frequency sequence of the abnormal segmentL / U :

[0019] P L / U = T ab / T t0

[0020] where Tab is the time length of the frequency sequence exceeding the upper control threshold UCL or the lower control threshold LCL, and Tt0 is the total time length of the abnormal segment;

[0021] Calculate Ts of the abnormal segment, where Ts is the time interval between two adjacent tabs;

[0022] When P L / U > P L / U ', and Ts < Ts', then determine that the time-frequency sequence is an abnormal sequence and go to step 6.3; otherwise, determine that the time-frequency sequence is normal;

[0023] Step 6.3, calculate the change difference M of the time series mean of the scour reference frequency in the time-frequency abnormal sequence s :

[0024] M s = |M1 - M2|

[0025] where M1 is the frequency mean of the time-frequency abnormal sequence, and M2 is the frequency mean in the healthy state of the previous same time interval of the abnormal segment;

[0026] When M s ≤ M s ', then determine that the abnormal sequence is a normal signal oscillation; when M s > M s ', then give a scour warning for the abnormal sequence;

[0027] Step 7, after the scour warning for the abnormal sequence is completed, repeat steps 5 - 6 to update the upper and lower control thresholds of the random fluctuation of the time-frequency characteristics of the bridge scour reference mode, and prepare for the next mutation identification and scour warning.

[0028] As a further preferred solution of the present invention, in step 1, it specifically includes the following steps:

[0029] Step 1.1, use the health monitoring system to collect the acceleration-time curve of each bridge foundation during scour, and use a filter and a signal detrending function to remove the high-order frequency signals in the acceleration-time curve;

[0030] Step 1.2, calculate and process to obtain the missing signal length in the frequency-time curve:

[0031] First, define the index structure missing:

[0032]

[0033] Among them, s m 、e m are the start and end indexes of the missing data of the mth segment respectively; k is the total number of missing data segments; the length of the missing signal of the mth segment is missing.longm=e m -s m ;

[0034] When the missing signal length is less than the length tolerance critical value, the extension filling method is used to fill the missing signal length;

[0035] When the missing signal length is greater than the length tolerance critical value, the missing signal length is discarded;

[0036] Step 1.3, identifying and removing outliers in the frequency-time curve, and using numerical interpolation to supplement the removed outliers;

[0037] Step 1.4, remove the temperature effect in the acceleration-time curve obtained in step 1.3 to obtain the bridge scour reference modal frequency-time curve: the bridge foundation structure frequency at a specific temperature is measured in advance by a temperature sensor, and the acceleration-time curve is decomposed by the EMD algorithm to obtain a multi-order sub-modal acceleration-time curve. The multi-order sub-modal acceleration-time curve is transformed by Fourier transform to obtain the main frequency, and the acceleration-time curve close to the bridge foundation structure frequency at a specific temperature is eliminated, and then reconstructed to obtain the scour reference modal acceleration-time curve.

[0038] As a further preferred embodiment of the present invention, the specific steps of step 3.1 are:

[0039] The frequency signal to be identified is taken as a one-dimensional continuous sample vector f i , and use the selected kernel density function to get the sample vector f i The kernel density estimation vector PDF(f i ),

[0040]

[0041]

[0042]

[0043] Among them, K(x) is the selected kernel density function, l is the data length of a set time series, h is the set time interval value, i is the order, and f is the time interval value. i,j is the modal frequency vector f i The jth data of

[0044] By kernel density estimation, the vector PDF (f i ), calculate the sample vector f i The cumulative probability estimate vector CDF(f i ):

[0045]

[0046] Finally, the cumulative probability estimate vector CDF(f i ) performs an inverse transformation of the standard normal distribution function and converts it into the Q statistic:

[0047] Q i =Φ -1 (CDF(f i ))

[0048] Among them, Φ -1 (·) is the inverse function of the standard normal distribution function, which completes the conversion of non-normal data into random variables that obey the standard normal distribution.

[0049] As a further preferred embodiment of the present invention, the specific steps of step 3.2 are: based on the random variable obeying the standard normal distribution, combined with the Shewhart mean control chart, preliminarily set the value of the significance level α, and obtain the probability distribution function corresponding to the significance level α:

[0050]

[0051] Among them, μ0 is the population mean, σ is the sample population standard deviation, α is the significance level, and Z α / 2 is the upper α / 2 quantile of the standard normal distribution, f is the parameter in the probability density function, and n is the total number of samples to be tested;

[0052] The normal distribution probability model is established through the probability distribution function corresponding to the significance level value α.

[0053] As a further preferred embodiment of the present invention, in step 4, the upper control threshold UCL is calculated as follows:

[0054]

[0055] The lower control threshold LCL is calculated as follows:

[0056]

[0057] As a further preferred embodiment of the present invention, it is characterized in that: in step 3.2, the value range of the initially set significance level value α is 0.05-0.15.

[0058] As a further preferred embodiment of the present invention, in step 6.1, let P L / U '=1%,Ts'=0.1s、M s '=0.01Hz.

[0059] The present invention has the following beneficial effects:

[0060] 1. The present invention discloses a bridge scour dynamics identification method based on health monitoring data. By collecting the change of acceleration of the bridge foundation structure under the scour state over time, the change vibration frequency of the bridge foundation structure under the scour state is obtained, thereby achieving the identification and early warning of the change of the bridge foundation structure, and finally achieving the rapid detection and diagnosis of the deformation of the bridge foundation structure, providing a theoretical basis for further selective underwater detection and regional bridge scour rapid screening, and providing important support for early warning of bridge structure safety. Compared with conventional scour detection technology, this method does not require underwater operations, does not require direct observation of the scour state, and only relies on on-site measurement and parameter tracking of the dynamic response of the bridge superstructure.

[0061] 2. By considering the influence of temperature effect on the frequency of vibration changes of the bridge foundation structure under scouring conditions, the accuracy of identifying the deformation degree of the bridge foundation structure can be improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] Figure 1 It is a scour warning implementation flow chart of a bridge scour dynamics identification method based on health monitoring data of the present invention;

[0063] Figure 2 It is a schematic diagram of the control limit setting of the Shewhart control chart used in the present invention when calculating the abnormal warning control threshold of the time-frequency characteristic change of the scour assessment benchmark mode;

[0064] Figure 3a This is a scour warning result diagram based on measured data of a large cable-stayed bridge when the significance level α is set to 0.05;

[0065] Figure 3b This is a scour warning result diagram based on measured data of a large cable-stayed bridge when the significance level α is set to 0.10;

[0066] Figure 3c This is a scour warning result diagram based on measured data of a large cable-stayed bridge when the significance level α is set to 0.15. DETAILED DESCRIPTION

[0067] The present invention will be further described in detail below in conjunction with the accompanying drawings and specific preferred embodiments.

[0068] In the description of the present invention, it should be understood that the terms "left side", "right side", "upper part", "lower part" and the like indicate the orientation or positional relationship based on the orientation or positional relationship shown in the drawings, which is only for the convenience of describing the present invention and simplifying the description, and does not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation. "First", "second" and the like do not indicate the importance of the components, and therefore cannot be understood as limiting the present invention. The specific dimensions used in this embodiment are only for illustrating the technical solution, and do not limit the scope of protection of the present invention.

[0069] The basic conditions that the present invention needs to meet are: the bridge structure is only subject to environmental noise or natural excitation, where natural excitation means that there are only environmental effects such as vehicle loads, wave scouring, wind loads, etc. without applying artificial excitation loads, and the external signal input to the structural system can be regarded as white noise. After that, the various excitation response outputs to the bridge structure system can be obtained by relevant algorithms and dynamic characteristics.

[0070] The algorithm of the present invention mainly serves the field of bridge scour health monitoring, and the application object is the field measured data of bridge structures. Therefore, the relevant algorithm steps can be completed by writing codes based on professional software toolkits. In this embodiment, the programming calculation of the relevant algorithm steps is completed based on the Matlab program.

[0071] A bridge scour dynamics identification method based on health monitoring data, the specific steps are as follows:

[0072] Step 1: Collect the acceleration-time curve of the bridge foundation structure when it vibrates: Use the health monitoring system to collect the acceleration-time curve of each bridge foundation structure when it vibrates under the scouring state, and pre-process the acceleration-time curve against interference factors.

[0073] Step 1.1, using the health monitoring system to collect data of each bridge foundation in the scouring state, obtain the acceleration-time curve of each bridge foundation during vibration, and use a filter and a signal detrending function to remove high-order frequency signals in the acceleration-time curve;

[0074] Step 1.2: Calculate and process the missing signal length in the frequency-time curve:

[0075] First, define the index structure missing:

[0076]

[0077] Among them, s m 、e mare the start and end indexes of the missing data of the mth segment respectively; k is the total number of missing data segments; the length of the missing signal of the mth segment is missing.longm=e m -s m ;

[0078] When the missing signal length is less than the length tolerance critical value, the extension filling method is used to fill the missing signal length;

[0079] When the missing signal length is greater than the length tolerance critical value, the missing signal length is discarded;

[0080] Step 1.3, identifying and removing outliers in the frequency-time curve, and using numerical interpolation to supplement the removed outliers;

[0081] Step 1.4, remove the temperature effect in the acceleration-time curve obtained in step 1.3 to obtain the bridge scour reference modal frequency-time curve: the bridge foundation structure frequency at a specific temperature is measured in advance by a temperature sensor, and the acceleration-time curve is decomposed by the EMD algorithm to obtain a multi-order sub-modal acceleration-time curve. The multi-order sub-modal acceleration-time curve is transformed by Fourier transform to obtain the main frequency, and the acceleration-time curve close to the bridge foundation structure frequency at a specific temperature is eliminated, and then reconstructed to obtain the scour reference modal acceleration-time curve.

[0082] Step 2: Obtain the bridge scour reference modal frequency-time curve: Perform Fourier transformation on the acceleration-time curve in step 1 to obtain the scour reference modal frequency-time curve.

[0083] Step 3: Determine the value of the significance level α

[0084] Step 3.1, using the kernel density estimation method to establish the probability distribution model of the bridge scour assessment benchmark modal time-frequency, converting the scour benchmark modal frequency into a random variable that obeys the standard normal distribution;

[0085] The frequency signal to be identified is taken as a one-dimensional continuous sample vector f i , and use the selected kernel density function to get the sample vector f i The kernel density estimation vector PDF(f i ),

[0086]

[0087]

[0088]

[0089] Among them, K(x) is the selected kernel density function, l is the data length of a set time series, h is the set time interval value, i is the order, and f is the time interval value. i,j is the modal frequency vector f i The jth data of

[0090] By kernel density estimation, the vector PDF (f i ), calculate the sample vector f i The cumulative probability estimate vector CDF(f i ):

[0091]

[0092] Finally, the cumulative probability estimate vector CDF(f i ) performs an inverse transformation of the standard normal distribution function and converts it into the Q statistic:

[0093] Q i =Φ -1 (CDF(f i ))

[0094] Among them, Φ -1 (·) is the inverse function of the standard normal distribution function, which completes the conversion of non-normal data into random variables that obey the standard normal distribution.

[0095] Step 3.2: According to the random variable obeying the standard normal distribution, combined with the Shewhart mean control chart, the value of the significance level α is preliminarily set. In this embodiment, the value range of the preliminarily set significance level α is 0.05-0.15, and the probability distribution function corresponding to the significance level α is obtained to establish a normal distribution probability model;

[0096] According to the random variables that obey the standard normal distribution, combined with the Shewhart mean control chart, the value of the significance level α is preliminarily set, and the probability distribution function corresponding to the significance level α is obtained:

[0097]

[0098] Among them, μ0 is the population mean, σ is the sample population standard deviation, α is the significance level, and Z α / 2 is the upper α / 2 quantile of the standard normal distribution, f is the parameter in the probability density function, and n is the total number of samples to be tested;

[0099] The normal distribution probability model is established through the probability distribution function corresponding to the significance level value α.

[0100] Step 3.3: Perform recognition sensitivity calibration based on the value range of the significance level α that is initially set.

[0101] Step 4: Substitute α’ into the normal distribution probability model to calculate the upper control threshold UCL and the lower control threshold LCL for the time-frequency change anomaly warning of the first-scour bridge evaluation reference mode:

[0102]

[0103]

[0104] Step 5: Identify the abnormal segments in the frequency segments of the bridge to be identified for scour:

[0105] Step 5.1: Determine the frequency segments of the bridge to be identified for scour that exceed the upper control threshold UCL or the lower control threshold LCL as time-frequency abnormal segments.

[0106] Step 6: Identify the abnormal time-frequency sequences in the time-frequency abnormal segments:

[0107] Step 6.1: The time-frequency abnormal segments contain multiple time-frequency sequences. Identify the time-frequency abnormal sequences in the multiple time-frequency sequences: Set the identification parameters for the time-frequency abnormal sequences. The identification parameters for the time-frequency abnormal sequences include the time length ratio parameter P L / U ' of the abnormal reference frequency sequence, the time interval parameter Ts' between two adjacent abnormal frequencies, and the mean change difference parameter M s ' of the scour reference frequency. In this embodiment, let P L / U ' = 1%, Ts' = 0.1 s, and M s ' = 0.01 Hz.

[0108] Step 6.2: Calculate the time length ratio parameter P L / U of the abnormal frequency sequence of the abnormal segment:

[0109] P L / U = T ab / T t0

[0110] where Tab is the time length of the frequency sequence that exceeds the upper control threshold UCL or the lower control threshold LCL, and Tt0 is the total time length of the abnormal segment;

[0111] Calculate Ts of the abnormal segment. Ts is the time interval between two adjacent tabs;

[0112] When P L / U > P L / U ', and Ts < Ts', then determine that this time-frequency sequence is an abnormal sequence and proceed to Step 6.3. Otherwise, determine that this time-frequency sequence is normal.

[0113] Step 6.3: Calculate the mean change difference M s of the scour reference frequency time series in the time-frequency abnormal sequence:

[0114] M s =|M1-M2|

[0115] Among them, M1 is the frequency mean of the time-frequency anomaly sequence, and M2 is the frequency mean of the healthy state in the same time interval before the anomaly segment;

[0116] When M s ≤M s ', then the abnormal sequence is judged to be a normal signal oscillation; when M s >M s ', then a flush warning will be issued for the abnormal sequence.

[0117] Step 7: After the abnormal sequence completes the scour warning, repeat steps 5 and 6 to update the upper and lower control thresholds of the random fluctuations of the bridge scour benchmark modal time-frequency characteristics themselves, in preparation for the next abnormality identification and scour warning.

[0118] The above method steps and basic formula principles can also be loaded onto a computer or other programmable data processing device, so that a series of operation steps are executed on the computer or other programmable device to realize computer processing, so that the instructions executed on the computer or other programmable device provide for realizing the process Figure 1 A function or step specified in a process or multiple processes.

[0119] The preferred embodiments of the present invention are described in detail above; however, the present invention is not limited to the specific details in the above embodiments. Within the technical concept of the present invention, various equivalent transformations can be made to the technical solutions of the present invention, and these equivalent transformations all belong to the protection scope of the present invention.

Claims

1. A bridge scour dynamics identification method based on health monitoring data, characterized by: The following steps are involved: Step 1, collecting the acceleration-time curve when the bridge foundation structure vibrates: using the health monitoring system to collect the acceleration-time curve of each bridge foundation structure when each bridge foundation structure vibrates under the scouring state, and pre-processing the acceleration-time curve with anti-interference factors; Step 2, obtaining a bridge scour reference modal frequency-time curve: the acceleration-time curve in step 1 is subjected to Fourier transformation to obtain a scour reference modal frequency-time curve; Step 3, determine the value of the significance level α; Step 3.1, using the kernel density estimation method to establish the probability distribution model of the bridge scour assessment benchmark modal time-frequency, converting the scour benchmark modal frequency into a random variable that obeys the standard normal distribution; Step 3.2: Based on the random variables that obey the standard normal distribution and in combination with the Shewhart mean control chart, preliminarily set the value of the significance level α, obtain the probability distribution function corresponding to the significance level α, and establish a normal distribution probability model; Step 3.3, according to the value range of the significance level α that is initially set, perform recognition sensitivity calibration; Step 4, bring α into the normal distribution probability model, and calculate the upper control threshold UCL and the lower control threshold LCL of the time-frequency change abnormal warning of the first scour bridge assessment benchmark mode; Step 5: Identify abnormal segments in the frequency segments of the bridge to be identified: Step 5.1, determining the frequency segment of the bridge to be identified that exceeds the upper control threshold UCL or the lower control threshold LCL as a time-frequency abnormal segment; Step 6: Identify abnormal time-frequency sequences in abnormal time-frequency segments: Step 6.1: The time-frequency anomaly segment includes multiple time-frequency sequences, and the time-frequency anomaly sequence among the multiple time-frequency sequences is identified: Set the recognition parameters of the time-frequency anomaly sequence. The recognition parameters of the time-frequency anomaly sequence include the time length ratio parameter P of the abnormal reference frequency sequence. L / U ', the time interval parameter Ts' between two adjacent abnormal frequencies, the mean change difference parameter M of the scour reference frequency s '; Step 6.2: Calculate the time length ratio parameter P of the abnormal frequency sequence of the abnormal segment L / U : P L / U =T ab / T t0 Among them, Tab is the time length of the frequency sequence exceeding the upper control threshold UCL or the lower control threshold LCL, and Tt0 is the total time length of the abnormal segment; Calculate Ts of the abnormal segment, where Ts is the time interval between two adjacent Tabs; When P L / U > P L / U ', and when Ts < Ts', then determine that this time-frequency sequence is an abnormal sequence and proceed to step 6.3; otherwise, determine that this time-frequency sequence has no abnormality. Step 6.3: Calculate the mean change difference M of the washout reference frequency time series in the time-frequency anomaly sequence s : M s =|M1-M2| Among them, M1 is the frequency mean of the time-frequency anomaly sequence, and M2 is the frequency mean of the healthy state in the same time interval before the anomaly segment; When M s ≤M s ', then the abnormal sequence is judged to be a normal signal oscillation; when M s >M s ', then the abnormal sequence will be flushed early warning; Step 7: After the abnormal sequence completes the scour warning, repeat steps 5 and 6 to update the upper and lower control thresholds of the random fluctuations of the bridge scour benchmark modal time-frequency characteristics themselves, in preparation for the next abnormality identification and scour warning.

2. The bridge scour dynamics identification method based on health monitoring data according to claim 1 is characterized by: The step 1 specifically includes the following steps: Step 1.1, using the health monitoring system to collect data of each bridge foundation in the scouring state, obtain the acceleration-time curve of each bridge foundation during vibration, and use a filter and a signal detrending function to remove high-order frequency signals in the acceleration-time curve; Step 1.2: Calculate and process the missing signal length in the frequency-time curve: First, define the index structure missing: Among them, s m 、e m are the start and end indexes of the missing data of the mth segment respectively; k is the total number of missing data segments; the length of the missing signal of the mth segment is missing.longm=e m -s m ; When the missing signal length is less than the length tolerance critical value, the extension filling method is used to fill the missing signal length; When the missing signal length is greater than the length tolerance critical value, the missing signal length is discarded; Step 1.3, identifying and removing outliers in the frequency-time curve, and using numerical interpolation to supplement the removed outliers; Step 1.4, remove the temperature effect in the acceleration-time curve obtained in step 1.3 to obtain the bridge scour reference modal frequency-time curve: the bridge foundation structure frequency at a specific temperature is measured in advance by a temperature sensor, and the acceleration-time curve is decomposed by the EMD algorithm to obtain a multi-order sub-modal acceleration-time curve. The multi-order sub-modal acceleration-time curve is transformed by Fourier transform to obtain the main frequency, and the acceleration-time curve close to the bridge foundation structure frequency at a specific temperature is eliminated, and then reconstructed to obtain the scour reference modal acceleration-time curve.

3. The bridge scour dynamics identification method based on health monitoring data according to claim 1 is characterized by: The specific steps of step 3.1 are: The frequency signal to be identified is taken as a one-dimensional continuous sample vector f i , and use the selected kernel density function to get the sample vector f i The kernel density estimation vector PDF(f i ), Among them, K(x) is the selected kernel density function, l is the data length of a set time series, h is the set time interval value, i is the order, and f is the time interval value. i,j is the modal frequency vector f i The jth data of By kernel density estimation, the vector PDF (f i ), calculate the sample vector f i The cumulative probability estimate vector CDF(f i ): Finally, the cumulative probability estimate vector CDF(f i ) performs an inverse transformation of the standard normal distribution function and converts it into the Q statistic: Q i =Φ -1 (CDF(f i )) Among them, Φ -1 (·) is the inverse function of the standard normal distribution function, which completes the conversion of non-normal data into random variables that obey the standard normal distribution.

4. The bridge scour dynamics identification method based on health monitoring data according to claim 3 is characterized by: The specific steps of step 3.2 are as follows: based on the random variable that obeys the standard normal distribution, combined with the Shewhart mean control chart, preliminarily set the value of the significance level α, and obtain the probability distribution function corresponding to the significance level α: Among them, μ0 is the population mean, σ is the sample population standard deviation, α is the significance level, and Z α / 2 is the upper α / 2 quantile of the standard normal distribution, f is the parameter in the probability density function, and n is the total number of samples to be tested; The normal distribution probability model is established through the probability distribution function corresponding to the significance level value α.

5. The bridge scour dynamics identification method based on health monitoring data according to claim 4 is characterized by: In step 4, the upper control threshold UCL is calculated as follows: The lower control threshold LCL is calculated as follows:

6. The bridge scour dynamics identification method based on health monitoring data according to claim 1 is characterized by: In step 3.2, the significance level α initially set ranges from 0.05 to 0.

15.

7. The bridge scour dynamics identification method based on health monitoring data according to claim 1 is characterized by: In step 6.1, let P L / U '=1%,Ts'=0.1s、M s '=0.01Hz.

Citation Information

Patent Citations

  • A method of bridge foundation erosion identification based on vehicle braking impact

    CN109002673A

  • Operation bridge vibration and health state abnormity early warning method based on three-dimensional diagram

    CN112834193A

  • Bridge Construction System and Method

    US20210025120A1

Cited By

  • Self-adaptive noise hybrid filtering method for single-pile scouring recognition

    CN122388354A