Failure monitoring and alarm method for stay cable damper of long-span cable-stayed bridge

Abstract

The present invention discloses a failure monitoring and alarming method for a stay cable damper of a long-span cable-stayed bridge, belonging to the technical field of bridge structural performance monitoring. The present invention includes the following steps: (1) Collect the acceleration response at the section where the measuring points are arranged; (2) Calculate the acceleration power spectrum of different sections; (3) Conduct Gaussian clustering analysis to obtain a Gaussian mixture model; (4) Obtain the maximum power spectrum value and form a time history curve of the maximum power spectrum value; (5) Establish a linear regression model between the maximum power spectrum value and the bridge ambient temperature, and a time history curve of the maximum power spectrum value normalized by temperature; (6) Conduct extreme value analysis to obtain the maximum value of the maximum power spectrum value with a 99% guarantee rate; (7) Monitor pulse mutations and give an alarm for the failure of the stay cable damper. The present invention utilizes the correlation between the acceleration response of the main girder and the acceleration response of the stay cable to realize the failure early warning of the damper, which is convenient for maintenance personnel to perceive and check in time.

Description

Technical Field

[0001] The invention relates to the field of bridge structure performance monitoring, and in particular to a cable damper state perception method based on a main beam acceleration power spectrum Gaussian mixture model. Background Art

[0002] The cables of a cable-stayed bridge are its key force transmission path. The working condition of the cables has a great influence on the performance of the cable-stayed bridge. Therefore, dampers are often installed on the cables to improve their mechanical characteristics and enhance the mechanical performance of the cable-stayed bridge.

[0003] The damper is in an open-air environment, directly affected by wind, rain and sun, and at the same time bears the alternating load of vehicles. Therefore, the cable damper is prone to disease and fatigue damage, and is even completely destroyed directly under the action of strong winds. In order to sense the working performance of the damper, the usual way is to install an acceleration sensor on it. When its dynamic performance parameters change, the damage of the damper can be judged. However, for a large-span cable-stayed bridge, the number of cable components is large, and there can be hundreds or even hundreds of cables on a large-span cable-stayed bridge. Simply installing sensors on the cables will be cost-unacceptable, so it is necessary to explore new technologies that use multi-source response data of bridges to sense the state of the cables. Summary of the invention

[0004] The object of the present invention is to provide a failure monitoring and alarm method for a cable damper of a long-span cable-stayed bridge based on a Gaussian mixture model of main beam acceleration power spectrum.

[0005] The technical solution of the present invention is: a failure monitoring and alarm method for a cable damper of a long-span cable-stayed bridge, characterized in that it specifically comprises the following steps:

[0006] Step 1: Set multiple measurement points on the main beam of the cable-stayed bridge. The measurement points have the function of expressing the potential energy of the main beam of the cable-stayed bridge. The acceleration response of the cross-sectional position of each measurement point is collected by an acceleration sensor;

[0007] Step 2: Calculate the acceleration power spectrum of the cross section at each measurement point under the same time step;

[0008] Step 3: Perform Gaussian cluster analysis on the acceleration power spectra of all measurement point sections to obtain a Gaussian mixture model of the acceleration power spectrum with the time step as the time interval;

[0009] Step 4: Obtain the maximum power spectrum value of the Gaussian mixture model and form a maximum power spectrum value time course curve;

[0010] Step 5: Establish a linear regression model between the maximum power spectrum value and the bridge ambient temperature, eliminate the influence of temperature, and establish a temperature-normalized maximum power spectrum value time history curve;

[0011] Step 6: Conduct an extreme value analysis on the time history curve of the maximum power spectral value after temperature normalization to obtain the maximum value of the maximum power spectral value with a 99% guarantee rate;

[0012] Step 7: When a pulse mutation occurs in the monitored time history curve of the maximum power spectral value and exceeds the maximum value obtained in Step 6, an alarm for the failure of the stay cable damper is given.

[0013] In a further technical solution, in Step 1, the acceleration responses at the mid-span of the main span of the cable-stayed bridge girder, at the 1 / 4 span of the main span, at the 1 / 8 span of the main span, and at the mid-span of the side span are collected.

[0014] In a further technical solution, the time step in Step 2 is taken as 1 min.

[0015] In a further technical solution, in Step 2, it is necessary to calculate the acceleration power spectrum of the data, and the calculation methods include algorithms such as PSD or LPSD or DPSD, etc.

[0016] In a further technical solution, the method for obtaining the regression model in Step 5 is regression techniques such as least squares or machine learning and deep learning.

[0017] In a further technical solution, set the maximum value in Step 6 as the warning threshold, and adjust the deviation range of the warning threshold according to the characteristics of the bridge.

[0018] Advantages of the present invention:

[0019] As a structure directly connected to the stay cable, the acceleration response of the cable-stayed bridge girder has a strong correlation with the acceleration response of the stay cable. When a sudden failure event occurs in the working state of the stay cable, it will surely cause abnormal vibration of the stay cable, and then be reflected in the dynamic parameters of the cable-stayed bridge girder. Therefore, the failure of the stay cable damper can be identified by using the abnormal movement of the dynamic parameters of the cable-stayed bridge girder, realizing the maintenance of the bridge.

[0020] In the present invention, the abnormal failure of the stay cable damper is identified by using the state of the cable-stayed bridge girder. By amplifying the abnormal vibration information of the bridge acceleration caused by the damper and eliminating the influence of environmental temperature on the natural frequency of the bridge components at the same time, the joint information performance characterization value is extracted, and finally the warning threshold characterizing the abnormal failure of the damper performance is obtained, realizing the alarm for the damage event of the bridge components.

[0021] Compared with the prior art, the present invention has the following advantages:

[0022] (1) Information multiplication: Using Gaussian mixture clustering, jointly using the information of all the main girder acceleration sensors, multiplying the smaller signal peaks of the abnormal vibration of the stay cable transmitted to the main girder, and ensuring the basic information of the abnormal event of the stay cable damper based on the vibration information of the main girder;

[0023] (2) High-precision characteristic data: Converting the acceleration information of the main girder into power spectral values can effectively reduce the uncertainty caused by the discreteness of acceleration data. By establishing a correlation model between the power spectral characteristic values and temperature information, the drift error caused by temperature effects can be removed. The threshold is determined with a 99% confidence level to ensure that the alarm system has a high level of reliability. Description of the Drawings

[0024] Figure 1 It is the algorithm flowchart of the method of the present invention.

[0025] Figure 2 It is the power spectrum of the main girder acceleration data within a 1-minute interval calculated by the method of the present invention.

[0026] Figure 3 It is the Gaussian mixture clustering model of the combined information of the power spectra of each main girder acceleration sensor obtained by the method of the present invention.

[0027] Figure 4 It is the peak time history curve of the Gaussian clustering model obtained by the method of the present invention.

[0028] Figure 5 It is the regression model obtained by using the peak time history data of Gaussian clustering and temperature data.

[0029] Figure 6 It is the power spectrum peak data after removing the temperature effect by using the regression value output by the regression model.

[0030] Figure 7 It is the probabilistic statistical extreme value analysis using the normalized peak time history data.

[0031] Figure 8 It is the failure alarm using the normalized peak time history data and the threshold obtained from the extreme value analysis. Detailed Embodiment

[0032] The present invention will be further described and understood through non-limiting embodiments below.

[0033] As Figure 1 shown is the flowchart of the method, which specifically includes the following steps:

[0034] Step 1: Use the acceleration sensors arranged by the bridge monitoring system to collect the acceleration responses at the mid-span of the main span, 1 / 4 span of the main span, 1 / 8 span of the main span, and mid-span of the side span of the cable-stayed bridge main girder.

[0035] Step 2: Segment the data of each main girder acceleration sensor collected with a 1-minute time interval, and then use the PSD algorithm to calculate the power spectrum for each segment of data to obtain the 1-minute interval power spectrum data of all sensors.

[0036] Step 3: Combine the acceleration power spectra of all cross-sections, and then perform Gaussian clustering analysis to obtain a Gaussian clustering model of the vibration power spectrum of the main girder with a time interval of 1 min;

[0037] Step 4: Take the maximum power spectrum value of the Gaussian mixture model per minute to form a time history curve of the maximum power spectrum value;

[0038] Step 5: Segment the bridge ambient temperature with a time length of 1 min as well, take the average value of each segment of data to obtain the time history data of the ambient temperature at 1-min intervals, perform linear regression analysis on the time history data of the ambient temperature and the time history data of the maximum power spectrum obtained in Step 4 to obtain a linear regression model expressing the correlation between the two; then subtract the regression value obtained from the linear regression model from the time history data of the maximum power spectrum value obtained in Step 4 to obtain a normalized time history curve of the maximum power spectrum value, and eliminate the effect of the ambient temperature on the maximum power spectrum value;

[0039] Step 6: Perform extreme value analysis on the time history curve of the temperature-normalized maximum power spectrum value to obtain the maximum value of the maximum power spectrum value with a 99% confidence level, and use this maximum value as the warning threshold;

[0040] Step 7: When a pulse mutation occurs in the monitored time history curve of the maximum power spectrum value and exceeds the maximum value, an alarm for the failure of the stay cable damper is given.

[0041] Example 1:

[0042] Based on the temperature monitoring data and the main girder deflection monitoring data of the Huanggang Road-Rail Cable-Stayed Bridge, the specific implementation process of the present invention is described.

[0043] (1) Establish a bridge structural health monitoring system installed on the Huanggang Road-Rail Cable-Stayed Bridge, and collect the accelerations of the main girder at the mid-span of the main span, 1 / 4 span of the main span, 1 / 8 span of the main span, and the mid-span of the side span. These acceleration data are the basis for data-driven;

[0044] (2) With a time interval of 1 min, segment the collected acceleration time history data, and perform a power spectrum analysis calculation on each segment of data. The schematic diagram of the power spectra of the four sensors calculated within a certain 1 min obtained by this method is as Figure 2 shown;

[0045] (3) Combine the power spectrum analysis data of the four cross-section acceleration sensors, and then perform Gaussian clustering analysis on the combined data. The clustering model is as Figure 3 described;

[0046] (4) Take Figure 3 the extreme value of the clustering model to obtain the power spectrum peak value of the combined information in each 1-min time interval, thereby obtaining a peak time history curve of the Gaussian clustering analysis of the main girder acceleration, asFigure 4 The peak time history curve within a period of time is shown as follows.

[0047] (5) The bridge environmental temperature obtained by the monitoring system is also processed in 1-minute intervals, and the average value per minute is taken as the representative value of the environmental temperature load to obtain the time series data of the environmental temperature at 1-minute intervals. Using Figure 4 the maximum power spectrum value data shown as follows and the bridge environmental temperature, a linear regression model is established, and then the influence of temperature is eliminated to establish the maximum power spectrum value time history curve with temperature normalization as shown in Figure 5 the following;

[0048] (6) As shown in Figure 6 , extreme value analysis is carried out on the maximum power spectrum value time history curve with temperature normalization to obtain the maximum value of the maximum power spectrum value with a 99% guarantee rate, and this maximum value is used as the warning threshold;

[0049] (7) When a pulse mutation occurs in the monitored maximum power spectrum value time history curve, it represents the sudden failure of the stay cable damper, which will cause the abnormal transmission of the stay cable to the main girder. Therefore, the Gaussian clustering extreme value of the vibration power spectrum of the main girder will exceed Figure 7 the obtained extreme value, and then the alarm of the failure of the stay cable damper can be realized; as shown in Figure 8 , it is an example of the alarm of the failure of a damper.

[0050] The above embodiments are only further specific descriptions and examples of the solution of the present invention. After reading the embodiments of the present invention, various equivalent forms of modification and replacement by those of ordinary skill in the art all fall within the scope of protection defined by the claims of the present invention application.

Claims

1. A failure monitoring and alarming method for a cable damper of a long-span cable-stayed bridge, characterized in that Specifically, it includes the following steps: Step 1: Set multiple measuring points on the main girder of the cable-stayed bridge. The measuring points have the function of expressing the potential energy of the main girder of the cable-stayed bridge, and collect the acceleration response at the cross-section position of each measuring point through an acceleration sensor; Step 2: Calculate the acceleration power spectrum of the cross-section at each measuring point under the same time step; Step 3: Conduct Gaussian clustering analysis on the acceleration power spectra of all the cross-sections of the measuring points to obtain a Gaussian mixture model of the acceleration power spectra with the time step as the time interval; Step 4: Obtain the maximum power spectrum value of the Gaussian mixture model to form a time history curve of the maximum power spectrum value; Step 5: Establish a linear regression model between the maximum power spectrum value and the bridge ambient temperature, and then eliminate the influence of temperature to establish a time history curve of the temperature-normalized maximum power spectrum value; Step 6: Conduct extreme value analysis on the time history curve of the temperature-normalized maximum power spectrum value to obtain the maximum value of the maximum power spectrum value with a 99% confidence level; Step 7: When a pulse mutation occurs in the monitored time history curve of the maximum power spectrum value and exceeds the maximum value obtained in Step 6, an alarm for the failure of the stay cable damper is given.

2. The failure monitoring and alarm method for the cable damper of a long-span cable-stayed bridge according to claim 1, characterized in that In Step 1, collect the acceleration responses at the mid-span of the main span, 1 / 4 span of the main span, 1 / 8 span of the main span, and the mid-span of the side span of the main girder of the cable-stayed bridge.

3. A failure monitoring and alarming method for a cable damper of a long-span cable-stayed bridge according to claim 1, characterized in that The time step in Step 2 is taken as 1 min.

4. A failure monitoring and alarm method for a stay cable damper of a long-span cable-stayed bridge according to claim 1, characterized in that, In Step 2, it is necessary to calculate the data acceleration power spectrum, and the calculation method is the PSD algorithm.

5. A failure monitoring and alarm method for a cable damper of a long-span cable-stayed bridge according to claim 1, characterized in that, The method for obtaining the regression model in Step 5 is the least squares regression technique.

6. The failure monitoring and alarm method for the cable damper of a long-span cable-stayed bridge according to claim 1, characterized in that, Set the maximum value in Step 6 as the warning threshold, and adjust the deviation range of the warning threshold according to the characteristics of the bridge.

Images