CN-122016200-A - Automatic extraction method of fundamental frequency of steel structure high-speed railway bridge under operation condition

CN122016200ACN 122016200 ACN122016200 ACN 122016200ACN-122016200-A

Abstract

The invention provides an automatic extraction method of the fundamental frequency of a steel structure high-speed railway bridge under an operation condition, which comprises the steps of sensor layout, response signal acquisition, working condition identification, quasi-constant load data segment segmentation, multi-measuring point coherent weighted superposition, rough frequency locking, fundamental frequency fine identification, statistical fusion output and the like. The method effectively solves the signal extraction problems of small vibration and large field noise of the steel bridge by a coherent weighted superposition technology in the signal extraction process, has stronger anti-interference capability, combines quasi-constant-load segment screening and high-resolution spectrum estimation, utilizes the strong excitation energy of live load, avoids the non-stationary effect and frequency error caused by short data, and has higher recognition precision.

Inventors

LEI LEI
WU WENHUA
YUAN MENGYANG
GUAN YANWEN
LI JUNCHENG
ZHANG YUESHUANG
LIU LIANGLIANG
JIAO FENG
ZHANG QINGNING
CUI ZHENYU
LIU MO
LUO YONG
GUO YANHAI
FENG YACHENG
ZHOU WEIYU

Assignees

中铁第一勘察设计院集团有限公司
鲁南高速铁路有限公司

Dates

Publication Date: 20260512
Application Date: 20251215

Claims (8)

1. The automatic extraction method of the fundamental frequency of the steel structure high-speed railway bridge under the operation condition is characterized by comprising the following steps of: 【1】 Sensor arrangement and response signal acquisition Arranging acceleration sensors at key positions of the bridge to be detected, continuously collecting acceleration responses in an operation state, and processing to obtain original displacement signals of the sensors; 【2】 Working condition identification and quasi-constant load data segment segmentation The energy envelope of the sensor signal is utilized to identify the bridge crossing state of the train, the moment that the train is fully loaded on the bridge and the speed is stably operated is obtained, and a quasi-constant load data segment with better stability is intercepted from an original displacement signal to be used as an effective signal; 【3】 Multi-measuring point coherent weighted superposition The method comprises the steps of carrying out weighted superposition on effective signals of a plurality of measuring points by taking coherence as a weight to obtain weighted superposition signals; 【4】 Coarse frequency locking Performing FFT (fast Fourier transform) on the weighted superposition signals to determine a fundamental frequency selection interval; 【5】 Fine identification of fundamental frequency Constructing a data covariance matrix in a fundamental frequency candidate interval, carrying out feature decomposition, constructing a pseudo spectrum function by using a MUSIC algorithm, and searching frequencies corresponding to pseudo spectrum peaks to obtain fundamental frequency values; 【6】 And (3) statistical fusion output: And carrying out statistical analysis on the multiple train number identification results, removing outliers which are interfered by accidental factors, and outputting a fundamental frequency result with statistical stability.
2. The method for automatically extracting the fundamental frequency of the steel-structured high-speed railway bridge under the operation condition of claim 1, wherein N acceleration sensors are distributed at the main girder mid-span position, the positions of 3L/4 and L/4 of the steel-structured high-speed railway bridge in the step [1 ], wherein L is the length of the bridge.
3. The method for automatically extracting the fundamental frequency of the steel-structured high-speed railway bridge according to claim 1, wherein the sampling frequency in the step [1] is 5-10 times of the estimated fundamental frequency of the bridge.
4. The automatic extraction method of the fundamental frequency of the steel-structured high-speed railway bridge under the operation condition of claim 1, wherein the identification step of the quasi-constant-load segment in the step [ 2] is as follows: [ 2.1 ] sliding window energy calculation Identifying train crossing states using an energy envelope of a sensor signal, defining energy within a sliding time window ; Time of day Reflecting the vibration intensity of the signal around the moment; Discrete time indexing; The number of sampling points within the window is calculated, The index of the sampling point in the window, Reference measuring point at time Is a function of the original displacement signal value of the sensor; The reference point number is used for the reference point, ; [ 2.2 ] Train bridge-entering judgment When (when) When the train is judged to pass through the bridge, For the train to enter the bridge energy threshold, the threshold is usually set according to the background noise energy level in the no-train state, and can be 3-5 times of the background noise energy, namely Wherein Average energy in no-vehicle state; [ 2.3 ] Signal variance calculation Further calculating intra-window signal variance For evaluating the stationarity of the signal: ; Time of day Reflecting the fluctuation degree of the signal near the moment; Sampling points in a window; Discrete time indexing; Sampling point index in window; The mean value of the original signal within the window, ; [ 2.4 ] Variance fluctuation ratio calculation The fluctuation rate of the variance is calculated to evaluate the stability of the signal: ; Time of day Variance fluctuation ratio of (a); In a time window Internal variance Is set in the standard deviation of (2), To evaluate window length, it is typically taken 10-20 seconds; In a time window Internal variance Average value of (2); [ 2.5 ] quasi-constant load fragment determination If the following conditions are satisfied, the time period is determined to be a quasi-constant load time period: 1) Energy conditions: Illustrating the train on the bridge; 2) Stability conditions: Wherein For a preset stability threshold, generally take ; 3) Duration condition quasi-constant load state duration Wherein For minimum duration requirements, generally take Second, to ensure that there is sufficient data length for frequency analysis; and intercepting the original signal data corresponding to the quasi-constant load time period meeting the condition as an effective signal.
5. The automatic extraction method of the fundamental frequency of the steel-structured high-speed railway bridge under the operation condition of claim 1, wherein the step [ 3 ] of multi-measuring-point coherent weighted superposition is as follows: [ 3.1 ] reference point selection Selecting the measuring point with the largest response energy as a reference measuring point , Reference measuring point at time Is a valid signal value of (2); The selection criteria of the reference points are: ; the duration of the quasi-constant load segment; a parameter that maximizes the objective function; The number of the measuring point is that, ; [ 3.2 ] Power spectral Density calculation Calculate each measuring point With reference points Cross-power spectral density and self-power spectral density; First of all Self-power spectral density of each measuring point; reference point A self-power spectral density of (2); reference point And the first Cross power spectral density between the individual stations; The power spectral density can be calculated by the Welch method or the periodogram method: ; : Is used for the fourier transform of (a), ; : Fourier transform of (a); : complex conjugate of (a); Expecting operators; the unit of the imaginary number is shown in the specification, ; [ 3.3 ] Coherence function calculation Calculating a coherence function : ; Frequency of The value of the coherence function is in the range of ; Modulus of cross power spectral density; First of all Self-power spectral density of each measuring point; the self-power spectrum density of the measuring point is referenced; [ 3.4 ] average coherence coefficient calculation Calculating average coherence coefficient within target frequency band : ; Average coherence coefficient in target frequency band with value range of ; The lower limit frequency of the target frequency band is expressed in Hz, which is usually taken Wherein To estimate the fundamental frequency; The upper limit frequency of the target frequency band is expressed in Hz and is usually taken ; Discretization implementation (frequency resolution is ): ; Frequency points within the target frequency band, ; The frequency index is used to determine the frequency index, ; The frequency resolution is used to determine the frequency resolution, Wherein Counting FFT points; [ 3.5 ] site screening If it is Then the measuring point is included in the superposition set , Coherence coefficient threshold, typically taken The method is used for screening out measuring points highly related to the reference measuring points; the method comprises the steps of participating in a superimposed measuring point set, And (2) and ; Weight calculation [ 3.6 ] ; First of all Normalized weight of each measuring point with the value range of And meet the following ; Superimposed collection A measuring point index in (2); The sum of the squares of the average coherence coefficients of all the participating superposition measuring points is used for normalization; [ 3.7 ] weighted superimposed signal construction ; Weighting the superimposed equivalent single-channel response signals; Summing all the measurement points participating in superposition.
6. The automatic extraction method of the fundamental frequency of the steel-structured high-speed railway bridge under the operation condition of claim 1, wherein the step [ 4] of rough frequency locking is as follows: For superimposed signals Performing Fast Fourier Transform (FFT) to obtain power spectral density: ; Frequency of A power spectral density value at; : is a Discrete Fourier Transform (DFT), ; Signal sampling points; : searching spectral peaks near the estimated fundamental frequency according to design data and the peak position of the power spectrum, and determining the candidate interval of the fundamental frequency , Is the lower frequency limit of the candidate frequency band, The candidate interval should contain the main peak value in the power spectrum, usually taking the peak value frequency Range.
7. The automatic extraction method of the fundamental frequency of the steel structure high-speed railway bridge under the operation condition of claim 1, wherein the step [5 ] of fine identification of the fundamental frequency comprises the following steps: [ 5.1 ] data covariance matrix construction Signal is sent to Segmentation processing, namely constructing a data matrix: ; : is a data matrix of the (c) data, For the number of lines, usually taken ; For the number of columns, usually take Or (b) ; Calculating a data covariance matrix: ; : Is a covariance matrix of (a); : is a conjugate transpose of (a); [ 5.2 ] pair covariance matrix And (3) performing characteristic decomposition: ; : is characterized by the feature vector matrix and column vector of (a) Is a feature vector of (1); : Diagonal matrix of (2), diagonal elements are Is arranged in descending order; : is a conjugate transpose of (a); [ 5.3 ] noise subspace extraction According to the magnitude of the characteristic value, the characteristic vector is divided into a signal subspace and a noise subspace, wherein the signal subspace is Feature vectors corresponding to larger feature values, the dimension is Noise subspace Feature vectors corresponding to smaller feature values, the dimension is Wherein For the number of signal sources, usually take ; [ 5.4 ] MUSIC pseudo-spectral function construction Constructing a pseudo spectrum function by using a MUSIC algorithm: ; Frequency of A MUSIC pseudo-spectrum value at the position; : is defined as the steering vector: ; The sampling interval is set to be equal to the sampling interval, ; Complex exponential terms representing frequency At the moment of time Is a phase of (2); Transpose operators; : Is the conjugate transpose of (1) Is a row vector of (2); : Is of the dimension of the conjugate transposed matrix of (a) ; : Is a noise subspace projection matrix of (2); [ 5.5 ] fundamental frequency fine recognition Candidate frequency bands In, search at high frequency resolution, typically 0.01 Hz The peak value of the frequency is the fundamental frequency of fine recognition : ; The obtained fundamental frequency estimated value is identified, and the unit is Hz. Parameters that maximize the objective function.
8. The automatic extraction method of the fundamental frequency of the steel structure high-speed railway bridge under the operation condition of claim 1, wherein the step [ 6 ] of statistical fusion output of the fundamental frequency is as follows: [ 6.1 ] Multi-train data collection Collecting The recognition result of the secondary train bridge crossing forms a fundamental frequency estimated value set: ; the number of train crossing times in the collected effective data; First of all The secondary train bridge crossing recognition obtains the basic frequency estimated value, The unit is Hz; [ 6.2 ] outlier rejection Removing outliers by using a statistical method, and marking the number of the remaining effective estimated values as Wherein ; [ 6.3 ] Statistical mean calculation Calculating the mean of the remaining valid estimates as the final fundamental frequency: ; Statistical average value of fundamental frequency; The number of effective estimated values remaining after eliminating abnormal values; Summing all valid estimates.

Description

Automatic extraction method of fundamental frequency of steel structure high-speed railway bridge under operation condition Technical Field The invention belongs to the technical field of bridge structure health monitoring, and particularly relates to an automatic extraction method of a fundamental frequency of a steel structure high-speed railway bridge under an operation condition. Background The steel-structured high-speed railway bridge has the characteristics of large span, high rigidity and great self-weight. Monitoring changes in the fundamental frequency (first order natural frequency) thereof during long-term operation is an important means of assessing the overall stiffness and structural health of the bridge. However, in a practical operating environment, accurate extraction of the fundamental frequency faces a great challenge: 1. the rigidity is high, the vibration is small, the rigidity of the steel bridge is extremely high, the structural vibration response is weak under the effect of random excitation (such as wind and ground pulsation) of the environment, the signal to noise ratio is extremely low, and the effective mode is difficult to identify by the conventional method. 2. The excitation source is complex, and strong excitation (live load) generated by passing a bridge of a train is usually required to excite structural vibration. However, huge additional mass (axle coupling effect) and non-stationary impact are introduced when a train passes through the bridge, so that the identified frequency is often 'axle coupling system frequency' instead of 'structural fundamental frequency' of the bridge, the data is non-stationary, and the traditional FFT analysis is easy to generate spectrum ambiguity. 3. The frequency resolution is insufficient, and the effective data length is limited due to short train crossing time, so that the frequency resolution of the traditional spectrum analysis method is low, and dense modal frequencies are difficult to distinguish. Disclosure of Invention The invention aims to provide a method for extracting the inherent fundamental frequency of a bridge structure with high precision by utilizing the live load of a train as effective excitation and eliminating non-stationary interference under the operating condition. The technical scheme of the invention is as follows: 1. The automatic extraction method of the fundamental frequency of the steel structure high-speed railway bridge under the operation condition is characterized by comprising the following steps of: 【1】 Sensor arrangement and response signal acquisition Arranging acceleration sensors at key positions of the bridge to be detected, continuously collecting acceleration responses in an operation state, and processing to obtain original displacement signals of the sensors; 【2】 Working condition identification and quasi-constant load data segment segmentation The energy envelope of the sensor signal is utilized to identify the bridge crossing state of the train, the moment that the train is fully loaded on the bridge and the speed is stably operated is obtained, and a quasi-constant load data segment with better stability is intercepted from an original displacement signal to be used as an effective signal; 【3】 Multi-measuring point coherent weighted superposition The method comprises the steps of carrying out weighted superposition on effective signals of a plurality of measuring points by taking coherence as a weight to obtain weighted superposition signals; 【4】 Coarse frequency locking Performing FFT (fast Fourier transform) on the weighted superposition signals to determine a fundamental frequency selection interval; 【5】 Fine identification of fundamental frequency Constructing a data covariance matrix in a fundamental frequency candidate interval, carrying out feature decomposition, constructing a pseudo spectrum function by using a MUSIC algorithm, and searching frequencies corresponding to pseudo spectrum peaks to obtain fundamental frequency values; 【6】 And (3) statistical fusion output: And carrying out statistical analysis on the multiple train number identification results, removing outliers which are interfered by accidental factors, and outputting a fundamental frequency result with statistical stability. The invention provides an automatic extraction method of the fundamental frequency of a steel structure high-speed railway bridge under the operation condition, which is used for carrying out a manual excitation test under the condition that the transportation of a train is not required to be interrupted, and the extracted data completely depend on the data of the operation train to realize automatic monitoring, so that the application is wide. The method effectively solves the signal extraction problems of small vibration and large field noise of the steel bridge by a coherent weighted superposition technology in the signal extraction process, has stronger anti-interference capability, combines