Signal denoising method for bridge dynamic load test

Abstract

The invention discloses a signal denoising method for a bridge dynamic load test, and belongs to the field of bridge data denoising. The invention discloses a signal denoising method for a bridge dynamic load test. The signal denoising method comprises the following steps: step 1, acquiring a bridge signal; 2, optimizing parameters of variational mode decomposition by using a Hemma optimization algorithm, and carrying out first denoising on the bridge signal; 3, singular spectrum analysis is introduced for second denoising; and 4, eliminating low-frequency oscillation in the bridge signal, and generating a denoised signal. According to the method, the problems that in the prior art, denoising is incomplete, so that an analysis result is distorted, feature information is lost, the denoising and feature extraction effects are affected, and even the state of a bridge structure is misjudged are solved, and parameter setting of variational mode decomposition is successfully achieved; the reasonable parameter setting is ensured by the excellent optimization capacity of the Hemma optimization algorithm, the low-frequency oscillation of variational mode decomposition is eliminated by singular spectrum analysis, the method can be directly used, and the precision is higher.

Description

Technical Field

[0001] The present invention relates to the technical field of bridge data noise reduction, and specifically to a signal denoising method for bridge dynamic load tests. Background Technique

[0002] As an important transportation infrastructure, bridges need to be regularly inspected. The monitoring data of bridges is affected by the environment and the errors of the instruments themselves, and will inevitably contain noise information, which affects the accuracy of the monitoring results. Therefore, it is very necessary to reduce the noise of bridge monitoring data.

[0003] Chinese Patent with the publication number CN117473232A discloses a method for noise reduction and reconstruction of bridge monitoring data, which acquires bridge monitoring data and performs modal decomposition on the bridge monitoring data; uses adaptive mutual information entropy for high-frequency signal extraction; uses wavelet synchronous compression transform to perform noise reduction processing on the obtained high-frequency sub-signals; and reconstructs the denoised bridge monitoring signal according to the superposition of the denoised high-frequency sub-signals and low-frequency sub-signals. This patent is widely used in the field of signal denoising and can identify high-frequency changes in short time windows in the case of sudden changes in environmental noise.

[0004] In the actual use process of the above patent, modal aliasing is likely to occur, resulting in incomplete denoising. The introduction of noise also amplifies the noise in the original signal, which may lead to the distortion of the analysis results and the loss of characteristic information, affecting the denoising and feature extraction effects, and easily causing local low-frequency oscillations in the signal, and even misjudging the state of the bridge structure. When the noise level is high, the denoising effect is poor. Therefore, it does not meet the existing requirements. For this reason, we propose a signal denoising method for bridge dynamic load tests. Summary of the Invention

[0005] The purpose of the present invention is to provide a signal denoising method for bridge dynamic load tests, improve the efficiency and accuracy of variational mode decomposition for bridge signal decomposition, and further refine the denoised signal. By recombining the main components identified by singular spectrum analysis, the denoised signal is generated. The signal completed in the entire denoising process is smoother, can remove most of the noise, and can more accurately reflect the true dynamic response of the bridge under load, ensuring that the data obtained from the bridge dynamic load test is both accurate and reliable, can obtain a signal closer to the real situation, thereby calculating the bridge vibration frequency and impact coefficient, which is beneficial to the structural identification and evaluation of the bridge, and solves the problems raised in the above background technique.

[0006] To achieve the above purpose, the present invention provides the following technical solution: A signal denoising method for bridge dynamic load tests, including the following steps:

[0007] Step 1: Collect bridge signals using a bridge dynamic load test detection system;

[0008] Step 2: Use the hippopotamus optimization algorithm to optimize the parameters of variational mode decomposition. The parameters are the decomposition scale K and the penalty factor alpha. The variational mode decomposition performs the first denoising on the bridge signal;

[0009] Step 3: After the variational mode decomposition performs the first denoising, introduce singular spectrum analysis for the second denoising;

[0010] Step 4: Use the second denoising to eliminate the low-frequency oscillation in the bridge signal and generate the denoised signal.

[0011] Preferably, the use of the hippopotamus optimization algorithm to optimize the parameters of variational mode decomposition specifically includes:

[0012] Set the initial parameters of the hippopotamus optimization algorithm. The initial parameters include the population size, the number of iterations, and the search space boundary;

[0013] Input the noise signal collected by the bridge dynamic load test detection system into the hippopotamus optimization algorithm;

[0014] Output the optimal solution found by the hippopotamus optimization algorithm, and save the currently found dominant hippopotamus. The dominant hippopotamus corresponding to the optimal position and the optimal solution is the parameter optimization result of the variational mode decomposition;

[0015] Preferably, the output of the optimal solution found by the hippopotamus optimization algorithm and saving the currently found dominant hippopotamus specifically includes:

[0016] Perform population initialization;

[0017] Set the maximum number of iterations T and the number of hippopotamus populations N, set the initial hippopotamus population, set i = 1, and t = 1;

[0018] Calculate the objective function and update the dominant hippopotamus according to the fitness function;

[0019] The first stage: Expand the search space, specifically including:

[0020] Hippopotamuses refresh in ponds or rivers;

[0021] Take half of the population to calculate the position of the dominant hippopotamus and the position of the immature hippopotamus and update the position of each hippopotamus X i . Each hippopotamus represents a solution;

[0022] The second stage: Fine-tune the current solution, specifically including:

[0023] Starting from 1 + N / 2, randomly generate the position of the predator and calculate the position of the hippopotamus to resist the predator and update the position of each hippopotamus X i ;

[0024] Set \(i = i + 1\), calculate the objective function and update the dominant hippopotamus;

[0025] If it is not less than, set \(i = 1\), calculate the new boundary and the hippopotamus escaping from the predator Enter the new pond location and update the position \(X\) of each hippopotamus i ;

[0026] The third stage: Find the local optimal solution, specifically including:

[0027] Calculate the new boundary and traverse from the first hippopotamus to the \(N\)th hippopotamus;

[0028] The hippopotamus escapes from the predator and enters a new pond or river;

[0029] Calculate the hippopotamus escaping from the predator Enter the new pond location and update the position \(X\) of each hippopotamus i ;

[0030] Save the currently found dominant hippopotamus, the corresponding optimal position and the optimal solution;

[0031] Output the optimal solution found by the hippopotamus optimization algorithm and end the hippopotamus optimization algorithm.

[0032] Preferably, the variational mode decomposition performs the first denoising on the bridge signal, specifically including:

[0033] Receive the optimization result of the hippopotamus optimization algorithm to obtain the decomposition scale \(K\) and the penalty factor \(\alpha\) of the variational mode decomposition;

[0034] Initialize the parameters of the variational mode decomposition. The parameters for initializing the variational mode decomposition include the decomposition scale \(K\), the penalty factor \(\alpha\), the step size \(\tau\), and the maximum number of iterations \(T\);

[0035] Assign an initial center frequency \(\omega\) and an initial mode signal \(u\) to each mode. For each mode component, use the alternating direction multiplier method to update the mode signal \(u\) and the center frequency \(\omega\);

[0036] Decompose the signal using the variational mode decomposition, calculate the correlation coefficient \(CC\) of the decomposed IMFs, and set the correlation coefficient \(CC\) threshold of the IMFs to 1 / 10 of the maximum correlation coefficient \(CC\) max ;

[0037] Judge whether the correlation coefficient \(CC\) of the IMFs reaches the threshold. If it does not reach the set threshold, remove the IMFs;

[0038] If the correlation coefficient \(CC\) of the IMFs reaches the set threshold, accumulate the IMFs to obtain the reconstructed signal.

[0039] Preferably, the variational mode decomposition for the first denoising of the bridge signal further includes:

[0040] Decompose the signal into multiple bandwidth-limited IMFs, where IMFs represent different frequency components of the signal, and each IMF is associated with a specific frequency band;

[0041] After the variational mode decomposition decomposes the bridge signal, calculate the correlation coefficient between the IMFs and the original signal;

[0042] Judge whether the bridge signal contains noise according to the calculation result. If the correlation coefficient is high, it means that the IMF is the main component of the signal and there is no noise;

[0043] If the correlation coefficient is low, it means that the IMF contains noise or irrelevant information, and remove the IMFs with low correlation coefficients.

[0044] Preferably, the introduction of singular spectrum analysis for the second denoising specifically includes:

[0045] Introduce singular spectrum analysis to calculate the trajectory matrix and decompose the singular values;

[0046] Define the signal y to be decomposed and set the window length L of the singular spectrum analysis;

[0047] Generate the initial trajectory matrix X, where the number of rows of X is the window length L and the number of columns is K = N - L + 1, and calculate each column of the trajectory matrix X to obtain the eigenvalues λ and eigenvectors u;

[0048] Calculate the covariance matrix R, perform eigenvalue decomposition on the covariance matrix R to obtain the eigenvalues λ and eigenvectors u;

[0049] Calculate the singular spectrum according to the eigenvalues λ, that is, the singular spectrum matrix Y, and calculate the singular spectrum Y_i at each time point;

[0050] According to the singular spectrum matrix Y and the eigenvectors u, reconstruct the components of the signal;

[0051] Calculate the reconstructed signal y_i of the i-th mode and output the reconstructed signals y_1, y_2,..., y_r;

[0052] Generate the denoised signal by recombining the main components identified by the singular spectrum analysis.

[0053] Preferably, the decomposition of the singular values to identify and separate the main patterns in the signal specifically includes:

[0054] Decompose the singular values, group them according to the descending order of the singular values, and set the threshold of the cumulative contribution rate;

[0055] Calculate the cumulative contribution rate using the singular value of the \(i\)-th component and the singular values of all components, and determine whether the contribution rate is greater than 0.99. If it is greater than 0.99, recombine the signals of the first \(i\) components to form the final signal for the analysis of the bridge structure;

[0056] If it is not greater than 0.99, continue to recalculate the cumulative contribution rate using the singular value of the \((i + 1)\)-th component and the singular values of all components until it is greater than 0.99 and then end.

[0057] Compared with the prior art, the beneficial effects of the present invention are:

[0058] The present invention uses the hippopotamus optimization algorithm to optimize the variational mode decomposition to improve the efficiency and accuracy of the variational mode decomposition for decomposing bridge signals. The singular spectrum analysis reveals the potential dynamic characteristics by decomposing the time series into different components, thereby further refining the denoised signal. By recombining the main components identified by the singular spectrum analysis, the denoised signal is generated. The signal completed in the entire denoising process is smoother, can remove most of the noise, and can more accurately reflect the true dynamic response of the bridge under the action of the load, which is crucial for the health monitoring, damage assessment, and maintenance strategy formulation of the bridge. It can ensure that the data obtained from the bridge dynamic load test is both accurate and reliable, can obtain a signal closer to the real situation, thereby calculating the bridge vibration frequency and impact coefficient, which is beneficial to the structural identification and evaluation of the bridge. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] Figure 1 It is the algorithm flow chart of the hippopotamus optimization algorithm for the signal denoising method of a bridge dynamic load test of the present invention;

[0060] Figure 2 It is the denoising flow chart of the variational mode decomposition optimized by the hippopotamus optimization algorithm for the signal denoising method of a bridge dynamic load test of the present invention;

[0061] Figure 3 It is the total denoising flow chart of the signal denoising method of a bridge dynamic load test of the present invention;

[0062] Figure 4 It is the fundamental frequency denoising diagram of the present invention;

[0063] Figure 5 It is the driving signal denoising diagram of the present invention;

[0064] Figure 6 It is the braking signal denoising diagram of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0065] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0066] To solve the problems that in the existing technology, modal aliasing is likely to occur, resulting in incomplete denoising, the introduction of noise also amplifies the noise in the original signal, which may lead to the distortion of the analysis result and the loss of feature information, affecting the denoising and feature extraction effects, easily causing local low-frequency oscillations in the signal, and even misjudging the state of the bridge structure, and the denoising effect is poor when the noise level is high, please refer to Figures 1 - 3 , the following technical solutions are provided in this embodiment:

[0067] A signal denoising method for bridge dynamic load tests, comprising the following steps:

[0068] Step 1: Collect bridge signals using a bridge dynamic load test detection system;

[0069] Step 2: Optimize the parameters of variational mode decomposition using the hippopotamus optimization algorithm. The parameters are the decomposition scale K and the penalty factor alpha, and variational mode decomposition performs the first denoising on the bridge signals;

[0070] Step 3: After variational mode decomposition performs the first denoising, introduce singular spectrum analysis for the second denoising;

[0071] Step 4: Use the second denoising to eliminate the low-frequency oscillations in the bridge signals and generate the denoised signals.

[0072] First, use the hippopotamus optimization algorithm to optimize the parameters of variational mode decomposition. After variational mode decomposition performs the first denoising and completes the selection of decomposition modes, introduce singular spectrum analysis for the second denoising to eliminate low-frequency oscillations and form the final signals for bridge structure analysis.

[0073] Collect bridge signals, specifically including:

[0074] Optimize the parameters of variational mode decomposition using the hippopotamus optimization algorithm, specifically including:

[0075] Set the initial parameters of the hippopotamus optimization algorithm. The initial parameters include the population size, the number of iterations, and the search space boundary;

[0076] Input the noise signals collected by the bridge dynamic load test detection system into the hippopotamus optimization algorithm;

[0077] Output the optimal solution found by the hippopotamus optimization algorithm, and save the currently found dominant hippopotamus. The optimal position and optimal solution corresponding to the dominant hippopotamus are the parameter optimization results of variational mode decomposition;

[0078] Output the optimal solution found by the hippopotamus optimization algorithm, and save the currently found dominant hippopotamus, specifically including:

[0079] Perform population initialization;

[0080] Set the maximum number of iterations T and the number of hippopotamus populations N, set the initial hippopotamus population, set i = 1, and t = 1;

[0081] Calculate the objective function and update the dominant hippopotamus according to the fitness function;

[0082] The first stage: Expand the search space, specifically including:

[0083] Hippopotamuses are refreshed in ponds or rivers;

[0084] Take half of the population to calculate the position of the dominant hippopotamus and the position of immature hippopotamuses and update the position of each hippopotamus X i The position of each hippopotamus represents a solution;

[0085] The second stage: Fine-tune the current solution, specifically including:

[0086] Starting from 1 + N / 2, randomly generate the position of the predator and calculate the position of the hippopotamus defending against the predator and update the position of each hippopotamus X i ;

[0087] Set i = i + 1, calculate the objective function and update the dominant hippopotamus;

[0088] If it is not less than, set i = 1, calculate the new boundary and the position of the hippopotamus escaping from the predator Enter the new pond position and update the position of each hippopotamus X i ;

[0089] The third stage: Find the local optimal solution, specifically including:

[0090] Calculate the new boundary and traverse from the first hippopotamus to the Nth hippopotamus;

[0091] The hippopotamus escapes from the predator and enters a new pond or river;

[0092] Calculate the position of the hippopotamus escaping from the predator Enter the new pond position and update the position of each hippopotamus X i ;

[0093] Save the currently found dominant hippopotamus, corresponding to the optimal position and the optimal solution;

[0094] Output the optimal solution found by the hippopotamus optimization algorithm and end the hippopotamus optimization algorithm.

[0095] Variational mode decomposition performs the first denoising on the bridge signal, specifically including:

[0096] Receive the optimization result of the hippopotamus optimization algorithm to obtain the decomposition scale K and the penalty factor alpha of the variational mode decomposition;

[0097] Initialize the parameters of the variational mode decomposition. The parameters for initializing the variational mode decomposition include the decomposition scale K, the penalty factor alpha, the step size tau, and the maximum number of iterations T;

[0098] Assign an initial center frequency omega and an initial modal signal u to each mode. For each modal component, use the alternating direction multiplier method to update the modal signal u and the center frequency omega;

[0099] Use variational mode decomposition to decompose the signal, calculate the correlation coefficient CC of the decomposed IMFs. The threshold of the correlation coefficient CC of the IMFs is set to 1 / 10 of the maximum correlation coefficient CC max ;

[0100] Judge whether the correlation coefficient CC of the IMFs reaches the threshold. If it does not reach the set threshold, remove the IMFs;

[0101] If the correlation coefficient CC of the IMFs reaches the set threshold, accumulate the IMFs to obtain the reconstructed signal;

[0102] Optimize the variational mode decomposition using the hippopotamus optimization algorithm. In the denoising algorithm, the hippopotamus optimization algorithm is a heuristic search algorithm that simulates the behavior of hippopotamuses. By simulating the natural behaviors of hippopotamuses: exploration, defense, and evading predators, it searches for the best parameter combination to improve the efficiency and accuracy of the variational mode decomposition for decomposing bridge signals.

[0103] The variational mode decomposition for the first denoising of the bridge signal also includes:

[0104] Decompose the signal into multiple bandwidth-limited IMFs. The IMFs represent different frequency components of the signal, and each IMF is associated with a specific frequency band. In this way, the variational mode decomposition can reveal the internal frequency structure of the signal and provide a basis for subsequent denoising steps;

[0105] After the variational mode decomposition decomposes the bridge signal, calculate the correlation coefficient between the IMFs and the original signal;

[0106] Judge whether the bridge signal contains noise according to the calculation results. If the correlation coefficient is high, it means that the IMFs are the main components of the signal and there is no noise.

[0107] If the correlation coefficient is low, it means that the IMFs contain noise or irrelevant information, and the IMFs with low correlation coefficients are removed.

[0108] After the variational mode decomposition decomposes the bridge signal, perform correlation analysis on each IMF to evaluate the relationship between each IMF and the original signal. By calculating the correlation coefficient between the IMFs and the original signal, the IMFs with high correlation coefficients are considered to be the main components of the signal, while those with low correlation coefficients may contain noise or irrelevant information and will therefore be removed.

[0109] Introduce singular spectrum analysis for the second denoising, specifically including:

[0110] Introduce singular spectrum analysis to calculate the trajectory matrix and perform singular value decomposition;

[0111] Define the signal y to be decomposed and set the window length L of the singular spectrum analysis;

[0112] Generate the initial trajectory matrix X, where the number of rows of X is the window length L and the number of columns is K = N - L + 1, and calculate each column of the trajectory matrix X to obtain the eigenvalues λ and eigenvectors u;

[0113] Calculate the covariance matrix R and perform eigenvalue decomposition on the covariance matrix R to obtain the eigenvalues λ and eigenvectors u;

[0114] Calculate the singular spectrum according to the eigenvalues λ, that is, the singular spectrum matrix Y, and calculate the singular spectrum Y_i at each time point;

[0115] According to the singular spectrum matrix Y and the eigenvectors u, reconstruct the components of the signal;

[0116] Calculate the reconstructed signal y_i of the i-th mode and output the reconstructed signals y_1, y_2,..., y_r;

[0117] Finally, generate the denoised signal by recombining the main components identified by the singular spectrum analysis.

[0118] Singular spectrum analysis is applied to the remaining IMFs. Singular spectrum analysis is a time series analysis method that reveals the underlying dynamic characteristics of a time series by decomposing it into different components. During the denoising process, singular spectrum analysis first calculates the trajectory matrix and then performs singular value decomposition to identify and separate the main patterns in the signal. By setting a threshold for the cumulative contribution rate, it is possible to determine which components contribute the most to the overall characteristics of the signal, thereby further refining the denoised signal. Finally, by recombining the main components identified by singular spectrum analysis, the denoised signal is generated;

[0119] The signal after completing the entire denoising process is smoother, with most of the noise removed, and can more accurately reflect the true dynamic response of the bridge under load. Such a signal is crucial for the health monitoring, damage assessment, and maintenance strategy formulation of the bridge. Through this series of in-depth signal processing steps, it is possible to ensure that the data obtained from the bridge dynamic load test is both accurate and reliable, and a signal closer to the actual situation can be obtained, thereby calculating the bridge vibration frequency and impact factor, which is beneficial for the structural identification and assessment of the bridge.

[0120] Perform singular value decomposition to identify and separate the main patterns in the signal, specifically including:

[0121] Perform singular value decomposition, group according to the descending order of singular values, and set the threshold for the cumulative contribution rate;

[0122] Calculate the cumulative contribution rate using the singular value of the i-th component and the singular values of all components, and judge whether the contribution rate is greater than 0.99. If it is greater than 0.99, recombine the first i components of the signal to form the final signal for the analysis of the bridge structure;

[0123] If it is not greater than 0.99, continue to recalculate the cumulative contribution rate using the singular value of the (i + 1)-th component and the singular values of all components until it is greater than 0.99 and then end.

[0124] Collect bridge signals using the bridge dynamic load test detection system and extract the characteristics of the bridge signals;

[0125] Draw a simulated signal based on the characteristics of the bridge signal. The composition of the simulated signal y function is as follows:

[0126]

[0127] In the formula, t is the time vector, f 1 、f 2 、f 3 、f 4 and f 5 are the signal frequencies, taking values 1, 3, 5, 7, and 9 respectively, and the levels of the noise n are set to 5dB, 10dB, 15dB, and 20dB respectively.

[0128] The effectiveness of denoising the analog signal using the denoising method of the present invention is verified, and the denoising result is as Figures 4 - 6 shown. Using the denoising method of the present invention can remove most of the noise.

[0129] In summary, a signal denoising method for bridge dynamic load tests of the present invention uses the hippopotamus optimization algorithm to optimize variational mode decomposition. In the denoising algorithm, the hippopotamus optimization algorithm is a heuristic search algorithm that simulates the behavior of hippopotamuses. By simulating the natural behavior of hippopotamuses, exploring, defending, and evading predators, it searches for the best parameter combination to improve the efficiency and accuracy of variational mode decomposition for decomposing bridge signals. Subsequently, singular spectrum analysis is applied to the remaining IMFs. Singular spectrum analysis is a time series analysis method that reveals the underlying dynamic characteristics of a time series by decomposing it into different components. During the denoising process, singular spectrum analysis first calculates the trajectory matrix and then performs singular value decomposition to identify and separate the main patterns in the signal. By setting a threshold for the cumulative contribution rate, it is possible to determine which components contribute the most to the overall characteristics of the signal, thereby further refining the denoised signal. Finally, by recombining the main components identified by singular spectrum analysis, the denoised signal is generated. The signal completed by the entire denoising process is smoother, removes most of the noise, and can more accurately reflect the true dynamic response of the bridge under load. Such a signal is crucial for the health monitoring, damage assessment, and formulation of maintenance strategies of the bridge. Through in-depth signal processing steps, it is possible to ensure that the data obtained from bridge dynamic load tests is both accurate and reliable, and a signal closer to the actual situation can be obtained, thereby calculating the bridge vibration frequency and impact factor, which is beneficial to the structural identification and assessment of the bridge.

[0130] It should be noted that in this article, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the term "comprising", "including" or any other variation thereof is intended to cover non-exclusive inclusion, so that a process, method, article or device comprising a series of elements not only includes those elements, but also includes other elements not expressly listed, or also includes elements inherent to such process, method, article or device.

[0131] Although the embodiments of the present invention have been shown and described, it will be understood by those of ordinary skill in the art that various changes, modifications, substitutions, and variations can be made to these embodiments without departing from the principles and spirit of the present invention.

Claims

1. A signal denoising method for bridge dynamic load test, characterized in that: The following steps are involved: Step 1: Collect bridge signals using the bridge dynamic load test detection system; Step 2: Use the Hippo optimization algorithm to optimize the parameters of variational mode decomposition, the parameters are the decomposition scale K and the penalty factor alpha, and the variational mode decomposition is used to perform the first denoising on the bridge signal; Step 3: After the first denoising by variational mode decomposition, singular spectrum analysis is introduced for the second denoising; Step 4: Use the second denoising to eliminate the low-frequency oscillation in the bridge signal and generate a denoised signal.

2. The signal denoising method for bridge dynamic load test according to claim 1 is characterized in that: The method of optimizing the parameters of variational mode decomposition using the Hippo optimization algorithm specifically includes: Set the initial parameters of the Hippo optimization algorithm, including population size, number of iterations, and search space boundaries; The noise signal collected by the bridge dynamic load test detection system is input into the Hippo optimization algorithm; Output the optimal solution found by the Hippopotamus optimization algorithm, save the dominant Hippopotamus currently found, and the optimal position and optimal solution corresponding to the dominant Hippopotamus are the parameter optimization results of variational mode decomposition.

3. The signal denoising method for bridge dynamic load test according to claim 2 is characterized in that: The output is the optimal solution found by the Hippo optimization algorithm, and the currently found dominant Hippo is saved, specifically including: Initialize the population; Set the maximum number of iterations T and the number of hippo populations N, set the initial hippo population, set i = 1, and t = 1; Calculate the objective function and update the dominant hippo according to the fitness function; Phase 1: Expanding the search space, including: Hippos spawn in ponds or rivers; Take half of the population and calculate the dominant hippopotamus The location and immature hippopotamus The position of each hippopotamus X i , each hippo represents a solution; Phase 2: Fine-tune the current solution, including: Starting from 1+N / 2, randomly generate the position of the predator and calculate the hippopotamus's defense against the predator position, update each hippopotamus position X i ; Set i=i+1, calculate the objective function and update the dominant hippo; If not, set i = 1, calculate the new boundary and the hippopotamus escape from the predator Enter the new pond location and update each hippo's position X i ; The third stage: Find the local optimal solution, which includes: Calculate the new boundary, traversing from the first hippopotamus to the Nth hippopotamus; Hippos flee from predators and enter new ponds or rivers; Counting hippos escaping predators Enter the new pond location and update each hippo's position X i ; Save the currently found dominant hippopotamus, corresponding to the optimal position and optimal solution; Output the optimal solution found by the Hippo optimization algorithm and end the Hippo optimization algorithm.

4. The signal denoising method for bridge dynamic load test according to claim 1 is characterized in that: The variational mode decomposition performs the first denoising on the bridge signal, specifically including: Receive the optimization result of the Hippo optimization algorithm to obtain the decomposition scale K and penalty factor alpha of the variational mode decomposition; Initialize the parameters of variational mode decomposition, including the decomposition scale K, penalty factor alpha, step size tau, and maximum number of iterations T; Assign an initial center frequency omega and an initial modal signal u to each mode. For each modal component, use the alternating direction multiplier method to update the modal signal u and the center frequency omega. The signal is decomposed by using variational mode decomposition, and the correlation coefficient CC of the IMFs obtained by decomposition is calculated. The correlation coefficient CC threshold of IMFs is set to the maximum correlation coefficient CC max 1 / 10 of Determine whether the correlation coefficient CC of IMFs reaches the threshold. If it does not reach the set threshold, remove the IMFs; If the correlation coefficient CC of IMFs reaches the set threshold, the IMFs are accumulated to obtain the reconstructed signal.

5. The signal denoising method for bridge dynamic load test according to claim 4 is characterized in that: The variational mode decomposition performs a first denoising on the bridge signal, and further comprises: Decompose the signal into multiple bandwidth-limited IMFs, which represent different frequency components of the signal and each IMF is associated with a specific frequency band; After decomposing the bridge signal by variational mode decomposition, the correlation coefficient between IMFs and the original signal is calculated; Based on the calculation results, it is determined whether the bridge signal contains noise. If the correlation coefficient is high, it means that the IMFs are the main component of the signal and there is no noise; If the correlation coefficient is low, it means that the IMFs contain noise or irrelevant information, and the IMFs with low correlation coefficients are removed.

6. The signal denoising method for bridge dynamic load test according to claim 1 is characterized in that: The introducing of singular spectrum analysis for the second denoising specifically includes: Singular spectrum analysis is introduced to calculate the trajectory matrix and decompose the singular values; Define the signal y to be decomposed and set the singular spectrum analysis window length L; Generate an initial trajectory matrix X, where the number of rows of X is the window length L, the number of columns is K = N - L + 1, and calculate each column of the trajectory matrix X to obtain the eigenvalue λ and eigenvector u; Calculate the covariance matrix R, perform eigenvalue decomposition on the covariance matrix R, and obtain the eigenvalue λ and eigenvector u; Calculate the singular spectrum according to the eigenvalue λ, that is, the singular spectrum matrix Y, and calculate the singular spectrum Yi at each time point; Reconstruct the components of the signal according to the singular spectrum matrix Y and the eigenvector u; Calculate the reconstructed signal y_i of the i-th mode and output the reconstructed signals y_1, y_2, ..., y_r; The denoised signal is generated by recombining the principal components identified by singular spectrum analysis.

7. The signal denoising method for bridge dynamic load test according to claim 6 is characterized by: Decomposing the singular values ​​to identify and separate the main modes in the signal specifically includes: Decompose the singular values, arrange them in descending order, and set the threshold of the cumulative contribution rate; The cumulative contribution rate is calculated using the singular value of the i-th component and the singular values ​​of all components to determine whether the contribution rate is greater than 0.

99. If it is greater than 0.99, the first i components are reorganized into signals to form the final signal for bridge structure analysis. If it is not greater than 0.99, continue to use the singular values ​​of the i+1th component and the singular values ​​of all components to recalculate the cumulative contribution rate until it is greater than 0.99.

Images