Differentiated regulation method for track irregularity of high-speed railway large-span bridge

Abstract

The high-speed railway large-span bridge track irregularity differential regulation method comprises the following steps: integrating track irregularity historical detection data under different environmental temperatures into multi-period track irregularity historical detection data with an environmental temperature label; a multivariate empirical wavelet transform method is established to obtain a plurality of intrinsic modal components; the intrinsic modal components are defined as an intrinsic mode, a multi-dimensional physical characteristic vector of all intrinsic modes is extracted to construct a characteristic vector, and an unsupervised classification is performed on all intrinsic modes by using a condensed hierarchical clustering algorithm; according to the classification result, a classified division is performed, and different differential regulation strategies are formulated for different classified categories; when the track adjustment amount required by the classified category exceeds the limit, an optimization model containing a physical constraint of the adjustment amount and a smoothness expectation target is established, and the optimal track adjustment amount considering the current state and future temperature adaptability is solved.

Description

Technical Field

[0001] This invention belongs to the field of track engineering technology, and in particular relates to a method for differentiated control of track irregularities in long-span bridges of high-speed railways. Background Technology

[0002] Maintaining the track alignment and track smoothness of long-span high-speed railway bridges presents unique challenges. Static monitoring data from the bridge track is actually a mixture of track geometric deviations and macroscopic temperature deformation of the bridge structure. Traditional track smoothness evaluation methods struggle to effectively separate these two factors, easily leading to misjudgments. Existing signal decomposition methods based on time-frequency analysis suffer from modal heterogeneity and characteristic wavelength drift when processing multi-period, wide-temperature-range data, failing to establish stable correlation models between each modal component and temperature changes. Consequently, they struggle to support differentiated and precise control based on multi-period evolution patterns. Therefore, there is an urgent need to develop a control method capable of synchronous decomposition of multi-period data, automatic modal alignment, and component identification and classification based on temperature correlations to improve the long-term track smoothness maintenance capability of long-span bridges under complex temperature conditions. Summary of the Invention

[0003] To address the problems of uneven mode decomposition, difficulty in identifying temperature-sensitive components, and lack of long-term maintenance capability of control schemes in traditional technologies for processing multi-phase track irregularity data of long-span high-speed railway bridges, this invention provides a differentiated control method for track irregularities of long-span high-speed railway bridges. To achieve the above objectives, this invention is implemented through the following technical solution:

[0004] A differentiated control method for track irregularities in long-span bridges of high-speed railways includes the following steps:

[0005] Step S1: Obtain historical track irregularity detection data of long-span bridges under different ambient temperatures, and integrate the historical track irregularity detection data into multi-period historical track irregularity detection data with ambient temperature labels.

[0006] Step S2: Establish a multivariate empirical wavelet transform method to calculate the average normalized spectrum of historical track irregularity detection data from multiple periods, and obtain several intrinsic mode components with consistent modal quantity and characteristic wavelength.

[0007] Step S3: Define an intrinsic mode component with the same number of modes and the same characteristic wavelength as an intrinsic mode. Extract the multidimensional physical feature vectors of all intrinsic modes to construct feature vectors. Use the agglomerative hierarchical clustering algorithm to perform unsupervised classification of all intrinsic modes.

[0008] Step S4: Based on the classification results made by the agglomerative hierarchical clustering algorithm, the intrinsic modes are divided into component categories, and different differentiated control strategies are formulated for different component categories.

[0009] Step S5: When the required orbital adjustment amount for a component category exceeds the limit, based on the differentiated control strategy, an optimization model is established that includes physical constraints on the adjustment amount and the expected smoothness objective, and the optimal orbital adjustment amount that takes into account both the current state and future temperature adaptability is solved.

[0010] Preferably, the historical track irregularity detection data in step S1 is the amplitude-mileage curve obtained by subtracting the design elevation from the absolute track elevation collected by the track measuring trolley.

[0011] Preferably, the specific implementation of step S2 includes:

[0012] Step S21: Perform Fourier transform on each period of historical track irregularity detection data to obtain the frequency and spectral amplitude of each period of historical track irregularity detection data, and use Formula 1 to calculate the average normalized spectrum of all periods of historical track irregularity detection data.

[0013] Formula 1

[0014] In the formula, For the average normalized spectrum, ω For frequency, N This represents the total number of periods for historical monitoring data on track irregularities. For the first n The average value of the spectrum amplitude during the period For the first n The standard deviation of the periodic spectrum amplitude, where n is the number of periods in the current execution of historical track irregularity detection data. This refers to the spectral amplitude of the historical detection data on track irregularities currently being performed.

[0015] Step S22: Based on the average normalized spectrum A unified spectrum segmentation boundary is obtained by using the Localmax segmentation strategy;

[0016] Step S23: Based on a unified spectrum segmentation boundary, perform a univariate empirical wavelet transform on each period of historical track irregularity detection data to obtain several intrinsic mode components with consistent modal quantity and characteristic wavelength.

[0017] Preferably, the specific implementation of step S3 includes:

[0018] Modality clustering identification based on multidimensional feature indicators defines an intrinsic modality as an intrinsic modality with the same number of modalities and the same feature wavelength. Multidimensional physical feature vectors of all intrinsic modalities are extracted to construct feature vectors. The average link distance is used as a similarity measure between samples. An agglomerative hierarchical clustering algorithm is used to perform unsupervised classification of all intrinsic modalities.

[0019] Preferably, the multidimensional physical feature vector includes temperature correlation coefficient, energy fluctuation standard deviation, and driving performance impact index.

[0020] Preferably, the temperature correlation coefficient is the Pearson correlation coefficient between the root mean square energy of the intrinsic mode at any temperature and the temperature; the standard deviation of energy fluctuation is the standard deviation of the root mean square energy of the intrinsic mode at any temperature; and the driving performance impact index is the maximum amplitude of the chord measurement at 60m midpoint of the intrinsic mode at any temperature.

[0021] Preferably, the specific implementation of step S4 includes:

[0022] Based on the classification results of the agglomerative hierarchical clustering algorithm, the inherent modes are divided into four component categories: longitudinal profile components, construction deviation components, high-fluctuation temperature-sensitive components, and low-fluctuation temperature-sensitive components. Corresponding differentiated control strategies are formulated for each of the four component categories.

[0023] Preferably, the implementation method of formulating corresponding differentiated control strategies for the four component categories includes:

[0024] The longitudinal profile components of the line are defined as Category 1. The differentiated control strategy corresponding to Category 1 is as follows: the longitudinal profile components of the line are regarded as the overall expansion and contraction and flexural deformation of the bridge structure under temperature load, and no adjustment is made; the scope of the longitudinal profile components of the line is the inherent mode with wavelength greater than or equal to 200m, temperature correlation coefficient greater than or equal to 0.8 and energy fluctuation standard deviation greater than or equal to 5mm.

[0025] The construction deviation component is defined as Category 2. The differentiated control strategy corresponding to Category 2 is as follows: the construction deviation component is regarded as a fixed irregularity, and the adjustment goal is to completely eliminate it. The scope of the construction deviation component is the intrinsic mode with a temperature correlation coefficient of less than or equal to 0.4 and an energy fluctuation standard deviation of less than 5 mm.

[0026] High-fluctuation temperature-sensitive components are defined as Category 3. The differentiated control strategy corresponding to Category 3 is as follows: calculate the average value of all high-fluctuation temperature-sensitive components in Category 3, and use the average value among the high-fluctuation temperature-sensitive components as the adjustment target. Then, evaluate the magnitude of the driving performance impact index in the adjusted high-fluctuation temperature-sensitive components. Finally, with the goal of eliminating the part of each high-fluctuation temperature-sensitive component that causes the driving performance impact index to exceed the limit, make another adjustment. The scope of the high-fluctuation temperature-sensitive components is the intrinsic modes with wavelength less than 200m, temperature correlation coefficient greater than 0.4, and energy fluctuation standard deviation greater than or equal to 5mm.

[0027] Low-fluctuation temperature-sensitive components are defined as category 4. The differentiated control strategy corresponding to category 4 is as follows: calculate the average value of all low-fluctuation temperature-sensitive components in category 4, and use the average value among the low-fluctuation temperature-sensitive components as the adjustment target. The category of low-fluctuation temperature-sensitive components is the intrinsic modes with wavelength less than 200m, temperature correlation coefficient greater than 0.4 and energy fluctuation standard deviation less than 5mm.

[0028] Preferably, the specific implementation of establishing the optimization model in step S5 includes:

[0029] Step S51: Let the first component category be... k The intrinsic mode at mileage position x The original amplitude is I k ( x The original amplitude is... I k ( x The corresponding track adjustment amount is C k ( x The residual amplitude of this intrinsic mode after modulation is R k ( x )= I k ( x ) + C k ( x );

[0030] Step S52: Establish the objective function of the optimization model:

[0031] Formula 2

[0032] In the formula, min F ω represents the objective function, k represents the index of the intrinsic mode, and M is the total number of intrinsic modes participating in the regulation; 1,k With ω 2,k For the first k The two weight coefficients of each intrinsic mode represent the penalty for the k-th intrinsic mode being poor in the current smoothness state and the penalty for the k-th intrinsic mode being sensitive to temperature fluctuations, respectively. E T Indicates temperature T The mathematical expectation (Δ) R k ( x,T )) 2 Represents the residual amplitude of the k-th intrinsic mode R k ( x The change at temperature T;

[0033] Step S53: Establish constraints for the objective function of the optimization model.

[0034] Preferably, the constraints include adjustment capability constraints and modal difference constraints; the adjustment capability constraint stipulates that the total adjustment at any mileage point x must not exceed the physical limits of the fastener and the track bed. C max ( x The modal difference constraint is designed based on the evolution law of different intrinsic modes. For the intrinsic modes in category 1, the adjustment amount is constrained. C k ( x =0; For the intrinsic modes in category 2, constrain their residual amplitude | R k ( x | Close to 0; For intrinsic modes in categories 3 and 4, constrain their residual amplitudes to fall within the allowable fluctuation range within the expected temperature variation range [- d , d ]Inside.

[0035] The present invention has the following advantages over the prior art:

[0036] 1. Modal homogeneity: The multivariate empirical wavelet transform method proposed in this invention achieves adaptive synchronous decomposition of multi-period non-stationary signals by averaging and normalizing the spectrum, overcoming the problems of mode aliasing and wavelength drift in traditional methods.

[0037] 2. Precise classification: By introducing agglomerative hierarchical clustering algorithm, this invention comprehensively considers the effects of temperature correlation, energy fluctuation and driving performance, avoiding the blindness of manually setting thresholds, and accurately identifying the disease modes that need to be adjusted and the elastic deformation modes that need to be retained.

[0038] 3. Long-term performance: The differentiated control and optimization algorithm proposed in this invention not only improves the smoothness of operation at the moment, but also incorporates the expected temperature change into the objective function, ensuring that the line can maintain a good smoothness under extreme temperature rise and fall conditions.

[0039] 4. Cost-effectiveness: By identifying and retaining long-wave temperature deformation modes, this invention significantly reduces the amount of unnecessary track adjustment work. Attached Figure Description

[0040] Figure 1 This is a flowchart illustrating the execution of the control method of the present invention;

[0041] Figure 2 This is a comparison chart showing the improvement in track smoothness after implementing the control method based on the present invention. Detailed Implementation

[0042] The present invention will now be further described with reference to the accompanying drawings and specific embodiments:

[0043] To make the objectives, technical solutions, and advantages of the present invention clearer and more explicit, the present invention will be further described below with reference to the accompanying drawings and embodiments.

[0044] Example 1:

[0045] like Figure 1 As shown, a differentiated control method for track irregularities in a long-span high-speed railway bridge includes the following steps:

[0046] Step S1: Obtain historical track irregularity monitoring data for long-span bridges under different ambient temperatures, and integrate this data into multi-period historical track irregularity monitoring data with ambient temperature labels. The historical track irregularity monitoring data in this step is the amplitude-mileage curve obtained by subtracting the design elevation from the absolute track elevation collected by the track measurement trolley. Since bridge deformation is significantly affected by temperature, the selected historical track irregularity monitoring data should cover a wide range of temperature variations, such as seasonal temperature differences, to include sufficient information on structural deformation.

[0047] Step S2: Establish a multivariate empirical wavelet transform method to obtain several intrinsic modal components with consistent modal numbers and characteristic wavelengths. Specifically, a multivariate empirical wavelet transform method is established. By calculating the average normalized spectrum of multi-period track irregularity historical detection data, a consistent spectral segmentation boundary is obtained. Based on the spectral segmentation boundary, a univariate empirical wavelet transform is performed on each period of track irregularity historical detection data. This achieves synchronous decomposition and modal wavelength alignment of the non-stationary multi-period track irregularity historical detection data, thereby obtaining several intrinsic modal components with consistent modal numbers and characteristic wavelengths.

[0048] Specifically, the specific implementation of step S2 above includes:

[0049] Step S21: Perform a Fourier transform on each period of historical track irregularity detection data to obtain the frequency and spectral amplitude of each period of historical track irregularity detection data. Calculate the average normalized spectrum of all periods of historical track irregularity detection data using Formula 1. This average normalized spectrum can simultaneously characterize all potential periodic components at different temperatures. By normalizing the obtained frequency and standardizing the obtained spectral amplitude using Z-Score, the energy dimension differences between different detection batches can be eliminated.

[0050] Formula 1

[0051] In the formula, For the average normalized spectrum, ω For frequency, NThis represents the total number of periods for historical monitoring data on track irregularities. For the first n The average value of the spectrum amplitude during the period For the first n The standard deviation of the periodic spectrum amplitude, where n is the number of periods in the current execution of historical track irregularity detection data. This refers to the spectral amplitude of the historical detection data on track irregularities currently being performed.

[0052] Step S22: Based on the average normalized spectrum A unified spectrum segmentation boundary is obtained by using the Localmax segmentation strategy;

[0053] Step S23: Based on a unified spectral segmentation boundary, perform a univariate empirical wavelet transform on each period of historical track irregularity detection data to obtain several intrinsic mode components with consistent modal quantity and characteristic wavelength. Through the above process, this invention achieves synchronous decomposition of multi-period data, ensuring that the several intrinsic mode components obtained by decomposition have consistent quantity and characteristic wavelength at different temperatures.

[0054] To be precise, the above steps S21, S22, and S23 together constitute the multivariate empirical wavelet transform method. The multivariate empirical wavelet transform method is an extension of the traditional univariate empirical wavelet transform method. The traditional univariate empirical wavelet transform method can only segment a single spectrum and has problems such as mode inhomogeneity and characteristic wavelength drift. However, the average normalized spectrum operation in step S21 can effectively solve the limitations of the traditional univariate empirical wavelet transform method, thereby obtaining several intrinsic mode components with the same number of modes and the same characteristic wavelength.

[0055] Step S3: Define an intrinsic mode component with the same number of modes and the same characteristic wavelength as an intrinsic mode. Extract the multidimensional physical feature vectors of all intrinsic modes to construct feature vectors. Use the agglomerative hierarchical clustering algorithm to perform unsupervised classification of all intrinsic modes.

[0056] The specific implementation method of this step includes:

[0057] Modal clustering identification based on multidimensional feature indicators defines an intrinsic mode as an intrinsic mode with consistent modal quantity and characteristic wavelength. The intrinsic mode represents the performance results of historical detection data on track irregularities in different periods. Multidimensional physical feature vectors of all intrinsic modes are extracted to construct feature vectors. The average link distance is used as a similarity measure between samples. An agglomerative hierarchical clustering algorithm is used to perform unsupervised classification of all intrinsic modes, thereby dividing all intrinsic modes into clusters with similar physical properties. This enables the classification of temperature-related modes and the identification of temperature-sensitive modes with different properties.

[0058] The multidimensional physical feature vector includes the temperature correlation coefficient, the standard deviation of energy fluctuation, and the performance impact index. The temperature correlation coefficient is the Pearson correlation coefficient between the root mean square energy of the intrinsic mode at any temperature and the temperature. The standard deviation of energy fluctuation is the standard deviation of the root mean square energy of the intrinsic mode at any temperature. The performance impact index is the maximum amplitude of the chord measurement at 60m midpoint of the intrinsic mode at any temperature.

[0059] Step S4: Based on the classification results made by the agglomerative hierarchical clustering algorithm, the intrinsic modes are divided into component categories, and different differentiated control strategies are formulated for different component categories.

[0060] The specific implementation method of this step includes:

[0061] Based on the classification results made by the agglomerative hierarchical clustering algorithm, the inherent modes are divided into four component categories: longitudinal profile components, construction deviation components, high-fluctuation temperature-sensitive components, and low-fluctuation temperature-sensitive components. Corresponding differentiated control strategies are formulated for each of the four component categories.

[0062] The implementation methods for developing differentiated control strategies for the four component categories include:

[0063] The longitudinal profile components of the line are defined as Category 1, where the scope of the longitudinal profile components includes intrinsic modes with wavelength greater than or equal to 200m, temperature correlation coefficient greater than or equal to 0.8, and energy fluctuation standard deviation greater than or equal to 5mm. The differentiated control strategy corresponding to Category 1 is to regard the longitudinal profile components of the line as the overall expansion and contraction and deflection deformation of the bridge structure under temperature load, and not to adjust them.

[0064] The construction deviation component is defined as Category 2, where the scope of the construction deviation component is the intrinsic mode with a temperature correlation coefficient less than or equal to 0.4 and an energy fluctuation standard deviation less than 5 mm; the differentiated control strategy corresponding to Category 2 is to treat the construction deviation component as a fixed non-compliance and aim to completely eliminate it.

[0065] High-fluctuation temperature-sensitive components are defined as Category 3. The scope of high-fluctuation temperature-sensitive components includes intrinsic modes with wavelengths less than 200m, temperature correlation coefficients greater than 0.4, and energy fluctuation standard deviations greater than or equal to 5mm. The differentiated control strategy corresponding to Category 3 is as follows: calculate the average value of all high-fluctuation temperature-sensitive components in Category 3, and use the average value among the high-fluctuation temperature-sensitive components as the adjustment target. Then, evaluate the magnitude of the driving performance impact index in the adjusted high-fluctuation temperature-sensitive components. Finally, with the goal of eliminating the part of the adjusted high-fluctuation temperature-sensitive components that causes the driving performance impact index to exceed the limit, make another adjustment.

[0066] Low-fluctuation temperature-sensitive components are defined as Category 4. The scope of low-fluctuation temperature-sensitive components includes intrinsic modes with wavelengths less than 200m, temperature correlation coefficients greater than 0.4, and energy fluctuation standard deviations less than 5mm. The differentiated control strategy corresponding to Category 4 is to calculate the average value of all low-fluctuation temperature-sensitive components in Category 4, and use the average value among low-fluctuation temperature-sensitive components as the adjustment target.

[0067] Step S5: Constrained by the adjustment limit of the ballasted track bed, when the track adjustment amount required by the component category exceeds the limit, based on the differentiated control strategy, an optimization model is established that includes the physical constraints of the adjustment amount and the expected smoothness target, and the optimal track adjustment amount that takes into account both the current state and future temperature adaptability is solved.

[0068] The specific implementation methods for establishing the optimization model in step S5 include:

[0069] Step S51: Let the first component category be... k The intrinsic mode at mileage position x The original amplitude is I k ( x The original amplitude is... I k ( x The corresponding track adjustment amount is C k ( x The residual amplitude of this intrinsic mode after modulation is R k ( x )= I k ( x ) + C k ( x );

[0070] Step S52: Establish the objective function of the optimization model:

[0071] Formula 2

[0072] In the formula, min F ω represents the objective function, k represents the index of the intrinsic mode, and M is the total number of intrinsic modes participating in the regulation; 1,k With ω 2,k For the first k The two weight coefficients of each intrinsic mode represent the penalty for the k-th intrinsic mode's poor performance in the current smoothness state and the penalty for the k-th intrinsic mode's sensitivity to temperature fluctuations, respectively. Based on the differential control strategy, ω under temperature fluctuation insensitivity... 1,k Higher and ω 2,k At lower temperatures, when temperature is sensitive and fluctuates drastically, then ω2,k It was assigned a higher value; E T Indicates temperature T The mathematical expectation (Δ) R k ( x,T )) 2 Represents the residual amplitude of the k-th intrinsic mode R k ( x The change at temperature T;

[0073] Step S53: Establish constraints for the objective function of the optimization model. These constraints include adjustment capability constraints and modal difference constraints; the adjustment capability constraint stipulates that the total adjustment at any mileage point x must not exceed the physical limits of the fastener and track bed. C max ( x The modal difference constraint is designed based on the evolution law of different intrinsic modes. For the intrinsic modes in category 1, the adjustment amount is constrained. C k ( x =0; For the intrinsic modes in category 2, constrain their residual amplitude | R k ( x | Close to 0; For intrinsic modes in categories 3 and 4, constrain their residual amplitudes to fall within the allowable fluctuation range within the expected temperature variation range [- d , d Finally, by solving the constrained optimization problem described above, the optimal set of adjustment sequences can be obtained. C ( x As a basis for on-site adjustment work, the changes in track smoothness before and after adjustment are as follows: Figure 2 As shown.

Claims

1. A method for differentiated control of track irregularities in long-span high-speed railway bridges, characterized in that, Includes the following steps: Step S1: Obtain historical track irregularity detection data of long-span bridges under different ambient temperatures, and integrate the historical track irregularity detection data into multi-period historical track irregularity detection data with ambient temperature labels. Step S2: Establish a multivariate empirical wavelet transform method to calculate the average normalized spectrum of historical track irregularity detection data from multiple periods, and obtain several intrinsic mode components with consistent modal quantity and characteristic wavelength. Step S3: Define an intrinsic mode component with the same number of modes and the same characteristic wavelength as an intrinsic mode. Extract the multidimensional physical feature vectors of all intrinsic modes to construct feature vectors. Use the agglomerative hierarchical clustering algorithm to perform unsupervised classification of all intrinsic modes. Step S4: Based on the classification results made by the agglomerative hierarchical clustering algorithm, the intrinsic modes are divided into component categories, and different differentiated control strategies are formulated for different component categories. Step S5: When the required orbital adjustment amount for a component category exceeds the limit, based on the differentiated control strategy, an optimization model is established that includes physical constraints on the adjustment amount and the expected smoothness objective, and the optimal orbital adjustment amount that takes into account both the current state and future temperature adaptability is solved.

2. The method for differentiated control of track irregularities in a long-span high-speed railway bridge according to claim 1, characterized in that, The historical track irregularity detection data in step S1 is the amplitude-mileage curve obtained by subtracting the design elevation from the absolute track elevation collected by the track measurement trolley.

3. The method for differentiated control of track irregularities in a long-span high-speed railway bridge according to claim 1, characterized in that, The specific implementation of step S2 includes: Step S21: Perform Fourier transform on each period of historical track irregularity detection data to obtain the frequency and spectral amplitude of each period of historical track irregularity detection data, and use Formula 1 to calculate the average normalized spectrum of all periods of historical track irregularity detection data. Official 1 In the formula, For the average normalized spectrum, ω For frequency, N This represents the total number of periods for historical monitoring data on track irregularities. For the first n The average value of the periodic spectrum amplitude For the first n The standard deviation of the periodic spectrum amplitude, where n is the number of periods in the current execution of historical track irregularity detection data. The spectral amplitude of the historical data on track irregularities being currently being processed; Step S22: Based on the average normalized spectrum A unified spectrum segmentation boundary is obtained by using the Localmax segmentation strategy; Step S23: Based on a unified spectrum segmentation boundary, perform a univariate empirical wavelet transform on each period of historical track irregularity detection data to obtain several intrinsic mode components with consistent modal quantity and characteristic wavelength.

4. The method for differentiated control of track irregularities in a long-span high-speed railway bridge according to claim 1, characterized in that, The specific implementation of step S3 includes: Modality clustering identification based on multidimensional feature indicators defines an intrinsic modality as an intrinsic modality with the same number of modalities and the same feature wavelength. Multidimensional physical feature vectors of all intrinsic modalities are extracted to construct feature vectors. The average link distance is used as a similarity measure between samples. An agglomerative hierarchical clustering algorithm is used to perform unsupervised classification of all intrinsic modalities.

5. The method for differentiated control of track irregularities in a long-span high-speed railway bridge according to claim 4, characterized in that, The multidimensional physical feature vector includes temperature correlation coefficient, energy fluctuation standard deviation, and driving performance impact index.

6. The method for differentiated control of track irregularities in a long-span high-speed railway bridge according to claim 5, characterized in that, The temperature correlation coefficient is the Pearson correlation coefficient between the root mean square energy of the intrinsic mode at any temperature and the temperature; the standard deviation of energy fluctuation is the standard deviation of the root mean square energy of the intrinsic mode at any temperature; the driving performance impact index is the maximum amplitude of the chord measurement at 60m midpoint of the intrinsic mode at any temperature.

7. The method for differentiated control of track irregularities in a long-span high-speed railway bridge according to claim 1, characterized in that, The specific implementation of step S4 includes: Based on the classification results of the agglomerative hierarchical clustering algorithm, the inherent modes are divided into four component categories: longitudinal profile components, construction deviation components, high-fluctuation temperature-sensitive components, and low-fluctuation temperature-sensitive components. Corresponding differentiated control strategies are formulated for each of the four component categories.

8. The method for differentiated control of track irregularities in a long-span high-speed railway bridge according to claim 7, characterized in that, The implementation methods for developing differentiated control strategies for the four component categories include: The longitudinal profile components of the line are defined as Category 1. The differentiated control strategy corresponding to Category 1 is as follows: the longitudinal profile components of the line are regarded as the overall expansion and contraction and flexural deformation of the bridge structure under temperature load, and no adjustment is made; the scope of the longitudinal profile components of the line is the inherent mode with wavelength greater than or equal to 200m, temperature correlation coefficient greater than or equal to 0.8 and energy fluctuation standard deviation greater than or equal to 5mm. The construction deviation component is defined as Category 2. The differentiated control strategy corresponding to Category 2 is as follows: the construction deviation component is regarded as a fixed irregularity, and the adjustment goal is to completely eliminate it. The scope of the construction deviation component is the intrinsic mode with a temperature correlation coefficient of less than or equal to 0.4 and an energy fluctuation standard deviation of less than 5 mm. High-fluctuation temperature-sensitive components are defined as Category 3. The differentiated control strategy corresponding to Category 3 is as follows: calculate the average value of all high-fluctuation temperature-sensitive components in Category 3, and use the average value among the high-fluctuation temperature-sensitive components as the adjustment target. Then, evaluate the magnitude of the driving performance impact index in the adjusted high-fluctuation temperature-sensitive components. Finally, with the goal of eliminating the part of each high-fluctuation temperature-sensitive component that causes the driving performance impact index to exceed the limit, make another adjustment. The scope of the high-fluctuation temperature-sensitive components is the intrinsic modes with wavelength less than 200m, temperature correlation coefficient greater than 0.4, and energy fluctuation standard deviation greater than or equal to 5mm. Low-fluctuation temperature-sensitive components are defined as category 4. The differentiated control strategy corresponding to category 4 is as follows: calculate the average value of all low-fluctuation temperature-sensitive components in category 4, and use the average value among the low-fluctuation temperature-sensitive components as the adjustment target. The category of low-fluctuation temperature-sensitive components is the intrinsic modes with wavelength less than 200m, temperature correlation coefficient greater than 0.4 and energy fluctuation standard deviation less than 5mm.

9. A method for differentiated control of track irregularities in a long-span high-speed railway bridge according to claim 8, characterized in that, The specific implementation methods for establishing the optimization model in step S5 include: Step S51: Let the first component category be... k The intrinsic mode at mileage position x The original amplitude is I k ( x The original amplitude is... I k ( x The corresponding track adjustment amount is C k ( x The residual amplitude of this intrinsic mode after modulation is R k ( x )= I k ( x ) + C k ( x ); Step S52: Establish the objective function of the optimization model: Formula 2 In the formula, min F ω represents the objective function, k represents the index of the intrinsic mode, and M is the total number of intrinsic modes participating in the regulation; 1,k With ω 2,k For the first k The two weight coefficients of each intrinsic mode represent the penalty for the k-th intrinsic mode being poor in the current smoothness state and the penalty for the k-th intrinsic mode being sensitive to temperature fluctuations, respectively. E T Indicates temperature T The mathematical expectation (Δ) R k ( x,T )) 2 Represents the residual amplitude of the k-th intrinsic mode R k ( x The change at temperature T; Step S53: Establish constraints for the objective function of the optimization model.

10. A method for differentiated control of track irregularities in a long-span high-speed railway bridge according to claim 9, characterized in that, The constraints include adjustment capability constraints and modal difference constraints; the adjustment capability constraint stipulates that the total adjustment at any mileage point x must not exceed the physical limits of the fastener and the track bed. C max ( x The modal difference constraint is designed based on the evolution law of different intrinsic modes, and for the intrinsic modes in category 1, it constrains their adjustment amount. C k ( x =0; For the intrinsic modes in category 2, constrain their residual amplitude | R k ( x | Close to 0; For intrinsic modes in categories 3 and 4, constrain their residual amplitude to fall within the fluctuation range of the expected temperature variation range [- d , d ]Inside.

Images