Railway screw pile dynamic damage deformation identification method suitable for indoor model test

By using wavelet packet analysis method in the composite foundation of the railway screw pile, combining adaptive filtering and WPT technology, the power damage deformation of the screw pile is identified, and the problem of failure to effectively utilize the characteristic parameters of the screw pile in the existing technology is solved, and the problem of realizing the Accurate identification of screw pile dynamic damage and analysis of evolutionary process of composite foundation deformation.

CN118940018APending Publication Date: 2024-11-12NORTHWEST RES INST CO LTD OF C R E C +3
View PDF 3 Cites 0 Cited by

Patent Information

Application Number
CN202411399393.8
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-10-09
Publication Date
2024-11-12

AI Technical Summary

Technical Problem

When studying the dynamic response and damage deformation of composite foundations of railway screw piles, the existing technology failed to effectively utilize the characteristic parameters of the screw piles themselves and the physical quantity monitored internally in the pile body. Especially under the dynamic load of the train, it is difficult to accurately identify the power of the screw piles. Damage deformation.

Method used

By using wavelet packet analysis method, the acceleration sensor is arranged in indoor model experiments, the acceleration data of the screw pile is obtained, adaptive filtering and wavelet packet transformation (WPT) are performed, the damage index evaluation method is established, and the subband of the signal is decomposed using wavelet packets. The change rate of energy proportion is used to establish structural vulnerability discrimination indicators and identify the power damage deformation of screw piles.

Benefits of technology

The detailed identification of the power damage deformation of screw piles under train load is achieved, and the deformation settlement evolution process of screw pile composite foundation is inferred, which makes up for the shortcomings of traditional signal processing methods and provides a more accurate damage evaluation method.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118940018A_ABST
    Figure CN118940018A_ABST
Patent Text Reader

Abstract

The invention discloses a railway screw pile dynamic damage deformation identification method suitable for an indoor model test, belongs to the technical field of model tests, and establishes a railway screw pile dynamic cumulative damage evaluation method by using an adaptive filtering and denoising technology and a wavelet packet transform (WPT) theory. In the prior art, the research on the damage and deformation of the railway composite foundation pile body is mostly based on the displacement and deformation monitoring of the pile body in a model test or numerical analysis, and cannot start from the characteristic parameters of the pile body and the physical quantity monitored in the pile body based on dynamic model test data. And the power damage deformation of the screw pile is identified from the angle of the energy transfer characteristic of the pile body. Besides, noise signals are removed through LMS self-adaptive filtering, baseline correction is carried out on acceleration data through a B-spline fitting algorithm, and the defects of traditional signal processing methods such as FIR filtering and least square fitting are overcome.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention belongs to the technical field of model tests, and specifically relates to a method for identifying dynamic damage of screw piles; more specifically, it relates to a method for identifying dynamic damage and deformation of railway screw piles suitable for indoor model tests. Background Art

[0002] Ground treatment plays a very critical role in railway engineering, which is directly related to the stability and safety of train operation. Pile-soil composite ground has a good effect on strengthening soft soil layers and controlling settlement, and has been widely used in the field of railway engineering.

[0003] At present, the train speeds planned for many railway lines in my country have reached 350km / h. The vertical dynamic loads generated during train operation have also increased, and the vertical dynamic response to the roadbed has become more and more obvious, especially in low embankment (embankment height <3m) sections. The dynamic response of the train dynamic load to the foundation below the embankment will also increase. Since the pile-soil composite foundation is subjected to train cyclic loads for a long time during the service of high-speed railways, it is very likely to cause local damage and deformation, resulting in the loss of pile stiffness and strength, and ultimately affecting the overall bearing capacity of the composite foundation and causing permanent deformation and damage.

[0004] The screw pile is a new type of special-shaped cross-section pile, with a straight rod type on the upper part and a screw type on the lower part. It is developed on the basis of a series of special-shaped piles with threads, such as long spiral bored piles and Japanese steel fiber fully threaded prefabricated piles. Since 2003, screw pile composite foundations have been widely used in construction engineering, railway engineering and other fields due to their high bearing capacity, small settlement, and economic rationality, and have achieved good results. However, there is little engineering experience in screw pile composite foundations in the prior art, especially the dynamic response of railway screw pile composite foundation systems under train dynamic loads has not been addressed. Research on the damage and deformation of screw piles is mostly based on model tests or deformation monitoring of the pile body in numerical analysis, and has failed to start from the characteristic parameters of the screw pile itself and the physical quantities monitored inside the pile body.

[0005] Damage identification methods are mainly divided into pattern recognition method, optimization identification method and damage index method. Among them, pattern recognition method and optimization identification method have defects such as complex calculation and unsuitability for complex structures; damage index method is favored by researchers because of its sensitivity to damage, good noise resistance and simple calculation. In recent years, the method of establishing damage index through wavelet packet analysis has been increasingly applied to the field of structural damage identification. Wavelet packet transform is a vibration signal processing method developed on the basis of wavelet transform. Wavelet packet transform not only has the advantage of accurate decomposition of the low-frequency part of the signal of wavelet transform, but also can accurately decompose the high-frequency part at the same time, thereby greatly improving the integrity of signal decomposition. When the structural model is stimulated by the environment, the damage of the structure can be tested from the detailed mutations after the wavelet packet decomposition of the corresponding data. The location of these mutations can accurately indicate the time when the damage occurs. Wavelet packet analysis is very suitable for analyzing stationary signals, so it can be used as a more ideal tool for signal processing in damage identification. It can be used to construct the characteristic factors required in damage identification, or directly extract useful information about the damage. Due to its outstanding time-frequency positioning ability, WPT has been widely used in the fields of damage identification in house construction, bridge engineering, tunnel engineering, etc. In general, the WPT method has good adaptability and high efficiency in processing vibration signals. However, the research on the dynamic damage and deformation identification of railway screw piles using the energy transfer characteristic method is limited, especially the selection of the optimal wavelet basis and the research on damage characteristic factors need further discussion. Summary of the invention

[0006] The present invention provides a railway screw pile dynamic damage deformation identification method suitable for indoor model tests, the purpose of which is to conduct research based on the characteristic parameters of the screw pile itself and the physical quantities monitored inside the pile body, and to establish a damage index evaluation method through wavelet packet analysis.

[0007] To this end, the present invention adopts the following technical solutions: A method for identifying dynamic damage and deformation of railway screw piles suitable for indoor model tests is specifically performed in the following steps: Step 1: Design a layout scheme for the acceleration sensor on each screw pile in the indoor train simulation vibration test model. Design it according to the principle of arranging the acceleration sensors at equal intervals in the same cross-section pile section, and arrange the acceleration sensors on each screw pile of the test model according to the design scheme; Step 2: fill the train simulation vibration test model according to the test requirements, and after the filling is completed, load the sine wave test model according to the loading condition requirements, and obtain the acceleration data of the screw pile under the action of the sine wave through each acceleration sensor; Step 3: Process the acquired screw pile acceleration data 1) Filtering: Perform LMS adaptive filtering on the original acceleration data obtained in step 2) through an adaptive filter; 2) Write a B-spline fitting algorithm in MATLAB software to remove the baseline drift of the acceleration data after adaptive filtering, that is, to correct the baseline of the acceleration data after adaptive filtering in 1) and obtain the acceleration data signal after baseline calibration. x ( t ); 3) Based on the acceleration data signal in 2) x ( t )'s spectrum characteristics, time domain characteristics, and smoothness characteristics are used to preliminarily select a wavelet basis that meets the requirements, and then select the optimal order and optimal decomposition level of the wavelet basis according to the cost function, where the cost function is the Shannon entropy; that is, first calculate the Shannon entropy values ​​of the preliminarily selected different wavelet bases and different decomposition orders, and then select the wavelet basis and decomposition level that minimize the Shannon entropy; 4) The original acceleration data obtained in step 2 is j Layer wavelet packet decomposition WPT, get 2 j Wavelet packet coefficients Where: i , j , k are the modulation, scale and translation parameters respectively; is the wavelet packet function; t For time; 5) Reconstruct each wavelet packet coefficient to obtain the corresponding baseline-calibrated acceleration data signal x ( t )'s wavelet packet component; In this way, the original acceleration data signal obtained in step 2 x ( t ) is decomposed into 2 j The sum of wavelet packet components: 6) Find the energy proportion of the sub-band of the wavelet packet decomposition signal: The original acceleration data signal obtained in step 2 x ( t )energy E e Defined as: Considering the orthogonality of the wavelet packet function, we get: The sub-band energy of each measuring point of the screw pile is normalized and the first i The distribution of sub-band energy along the structure is: Where M is the number of measuring points, E n i is the energy of the ith sub-band at the nth measurement point of the structure; Considering the influence of noise energy, the energy ratio change rate of the i-th sub-band of the screw pile before and after damage is defined as ERS i for: In the formula: superscript N represents static state, superscript Y represents loaded state; 7) Considering that some sub-bands have no energy distribution or account for a small proportion, they have little effect on the energy characteristics of the vibration signal; therefore, the sub-band energy proportion greater than 5% is selected as the effective sub-band characteristic value, and then the structural vulnerability judgment index Δ is established according to the root mean square value of the change rate of the effective sub-band energy ratio of the screw pile ERMS : In the formula, m is the number of effective sub-bands, ERS for m The average value of the energy ratio change rate of each sub-band; wherein the superscript i and the subscript i in steps 4) to 7) have the same meaning.

[0008] The beneficial effects of the present invention are: The identification method of the present invention is based on the acceleration time history curve of railway screw piles obtained by excitation test, and the dynamic cumulative damage evaluation method of railway screw piles is established by using adaptive filtering denoising technology and wavelet packet transform (WPT) theory. The damage and deformation research of railway composite foundation piles in the prior art is mostly based on the displacement and deformation monitoring of the pile body in model tests or numerical analysis, and fails to identify the dynamic damage and deformation of screw piles from the perspective of the energy transfer characteristics of the pile body itself based on the dynamic model test data based on the characteristic parameters of the pile body itself and the physical quantities monitored inside the pile body. In addition, the noise signal is removed by LMS adaptive filtering, and the acceleration data is baseline corrected by the B-spline fitting algorithm, which makes up for the shortcomings of traditional signal processing methods such as FIR filtering and least squares fitting.

[0009] The identification method of the present invention is based on the analysis of the acceleration data of the screw pile test in the indoor dynamic model test, combined with the adaptive filtering denoising technology and the wavelet packet transform (WPT) method, to fully identify the dynamic damage deformation of the screw pile under the train load and infer the deformation and settlement evolution process of the screw pile composite foundation. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] Figure 1 It is a flow chart of the identification method of the present invention; Figure 2 This is the block diagram of the LMS adaptive filter structure; Figure 3 It is a schematic diagram of wavelet packet decomposition; Figure 4 It is a three-dimensional schematic diagram of the vibration simulation test model used in the present invention; Figure 5 yes Figure 4 Cross-sectional view of the model in Figure 2; Figure 6 yes Figure 4 Vertical section of the model in the middle; Figure 7 is the acceleration time history graph; Figure 8 It is the Fourier spectrum curve diagram; Fig. 9 It is the db15 wavelet time history graph; Fig.10 This is a wavelet packet decomposition result diagram of the PA1 measuring point under the action of a train speed of 240 km / h in the embodiment; Fig.11 It is the energy proportion of sub-bands at each measuring point of the pile body under different train speeds; Fig.12 It is the energy proportion of sub-bands at each measuring point of the pile body under different train speeds; Fig.13 It is the energy proportion of sub-bands at each measuring point of the pile body under different train speeds; Fig.14 is the variation curve of the vulnerability index ΔERMS of the pile body measuring point under different speed conditions; Figure 4-6 Middle: 1. Reaction steel frame, 2. Vibrator, 3. Load-bearing plate, 4. Screw pile, 5. Model box, 6. Foam board, 7. Connecting frame, 8. Film layer, 9. Acceleration sensor, 10. Soil foundation, 11. Cushion. DETAILED DESCRIPTION

[0011] The present invention will be further described below in conjunction with the accompanying drawings and specific embodiments: The present invention provides a method for identifying dynamic damage and deformation of railway screw piles suitable for indoor model tests. Starting from vibration acceleration test data, the invention adopts adaptive filtering denoising technology and WPT means, establishes damage characteristic indicators from the perspective of energy transfer characteristics, identifies the damage and deformation of screw piles under the dynamic load of reciprocating trains, and clarifies the dynamic deformation evolution process of screw pile composite foundations in combination with the macroscopic settlement deformation phenomenon of composite foundations, providing a refined analysis approach for identifying damage and deformation of screw piles under train load excitation. The flowchart of the identification method is shown in the figure below. Figure 1 .

[0012] The identification method is specifically carried out in the following steps: Step 1: Design a layout scheme for the acceleration sensor on each screw pile in the indoor train simulation vibration test model. Design is performed according to the principle of arranging the acceleration sensors at equal intervals in the same cross-section pile section, and arrange the acceleration sensors on each screw pile in the test model according to the design scheme; Step 2: Construct a train simulation vibration test model, load a sine wave test, and obtain the acceleration data of the screw pile through the acceleration sensor; the loading conditions are shown in Table 1: Table 1 - Test loading conditions Step 3: Process the acquired screw pile acceleration data 1) Filtering: Perform LMS adaptive filtering on the acquired acceleration data through an adaptive filter; The block diagram of LMS adaptive filtering is as follows Figure 2 :If the order of adaptive filtering is N , the adaptive filter coefficients are w , the input signal sequence is X , then the output y ( n )for: e ( n )= d ( n )- y ( n ) In the formula, d ( n ) is the expected signal; e ( n ) is the error signal; n Indicates a moment, w i ( n )express n The adaptive filter coefficients at time , x ( n - i ) represents the sampling value of the input signal at time ni.

[0013] make W ( n )=[ w 0 , w 1 ,…, w i-1 ]T , X ( n )=[ x 1 , x 2 ,…, x i-1 ] T The output of the adaptive filter is y ( n ) can be written in matrix form: y ( n )= X ( n ) T W ( n )= W ( n ) T X ( n ) The error signal can be written as: e ( n )= d ( n )- y ( n )= d ( n )- W T ( n ) X ( n ) Define the cost function: J ( w )= E [| e ( n )| 2 ]= E [| d ( n )- y ( n )| 2 ]= E [|( d ( n )- W T ( n ) X ( n )| 2 ] E is a function symbol, indicating the dependent variable; e(n) is the independent variable;y = J ( w ), f = E , x=e(n) ; When the cost function reaches its minimum value, it is considered that optimal filtering is achieved. Such adaptive filtering is called least mean square adaptive filtering (LMS).

[0014] For the least mean square adaptive filtering, it is necessary to determine the filter coefficients that minimize the mean square error, which is generally solved using the gradient descent method. The iterative formula for the filter coefficient vector is: W ( n+ 1)= W ( n )+{[ μ (-▽ J ( w )] / 2} Where: μ is the step size factor; J ( w ) is the gradient of the cost function; W ( n+ 1) is the filter coefficient after iterative update, W ( n ) is the filter coefficient before iterative update.

[0015] Since the instantaneous gradient -2 X ( n ) e ( n ) is an unbiased estimate of the true gradient value. In practical applications, the instantaneous gradient can be used instead of the true gradient, that is: ▽ J ( w )=-2 X ( n )[ d ( n )- W T ( n ) X ( n ) ]=-2 X ( n ) e ( n ) W ( n+ 1)= W ( n )+ μe ( n ) X ( n ) Through step-by-step iteration, the optimal filter coefficients are obtained to achieve adaptive filtering of the input signal.

[0016] 2) Write a B-spline fitting algorithm in MATLAB software to remove the baseline drift of the acceleration data after adaptive filtering, that is, to correct the baseline of the acceleration data after adaptive filtering, and obtain the acceleration signal after baseline calibration. x ( t ); x ( t ) is also called the original signal; 3) According to the spectrum characteristics, time domain characteristics, and smoothness characteristics of the acceleration signal after baseline calibration, a suitable wavelet basis is preliminarily selected (as shown in Table 2), and then the optimal order and optimal decomposition level of the wavelet basis are selected according to the cost function (Shannon entropy); that is, the Shannon entropy values ​​of different preliminarily selected wavelet bases and different decomposition orders are first calculated, and then the wavelet basis and decomposition level that minimize the Shannon entropy are selected; Defining a sequence x ={ x i The entropy of} is: M ( x )=-Σ p i log2( p i ) in, p i =(| x i | 2 ) / (|| x || 2 ), and when p =0, p log2( p )=0. Information is semi-additive, so we introduce an additive function λ ( x )=-Σ| x k | 2 log2| x k | 2 ,but M ( x ) can be expressed as: M ( x )=[ λ ( x ) / || x || 2 ]+log2|| x || 2 so λ ( x ) is minimum, M ( x ) is also the smallest. The smaller the cost function value is, the more effective it is, and the corresponding wavelet basis is the optimal wavelet basis.

[0017] The types of wavelet basis functions commonly used in the prior art are shown in Table 2.

[0018] Table 2 - Common wavelet basis function types 4) The original signal x ( t )conduct j Layer wavelet packet decomposition (WPT), such as Figure 3 , we get 2 j Wavelet packet coefficients : 5) Reconstruct each wavelet packet coefficient to obtain the corresponding baseline-calibrated acceleration signal x ( t )'s wavelet packet component; represents the wavelet basis function After the signal is decomposed by wavelet packets, the wavelet basis functions of each layer are orthogonal to each other, that is, when m≠n, there is Thus, the original signal x ( t ) is decomposed into 2 j The sum of wavelet packet components: 6) Find the energy proportion of the sub-band of the wavelet packet decomposition signal: Original signal x ( t )energy E e Defined as: x j m (t) and x j n (t) Represents a set of wavelet packet components decomposed from mutually orthogonal wavelet basis functions Considering the orthogonality of the wavelet packet function, we get: The sub-band energy of each measuring point of the screw pile is normalized and the first i [The meanings of superscript i and subscript i in steps 4) to 7) are consistent] The distribution of sub-band energy along the structure is: Considering the influence of noise energy, the energy ratio change rate of the i-th sub-band of the screw pile before and after damage is defined as ERS i for: The superscript N represents the ERS before pile damage. i The superscript Y indicates the ERS after the pile is damaged. i value 7) Considering that some sub-bands have no energy distribution or account for a small proportion, they have little impact on the energy characteristics of the vibration signal. Therefore, only the sub-band energy proportion greater than 5% is selected as the effective sub-band characteristic value, and then the structural vulnerability judgment index Δ is established according to the root mean square value of the effective sub-band energy ratio change rate of the screw pile. ERMS : Example

[0019] Dynamic damage identification of railway screw pile composite foundation based on wavelet packet energy analysis.

[0020] Step 1: This embodiment uses the method of excitation simulation test to study the dynamic damage deformation identification of railway screw pile composite foundation under train load, designs the screw pile acceleration sensor layout plan, and arranges the acceleration sensor 9 according to the design plan. The acceleration sensor 9 is attached to the pile body of the screw pile 4, and the excitation simulation test model and the layout of the acceleration sensor therein are as follows. Figure 4 , Figure 5 and Figure 6As shown. The excitation simulation test model includes a door frame-shaped reaction steel frame 1 and an assembled model box 5. The model box 5 includes four side walls with a length and height of 1m and a certain thickness. Two adjacent side walls are fixedly connected by a connecting frame 7. The interior of the model box 5 is cylindrical, and the shape of the cylinder is a regular quadrangular prism. All inner walls of the model box 5 are covered with foam boards 6 with a thickness of 5cm. The inner walls of all foam boards 6 are covered with polyvinyl chloride films to form film layers 8. The distance L between the film layers 8 covered on the inner walls of the model box 5 is 90cm. The space surrounded by the film layers 8 is filled with soil to form a soil foundation 10. A plurality of screw piles 4 are arranged in the soil foundation 10. The top surfaces of all screw piles 4 are flush with the top surface of the soil foundation 10. The soil foundation 10 and all screw piles 4 constitute a pile-soil foundation model. A cushion layer 11 with a thickness of 5 cm is provided on the soil foundation 10; the model box 5 is located below the crossbeam in the reaction steel frame 1, and an exciter 2 is installed on the side wall of the crossbeam facing the model box 5, and a bearing plate 3 is provided between the exciter 2 and the cushion layer 11.

[0021] Step 2: The soil foundation 10 is composed of a fine sand layer 12 and a silty clay layer 13 arranged in sequence from bottom to top. The fine sand layer 12 is filled with fine sand and has a thickness of 70 cm; the silty clay layer 13 is filled with silty clay and has a thickness of 20 cm.

[0022] The screw piles 4 form a matrix of three rows and three columns in the soil foundation 10, and each row and each column are provided with three screw piles 4; the axis of the screw pile 4 located at the center of the matrix coincides with the central axis of the regular quadrangular prism, and the distance between two adjacent screw piles 4 in the same row and the distance between two adjacent screw piles 4 in the same column are both L2, L2=20cm; the distance between the side wall of the model box 5 and the screw pile 4 adjacent to the side wall is L1, L1=25cm. The pitch of the threaded part on the screw pile 4 is 4cm, and the threaded part is entirely located in the fine sand layer 12.

[0023] Six acceleration sensors 9 are attached to the screw pile 4 located at the center of the matrix. The six acceleration sensors 9 are numbered PA1, PA2, PA3, PA4, PA5 and PA6 from top to bottom.

[0024] All acceleration sensors 9 are connected to the DHDAS data acquisition system signal, the vibrator 2 is connected to the hydraulic control system, and the hydraulic control system is connected to the PC dynamic load input system and the hydraulic system.

[0025] Before the formal test begins, check whether each test system is operating normally and test whether each acceleration sensor 9 functions normally. After confirming that each instrument is normal, set the loading conditions in the PC dynamic load input system according to Table 1. Start the hydraulic system, turn on the DHDAS data acquisition system, and finally apply the load through the PC dynamic load input system to start the test.

[0026] Step 3: 1) Initial processing of the acceleration data collected by the DHDAS data acquisition system, mainly including data filtering and baseline correction.

[0027] 2) The filtering process is implemented by programming the LMS adaptive filtering program in MATLAB software, and the baseline correction is implemented by programming the B-spline fitting algorithm in MATLAB software. After filtering and baseline correction of the collected acceleration data, the time domain of the acquired signal is obtained. Figure 7 The time course graph shown and the spectrum characteristics of the obtained signal Figure 8 The Fourier spectrum curve is shown.

[0028] 3) According to the spectrum characteristics, time domain characteristics and smoothness characteristics of the pile acceleration signal, a suitable wavelet basis is preliminarily selected (as shown in Table 2), and then the optimal order and optimal decomposition level of the wavelet basis are selected according to the cost function (Shannon entropy).

[0029] 4) Daubechies (db wavelet) is selected as the wavelet basis through calculation, N is 15, that is, db15 wavelet basis function is selected to perform 3-layer wavelet packet decomposition. The time course curve of db15 wavelet is as follows: Fig. 9 .

[0030] 5) Perform WPT decomposition on the filtered and baseline-corrected data of the time-history curve of the pile acceleration measurement point, and decompose the original signal X(t) into several wavelet packet components.

[0031] 6) Integrate the wavelet packet components in the time domain to obtain the energy spectrum of the signal of different sub-band components. Normalize the sub-band energy of each measuring point to obtain the sub-band energy ratio of the wavelet packet decomposition signal as follows: Fig. 9 As shown: Extract the effective sub-band energy ratio root mean square change rate index (Δ ERMS ), the PMSA of each measuring point is plotted under different train speed conditions with the relative elevation of the pile body as follows Fig.10 shown.

[0032] according to Fig.14 It can be seen that the various parts of the pile body are in Δ ERMS The index value increases with the increase of train speed, and there is a critical speed of about 280km / h. ERMS The index value is the largest, and the vibration of trains at this speed is more destructive to the composite foundation pile. ERMS The distribution of damage index values ​​along the pile body can divide the pile body into three different damage segments: low-vulnerability segment, medium-vulnerability segment and vulnerable segment. ERMSThe maximum index value exceeds 0.6, indicating that under the same dynamic load conditions, the area near the variable cross-section of the screw pile is a vulnerable and weak section; at the pile top (PA1) and the middle of the threaded section (PA5), Δ ERMS The maximum value of the index is in the range of 0.3-0.5, which is a moderately vulnerable section; the maximum EDIS index of the middle and lower part of the straight rod section (PA2, PA3) and the pile bottom (PA6) does not exceed 0.3, which is a low vulnerable section. The screw pile is a variable-section special-shaped pile. The change in the cross-section of the pile body causes a sudden change in the vibration energy transmitted from the pile top. This energy variation in a certain part of the structure can easily cause structural damage. Therefore, the variable cross-section of the screw pile is more susceptible to damage under the action of train loads, and optimization can be considered when designing the railway screw pile composite foundation.

Claims

1. A method for identifying dynamic damage and deformation of railway screw piles suitable for indoor model tests, characterized in that: The identification method is specifically carried out in the following steps: Step 1: Design a layout scheme for the acceleration sensor on each screw pile in the indoor train simulation vibration test model. Design it according to the principle of arranging the acceleration sensors at equal intervals in the same cross-section pile section, and arrange the acceleration sensor on each screw pile of the test model according to the design scheme; Step 2: fill the train simulation vibration test model according to the test requirements, and after the filling is completed, load the sine wave test model according to the loading condition requirements, and obtain the acceleration data of the screw pile under the action of the sine wave through each acceleration sensor; The acceleration sensor is attached to the pile body of the screw pile. The indoor train simulation vibration test model also includes a door frame-shaped reaction steel frame and an assembled model box. The model box is a regular quadrangular prism. All inner walls of the model box are covered with foam boards. The inner walls of the foam boards are covered with polyvinyl chloride films to form film layers. The space surrounded by the film layers is filled with soil to form a land foundation. A plurality of screw piles are arranged in the land foundation. The top surfaces of all the screw piles are flush with the top surface of the land foundation. The land foundation and all the screw piles constitute a pile-land foundation model. A cushion layer with a thickness of 5-7 cm is laid on the soil foundation; the model box is located below the crossbeam in the reaction steel frame, and a vibrator is installed on the side wall of the crossbeam facing the model box, and a bearing plate is provided between the vibrator and the cushion layer; the soil foundation includes a fine sand layer and a silty clay layer from bottom to top, the fine sand layer is filled with fine sand and has a thickness of 65-75 cm; the silty clay layer is filled with silty clay and has a thickness of 15-25 cm; The screw piles form a matrix of three rows and three columns in the soil foundation, and each row and each column are provided with three screw piles; the axis of the screw pile located at the center of the matrix coincides with the central axis of the model box; Step 3: Process the acquired screw pile acceleration data 1) Filtering: Perform LMS adaptive filtering on the original acceleration data obtained in step 2) through an adaptive filter; 2) Write a B-spline fitting algorithm in MATLAB software to remove the baseline drift of the acceleration data after adaptive filtering, that is, to correct the baseline of the acceleration data after adaptive filtering in 1) and obtain the acceleration data signal after baseline calibration. x ( t ); 3) Based on the acceleration data signal in 2) x ( t )'s spectrum characteristics, time domain characteristics, and smoothness characteristics are used to preliminarily select a wavelet basis that meets the requirements, and then select the optimal order and optimal decomposition level of the wavelet basis according to the cost function, where the cost function is the Shannon entropy; that is, first calculate the Shannon entropy values ​​of the preliminarily selected different wavelet bases and different decomposition orders, and then select the wavelet basis and decomposition level that minimize the Shannon entropy; 4) The original acceleration data obtained in step 2 is j Layer wavelet packet decomposition WPT, get 2 j Wavelet packet coefficients Where: i , j , k are modulation, scale and translation parameters respectively; is the wavelet packet function; t For time; 5) Reconstruct each wavelet packet coefficient to obtain the corresponding baseline-calibrated acceleration data signal x ( t )'s wavelet packet component; In this way, the original acceleration data signal obtained in step 2 x ( t ) is decomposed into 2 j The sum of wavelet packet components: 6) Find the energy proportion of the sub-band of the wavelet packet decomposition signal: The original acceleration data signal obtained in step 2 x ( t )energy E e Defined as: Considering the orthogonality of the wavelet packet function, we get: The sub-band energy of each measuring point of the screw pile is normalized and the first i The distribution of sub-band energy along the structure is: Where M is the number of measuring points, E n i is the energy of the ith sub-band at the nth measurement point of the structure; Considering the influence of noise energy, the energy ratio change rate of the i-th sub-band of the screw pile before and after damage is defined as ERS i for: In the formula: superscript N represents static state, superscript Y represents loaded state; 7) Considering that some sub-bands have no energy distribution or account for a small proportion, they have little effect on the energy characteristics of the vibration signal; therefore, the sub-band energy proportion greater than 5% is selected as the effective sub-band characteristic value, and then the structural vulnerability judgment index Δ is established according to the root mean square value of the change rate of the effective sub-band energy ratio of the screw pile ERMS : Where: m is the number of effective sub-bands, ERS for m The average value of the energy ratio change rate of each sub-band; wherein the superscript i and the subscript i in steps 4) to 7) have the same meaning.

2. The method for identifying dynamic damage and deformation of railway screw piles suitable for indoor model testing according to claim 1 is characterized in that: The processing method of 2) adaptive filtering in step 3 is as follows: If the order of adaptive filtering is N , the adaptive filter coefficients are w , the input signal sequence is X , then the output y ( n )for: e ( n )= d ( n )- y ( n ) In the formula, d ( n ) is the expected signal; e ( n ) is the error signal; n Indicates a moment, w i ( n )express n The adaptive filter coefficients at time , x ( n - i ) represents the sampling value of the input signal at time ni; make W ( n )=[ w 0, w 1,…, w i-1 ] T , X ( n )=[ x 1, x 2,…, x i-1 ] T The output of the adaptive filter is y ( n ) is written in matrix form: y ( n )= X ( n ) T W ( n )= W ( n ) T X ( n ) The error signal is written as: e ( n )= d ( n )- y ( n )= d ( n )- W T ( n ) X ( n ) Define the cost function: J ( w )= E [| e ( n )| 2 ]= E [| d ( n )- y ( n )| 2 ]= E [|( d ( n )- W T ( n ) X ( n )| 2 ] When the cost function reaches the minimum value, it is considered that the optimal filtering is achieved. The adaptive filtering that meets this condition is called the least mean square adaptive filtering LMS. For minimum mean square adaptive filtering, it is necessary to determine the filter coefficients that minimize the mean square error, and use the gradient descent method to solve it; The iterative formula for the filter coefficient vector is: W ( n+ 1)= W ( n )+{[ μ (-▽ J ( w )] / 2} Where: μ is the step size factor; J ( w ) is the gradient of the cost function; W ( n+ 1) is the filter coefficient after iterative update, W ( n ) is the filter coefficient before iterative update; Since the instantaneous gradient -2 X ( n ) e ( n ) is an unbiased estimate of the true gradient value. In practical applications, the instantaneous gradient is used instead of the true gradient, that is, ▽ J ( w )=-2 X ( n )[ d ( n )- W T ( n ) X ( n ) ]=-2 X ( n ) e ( n ) W ( n+ 1)= W ( n )+ μe ( n ) X ( n ) Through step-by-step iteration, the optimal filter coefficients are obtained to achieve adaptive filtering of the input signal.

3. The method for identifying dynamic damage and deformation of railway screw piles suitable for indoor model testing according to claim 2 is characterized in that: The method for the optimal order and optimal decomposition level of the wavelet basis in step 2 is as follows: Defining a sequence x ={ x i The entropy of} is: M ( x )=-Σ p i log2( p i ) in, p i =(| x i | 2 ) / (|| x || 2 ), and when p =0, p log2( p )=0; information is semi-additive, introducing additive functions λ ( x )=-Σ| x k | 2 log2| x k | 2 ,but M ( x ) can be expressed as: M ( x )=[ λ ( x ) / || x || 2 ]+log2|| x || 2 so λ ( x ) is minimum, M ( x ) is also the smallest; the smaller the cost function value is, the more effective it is, and the corresponding wavelet basis is the optimal wavelet basis.

Citation Information

Patent Citations

  • A wavelet packet energy spectrum damage identification method

    CN109815940A

  • Ammunition firing vibration signal feature extraction and detection device and method

    CN116595357A

  • A method for automatically detecting free vibration response of high-speed railway bridge for modal identification

    US20200284687A1