A method for refined extraction of time-frequency features of rotating components of the running gear of a rail vehicle

Abstract

This invention discloses a method for refined extraction of time-frequency features of rotating components of a rail vehicle's running gear, comprising the following steps: signal acquisition; high-order polynomial-ratio tracking signal resampling; Fourier intrinsic mode decomposition based on energy entropy optimization: obtaining M Fourier intrinsic modes from the Fourier intrinsic mode decomposition; evaluating the signal using energy entropy; selecting the algorithm with the lowest energy entropy as the preferred result; refined extraction of non-stationary features based on sparse measure; setting a number K for optimizing Fourier intrinsic modes, and summing the first K Fourier intrinsic mode functions to achieve refined extraction of non-stationary features. The high-order polynomial-ratio tracking method proposed in this invention can achieve equal-angle resampling of signals with varying frequency characteristics; it uses information entropy to optimize the HTL and LTH algorithms of the Fourier intrinsic mode decomposition method; and it refines and accurately extracts the non-stationary features of rotating components of the rail vehicle's running gear.

Description

Technical Field

[0001] This invention relates to the field of rail transit technology, specifically a method for refined extraction of time-frequency features of rotating components of the running gear of rail vehicles. Background Technology

[0002] Non-stationary processes are a typical characteristic of vibration signals from rotating components of the running gear in rail vehicles, and effectively extracting their feature information is of great significance for dynamic control and fault diagnosis. Traditional feature extraction methods use quadratic or cubic polynomials for order tracking to achieve signal resampling, which easily leads to a decrease in the accuracy of order spectrum calculation for rapidly changing signals. At the same time, methods such as empirical mode decomposition and variational mode decomposition are limited by the lack of sufficient mathematical theory, mode mixing, and insufficient adaptive ability, and are not effective in the refined extraction of non-stationary features. Summary of the Invention

[0003] The purpose of this invention is to provide a method for refined extraction of time-frequency features of rotating components of the running gear of a rail vehicle, so as to solve the problems mentioned in the background art.

[0004] To achieve the above objectives, the present invention provides the following technical solution:

[0005] A method for refined extraction of time-frequency features of rotating components of a rail vehicle's running gear includes the following steps:

[0006] (1) Signal acquisition: The vibration acceleration of the rotating parts of the running section is acquired, and the traction motor speed information is acquired by photoelectric encoder;

[0007] (2) High-order polynomial order tracking signal resampling: reconstruct the equal-phase resampling according to the relationship of the nth order polynomial. Based on the phase relationship, linear interpolation is used to obtain the equal-phase reconstructed signal.

[0008] (3) Fourier intrinsic mode decomposition based on energy entropy optimization: The Fourier intrinsic mode decomposition yields M Fourier intrinsic modes. According to the different order of solving the intrinsic modes, the Fourier decomposition can be divided into the LTH algorithm calculated from low frequency to high frequency and the HTL algorithm calculated from high frequency to low frequency. The signal is evaluated using energy entropy. The algorithm with the smaller energy entropy is selected as the optimal result.

[0009] (4) Refined extraction of non-stationary features based on sparse measure: Based on the multiple single-component Fourier intrinsic mode functions obtained by Fourier intrinsic mode decomposition, calculate the sparse measure of each mode and sort them in descending order according to the sparse measure values.

[0010] (5) Set the number of Fourier intrinsic modes K to be selected, and select the first K Fourier intrinsic mode functions to be added together to achieve refined extraction of non-stationary features.

[0011] Furthermore: the phase relationship is expressed as θ(t) = b0 + b1t + b2t 2 +…+b n t n The coefficients are expressed by the following formula:

[0012] calculate:

[0013]

[0014] In the formula, (θ1,t1), (θ2,t2),…, (θ n ,t n ) are n different sampling points.

[0015] Furthermore: Following the Fourier intrinsic mode decomposition method, the analytic function... Represented as:

[0016]

[0017] In the formula, a i (t) is the instantaneous amplitude function, φ i (t) is the instantaneous phase function;

[0018] Based on the Fourier series, the resampled signal Represented as:

[0019]

[0020] In the formula, c k =(a k -jb k ); a k and b k These are the Fourier expansion coefficients.

[0021] Furthermore: M Fourier intrinsic mode FIMFs obtained from Fourier intrinsic mode decomposition i (t) can be represented as:

[0022]

[0023] Furthermore: the energy entropy is used to evaluate the signal as follows:

[0024]

[0025] In the formula, p i p represents the percentage of the energy of the i-th intrinsic mode component in the total signal energy. i =E i / E,

[0026] Compared with the prior art, the beneficial effects of the present invention are:

[0027] 1) A high-order polynomial order ratio tracking method is proposed, which can realize the resampling of signals with variable frequency characteristics at the same angle.

[0028] 2) The HTL and LTH algorithms of the Fourier intrinsic mode decomposition method are optimized using information entropy.

[0029] 3) Extract the non-stationary characteristics of rotating components of the running gear of rail vehicles in a precise and accurate manner. Attached Figure Description

[0030] To more clearly illustrate the technical solutions of the embodiments of this disclosure, the accompanying drawings used in the embodiments will be briefly described below. These drawings are incorporated in and constitute a part of this specification. They illustrate embodiments conforming to this disclosure and, together with the specification, serve to explain the technical solutions of this disclosure. It should be understood that the following drawings only show some embodiments of this disclosure and should not be considered as limiting the scope. Those skilled in the art can obtain other related drawings based on these drawings without creative effort.

[0031] Figure 1 This is a flowchart of a method for refined extraction of time-frequency features of rotating components of the running gear of a rail vehicle.

[0032] Figure 2 This is a waveform diagram of the reconstructed time-domain signal in an embodiment of the present invention.

[0033] Figure 3 This is a comparison chart of the refined feature extraction effects in embodiments of the present invention. Detailed Implementation

[0034] The technical solution of this patent will be further described in detail below with reference to specific embodiments. The detailed description of the embodiments of this disclosure provided in the accompanying drawings is not intended to limit the scope of the claimed disclosure, but merely to illustrate selected embodiments of this disclosure. All other embodiments obtained by those skilled in the art based on the embodiments of this disclosure without inventive effort are within the scope of protection of this disclosure.

[0035] Please see Figure 1 A method for refined extraction of time-frequency features of rotating components of the running gear of a rail vehicle, comprising the following steps:

[0036] 1) Signal acquisition: The vibration acceleration x(t) of the rotating part of the traveling section is acquired, and the traction motor speed information ω(t) is acquired using a photoelectric encoder.

[0037] 2) High-order polynomial tracking signal resampling: Reconstructing equal-phase resampling according to the relationship of an nth-order high-order polynomial, the phase relationship can be expressed as θ(t)=b0+b1t+b2t 2 +…+b n t n The coefficients in the formula can be calculated using the following formula.

[0038]

[0039] In the formula, (θ1,t1), (θ2,t2),…, (θ n ,t n Let n be the n distinct sampling points. Using matrix operations, the solutions for each coefficient can be obtained. Based on the phase relationship, linear interpolation is used to obtain the equal-phase reconstructed signal.

[0040] 3) Fourier intrinsic mode decomposition based on energy entropy optimization: According to the Fourier intrinsic mode decomposition method, the analytical function is... Represented as

[0041]

[0042] In the formula, a i (t) is the instantaneous amplitude function, φ i (t) is the instantaneous phase function.

[0043] Based on the Fourier series, the resampled signal Represented as

[0044]

[0045] In the formula, c k =(a k -jb k ); a k and b k These are the Fourier expansion coefficients.

[0046] The M Fourier intrinsic modes FIMF obtained by Fourier intrinsic mode decomposition i (t) can be represented as

[0047]

[0048] Depending on the order in which intrinsic modes are solved, Fourier decomposition can be divided into the LTH algorithm (calculating from low to high frequencies) and the HTL algorithm (calculating from high to low frequencies). To optimize the algorithm, energy entropy is used to evaluate the signal.

[0049]

[0050] In the formula, p i p represents the percentage of the energy of the i-th intrinsic mode component in the total signal energy. i =E i / E,

[0051] Ultimately, the algorithm with the lowest energy entropy was chosen as the optimal result.

[0052] 4) Refined extraction of non-stationary features based on sparse measure: Based on multiple single-component Fourier intrinsic mode functions (FIMFs) obtained from Fourier intrinsic mode decomposition. i (t), calculate the sparsity measure of each mode, and sort them in descending order according to the sparsity measure values.

[0053] 5) Set the number K of Fourier intrinsic modes to be optimized. Finally, select the first K Fourier intrinsic mode functions and sum them to achieve refined extraction of non-stationary features.

[0054] For example, such as Figure 2 The order spectrum waveform of the reconstructed signal shown shows that the first three intrinsic modes extracted are exactly f. m 2f m and 3f m The order spectrum of the reconstructed signal, consisting of the three meshing frequencies and their nearby sidebands, contains the main characteristic information of gear operation and effectively filters out irrelevant information such as noise in the vibration signal.

[0055] like Figure 3 As shown, by comparing the analysis results of the original signal and the reconstructed signal, it can be found that the axis frequencies fo and 2f were extracted from the reconstructed signal. o This indicates that the periodic impact caused by localized fault spalling has been effectively extracted.

Claims

1. A method for fine extraction of time-frequency features of a rolling stock running gear rotating component, characterized in that, Comprise the following steps: (1) Signal acquisition: collect the vibration acceleration of the rotating part of the running part, and collect the speed information of the traction motor by using an optical encoder; (2) High-order polynomial order tracking signal resampling: reconstruct the equal-phase resampling according to the relationship of the n-order high-order polynomial, and obtain the equal-phase reconstruction signal by using the linear interpolation method based on the phase relationship; (3) Fourier intrinsic mode decomposition based on energy entropy: M Fourier intrinsic modes obtained by Fourier intrinsic mode decomposition, according to the difference of the solution sequence of the intrinsic mode, the Fourier decomposition is divided into LTH algorithm from low frequency to high frequency calculation and HTL algorithm from high frequency to low frequency calculation; the energy entropy is used to evaluate the signal; the algorithm with small energy entropy is selected as the result; (4) Non-stationary feature fine extraction based on sparse measure: based on the multiple single-component Fourier intrinsic mode functions obtained by Fourier intrinsic mode decomposition, the sparse measure of each mode is calculated, and the sparse measure is sorted in descending order according to the numerical value; (5) Set the number of Fourier intrinsic modes K, select the first K Fourier intrinsic mode functions to add, and realize the fine extraction of non-stationary features.

2. The method according to claim 1, characterized in that, The phase relationship is expressed as θ(t) = b0+ b1t + b2t2 2 +L+b n t n , each coefficient being calculated by the following formula: In the formula, (θ1, t1), (θ2, t2), …, (θn, tn) are respectively n different sampling points. n n ) are respectively n different sampling points.​ 3. The method according to claim 1, characterized in that, According to the Fourier intrinsic mode decomposition method, the analytical function is expressed as: where a i (t) is an instantaneous amplitude function, and φ i (t) is an instantaneous phase function. According to the Fourier series, the resampled signal is expressed as: where c k = (a k - jb k ) ; a k and b k are Fourier expansion coefficients.

4. The method according to claim 1, characterized in that, M Fourier intrinsic modes, FIMF, obtained by Fourier intrinsic mode decomposition i (t) is represented by:

5. The method according to claim 1, characterized in that, The energy entropy for evaluating the signal is represented as where p i is the percentage of the energy of the i-th eigenmode component in the total signal energy, p i = E i / E,

Images