Turnout vibration time delay analysis method and device

Abstract

The present invention discloses a method and device for analyzing turnout vibration delay, relating to the field of railway engineering technology. The method comprises: performing EEMD decomposition on an ideal simulation signal and an actual turnout signal to extract the first intrinsic modal component of the ideal simulation signal and the first intrinsic modal component of the actual turnout signal; performing time-redistribution multi-synchronous compression transformation on the first intrinsic modal components of the ideal simulation signal and the actual turnout signal to obtain a first time-frequency spectrum of the first intrinsic modal component of the ideal simulation signal and a second time-frequency spectrum of the first intrinsic modal component of the actual turnout signal; performing time-frequency slicing on the first time-frequency spectrum and the second time-frequency spectrum at the frequency where the vibration impact is strongest and most concentrated, respectively, to obtain a first time-frequency slice of the first time-frequency spectrum and a second time-frequency slice of the second time-frequency spectrum; and calculating the cross-correlation between the first time-frequency slice and the second time-frequency slice to obtain time delay information of the actual turnout signal. The present invention can improve the accuracy of turnout vibration delay analysis.

Description

Technical Field

[0001] The present invention relates to the technical field of railway engineering, and in particular to a method and device for analyzing turnout vibration delay. Background Art

[0002] This section is intended to provide a background or context to the embodiments of the invention that are recited in the claims. No statement herein is admitted to be prior art by virtue of its inclusion in this section.

[0003] Railway turnouts are connecting devices that allow trains to transfer from one track to another or cross over to another. They are the only way for trains to change tracks and are often located where trains enter and exit stations. At turnouts, including welded joints, insulated joints, heart rails, and point rails, the inherent unevenness of the structure causes high-frequency vibration shocks between the train wheels and rails when a train passes. Therefore, compared to the mainline track, the wheel-rail interaction in the turnout is more intense, specifically manifested in a higher frequency of wheel-rail impacts per unit length and a greater amplitude of impact force, which causes friction and wear on the rail components in the turnout. Therefore, the rails in the turnout section are more prone to deterioration or damage than those in the mainline section, leading to problems such as poor joints, abrasions and chipping, and fish-scale marks. Therefore, the analysis and evaluation of the short-wave impact state of the rails in the turnout section is an important part of the comprehensive turnout status assessment system. Currently, the high-speed integrated train wheel-rail force and axlebox acceleration detection system can provide targeted evaluation of the short-wave impact state of the rail surface. However, due to sensor signal transmission delays or deviations between the actual wheel diameter value and the system-set wheel diameter value due to wheel turning and repair, there is a deviation between the mileage of the sensor-collected signal and the actual mileage. In the turnout section, it is impossible to match the track structure with the above dynamic detection data one-to-one, resulting in the application effect of the above data in the turnout section being less than ideal. Therefore, accurately locating the actual position of the turnout in the train detection mileage is a prerequisite for the detailed analysis of the short-wave state of the track in the turnout section and the subsequent formulation of targeted maintenance and repair strategies.

[0004] The vibration response analysis within a turnout requires the precise extraction of impacts caused by structural irregularities such as joints and switch cores, against a backdrop of complex impacts and noise interference. Traditional time-frequency analysis methods, such as the short-time Fourier transform (STFT), synchronized compression transform (SST), and synchronized extraction transform (SET), are limited by window length. Even if an impact is observed on the time-frequency spectrum, the time resolution of the impact is still limited, making it difficult to accurately locate the mileage of the impact within the turnout. Other technicians have proposed the time-redistributed compression transform (TSST) based on the STFT, which rearranges the time-frequency coefficients in the time direction and improves the time resolution of the time-frequency representation. However, the time-frequency results of the TSST transform still require manual identification to represent the impact within the turnout. Moreover, the time-frequency representation cannot yet determine the deviation between the mileage of the sensor-collected signal and the actual mileage. Summary of the Invention

[0005] An embodiment of the present invention provides a turnout vibration time delay analysis method for improving the accuracy of turnout vibration time delay analysis. The method includes:

[0006] Performing EEMD decomposition on the ideal simulation signal and the collected actual turnout signal to extract the first intrinsic modal component of the ideal simulation signal and the first intrinsic modal component of the actual turnout signal;

[0007] Performing time-redistribution multi-synchronous compression transformation on the first connotation modal component of the ideal simulation signal and the first connotation modal component of the actual turnout signal to obtain a first time-frequency spectrum of the first connotation modal component of the ideal simulation signal and a second time-frequency spectrum of the first connotation modal component of the collected actual turnout signal;

[0008] At the frequency where the vibration impact is the strongest and most concentrated, time-frequency slicing is performed on the first time-frequency spectrum and the second time-frequency spectrum to obtain a first time-frequency slice of the first time-frequency spectrum and a second time-frequency slice of the second time-frequency spectrum;

[0009] The cross-correlation between the first time-frequency slice and the second time-frequency slice is calculated to obtain the time delay information of the actual turnout signal.

[0010] An embodiment of the present invention further provides a turnout vibration time delay analysis device for improving the accuracy of turnout vibration time delay analysis, the device comprising:

[0011] A first processing module is used for performing EEMD decomposition on the ideal simulation signal and the collected actual turnout signal to extract the first intrinsic modal component of the ideal simulation signal and the first intrinsic modal component of the actual turnout signal;

[0012] The second processing module is used to perform time redistribution multi-synchronous compression transformation on the first intrinsic modal component of the ideal simulation signal and the first intrinsic modal component of the actual turnout signal to obtain a first time-frequency spectrum of the first intrinsic modal component of the ideal simulation signal and a second time-frequency spectrum of the first intrinsic modal component of the collected actual turnout signal;

[0013] The third processing module is used to perform time-frequency slicing on the first time-frequency spectrum and the second time-frequency spectrum at the frequency where the vibration impact is the strongest and most concentrated, to obtain a first time-frequency slice of the first time-frequency spectrum and a second time-frequency slice of the second time-frequency spectrum;

[0014] The fourth processing module is used to calculate the cross-correlation between the first time-frequency slice and the second time-frequency slice to obtain the time delay information of the actual turnout signal.

[0015] An embodiment of the present invention further provides a computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the above-mentioned turnout vibration delay analysis method when executing the computer program.

[0016] An embodiment of the present invention further provides a computer-readable storage medium storing a computer program. When the computer program is executed by a processor, the above-mentioned turnout vibration delay analysis method is implemented.

[0017] An embodiment of the present invention further provides a computer program product, which includes a computer program. When the computer program is executed by a processor, the above-mentioned turnout vibration delay analysis method is implemented.

[0018] In an embodiment of the present invention, an ensemble empirical mode decomposition (EEMD) is performed on an ideal simulation signal and a collected actual turnout signal to extract a first intrinsic modal component of the ideal simulation signal and a first intrinsic modal component of the actual turnout signal; a time redistribution multi-synchronous compression transformation is performed on the first intrinsic modal component of the ideal simulation signal and the first intrinsic modal component of the actual turnout signal to obtain a first time-frequency spectrum of the first intrinsic modal component of the ideal simulation signal and a second time-frequency spectrum of the first intrinsic modal component of the collected actual turnout signal; time-frequency slices are performed on the first time-frequency spectrum and the second time-frequency spectrum at the frequency where the vibration impact is strongest and most concentrated, respectively, to obtain a first time-frequency slice of the first time-frequency spectrum and a second time-frequency slice of the second time-frequency spectrum; the cross-correlation between the first time-frequency slice and the second time-frequency slice is calculated to obtain time delay information of the actual turnout signal, thereby improving the accuracy of turnout vibration delay analysis. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative work. In the drawings:

[0020] Figure 1 This is a flow chart of a turnout vibration delay analysis method provided in an embodiment of the present invention;

[0021] Figure 2 A flow chart of a method for performing time-redistribution multi-synchronous compression transformation on a first intrinsic modal component of an ideal simulation signal and a first intrinsic modal component of an actual turnout signal, respectively, to obtain a first time-frequency spectrum of the first intrinsic modal component of the ideal simulation signal and a second time-frequency spectrum of the first intrinsic modal component of the acquired actual turnout signal, provided in an embodiment of the present invention;

[0022] Figure 3 A comparison diagram of an actual turnout signal collected in the first half of an up line provided in an embodiment of the present invention and the first intrinsic modal component IMF1 after EEMD decomposition of the actual turnout signal collected in the first half of an up line;

[0023] Figure 4 A comparison diagram of the first intrinsic modal component IMF1 of an actual turnout signal collected in the first half of an up line and an ideal simulation signal provided in an embodiment of the present invention;

[0024] Figure 5 A comparison diagram of a noisy simulation signal and an ideal simulation signal provided in an embodiment of the present invention;

[0025] Figure 6 A comparison diagram of a time-frequency slice of a noisy simulated signal and a time-frequency slice of an ideal simulated signal provided in an embodiment of the present invention;

[0026] Figure 7 This is an example diagram of the cross-correlation between a noisy simulated signal and an ideal simulated signal time-frequency slice provided in an embodiment of the present invention;

[0027] Figure 8 A comparison diagram of a time-frequency slice of a noisy simulated signal after delay correction and a time-frequency slice of an ideal simulated signal provided in an embodiment of the present invention;

[0028] Figure 9 A comparison diagram of the time-frequency slices of an actual turnout signal collected in the first half of an up line and the time-frequency slices of an ideal simulation signal provided in an embodiment of the present invention;

[0029] Figure 10This is an example diagram of the cross-correlation between the time-frequency slices of an actual turnout signal collected in the first half of an up line and the time-frequency slices of an ideal simulation signal provided in an embodiment of the present invention;

[0030] Figure 11 A comparison diagram of the time-frequency slices of an actual turnout signal collected in the first half of an up line provided in an embodiment of the present invention after time delay correction and the time-frequency slices of an ideal simulation signal;

[0031] Figure 12 This is an example diagram of the cross-correlation between the time-frequency slices of an actual turnout signal collected in the second half of an up line and the time-frequency slices of an ideal simulation signal provided in an embodiment of the present invention;

[0032] Figure 13 A comparison diagram of the time-frequency slices of an actual turnout signal collected in the second half of an up line provided in an embodiment of the present invention after time delay correction and the time-frequency slices of an ideal simulation signal;

[0033] Figure 14 Schematic diagram of a turnout vibration time delay analysis method and device provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0034] To make the purpose, technical solutions and advantages of the embodiments of the present invention more clear, the embodiments of the present invention are further described in detail below with reference to the accompanying drawings. Here, the exemplary embodiments of the present invention and their descriptions are used to explain the present invention, but are not intended to limit the present invention.

[0035] The acquisition, storage, use, and processing of data in this application's technical solution comply with relevant national laws and regulations.

[0036] The term "and / or" herein simply describes an association relationship, indicating that three relationships can exist. For example, A and / or B can represent the existence of A alone, the simultaneous existence of A and B, and the existence of B alone. In addition, the term "at least one" herein refers to any combination of at least two of any one or more of a plurality of items. For example, "at least one of A, B, and C" can represent any one or more elements selected from the set consisting of A, B, and C.

[0037] In the description of this specification, the terms "include", "including", "have", "contain", etc. are all open terms, which mean including but not limited to. The descriptions with reference to the terms "one embodiment", "a specific embodiment", "some embodiments", "for example", etc. mean that the specific features, structures or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present application. In this specification, the schematic expressions of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures or characteristics described can be combined in a suitable manner in any one or more embodiments or examples. The order of steps involved in each embodiment is used to schematically illustrate the implementation of the present application, and the order of steps therein is not limited and can be appropriately adjusted as needed.

[0038] Research has found that railway turnouts are connecting devices that allow trains to transfer from one track to or cross over to another. They are the only way for trains to change tracks and are often located where trains enter and exit stations. At turnouts, including welded joints, insulated joints, heart rails, and point rails, due to the inherent unevenness of the structure, when a train passes, it will cause high-frequency vibration impacts between the train wheels and rails. Therefore, compared to the mainline track, the wheel-rail interaction in the turnout is more intense, specifically manifested in a higher frequency of wheel-rail impacts per unit length and a greater amplitude of impact force, which causes friction and wear of the rail parts in the turnout. Therefore, the rails in the turnout section are more prone to deterioration or damage than those in the mainline section, leading to problems such as poor joints, abrasions and chipping, and fish scale marks. Therefore, the analysis and evaluation of the short-wave impact state of the rails in the turnout section is an important part of the comprehensive turnout status assessment system. Currently, the high-speed integrated train wheel-rail force and axlebox acceleration detection system can provide targeted evaluation of the short-wave impact state of the rail surface. However, due to sensor signal transmission delays or deviations between the actual wheel diameter value and the system-set wheel diameter value due to wheel turning and repair, there is a deviation between the mileage of the sensor-collected signal and the actual mileage. In the turnout section, it is impossible to match the track structure with the above dynamic detection data one-to-one, resulting in the application effect of the above data in the turnout section being less than ideal. Therefore, accurately locating the actual position of the turnout in the train detection mileage is a prerequisite for the detailed analysis of the short-wave state of the track in the turnout section and the subsequent formulation of targeted maintenance and repair strategies.

[0039] Accurately locating the actual position of a turnout within the train's inspection mileage requires performing turnout vibration time-delay analysis on the turnout signal. Analysis of the vibration response within the turnout requires precisely extracting the impacts caused by structural irregularities such as joints and switch cores, despite the complex impacts and noise interference. Traditional time-frequency analysis methods, such as SST and SET, are limited by window length. Even if the impact is visible on the time-frequency spectrum, the temporal resolution of the impact is still limited, making it difficult to accurately locate the impact mileage within the turnout. Other researchers have proposed the STFT-based time-redistributed compression transform (TSST), which rearranges the time-frequency coefficients in the temporal direction and improves the temporal resolution of the time-frequency representation. However, the time-frequency results of the TSST transform still require manual identification to represent the impact within the turnout. Furthermore, calculating the time delay between the actual acquired signal and the ideal simulated signal and calculating the correlation between their time-frequency spectra does not directly reflect the time delay between the impacts, resulting in errors in the delay results. Therefore, current turnout vibration time-delay analysis methods have low accuracy.

[0040] In view of the above research, the embodiment of the present invention provides a method for analyzing the vibration delay of a turnout. Figure 1 Shown, including:

[0041] S101: performing EEMD decomposition on the ideal simulation signal and the collected actual turnout signal to extract the first intrinsic modal component of the ideal simulation signal and the first intrinsic modal component of the actual turnout signal;

[0042] S102: performing a time-redistribution multi-synchronous compression transform on the first intrinsic modal component of the ideal simulation signal and the first intrinsic modal component of the actual turnout signal to obtain a first time-frequency spectrum of the first intrinsic modal component of the ideal simulation signal and a second time-frequency spectrum of the first intrinsic modal component of the collected actual turnout signal;

[0043] S103: performing time-frequency slicing on the first time-frequency spectrum and the second time-frequency spectrum at the frequency where the vibration impact is strongest and most concentrated, respectively, to obtain a first time-frequency slice of the first time-frequency spectrum and a second time-frequency slice of the second time-frequency spectrum;

[0044] S104: Calculate the cross-correlation between the first time-frequency slice and the second time-frequency slice to obtain the time delay information of the actual turnout signal.

[0045] In an embodiment of the present invention, an ensemble empirical mode decomposition (EEMD) is performed on an ideal simulation signal and a collected actual turnout signal to extract a first intrinsic modal component of the ideal simulation signal and a first intrinsic modal component of the actual turnout signal; a time redistribution multi-synchronous compression transformation is performed on the first intrinsic modal component of the ideal simulation signal and the first intrinsic modal component of the actual turnout signal to obtain a first time-frequency spectrum of the first intrinsic modal component of the ideal simulation signal and a second time-frequency spectrum of the first intrinsic modal component of the collected actual turnout signal; time-frequency slices are performed on the first time-frequency spectrum and the second time-frequency spectrum at the frequency where the vibration impact is strongest and most concentrated, respectively, to obtain a first time-frequency slice of the first time-frequency spectrum and a second time-frequency slice of the second time-frequency spectrum; the cross-correlation between the first time-frequency slice and the second time-frequency slice is calculated to obtain time delay information of the actual turnout signal, thereby improving the accuracy of turnout vibration delay analysis.

[0046] The above-mentioned turnout vibration time delay analysis method is described in detail below.

[0047] With respect to the above S101, the ideal simulation signal and the collected actual switch signal are subjected to collective empirical mode EEMD decomposition to extract the first intrinsic modal component of the ideal simulation signal and the first intrinsic modal component of the collected actual switch signal, for example, including: pre-configuring a first number of Gaussian white noise additions; copying a first number of ideal signals and a first number of actual switch signals; adding a type of Gaussian white noise to each ideal signal to obtain multiple first new signals; adding a type of Gaussian white noise to each actual switch signal to obtain multiple second new signals; performing empirical mode EMD decomposition on each first new signal to obtain an initial first intrinsic modal component (IMF1) of each first new signal; summing and averaging the initial first intrinsic modal component of each first new signal to obtain the first intrinsic modal component IMF11 of the ideal signal; performing empirical mode EMD decomposition on each second new signal to obtain an initial first intrinsic modal component (IMF12) of each second new signal; summing and averaging the initial first intrinsic modal component of each second new signal to obtain the first intrinsic modal component of the actual switch signal.

[0048] Here, the first number of pre-configured Gaussian white noises to be added is configured according to the type of Gaussian white noise to be added, and the first number is equal to the number of types of Gaussian white noise to be added.

[0049] For example, it is predetermined that q (q is a positive integer) types of Gaussian white noise are to be added, then the first number is determined to be q, and q ideal signals and q actual switch signals are copied respectively. A type of Gaussian white noise is added to each ideal signal to obtain q first new signals, and a type of Gaussian white noise is added to each actual switch signal to obtain q second new signals.

[0050] Regarding the above S102, if Figure 2 FIG. 1 is a flowchart of a method for performing time-redistribution multi-synchronous compression transformation on a first intrinsic modal component of an ideal simulation signal and a first intrinsic modal component of an actual turnout signal, respectively, to obtain a first time-frequency spectrum of the first intrinsic modal component of the ideal simulation signal and a second time-frequency spectrum of the first intrinsic modal component of the acquired actual turnout signal, provided by an embodiment of the present invention. The method includes:

[0051] S201: Performing spectrum conversion on the first intrinsic modal component of the ideal simulation signal to obtain a spectrum expression of the first intrinsic modal component of the ideal simulation signal.

[0052] S202: Performing a short-time Fourier transform (STFT) on the frequency spectrum expression of the first intrinsic modal component of the ideal simulation signal in the frequency domain to obtain an initial first time-frequency spectrum of the first intrinsic modal component of the ideal simulation signal, wherein the initial first time-frequency spectrum includes a first strong frequency-converting signal.

[0053] Specifically, the initial first time-frequency spectrum is, for example, the following formula (1):

[0054]

[0055] in, is the initial first time spectrum, It is the first strong frequency conversion signal.

[0056] S203: Substitute the frequency domain expression of the Gaussian window function in the STFT and the second-order Taylor expansion of the first strong frequency-converted signal into the initial first time-frequency spectrum to obtain a first intermediate transformation result.

[0057] Specifically, the second-order Taylor expansion of the first strong frequency conversion signal is expressed as the following formula (2):

[0058]

[0059] Among them, A1(ω) is the amplitude of the first strong frequency conversion signal, is the phase of the first strong frequency conversion signal, is the first-order derivative of the phase of the first strong frequency conversion signal, is the second-order phase derivative of the first strong frequency conversion signal, and i is an imaginary odd number.

[0060] The Gaussian window function in STFT can be expressed as:

[0061]

[0062] Therefore, the frequency domain expression of the Gaussian window function in STFT is the following formula (3):

[0063]

[0064] ω is the frequency, σ is the window length parameter, and t is the time.

[0065] Substitute the frequency domain expression formula (3) of the Gaussian window function in the STFT and the second-order Taylor expansion formula (2) of the first strong frequency conversion signal into the initial first time-frequency spectrum formula (1) to obtain the first intermediate transformation result (as shown in formula (4)):

[0066]

[0067] Where t is time, is the first intermediate transformation result.

[0068] S204: Integrate the first intermediate transformation result along the time direction based on the two-dimensional GD estimation and the ideal time-frequency spectrum of the first intrinsic modal component to compress the energy of the first STFT time-frequency spectrum into the GD trajectory, thereby obtaining a first time-frequency spectrum of the first intrinsic modal component of the ideal simulation signal.

[0069] In the embodiment of the present invention, for example, the first time-frequency spectrum of the first intrinsic modal component of the ideal simulation signal is obtained through the following steps 1 and 2.

[0070] Step 1: Substitute the first intermediate transformation result into the two-dimensional GD estimation and perform multiple iterations to obtain the first two-dimensional GD estimation result formula (5) after iteration:

[0071]

[0072] in, is the first-order derivative of the phase of the first strong frequency conversion signal; N is the number of iterations, is the first two-dimensional GD estimation result after iteration.

[0073] Here, in order to improve the energy concentration of the time-frequency spectrum representation, TSST proposes a two-dimensional GD estimation, which is expressed as the following formula (6):

[0074]

[0075] Substituting the first intermediate transformation result formula (4) into the two-dimensional GD formula (6) for estimation, we can obtain formula (7):

[0076]

[0077] because is a function Fixed point, you can use fixed point iteration to reduce and The error between Substituting into formula (7), the result of the first fixed point iteration is formula (8):

[0078]

[0079] New 2D GD estimation Already Closer If N iterations are performed, will approach The result is as follows:

[0080]

[0081] Step 2: Based on the iterative first two-dimensional GD estimation result and the ideal time-frequency spectrum of the first intrinsic modal component, the first intermediate transformation result is integrated to compress the energy of the first STFT time-frequency spectrum into the GD trajectory to obtain the first time-frequency spectrum of the first intrinsic modal component of the ideal simulation signal.

[0082] Specifically, the ideal time-frequency spectrum of the first intrinsic modal component is, for example:

[0083]

[0084] Where δ(t) is the Dirac function.

[0085] In one embodiment of the present invention, for example, the following formula (9) is used to integrate the first intermediate transformation result based on the iterative first two-dimensional GD estimation result and the ideal time-frequency spectrum of the first intrinsic modal component, so as to compress the energy of the first STFT time-frequency spectrum into the GD trajectory, thereby obtaining the first time-frequency spectrum after compression of the first intrinsic modal component of the ideal simulation signal:

[0086]

[0087] Among them, Ts1 [N] (u,ω) is the first time-frequency spectrum of the first intrinsic modal component of the ideal simulation signal after compression, is the synchronous compression operator, G1(t,ω) is the first intermediate transformation result, and u is the recompressed time axis.

[0088] S205: Performing spectrum conversion on the first intrinsic modal component of the actual turnout signal to obtain a spectrum expression of the first intrinsic modal component of the actual turnout signal.

[0089] S206: performing STFT on the frequency spectrum expression of the first intrinsic modal component of the actual turnout signal in the frequency domain to obtain an initial second time-frequency spectrum of the first intrinsic modal component of the actual turnout signal, wherein the initial second time-frequency spectrum includes a second strong frequency-varying signal.

[0090] Specifically, the initial second time-frequency spectrum is, for example, the following formula (10):

[0091]

[0092] in, is the initial second time spectrum, It is the second strongest frequency conversion signal.

[0093] S207: Substitute the frequency domain expression of the Gaussian window function in the STFT and the second-order Taylor expansion of the second strong frequency-converted signal into the initial second time-frequency spectrum to obtain a second intermediate transformation result.

[0094] Specifically, the second-order Taylor expansion of the second strong frequency conversion signal is expressed as the following formula (11):

[0095]

[0096] Where A2(ω) is the amplitude of the second strongest frequency conversion signal, is the phase of the second strongest frequency conversion signal, is the first-order derivative of the phase of the second strongest frequency conversion signal, is the second-order phase derivative of the second strongest frequency-converted signal, and i is an imaginary odd number.

[0097] Substitute the frequency domain expression formula (3) of the Gaussian window function in the STFT and the second-order Taylor expansion formula (11) of the second strong frequency conversion signal into the initial second time-frequency spectrum formula (10) to obtain the second intermediate transformation result (as shown in formula (12)):

[0098]

[0099] Where t is time, is the second intermediate transformation result.

[0100] S208: Based on the two-dimensional GD estimation and the ideal time-frequency spectrum of the first intrinsic modal component of the actual turnout signal, the second intermediate transformation result is integrated along the time direction to compress the energy of the second STFT time-frequency spectrum into the GD trajectory, thereby obtaining the second time-frequency spectrum of the first intrinsic modal component of the actual turnout signal.

[0101] In the embodiment of the present invention, for example, the first time-frequency spectrum of the first intrinsic modal component of the actual turnout signal is obtained through the following steps A and B.

[0102] Step A: Substitute the second intermediate transformation result into the two-dimensional GD estimation and perform multiple iterations to obtain the iterated two-dimensional GD estimation result formula (13):

[0103]

[0104] in, is the first-order derivative of the phase of the second strong frequency conversion signal; N is the number of iterations, is the two-dimensional GD estimation result after iteration.

[0105] Here, the second intermediate transformation result is substituted into the two-dimensional GD estimation and multiple iterations are performed to obtain the second two-dimensional GD estimation result after the iteration. This is similar to the principle of substituting the first intermediate transformation result into the two-dimensional GD estimation and multiple iterations to obtain the first two-dimensional GD estimation result after the iteration. The repeated parts will not be repeated here.

[0106] Step B: Based on the iterative second two-dimensional GD estimation result and the ideal time-frequency spectrum of the first intrinsic modal component of the actual turnout signal, the second intermediate transformation result is integrated along the time direction to compress the second STFT time-frequency spectrum energy into the GD trajectory, thereby obtaining the second time-frequency spectrum of the first intrinsic modal component of the actual turnout signal.

[0107] Specifically, the ideal time-frequency spectrum of the first intrinsic modal component of the actual turnout signal is, for example:

[0108]

[0109] Where δ(t) is the Dirac function.

[0110] In one embodiment of the present invention, the following formula (14) is used to integrate the second intermediate transformation result along the time direction based on the iterative second two-dimensional GD estimation result and the ideal time-frequency spectrum of the first intrinsic modal component of the actual turnout signal, to obtain the second STFT time-frequency spectrum energy, and to compress the second STFT time-frequency spectrum energy into the GD trajectory to obtain the second time-frequency spectrum of the first intrinsic modal component of the actual turnout signal after compression:

[0111]

[0112] Among them, Ts2 [N] (u,ω) is the second time-frequency spectrum of the actual turnout signal after the first intrinsic modal component is compressed. is the synchronous compression operator, G2(t,ω) is the second intermediate transformation result, and u is the re-compressed time axis.

[0113] Regarding S103 above, while the second time-frequency spectrum can reflect the vibration impact caused by the welded joints, point rails, and point rails within the turnout, it is still difficult to pinpoint the specific location of the impact within the turnout. To this end, this embodiment of the present invention employs a time-frequency slicing strategy, obtaining a time-frequency slice of the first intrinsic modal component of the actual turnout signal at the frequency with the highest energy concentration. This slice is then cross-correlated with the time-frequency slice of the first intrinsic modal component of the ideal simulation signal at the frequency with the highest energy concentration.

[0114] In one embodiment of the present invention, performing time-frequency slicing on the first time-frequency spectrum at the frequency where the vibration impact is the strongest and most concentrated to obtain a first time-frequency slice of the first time-frequency spectrum includes: performing time-frequency slicing on the first time-frequency spectrum at the frequency where the vibration impact is the strongest and most concentrated to obtain the first time-frequency slice of the first time-frequency spectrum using the following formula (15):

[0115]

[0116] Where ω1 is the ideal simulation signal vibration impact with the strongest and most concentrated frequency, f1(u) is is the first time-frequency slice of the first time-frequency spectrum.

[0117] Here, the time-frequency slice of the first intrinsic modal component of the ideal simulation signal at the frequency ω1 where the energy is most concentrated is selected That is, the frequency ω1 processes the spectrum of the first intrinsic modal component of the simulated signal The amplitude A(ω1) and phase Changes with time, but in actual analysis, we are generally concerned with the energy of the signal at a certain point in time, that is, the amplitude A(ω1) and the phase The module of TMSST time-frequency slice|Ts1 [N] (u,ω1)| can be expressed as formula (15):

[0118]

[0119] Time-frequency slicing can effectively extract the time-varying energy changes of the actual turnout signal at a specific frequency. The idea of ​​time-frequency slicing is to perform a specific frequency filtering operation on the obtained time-reassigned multi-synchronous compressed time-frequency spectrum, which retains the impact components in the actual turnout signal and removes the influence of noise and other interfering impacts in the time domain signal, laying the foundation for the subsequent identification of impacts in the turnout and the time delay of the actual turnout signal.

[0120] In one embodiment of the present invention, time-frequency slicing is performed on the second time-frequency spectrum at the frequency where the vibration impact is the strongest and most concentrated to obtain a second time-frequency slice of the second time-frequency spectrum, including: using the following formula (16) to perform time-frequency slicing on the second time-frequency spectrum at the frequency where the vibration impact is the strongest and most concentrated to obtain the second time-frequency slice of the second time-frequency spectrum:

[0121]

[0122] Where ω2 is the frequency at which the actual turnout signal vibration impact is the strongest and most concentrated, and f2(u) is also is the second time-frequency slice of the second time-frequency spectrum.

[0123] With respect to the above S104, in one embodiment of the present invention, calculating the cross-correlation between the first time-frequency slice and the second time-frequency slice to obtain the time delay information of the actual turnout signal includes: obtaining a cross-correlation function based on the first time-frequency slice and the second time-frequency slice:

[0124]

[0125]

[0126] Where f1(u) is the first time-frequency slice, is the conjugate of f1(u), f2(u) is the second time-frequency slice, C 12 (τ) is the cross-correlation function, Δu is the time delay information of the actual turnout signal; the maximum value of the cross-correlation function is substituted into the cross-correlation function to obtain the time delay information of the actual turnout signal.

[0127] In one embodiment of the present invention, after obtaining the time delay information, the method further includes: determining the course error of the actual turnout signal according to the time delay information of the actual turnout signal.

[0128] In addition, in order to further verify the accuracy of the turnout delay analysis method provided by an embodiment of the present invention, in another embodiment of the present invention, for example, noise, impact, and mileage error are added to the ideal simulation signal to obtain a noisy simulation signal to simulate the actual turnout signal. The time delay information between the noisy simulation signal and the ideal simulation signal is known. The above-mentioned turnout delay analysis method is used to calculate the time delay information between the noisy simulation signal and the ideal simulation signal. The error value between the calculated time delay information and the known time delay information is less than the preset error, thereby verifying that the turnout delay analysis method of the embodiment of the present invention has high accuracy.

[0129] For example, in order to verify the effectiveness and accuracy of the proposed time-frequency slicing and cross-correlation strategy for mileage correction, the ideal simulation signal and the noisy simulation signal time-frequency slices are constructed. As mentioned above, they are represented by f1(u) and f3(u) respectively, and the time delay information between the two is calculated using the cross-correlation theory. Then, the mileage error between the two is obtained based on the time delay information between the two. The cross-correlation function C 13 (τ) is defined as follows:

[0130]

[0131] in, is the conjugate of f1(u). If f3(u) is derived from f1(u) with a time delay of Δu1, i.e. f3(u)=f1(u-Δu1), then:

[0132]

[0133] It can be seen that when and only when τ=Δu1, C13 (τ) reaches its maximum value, and

[0134]

[0135] The cross-correlation function of the N-point discrete signals f1[n] and f3[n] after sampling is defined as follows:

[0136]

[0137] From the above analysis, we can see that the cross-correlation of two signals is equal to the linear convolution of the first signal after folding and conjugating it with the second signal. Therefore, when the lengths of the two signals are M points and K points respectively, the length of the final calculated cross-correlation function is M+K-1 points. When the length of both signals is M points, the length of their cross-correlation function is 2M-1 points.

[0138] The ideal simulation signal time-frequency slice f1(u) and the noisy simulation signal time-frequency slice f3(u) are discrete signals in the time domain, and the cross-correlation calculation of the present invention is the linear convolution of discrete signals. The ideal simulation signal time-frequency slice f1(u) (length is M) and the noisy simulation signal time-frequency slice f3(u) (length is K) are of different lengths. The correlation function fills the simulation signal with the shorter signal length with MK zeros, making it a linear convolution of two signals with a length of M. At this time, the length of the cross-correlation function is 2M-1 points. At the maximum cross-correlation point, when the time delay is a negative value, it indicates that the time-frequency slice f3(u) of the noisy simulation signal lags behind the ideal simulation signal time-frequency slice f1(u); when the time delay is a positive value, it indicates that the time-frequency slice f3(u) of the noisy simulation signal is ahead of the ideal simulation signal time-frequency slice f1(u).

[0139] From formula (19), we can see that the maximum value of the cross-correlation between f3(u) and f1(u) is C 13(max) (τ), that is, the time-frequency slice of the noisy simulation signal and the time-frequency slice of the ideal simulation signal are delayed by Δu1, that is, the time-frequency slice of the noisy simulation signal and the ideal simulation signal are delayed by Δu1.

[0140] The following is an explanation using a preferred example.

[0141] The actual turnout signal collected was the acceleration of the left axlebox at a turnout section on a high-speed railway. The sampling frequency of the high-speed train acceleration data (i.e., the actual turnout signal collected) was 5000 Hz. The section is divided into an uplink line and a downlink line. The uplink line is 181 km to 175 km long, and the downlink line is 176 km to 182 km long. Each of the uplink and downlink lines has four turnouts, for a total of eight turnouts.

[0142] In an embodiment of the present invention, the actual turnout signal collected on the upline is divided into the first half (179+420 to 179+070) and the second half (178+434 to 178+044). Since the railway mileage of the upline is always from large to small in actual railway operation, the present invention still draws the graph according to the actual mileage sequence. The time-frequency slices of the simulated signals of the front and back halves of the upline and the actual turnout signal are cross-correlated to obtain the time delay between the two. Specifically, the following steps are included:

[0143] Step (a): Perform EEMD decomposition on the actual turnout signal collected in the first half of the up line to obtain the first intrinsic modal component as follows: Figure 3 The first intrinsic modal component IMF1 after EEMD decomposition of the actual turnout signal collected in the first half of the up line is compared with the ideal simulation signal, as shown in Figure 4 shown.

[0144] Step (b): Add shock, noise and mileage error to the ideal simulation signal to obtain a noisy simulation signal. The noisy simulation signal is used to simulate the actual working conditions of railway vehicles, and a comparison diagram between the noisy simulation signal and the ideal simulation signal is obtained, such as Figure 5 Then, the ideal simulation signal and the noisy simulation signal are transformed by TMSST after two iterations and the impulse signal is extracted by time-frequency slicing at 555Hz for comparison, as shown in Figure 6 After that, the cross-correlation between the two is calculated to obtain the time delay of the ideal simulation signal and the noisy simulation signal, as shown in Figure 7 shown. Figure 7 The correlation between the noisy simulation signal and the ideal simulation signal is the maximum at -100m, indicating that the noisy simulation signal lags behind the ideal simulation signal. The accuracy of the cross-correlation function to calculate the time delay of the railway switch vibration signal is verified by moving the noisy simulation signal with a 100m delay, that is, the noisy simulation signal moves 100m to the right. Figure 8 shown.

[0145] Step (c): Figure 4 The actual turnout signal and the ideal simulation signal shown in the figure are compared by using TMSST transform after secondary iteration and extracting the impact signal using time-frequency slicing at the 555Hz frequency where the vibration impact is concentrated. Figure 9 shown.

[0146] Step (d): Use the cross-correlation function to calculate the cross-correlation value between the ideal simulation signal time-frequency slice f1(u) and the actual turnout signal time-frequency slice f2(u). The maximum cross-correlation corresponds to the time delay between the ideal simulation signal and the actual turnout signal, as shown in the figure. Figure 10 shown.

[0147] Step (e): Shift the time-frequency slice of the first intrinsic modal component of the actual turnout signal to the right by the lag delay calculated in step (d) of 70.98m to verify the impact caused by the joint and the center rail in the turnout vibration signal, thereby proving the accuracy of the cross-correlation delay. The comparison diagram is shown in the figure below. Figure 11 shown.

[0148] Repeat the above steps for the actual turnout signal in the second half of the up line to obtain Figure 12 The cross-correlation time delay results shown, and Figure 13 The time-frequency slice of the actual turnout signal IMF1 in the second half of the upline, shifted rightward by 73.11 meters, is compared with the time-frequency slice of the ideal simulated signal. The results show that the time delay error between the actual turnout signal and the ideal simulated signal in the first and second half of the upline is 2.13 meters, which is within the acceptable range, validating the accuracy of the proposed method for turnout vibration response analysis.

[0149] The present invention also provides a turnout vibration time delay analysis device, as described in the following embodiments. Since the principles of this device are similar to those of the turnout vibration time delay analysis method, the implementation of this device can refer to the implementation of the turnout vibration time delay analysis method, and the repeated parts will not be repeated here.

[0150] like Figure 14 FIG. 1 is a schematic diagram of a switch vibration time delay analysis device provided by an embodiment of the present invention, comprising:

[0151] The first processing module 1401 is configured to perform EEMD decomposition on the ideal simulation signal and the collected actual turnout signal to extract the first intrinsic modal component of the ideal simulation signal and the first intrinsic modal component of the actual turnout signal;

[0152] The second processing module 1402 is configured to perform a time-redistribution multi-synchronous compression transform on the first intrinsic modal component of the ideal simulation signal and the first intrinsic modal component of the actual turnout signal to obtain a first time-frequency spectrum of the first intrinsic modal component of the ideal simulation signal and a second time-frequency spectrum of the first intrinsic modal component of the acquired actual turnout signal;

[0153] The third processing module 1403 is configured to perform time-frequency slicing on the first time-frequency spectrum and the second time-frequency spectrum at the frequency where the vibration impact is strongest and most concentrated, to obtain a first time-frequency slice of the first time-frequency spectrum and a second time-frequency slice of the second time-frequency spectrum;

[0154] The fourth processing module 1404 is configured to calculate the cross-correlation between the first time-frequency slice and the second time-frequency slice to obtain the time delay information of the actual turnout signal.

[0155] In one possible implementation, the first processing module is specifically used to pre-configure a first number of added Gaussian white noises; copy a first number of ideal signals and a first number of actual switch signals; add a type of Gaussian white noise to each ideal signal to obtain multiple first new signals; add a type of Gaussian white noise to each actual switch signal to obtain multiple second new signals; perform empirical mode EMD decomposition on each first new signal to obtain an initial first intrinsic modal component of each first new signal; sum and average the initial first intrinsic modal components of each first new signal to obtain the first intrinsic modal component of the ideal signal; perform empirical mode EMD decomposition on each second new signal to obtain an initial first intrinsic modal component of each second new signal; sum and average the initial first intrinsic modal components of each second new signal to obtain the first intrinsic modal component of the actual switch signal.

[0156] In one possible implementation, the second processing module is specifically used to perform spectrum conversion on the first connotation modal component of the ideal simulation signal to obtain the spectrum expression of the first connotation modal component of the ideal simulation signal; perform short-time Fourier transform (STFT) on the spectrum expression of the first connotation modal component of the ideal simulation signal in the frequency domain to obtain the initial first time-frequency spectrum of the first connotation modal component of the ideal simulation signal, and the initial first time-frequency spectrum contains a strong frequency-converting signal; substitute the frequency domain expression of the Gaussian window function in the STFT and the second-order Taylor expansion of the strong frequency-converting signal into the initial first time-frequency spectrum to obtain a first intermediate transformation result; integrate the first intermediate transformation result along the time direction according to the two-dimensional GD estimation and the ideal time-frequency spectrum of the first connotation modal component of the ideal signal to compress the energy of the first STFT time-frequency spectrum into the GD trajectory to obtain the first The first time-frequency spectrum of the intrinsic modal component of the actual turnout signal is obtained; the spectrum conversion is performed on the first intrinsic modal component of the actual turnout signal to obtain the spectrum expression of the first intrinsic modal component of the actual turnout signal; the spectrum expression of the first intrinsic modal component of the actual turnout signal is subjected to STFT in the frequency domain to obtain the initial second time-frequency spectrum of the first intrinsic modal component of the actual turnout signal, and the initial second time-frequency spectrum contains a strong frequency-varying signal; the frequency-domain expression of the Gaussian window function in the STFT and the second-order Taylor expansion of the strong frequency-varying signal are substituted into the initial second time-frequency spectrum to obtain a second intermediate transformation result; according to the two-dimensional GD estimation and the ideal time-frequency spectrum of the first intrinsic modal component of the actual turnout signal, the second intermediate transformation result is integrated along the time direction to compress the energy of the second STFT time-frequency spectrum into the GD trajectory to obtain the second time-frequency spectrum of the first intrinsic modal component of the actual turnout signal.

[0157] In a possible implementation, the initial first time-frequency spectrum is expressed as follows:

[0158]

[0159] in, is the initial first time spectrum, is a strong frequency conversion signal. The second-order Taylor expansion of the first strong frequency conversion signal is expressed as

[0160]

[0161] A1(ω) is the amplitude of the first strong frequency conversion signal, is the phase of the first strong frequency conversion signal, is the first-order derivative of the phase of the first strong frequency conversion signal, is the second-order derivative of the phase of the first strong frequency conversion signal, i is an imaginary odd number;

[0162] The frequency domain expression of the Gaussian window function in STFT is:

[0163] ω is the frequency, σ is the window length parameter;

[0164] The result of the first intermediate transformation is:

[0165]

[0166] t is time, is the first intermediate transformation result.

[0167] In a possible implementation, the second processing module is specifically configured to substitute the first intermediate transformation result into the two-dimensional GD estimation and perform multiple iterations to obtain the first two-dimensional GD estimation result after iteration:

[0168] is the first-order derivative of the phase of the first strong frequency conversion signal; N is the number of iterations, is the first two-dimensional GD estimation result after iteration;

[0169] According to the iterative first two-dimensional GD estimation result and the ideal time-frequency spectrum of the first connotation modal component of the ideal simulation signal, multiple synchronous compression transformations are performed on the first intermediate transformation result to compress the energy of the first STFT time-frequency spectrum into the GD trajectory to obtain the first time-frequency spectrum after compression of the first connotation modal component;

[0170] The following formula is used to integrate the first intermediate transformation result based on the iterative first two-dimensional GD estimation result and the ideal time-frequency spectrum of the first intrinsic modal component to compress the energy of the first STFT time-frequency spectrum into the GD trajectory, thereby obtaining the first time-frequency spectrum after compression of the first intrinsic modal component:

[0171]

[0172] Among them, Ts1 [N](u,ω) is the first time-frequency spectrum of the first intrinsic modal component of the ideal simulation signal after compression, is the synchronous compression operator, G1(t,ω) is the first intermediate transformation result, and u is the recompressed time axis.

[0173] In a possible implementation, the third processing module is specifically configured to perform time-frequency slicing on the first time-frequency spectrum at the frequency where the vibration impact is strongest and most concentrated, to obtain a first time-frequency slice of the first time-frequency spectrum using the following formula:

[0174] ω1 is the ideal simulation signal vibration impact with the strongest and most concentrated frequency, f1(u) is is the first time-frequency slice of the first time-frequency spectrum.

[0175] In a possible implementation, the initial second time-frequency spectrum is expressed as follows:

[0176]

[0177] in, is the initial second time spectrum, is the second strongest frequency conversion signal, and the second-order Taylor expansion of the second strongest frequency conversion signal is expressed as

[0178]

[0179] A2(ω) is the amplitude of the second strongest frequency conversion signal, is the phase of the second strongest frequency conversion signal, is the first-order derivative of the phase of the second strongest frequency conversion signal, is the second-order derivative of the phase of the second strongest frequency-converted signal, i is an imaginary odd number;

[0180] The frequency domain expression of the Gaussian window function in STFT is:

[0181] ω is the frequency, σ is the window length parameter;

[0182] The result of the second intermediate transformation is:

[0183]

[0184] t is time, is the second intermediate transformation result.

[0185] In a possible implementation, the second processing module is specifically configured to substitute the second intermediate transformation result into the two-dimensional GD estimation and perform multiple iterations to obtain a second two-dimensional GD estimation result after iteration:

[0186] is the first-order derivative of the phase of the second strong frequency conversion signal; N is the number of iterations, is the second two-dimensional GD estimation result after iteration;

[0187] Based on the iterative second two-dimensional GD estimation result and the ideal time-frequency spectrum of the first intrinsic modal component of the actual turnout signal, the second intermediate transformation result is integrated along the time direction to obtain the second STFT time-frequency spectrum energy, and the second STFT time-frequency spectrum energy is compressed into the GD trajectory to obtain the second time-frequency spectrum of the first intrinsic modal component of the actual turnout signal;

[0188] The following formula is used to integrate the second intermediate transform result along the time direction based on the iterative second two-dimensional GD estimation result and the ideal time-frequency spectrum of the first intrinsic modal component of the actual turnout signal, so as to compress the energy of the second STFT time-frequency spectrum into the GD trajectory, thereby obtaining the second time-frequency spectrum after compression of the first intrinsic modal component of the actual turnout signal:

[0189]

[0190] Among them, Ts2 [N] (u,ω) is the second time-frequency spectrum of the actual turnout signal after the first intrinsic modal component is compressed. is the synchronous compression operator, G2(t,ω) is the second intermediate transformation result, and u is the re-compressed time axis.

[0191] In a possible implementation, the third processing module is specifically configured to perform time-frequency slicing on the second time-frequency spectrum at the frequency where the vibration impact is strongest and most concentrated, to obtain a second time-frequency slice of the second time-frequency spectrum using the following formula:

[0192] ω2 is the frequency at which the actual turnout signal vibration impact is the strongest and most concentrated, f2(u) is is the second time-frequency slice of the second time-frequency spectrum.

[0193] In a possible implementation, the fourth processing module is specifically configured to obtain a cross-correlation function according to the first time-frequency slice and the second time-frequency slice:

[0194]

[0195] Where f1(u) is the first time-frequency slice, is the conjugate of f1(u), f2(u) is the second time-frequency slice, C 12 (τ) is the cross-correlation function, Δu is the time delay information of the actual turnout signal; the maximum value of the cross-correlation function is substituted into the cross-correlation function to obtain the time delay information of the actual turnout signal.

[0196] An embodiment of the present invention further provides a computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the above-mentioned turnout vibration delay analysis method when executing the computer program.

[0197] An embodiment of the present invention further provides a computer-readable storage medium storing a computer program. When the computer program is executed by a processor, the above-mentioned turnout vibration delay analysis method is implemented.

[0198] An embodiment of the present invention further provides a computer program product, which includes a computer program. When the computer program is executed by a processor, the above-mentioned turnout vibration delay analysis method is implemented.

[0199] In an embodiment of the present invention, an ensemble empirical mode decomposition (EEMD) is performed on an ideal simulation signal and a collected actual turnout signal to extract a first intrinsic modal component of the ideal simulation signal and a first intrinsic modal component of the actual turnout signal; a time redistribution multi-synchronous compression transformation is performed on the first intrinsic modal component of the ideal simulation signal and the first intrinsic modal component of the actual turnout signal to obtain a first time-frequency spectrum of the first intrinsic modal component of the ideal simulation signal and a second time-frequency spectrum of the first intrinsic modal component of the collected actual turnout signal; time-frequency slices are performed on the first time-frequency spectrum and the second time-frequency spectrum at the frequency where the vibration impact is strongest and most concentrated, respectively, to obtain a first time-frequency slice of the first time-frequency spectrum and a second time-frequency slice of the second time-frequency spectrum; the cross-correlation between the first time-frequency slice and the second time-frequency slice is calculated to obtain time delay information of the actual turnout signal, thereby improving the accuracy of turnout vibration delay analysis.

[0200] It will be understood by those skilled in the art that embodiments of the present invention may be provided as methods, systems, or computer program products. Thus, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0201] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowcharts and / or block diagrams, as well as combinations of processes and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowcharts and / or block diagrams. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0202] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0203] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0204] The specific embodiments described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for analyzing turnout vibration delay, characterized in that: include: Performing EEMD decomposition on the ideal simulation signal and the collected actual turnout signal to extract the first intrinsic modal component of the ideal simulation signal and the first intrinsic modal component of the actual turnout signal; Performing time-redistribution multi-synchronous compression transformation on the first connotation modal component of the ideal simulation signal and the first connotation modal component of the actual turnout signal to obtain a first time-frequency spectrum of the first connotation modal component of the ideal simulation signal and a second time-frequency spectrum of the first connotation modal component of the collected actual turnout signal; At the frequency where the vibration impact is the strongest and most concentrated, time-frequency slicing is performed on the first time-frequency spectrum and the second time-frequency spectrum to obtain a first time-frequency slice of the first time-frequency spectrum and a second time-frequency slice of the second time-frequency spectrum; The cross-correlation between the first time-frequency slice and the second time-frequency slice is calculated to obtain the time delay information of the actual turnout signal.

2. The turnout vibration time delay analysis method according to claim 1, characterized in that: Performing EEMD decomposition on the ideal simulation signal and the collected actual turnout signal to extract the first intrinsic modal component of the ideal simulation signal and the first intrinsic modal component of the actual turnout signal, including: Preconfigure a first amount of Gaussian white noise to be added; replicating a first number of ideal signals and a first number of actual turnout signals; Add a kind of Gaussian white noise to each ideal signal to obtain multiple first new signals; Adding a Gaussian white noise to each actual turnout signal to obtain multiple second new signals; Performing empirical mode decomposition (EMD) on each first new signal to obtain an initial first connotation modal component of each first new signal; Summing and averaging the initial first connotation modal components of each first new signal to obtain the first connotation modal components of the ideal signal; Performing empirical mode EMD decomposition on each second new signal to obtain the initial first connotation mode component of each second new signal; The initial first intrinsic modal component of each second new signal is summed and averaged to obtain the first intrinsic modal component of the actual turnout signal.

3. The turnout vibration time delay analysis method according to claim 1, characterized in that: Performing a time redistribution multi-synchronous compression transformation on the first connotation modal component of the ideal simulation signal to obtain a first time-frequency spectrum of the first connotation modal component of the ideal simulation signal, including: Performing spectrum conversion on the first connotation modal component of the ideal simulation signal to obtain a spectrum expression of the first connotation modal component of the ideal simulation signal; Performing a short-time Fourier transform (STFT) on the spectrum expression of the first connotation modal component of the ideal simulation signal in the frequency domain to obtain an initial first time-frequency spectrum of the first connotation modal component of the ideal simulation signal, wherein the initial first time-frequency spectrum contains a strong frequency conversion signal; Substitute the frequency domain expression of the Gaussian window function in STFT and the second-order Taylor expansion of the strong frequency-converting signal into the initial first time-frequency spectrum to obtain the first intermediate transformation result; Integrating the first intermediate transform result along the time direction based on the two-dimensional GD estimation and the ideal time-frequency spectrum of the first connotation modal component of the ideal signal to compress the energy of the first STFT time-frequency spectrum into the GD trajectory, thereby obtaining the first time-frequency spectrum of the first connotation modal component of the ideal simulation signal; Performing spectrum conversion on the first intrinsic modal component of the collected actual turnout signal to obtain a spectrum expression of the first intrinsic modal component of the actual turnout signal; Perform STFT on the spectrum expression of the first connotation modal component of the actual turnout signal in the frequency domain to obtain the initial second time-frequency spectrum of the first connotation modal component of the actual turnout signal, which contains a strong frequency conversion signal. Substitute the frequency domain expression of the Gaussian window function in STFT and the second-order Taylor expansion of the strong frequency-converting signal into the initial second time-frequency spectrum to obtain the second intermediate transformation result; Based on the two-dimensional GD estimation and the ideal time-frequency spectrum of the first intrinsic modal component of the actual turnout signal, the second intermediate transformation result is integrated along the time direction to compress the second STFT time-frequency spectrum energy into the GD trajectory, thereby obtaining the second time-frequency spectrum of the first intrinsic modal component of the actual turnout signal.

4. The method for analyzing turnout vibration delay according to claim 3, wherein: The initial first time-frequency spectrum is expressed as follows: in, is the initial first time spectrum, is a strong frequency conversion signal. The second-order Taylor expansion of the first strong frequency conversion signal is expressed as A1(ω) is the amplitude of the first strong frequency conversion signal, is the phase of the first strong frequency conversion signal, is the first-order derivative of the phase of the first strong frequency conversion signal, is the second-order derivative of the phase of the first strong frequency conversion signal, i is an imaginary odd number; The frequency domain expression of the Gaussian window function in STFT is: ω is the frequency, σ is the window length parameter; The result of the first intermediate transformation is: t is time, is the first intermediate transformation result.

5. The turnout vibration time delay analysis method according to claim 4, characterized in that: Based on the two-dimensional GD estimation and the ideal first time-frequency spectrum of the first connotation modal component of the ideal simulation signal, the first intermediate transformation result is integrated along the time direction to compress the energy of the first STFT time-frequency spectrum into the GD trajectory, thereby obtaining the first time-frequency spectrum of the first connotation modal component of the ideal simulation signal, including: Substitute the first intermediate transformation result into the two-dimensional GD estimation and perform multiple iterations to obtain the first two-dimensional GD estimation result after iteration: is the first-order derivative of the phase of the first strong frequency conversion signal; N is the number of iterations, is the first two-dimensional GD estimation result after iteration; According to the iterative first two-dimensional GD estimation result and the ideal time-frequency spectrum of the first connotation modal component of the ideal simulation signal, multiple synchronous compression transformations are performed on the first intermediate transformation result to compress the energy of the first STFT time-frequency spectrum into the GD trajectory to obtain the first time-frequency spectrum after compression of the first connotation modal component; The following formula is used to integrate the first intermediate transformation result based on the iterative first two-dimensional GD estimation result and the ideal time-frequency spectrum of the first intrinsic modal component to compress the energy of the first STFT time-frequency spectrum into the GD trajectory, thereby obtaining the first time-frequency spectrum after compression of the first intrinsic modal component: Among them, Ts1 [N] (u,ω) is the first time-frequency spectrum of the first intrinsic modal component of the ideal simulation signal after compression, is the synchronous compression operator, G1(t,ω) is the first intermediate transformation result, and u is the recompressed time axis.

6. The method for analyzing turnout vibration delay according to claim 5, characterized in that: At the frequency where the vibration impact is the strongest and most concentrated, the first time-frequency slice of the first time-frequency spectrum is obtained, including: The following formula is used to slice the first time-frequency spectrum at the frequency where the vibration impact is the strongest and most concentrated, to obtain the first time-frequency slice of the first time-frequency spectrum: ω1 is the ideal simulation signal vibration impact with the strongest and most concentrated frequency, f1(u) is is the first time-frequency slice of the first time-frequency spectrum.

7. The turnout vibration time delay analysis method according to claim 3, characterized in that: The initial second time spectrum is expressed as follows: in, is the initial second time spectrum, is the second strongest frequency conversion signal, and the second-order Taylor expansion of the second strongest frequency conversion signal is expressed as A2(ω) is the amplitude of the second strongest frequency conversion signal, is the phase of the second strongest frequency conversion signal, is the first-order derivative of the phase of the second strongest frequency conversion signal, is the second-order derivative of the phase of the second strongest frequency-converted signal, i is an imaginary odd number; The frequency domain expression of the Gaussian window function in STFT is: ω is the frequency, σ is the window length parameter; The result of the second intermediate transformation is: t is time, is the second intermediate transformation result.

8. The method for analyzing turnout vibration delay according to claim 7, wherein: Based on the two-dimensional GD estimation and the ideal time-frequency spectrum of the first intrinsic modal component of the actual turnout signal, the second intermediate transformation result is integrated along the time direction to compress the energy of the second STFT time-frequency spectrum into the GD trajectory, thereby obtaining the second time-frequency spectrum of the first intrinsic modal component of the actual turnout signal, including: Substitute the second intermediate transformation result into the two-dimensional GD estimation and perform multiple iterations to obtain the second two-dimensional GD estimation result after iteration: is the first-order derivative of the phase of the second strong frequency conversion signal; N is the number of iterations, is the second two-dimensional GD estimation result after iteration; Based on the iterative second two-dimensional GD estimation result and the ideal time-frequency spectrum of the first intrinsic modal component of the actual turnout signal, the second intermediate transformation result is integrated along the time direction to obtain the second STFT time-frequency spectrum energy, and the second STFT time-frequency spectrum energy is compressed into the GD trajectory to obtain the second time-frequency spectrum of the first intrinsic modal component of the actual turnout signal; The following formula is used to integrate the second intermediate transform result along the time direction based on the iterative second two-dimensional GD estimation result and the ideal time-frequency spectrum of the first intrinsic modal component of the actual turnout signal, so as to compress the energy of the second STFT time-frequency spectrum into the GD trajectory, thereby obtaining the second time-frequency spectrum after compression of the first intrinsic modal component of the actual turnout signal: Among them, Ts2 [N] (u,ω) is the second time-frequency spectrum of the actual turnout signal after the first intrinsic modal component is compressed. is the synchronous compression operator, G2(t,ω) is the second intermediate transformation result, and u is the re-compressed time axis.

9. The method for analyzing turnout vibration delay according to claim 8, wherein: At the frequency where the vibration impact is the strongest and most concentrated, the second time-frequency spectrum is sliced ​​to obtain a second time-frequency slice of the second time-frequency spectrum, including: The following formula is used to slice the second time-frequency spectrum at the frequency where the vibration impact is the strongest and most concentrated, to obtain the second time-frequency slice of the second time-frequency spectrum: ω2 is the frequency at which the actual turnout signal vibration impact is the strongest and most concentrated, f2(u) is is the second time-frequency slice of the second time-frequency spectrum.

10. The turnout vibration time delay analysis method according to claim 1, characterized in that: Calculate the cross-correlation between the first time-frequency slice and the second time-frequency slice to obtain the time delay information of the actual turnout signal, including: The cross-correlation function is obtained based on the first time-frequency slice and the second time-frequency slice: Where f1(u) is the first time-frequency slice, is the conjugate of f1(u), f2(u) is the second time-frequency slice, C 12 (τ) is the cross-correlation function, Δu is the time delay information of the actual turnout signal; Substitute the maximum value of the cross-correlation function into the cross-correlation function to obtain the time delay information of the actual turnout signal.

11. A turnout vibration delay analysis device, characterized in that: include: A first processing module is used to perform EEMD decomposition on the ideal simulation signal and the collected actual turnout signal to extract the first intrinsic modal component of the ideal simulation signal and the first intrinsic modal component of the actual turnout signal; The second processing module is used to perform time redistribution multi-synchronous compression transformation on the first intrinsic modal component of the ideal simulation signal and the first intrinsic modal component of the actual turnout signal to obtain a first time-frequency spectrum of the first intrinsic modal component of the ideal simulation signal and a second time-frequency spectrum of the first intrinsic modal component of the collected actual turnout signal; The third processing module is used to perform time-frequency slicing on the first time-frequency spectrum and the second time-frequency spectrum at the frequency where the vibration impact is the strongest and most concentrated, to obtain a first time-frequency slice of the first time-frequency spectrum and a second time-frequency slice of the second time-frequency spectrum; The fourth processing module is used to calculate the cross-correlation between the first time-frequency slice and the second time-frequency slice to obtain the time delay information of the actual turnout signal.

12. The switch vibration time delay analysis device according to claim 11, characterized in that: A first processing module is specifically configured to pre-configure a first amount of added Gaussian white noise; replicating a first number of ideal signals and a first number of actual turnout signals; Add a kind of Gaussian white noise to each ideal signal to obtain multiple first new signals; Adding a Gaussian white noise to each actual turnout signal to obtain multiple second new signals; Performing empirical mode decomposition (EMD) on each first new signal to obtain an initial first connotation modal component of each first new signal; Summing and averaging the initial first connotation modal components of each first new signal to obtain the first connotation modal components of the ideal signal; Performing empirical mode EMD decomposition on each second new signal to obtain the initial first connotation mode component of each second new signal; The initial first intrinsic modal component of each second new signal is summed and averaged to obtain the first intrinsic modal component of the actual turnout signal.

13. The switch vibration time delay analysis device according to claim 11, characterized in that: The second processing module is specifically used for Performing spectrum conversion on the first connotation modal component of the ideal simulation signal to obtain a spectrum expression of the first connotation modal component of the ideal simulation signal; Performing a short-time Fourier transform (STFT) on the spectrum expression of the first connotation modal component of the ideal simulation signal in the frequency domain to obtain an initial first time-frequency spectrum of the first connotation modal component of the ideal simulation signal, wherein the initial first time-frequency spectrum contains a strong frequency conversion signal; Substitute the frequency domain expression of the Gaussian window function in STFT and the second-order Taylor expansion of the strong frequency-converting signal into the initial first time-frequency spectrum to obtain the first intermediate transformation result; Integrating the first intermediate transform result along the time direction based on the two-dimensional GD estimation and the ideal time-frequency spectrum of the first connotation modal component of the ideal signal to compress the energy of the first STFT time-frequency spectrum into the GD trajectory, thereby obtaining the first time-frequency spectrum of the first connotation modal component of the ideal simulation signal; Performing spectrum conversion on the first intrinsic modal component of the collected actual turnout signal to obtain a spectrum expression of the first intrinsic modal component of the actual turnout signal; Perform STFT on the spectrum expression of the first connotation modal component of the actual turnout signal in the frequency domain to obtain the initial second time-frequency spectrum of the first connotation modal component of the actual turnout signal, which contains a strong frequency conversion signal. Substitute the frequency domain expression of the Gaussian window function in STFT and the second-order Taylor expansion of the strong frequency-converting signal into the initial second time-frequency spectrum to obtain the second intermediate transformation result; Based on the two-dimensional GD estimation and the ideal time-frequency spectrum of the first intrinsic modal component of the actual turnout signal, the second intermediate transformation result is integrated along the time direction to compress the second STFT time-frequency spectrum energy into the GD trajectory, thereby obtaining the second time-frequency spectrum of the first intrinsic modal component of the actual turnout signal.

14. The switch vibration time delay analysis device according to claim 11, characterized in that: The fourth processing module is specifically configured to obtain a cross-correlation function based on the first time-frequency slice and the second time-frequency slice: Where f1(u) is the first time-frequency slice, is the conjugate of f1(u), f2(u) is the second time-frequency slice, C 12 (τ) is the cross-correlation function, Δu is the time delay information of the actual turnout signal; Substitute the maximum value of the cross-correlation function into the cross-correlation function to obtain the time delay information of the actual turnout signal.

15. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, the method according to any one of claims 1 to 10 is implemented.

16. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method according to any one of claims 1 to 10 is implemented.

17. A computer program product, characterized in that The computer program product comprises a computer program, and when the computer program is executed by a processor, the method according to any one of claims 1 to 10 is implemented.

Images