A method, device, equipment and medium for determining characteristic frequency of dynamic response data of a rail vehicle

Abstract

This application provides a method, apparatus, device, and medium for determining the characteristic frequencies of dynamic response data of rail vehicles, relating to the field of rail vehicle technology. The method determines the resonant frequency band in the dynamic response based on the known excitation frequency of the operating line, and separates the dynamic response data according to the energy change rate of the resonant frequency band to obtain vehicle vibration response data, thereby determining the characteristic frequencies of target vehicle components. This overcomes the problems of existing technologies where fixed-band filtering cannot adapt to the differences in excitation characteristics of different lines, easily leads to insufficient interference filtering or over-filtering that mistakenly deletes effective information, and manual frequency selection relies on experience, is inefficient, and has subjective judgment errors, making it difficult to adapt to scenarios with large changes in rail vehicle speed and complex operating conditions. It can accurately eliminate line excitation interference, accurately extract the characteristic frequencies of vehicle components, and improve the accuracy and reliability of characteristic frequency identification.

Description

Technical Field

[0001] This application relates to the field of rail vehicle technology, and in particular to a method, apparatus, equipment and medium for determining the characteristic frequencies of dynamic response data of rail vehicles. Background Technology

[0002] Currently, during long-term operation, key components of rail vehicles, such as axle boxes, wheelsets, and suspension systems, are susceptible to fatigue, impact, and wear, leading to structural abnormalities or even failures, directly threatening operational safety and transportation efficiency. To achieve early warning of component failures and health status assessment, the industry commonly uses sensors installed on the vehicles to collect dynamic response data during operation, then extracts the characteristic frequencies of target components from the data as a basis for judging the structural condition of the components. However, the dynamic response data of rail vehicles is a coupled signal of vehicle body vibration and periodic excitation vibration of the track. Periodic excitations on the track side, such as sleeper pitch, track corrugation, and wheel-rail contact, will form a resonant frequency band with strong energy and fixed frequency in the frequency domain. This type of resonant frequency band will severely mask the weak vibration response caused by vehicle component failures, making it difficult to accurately extract the component's characteristic frequency, thus becoming an interference factor in characteristic frequency identification.

[0003] In existing technologies, the elimination of line excitation interference often employs fixed-band filtering or manual frequency selection. However, fixed-band filtering cannot adapt to the differences in excitation characteristics of different lines, and is prone to insufficient interference filtering. The remaining line excitation components may still mask the component characteristic signals, or over-filtering may mistakenly delete the effective vibration information of the vehicle body, reducing the accuracy of characteristic frequency extraction. In addition, manual frequency selection relies on the experience of operators, which is not only inefficient but also subject to subjective judgment errors, making it difficult to adapt to the actual application scenarios of large speed variations and complex operating conditions of rail vehicles. Summary of the Invention

[0004] This application provides a method, apparatus, equipment, and medium for determining the characteristic frequencies of dynamic response data of rail vehicles, in order to solve the technical problems existing in the prior art.

[0005] In a first aspect, this application provides a method for determining the characteristic frequencies of dynamic response data of rail vehicles, including:

[0006] Acquire dynamic response data generated by the vehicle on the operating route;

[0007] Based on the known excitation frequency on the operating line, the resonant frequency band in the dynamic response that matches the known excitation frequency is determined;

[0008] Based on the energy change rate of the resonant frequency band, the dynamic response data is separated to obtain vehicle vibration response data;

[0009] Based on the vehicle vibration response data, the characteristic frequencies of the target components of the vehicle are determined.

[0010] In one possible design, acquiring the dynamic response data generated by the vehicle on the operating route includes:

[0011] Using a vibration acceleration sensor installed on the vehicle bogie, the original vibration signal generated by the target component of the vehicle during operation is collected, and the original vibration signal is used as the dynamic response data.

[0012] In one possible design, determining the resonant frequency band in the dynamic response that matches the known excitation frequency based on the known excitation frequency on the operating circuit includes:

[0013] Based on the known excitation frequency of the operating line, at least one analysis frequency band centered on an overtone of the known excitation frequency and having a preset bandwidth is determined, and the analysis frequency band is determined as a resonant frequency band that matches the known excitation frequency.

[0014] In one possible design, the separation and processing of the dynamic response data based on the energy change rate of the resonant frequency band to obtain vehicle vibration response data includes:

[0015] The energy change rate of each of the resonant frequency bands is compared with a preset sensitivity threshold;

[0016] If the energy change rate of any of the resonant frequency bands is less than the preset sensitivity threshold, the corresponding resonant frequency band will be determined as the dominant frequency band for line excitation.

[0017] A band-stop filter is used to filter out the frequency components corresponding to the dominant frequency band of the line excitation, and the filtered signal is determined as the vehicle vibration response data.

[0018] In one possible design, determining the characteristic frequencies of the target vehicle component based on the vehicle vibration response data includes:

[0019] Perform Hilbert transform on the vehicle vibration response data to extract the envelope signal of the vehicle vibration response data;

[0020] The envelope signal is subjected to spectral analysis to obtain the envelope spectrum;

[0021] From the set of frequency peaks of the envelope spectrum, determine the target frequency peak with the smallest difference from the preset theoretical fault characteristic frequency;

[0022] The frequency corresponding to the target spectral peak is determined as the characteristic frequency of the target vehicle component.

[0023] In one possible design, after the dynamic response data generated by the vehicle on the operating route, the system further includes:

[0024] The dynamic response data is processed to remove the DC component in order to obtain initial preprocessed data;

[0025] The initial preprocessed data is denoised using wavelet threshold denoising to obtain clean dynamic response data.

[0026] The pure dynamic response data is subjected to amplitude normalization to obtain standardized dynamic response data.

[0027] One possible design also includes:

[0028] The dynamic response data is calculated using windowed Fourier transform to obtain the frequency domain energy corresponding to different times within the resonance band;

[0029] Based on the frequency range of the resonant frequency band, the frequency domain energy corresponding to each moment is integrated to generate the frequency band energy sequence corresponding to the resonant frequency band.

[0030] Based on the frequency band energy sequence corresponding to the resonant frequency band, calculate the standard deviation and average value of the frequency band energy sequence respectively;

[0031] The ratio of the standard deviation to the average value is determined as the energy change rate of the resonant frequency band.

[0032] Secondly, this application provides a device for determining the characteristic frequency of dynamic response data of a rail vehicle, comprising:

[0033] The acquisition module is used to acquire dynamic response data generated by the vehicle on the operating route;

[0034] The resonant frequency band determination module is used to determine the resonant frequency band in the dynamic response that matches the known excitation frequency based on the known excitation frequency on the operating line.

[0035] The separation module is used to separate the dynamic response data according to the energy change rate of the resonant frequency band to obtain vehicle vibration response data.

[0036] The characteristic frequency determination module is used to determine the characteristic frequencies of the target components of the vehicle based on the vehicle vibration response data.

[0037] Thirdly, this application provides an electronic device, including: a processor, and a memory communicatively connected to the processor;

[0038] The memory stores computer-executed instructions;

[0039] The processor executes computer execution instructions stored in the memory to implement the method as described in any of the first aspects.

[0040] Fourthly, embodiments of this application provide a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, are used to implement the method as described in any of the first aspects.

[0041] Fifthly, embodiments of this application provide a computer program product, including a computer program that, when executed by a processor, implements the method described in any of the first aspects.

[0042] This application provides a method, apparatus, equipment, and medium for determining the characteristic frequencies of dynamic response data of rail vehicles. The method determines the resonant frequency band in the dynamic response based on the known excitation frequency of the operating line, and separates the dynamic response data according to the energy change rate of the resonant frequency band to obtain vehicle vibration response data, thereby determining the characteristic frequencies of target vehicle components. This overcomes the problems of existing technologies where fixed-band filtering cannot adapt to the differences in excitation characteristics of different lines, easily leads to insufficient or excessive filtering that mistakenly deletes effective information, and manual frequency selection relies on experience, is inefficient, and has subjective judgment errors, making it difficult to adapt to scenarios with large changes in rail vehicle speed and complex operating conditions. It can accurately eliminate line excitation interference, accurately extract the characteristic frequencies of vehicle components, and improve the accuracy and reliability of characteristic frequency identification. Attached Figure Description

[0043] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0044] Figure 1 This application provides an example of an method for determining the characteristic frequencies of dynamic response data of a rail vehicle, as shown in an embodiment of this application, and includes an application scenario diagram.

[0045] Figure 2 A flowchart illustrating a method for determining the characteristic frequencies of dynamic response data of a rail vehicle, provided in an embodiment of this application;

[0046] Figure 3 A flowchart illustrating a method for determining the characteristic frequencies of dynamic response data of a rail vehicle, provided in another embodiment of this application;

[0047] Figure 4 A flowchart illustrating a method for determining the characteristic frequencies of dynamic response data of a rail vehicle, provided in another embodiment of this application;

[0048] Figure 5A schematic diagram of the structure of a device for determining the characteristic frequency of dynamic response data of a rail vehicle provided in an embodiment of this application;

[0049] Figure 6 This is a structural example diagram of an electronic device provided in an embodiment of this application.

[0050] The accompanying drawings illustrate specific embodiments of this application, which will be described in more detail below. These drawings and descriptions are not intended to limit the scope of the concept in any way, but rather to illustrate the concept of this application to those skilled in the art through reference to particular embodiments. Detailed Implementation

[0051] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numbers in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this application as detailed in the appended claims.

[0052] To clearly understand the technical solution of this application, the solutions of the prior art will be described in detail first.

[0053] During long-term operation, key components of rail vehicles, such as axle boxes, wheelsets, and suspension systems, are susceptible to fatigue, impact, and wear, leading to structural abnormalities or even failures that directly threaten operational safety and transportation efficiency. To achieve early warning of component failures and health status assessment, the industry commonly uses sensors installed on the vehicle to collect dynamic response data during operation, then extracts the characteristic frequencies of target components from the data as a basis for judging the structural condition of the components. However, the dynamic response data of rail vehicles is a coupled signal of vehicle body vibration and periodic excitation vibration of the track. Periodic excitations on the track side, such as sleeper pitch, track corrugation, and wheel-rail contact, will form a high-energy, fixed-frequency resonance band in the frequency domain. This type of resonance band can severely mask the weak vibration response caused by vehicle component failures, making it difficult to accurately extract the component's characteristic frequencies and becoming an interference factor in characteristic frequency identification.

[0054] In existing technologies, the elimination of line excitation interference often employs fixed-band filtering or manual frequency selection. However, fixed-band filtering cannot adapt to the differences in excitation characteristics of different lines, and is prone to insufficient interference filtering. The remaining line excitation components may still mask the component characteristic signals, or over-filtering may mistakenly delete the effective vibration information of the vehicle body, reducing the accuracy of characteristic frequency extraction. In addition, manual frequency selection relies on the experience of operators, which is not only inefficient but also subject to subjective judgment errors, making it difficult to adapt to the actual application scenarios of large speed variations and complex operating conditions of rail vehicles.

[0055] In summary, how to accurately identify and eliminate the resonant frequency band dominated by line excitation from non-stationary dynamic response data coupled with line excitation and mixed with random noise, effectively highlight the vibration characteristics of vehicle components, and then accurately extract the characteristic frequencies of vehicle components has become an urgent technical problem to be solved in the field of rail vehicle operation monitoring.

[0056] Figure 1 An application scenario diagram corresponding to a method for determining the characteristic frequencies of dynamic response data of a rail vehicle provided in an embodiment of this application is shown, such as... Figure 1 As shown, the application scenario provided in this embodiment includes: a rail vehicle 10, a data processing device 11, and a track information database 12. The rail vehicle 10 and the data processing device 11 transmit collected data through a dedicated wireless communication link, and the data processing device 11 and the track information database 12 establish a real-time retrieval connection through an intranet. Specifically, the rail vehicle 11 is equipped with vibration sensors, which are deployed at key components such as axle boxes, wheelsets, and suspension systems to collect dynamic response data of vehicle operation in real time. The track information database 12 stores the known excitation frequencies of each operating line, including characteristic excitation frequencies corresponding to sleeper pitch, track corrugation, and wheel-rail contact.

[0057] Specifically, when the rail vehicle 10 is running on the operating line, the on-board vibration sensor collects dynamic response data of parts such as axle boxes and wheelsets in real time and uploads it to the data processing device 11. Subsequently, the data processing device 11 retrieves the corresponding known excitation frequency from the line information database 12 based on the current operating line information of the vehicle and performs frequency domain matching with the dynamic response data to locate the resonant frequency band in the data that matches the line excitation. Then, it calculates the energy change rate of the resonant frequency band and performs separation processing on the dynamic response data based on the energy change rate to accurately remove the interference signal brought by the line excitation and obtain response data containing only the vibration of the vehicle body. Finally, the data processing device 11 performs frequency domain analysis and feature extraction on the vehicle vibration response data to determine the characteristic frequencies of target components such as axle boxes, wheelsets, and suspension systems, providing data basis for subsequent early warning of component failures and health status assessment.

[0058] The technical solution of this application and how the technical solution of this application solves the above-mentioned technical problems are described in detail below with specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments. The embodiments of this application will now be described with reference to the accompanying drawings.

[0059] In the several embodiments provided in this application, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple modules may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be indirect coupling or communication connection through some interfaces, devices, or modules, and may be electrical, mechanical, or other forms.

[0060] Figure 2 This is a flowchart illustrating a method for determining the characteristic frequencies of dynamic response data of a rail vehicle according to an embodiment of this application, as shown below. Figure 2 As shown, the execution subject of this embodiment is a characteristic frequency determination device for dynamic response data of a rail vehicle. This device can be implemented by a computer program, or by a medium storing the relevant computer program, such as a USB flash drive and / or optical disc; alternatively, it can be implemented by a physical device integrating or installing the relevant computer program, such as a chip or electronic device. The electronic device may be a computer or a server, etc. The characteristic frequency determination method for dynamic response data of a rail vehicle provided in this embodiment includes the following steps:

[0061] S201. Obtain dynamic response data generated by the vehicle on the operating route.

[0062] Specifically, the purpose of this step is to obtain the raw data for subsequent feature frequency extraction, providing a foundation for interference removal, signal separation, and feature recognition.

[0063] Optionally, sensors pre-installed at key locations on the rail vehicle can be used to collect dynamic response data generated by the vehicle in real time during its operation on the current track. It should be noted that the sensor installation locations must correspond to the target components (such as axle boxes, wheelsets, and suspension systems). The collected dynamic response data includes, but is not limited to, vibration acceleration signals and displacement vibration signals. The sampling frequency must satisfy the Nyquist sampling theorem to ensure that all signal components in the frequency domain (including vehicle component vibration signals and track excitation resonance signals) are fully preserved, avoiding signal distortion due to insufficient sampling frequency.

[0064] Optionally, after acquiring the dynamic response data, preprocessing operations can be performed on the raw data. These operations may include denoising (removing random noise such as sensor errors and electromagnetic interference), signal smoothing, and normalization to reduce the impact of non-line excitation interference on subsequent processing and ensure the reliability of the raw data. During preprocessing, it is crucial to avoid damaging the line excitation resonant frequency band and the characteristic signals of vehicle components to provide accurate data support for subsequent resonant frequency band positioning.

[0065] S202. Based on the known excitation frequency on the operating line, determine the resonant frequency band in the dynamic response that matches the known excitation frequency.

[0066] Among them, the excitation frequency corresponding to the periodic excitation on the track side (such as sleeper pitch and track corrugation) can be calculated or measured in advance, that is, "known excitation frequency". This kind of known excitation frequency will form a corresponding resonance frequency band in the frequency domain of dynamic response data, and the center frequency of the resonance frequency band is consistent with the known excitation frequency or has a fixed proportional relationship (such as harmonic or frequency division). Based on this, the interference source that masks the characteristic signal of the component can be accurately located.

[0067] Optionally, the known excitation frequency of the current operating line can be obtained. For example, the excitation frequency corresponding to the sleeper pitch can be calculated by the vehicle speed and the sleeper pitch; the excitation frequency corresponding to track corrugation can be calculated by pre-measuring the wavelength of track corrugation and the vehicle speed; the frequency of periodic excitation of wheel-rail contact can be derived by parameters such as vehicle wheel diameter and rotational speed.

[0068] Optionally, the dynamic response data acquired and preprocessed in S201 is subjected to frequency domain transformation (such as Fourier transform) to obtain a frequency domain distribution map of the dynamic response data (the horizontal axis is frequency, and the vertical axis is signal energy). Then, the center frequencies of each frequency band in the frequency domain distribution map are matched with the known excitation frequencies mentioned above. By setting a reasonable frequency matching threshold (such as ±5% frequency deviation), the frequency bands whose center frequencies fall within the matching threshold range can be determined as the resonant frequency bands that match the known excitation frequencies of the line, i.e., the frequency bands corresponding to the line excitation interference.

[0069] It should be noted that by using the known excitation frequency of the line to directionally match the resonant frequency band, the problem of fixed filtering in existing technologies being unable to adapt to different lines is avoided. It also avoids the subjective error of manually selecting the frequency band to locate the interference band, ensuring that the interference band corresponding to the excitation of all lines can be accurately identified.

[0070] S203. Based on the energy change rate of the resonant frequency band, the dynamic response data is separated and processed to obtain vehicle vibration response data.

[0071] It should be noted that the resonant frequency band generated by the line excitation and the frequency band corresponding to the vibration of vehicle components differ significantly in their energy change rates. Specifically, the line excitation is a periodic and stable excitation, and the corresponding resonant frequency band exhibits a smooth energy change rate and small fluctuation range during vehicle operation. In contrast, the energy change rate of the frequency band corresponding to the vibration of vehicle components fluctuates significantly with changes in vehicle operating conditions and component status. Especially when an early failure occurs in a component, the energy change rate of the characteristic frequency band may exhibit abrupt changes. Based on this difference, precise separation of the line excitation resonant frequency band (interference signal) and the vehicle vibration response signal (effective signal) can be achieved.

[0072] Specifically, the energy change rate of each resonant frequency band determined in S202 is first calculated. Optionally, for the time-domain sequence of dynamic response data, the data is divided into segments according to fixed time windows (e.g., 1s~5s, which can be adjusted according to vehicle speed and operating conditions). The energy value of the resonant frequency band within each time window is calculated, and then the real-time energy change rate of the resonant frequency band is obtained by the ratio of the energy difference between adjacent time windows to the energy value of the previous time window. Next, the dynamic response data is separated. Signal components whose energy change rate falls within the interference judgment threshold range (i.e., the signal corresponding to the line excitation resonant frequency band) are removed, while signal components whose energy change rate exceeds the interference judgment threshold are retained. The retained signal components are the vehicle vibration response data after removing line excitation interference.

[0073] Secondly, the interference judgment threshold is pre-calibrated based on the energy change law of the line excitation resonant frequency band to ensure that the difference in energy change between the line excitation resonant frequency band and the vibration frequency band of vehicle components can be distinguished. For example, the energy change rate of the line excitation resonant frequency band is usually stable within ±3%, and this range can be set as the interference judgment threshold.

[0074] Optionally, during the separation process, an adaptive filtering algorithm can be used to adjust the filtering parameters in real time based on the energy change rate of the resonant frequency band, avoiding over-filtering or under-filtering. When the energy change rate of the resonant frequency band fluctuates slightly, the filtering intensity is adjusted appropriately to ensure that the interference signal is completely eliminated; when the energy change rate of a certain frequency band exceeds the interference judgment threshold, the filtering operation for that frequency band is stopped, and the valid signal is completely preserved.

[0075] S204. Based on vehicle vibration response data, determine the characteristic frequencies of the target components of the vehicle.

[0076] Specifically, after eliminating line excitation interference, the characteristic frequencies of the target components are accurately extracted from the pure vehicle vibration response data, providing a basis for component health status assessment.

[0077] Optionally, firstly, the vehicle vibration response data obtained in S203 is subjected to frequency domain transformation (such as wavelet transform or Fourier transform) to obtain the frequency domain distribution map of the vehicle vibration response data. At this time, there is no strong energy resonance frequency band corresponding to the line excitation in the frequency domain distribution map, and the frequency band corresponding to the vibration of the vehicle components can be clearly presented.

[0078] Secondly, by combining the structural parameters (such as stiffness, mass, and natural frequency) of the target component (such as axle box and wheelset), the theoretical range of the characteristic frequency of the target component is determined. For example, the characteristic frequency of the axle box bearing can be derived from parameters such as bearing type, number of rolling elements, and raceway diameter.

[0079] Finally, in the frequency domain distribution map, the frequency bands that fall within the theoretical range of the characteristic frequency of the target component are extracted. Amplitude analysis and peak extraction are performed on the frequency bands to determine the center frequency and amplitude stability of the frequency bands. The center frequency is used as the characteristic frequency of the vehicle target component. If the amplitude of the characteristic frequency shows abnormal deviation, splitting, or amplitude fluctuations exceeding the normal range, it can be determined that the target component has structural abnormalities or potential faults.

[0080] This application provides a method for determining the characteristic frequencies of dynamic response data of rail vehicles. It determines the resonant frequency band in the dynamic response based on the known excitation frequency of the operating line, and then separates the dynamic response data according to the energy change rate of the resonant frequency band to obtain vehicle vibration response data, thereby determining the characteristic frequencies of target vehicle components. This overcomes the problems of existing technologies where fixed-band filtering cannot adapt to the differences in excitation characteristics of different lines, easily leads to insufficient interference filtering or over-filtering that mistakenly deletes effective information, and manual frequency selection relies on experience, is inefficient, and has subjective judgment errors, making it difficult to adapt to scenarios with large changes in rail vehicle speed and complex operating conditions. This method can accurately eliminate line excitation interference, accurately extract the characteristic frequencies of vehicle components, and improve the accuracy and reliability of characteristic frequency identification.

[0081] As an optional implementation, based on any of the above embodiments, the acquisition of dynamic response data generated by the vehicle on the operating route includes the following steps: using a vibration acceleration sensor installed on the vehicle bogie to collect the original vibration signal generated by the target component of the vehicle during operation, and using the original vibration signal as dynamic response data.

[0082] The bogie is the core running gear of a rail vehicle. Target components such as axle boxes, wheelsets, and suspension systems are all directly mounted on the bogie. Vibration acceleration sensors are mounted on the bogie, and the installation position must correspond to the target component. For example, for axle box components, the sensor is installed near the axle box seat of the bogie; for wheelset components, the sensor is installed at the corresponding position on the wheelset bearing seat of the bogie. Priority is given to selecting the installation point on the bogie with the least vibration transmission loss and the closest distance to the target component. Avoid installing the sensor on parts of the bogie where vibration attenuation is obvious and which are easily interfered with by other non-target components, such as the ends of the bogie crossbeams or non-load-bearing areas.

[0083] Optionally, based on the operating characteristics of the rail vehicle and the data acquisition requirements, a high-precision, wide-bandwidth vibration acceleration sensor should be selected first. Its measurement range should be adapted to the vibration acceleration range during the operation of the rail vehicle, and its frequency response range should cover the entire range of the characteristic frequency of the target component and the excitation frequency of the line. This ensures that the weak vibration signal of the target component and the resonance signal generated by the line excitation can be completely captured, avoiding signal saturation due to insufficient sensor range or signal distortion due to insufficient frequency response range.

[0084] Optionally, after the vibration acceleration sensor is installed, a calibration operation is required to ensure the measurement accuracy of the sensor and eliminate the influence of installation deviation and the inherent error of the vibration acceleration sensor itself. During vehicle operation, the vibration acceleration sensor collects the vibration signal transmitted by the corresponding target component on the bogie in real time. This vibration signal is the original vibration signal generated by the target component during operation due to its own vibration, wheel-rail contact impact, track excitation, and other factors. During the acquisition process, the sampling frequency of the sensor must strictly follow the Nyquist sampling theorem, and the sampling frequency must not be less than twice the highest characteristic frequency of the target component to ensure that the original vibration signal can completely retain all signal components in the frequency domain.

[0085] This application provides a method for determining the characteristic frequency of dynamic response data of rail vehicles. By installing a vibration acceleration sensor on the vehicle bogie to collect the original vibration signal of the target component as dynamic response data, it can directly obtain real vibration information closely related to the state of the vehicle component. The data source is reliable and highly targeted.

[0086] As an optional implementation, based on any of the above embodiments, the resonant frequency band that matches the known excitation frequency in the dynamic response is determined according to the known excitation frequency on the operating circuit, including the following steps:

[0087] Based on the known excitation frequency of the operating line, at least one analysis frequency band centered on an overtone of the known excitation frequency and having a preset bandwidth is determined, and the analysis frequency band is determined as the resonant frequency band that matches the known excitation frequency.

[0088] It should be noted that the periodic excitation of the track is essentially a simple harmonic excitation. When applied to the rail vehicle, it not only creates a resonant frequency band at the fundamental frequency but also, due to the nonlinear characteristics of the vehicle structure and the periodic impact of wheel-rail contact, generates resonance at integer harmonics of the fundamental frequency, forming an overtone resonant frequency band with energy intensity second only to the fundamental frequency resonant frequency band. This type of overtone resonant frequency band can also mask the characteristic frequencies of vehicle components. If it is not identified and removed in time, it will still affect the accuracy of characteristic frequency extraction.

[0089] Therefore, it is necessary to determine the excitation frequency based on the known excitation frequency. At least one harmonic value needs to be determined. The selection of the harmonic value should be based on the vehicle's operating conditions, the line excitation intensity, and the characteristic frequency range of the target component. Optionally, a lower-order harmonic with higher energy intensity (such as...) can be selected. , If the lower-order harmonics overlap with the characteristic frequency range of the target component, they can be appropriately extended to higher-order harmonics (such as...). , This ensures that no key interference frequency bands are missed; in addition, it is necessary to rule out cases where the frequency multiplication value exceeds the frequency domain range of the dynamic response data. This can be done by combining the frequency response range of the sensor in S201 to avoid invalid frequency multiplication settings.

[0090] The preset bandwidth refers to the frequency range defined around each harmonic value for determining the resonant frequency band. The preset bandwidth setting must balance "interference capture integrity" and "effective signal protection," avoiding situations where excessively wide bandwidth leads to the mistaken inclusion of vehicle component characteristic signals in the interference band, or excessively narrow bandwidth leads to the omission of some energy components in the harmonic resonant frequency band. Optionally, the preset bandwidth value range is typically the known excitation frequency. The bandwidth can be dynamically adjusted from 5% to 15% based on the stability of the line excitation and the intensity of the harmonic energy. When the line excitation is highly stable and the harmonic energy is concentrated, a smaller bandwidth (e.g., 5% to 8%) can be selected to improve positioning accuracy; when the line excitation fluctuates significantly and the harmonic energy is dispersed, a larger bandwidth (e.g., 12% to 15%) can be selected to ensure complete capture of the harmonic resonance frequency band. For example, if the excitation frequency is known... If the bandwidth is selected as 10%, the corresponding multiplier is... The analysis frequency band is 19Hz~21Hz, and the harmonics are... The analysis frequency band is 28.5Hz~31.5Hz. In addition, the preset bandwidth can be calibrated through historical data, and combined with the excitation characteristics of different lines, a correspondence between the bandwidth and known excitation frequency and line type can be established to improve adaptability.

[0091] Finally, using each selected harmonic value as the center and combining it with the preset bandwidth, a corresponding analysis frequency band is defined. Each analysis frequency band is the resonant frequency band that matches the known excitation frequency. If multiple harmonic values ​​are selected simultaneously, multiple analysis frequency bands are determined accordingly. All analysis frequency bands together constitute the complete resonant frequency band corresponding to the line excitation. For example, given the excitation frequency... Select the multiplier , With a preset bandwidth of 10%, the determined resonant frequency band is 19Hz~21Hz. Corresponding), 28.5Hz~31.5Hz ( (Corresponding). Once the resonant frequency bands are determined, signal separation processing can be performed based on the energy change rate of each resonant frequency band.

[0092] This application provides a method for determining the characteristic frequencies of dynamic response data of rail vehicles. Based on the known excitation frequency of the operating line, an analysis frequency band with a preset bandwidth is set with the harmonic frequency as the center to determine the resonance frequency band. This method can comprehensively and accurately include the resonance range formed by the line excitation in the frequency domain, effectively avoiding the omission of key resonance frequency bands. This provides a reliable basis for the subsequent accurate separation of line excitation interference and acquisition of the vehicle's real vibration response data, thereby ensuring the accuracy of determining the characteristic frequencies of the vehicle's target components.

[0093] Figure 3 A flowchart illustrating a method for determining the characteristic frequencies of dynamic response data of a rail vehicle, provided in another embodiment of this application; as shown. Figure 3 As shown, as an optional implementation, based on any of the above embodiments, the dynamic response data is separated and processed according to the energy change rate of the resonant frequency band to obtain vehicle vibration response data, including the following steps:

[0094] S301. Compare the energy change rate of each resonant frequency band with the preset sensitivity threshold.

[0095] Secondly, the preset sensitivity threshold is a quantitative standard for distinguishing between the "line-excited dominant frequency band" and the "non-line-excited frequency band". The preset sensitivity threshold is set based on the difference in energy change rate between the line-excited resonance frequency band and the vehicle component vibration frequency band, and is pre-calibrated by combining historical monitoring data and on-site measured data.

[0096] Optionally, the formula for calculating the preset sensitivity threshold can be:

[0097]

[0098] in, The average energy change rate under normal circuit excitation. Standard deviation The empirical relaxation factor is set to 1 to 2.

[0099] Optionally, if the line excitation stability is high (e.g., newly built lines, slight track corrugation), the energy change rate of the line excitation resonance frequency band fluctuates very little, and the preset sensitivity threshold can be set to ±2% to improve the judgment sensitivity; if the line excitation fluctuation is large (e.g., old lines, severe sleeper wear), the preset sensitivity threshold can be set to ±4%~±5% to avoid misjudging the fluctuating line excitation frequency band as a non-line excitation frequency band.

[0100] Optionally, the calculated real-time energy change rate of each resonant frequency band is compared with a preset sensitivity threshold, and the relationship between the energy change rate of each resonant frequency band and the threshold is recorded. If the absolute value of the energy change rate of a certain resonant frequency band is less than the preset sensitivity threshold, then proceed to S302 for further judgment; if the absolute value of the energy change rate of a certain resonant frequency band is greater than or equal to the preset sensitivity threshold, then it is determined that the frequency band is not dominated by line excitation, and no further filtering operation is required, and its corresponding signal components are retained.

[0101] S302. When the energy change rate of any resonant frequency band is less than the preset sensitivity threshold, the corresponding resonant frequency band is determined as the dominant frequency band for line excitation.

[0102] It should be noted that the resonant frequency band corresponding to the line excitation has a smooth energy change rate and a very small fluctuation range during normal vehicle operation, and remains at a low level. However, the frequency band corresponding to the vibration of vehicle components (especially fault vibration) will have a significant fluctuation in energy change rate as the component status and operating conditions change, and its value usually exceeds the preset sensitivity threshold.

[0103] Therefore, when the energy change rate of any resonant frequency band is less than the preset sensitivity threshold, it indicates that the energy fluctuation of that resonant frequency band is within the stable range corresponding to the line excitation. Its signal components are mainly dominated by the periodic excitation of the line and contain almost no effective signals from vehicle component vibration. Based on this, the resonant frequency band can be accurately identified as the line excitation dominant frequency band. If the energy change rates of multiple resonant frequency bands (such as the fundamental frequency resonant frequency band and the harmonic resonant frequency band) are all less than the preset sensitivity threshold, then all such resonant frequency bands are identified as the line excitation dominant frequency bands, forming a complete set of interference frequency bands, providing a clear basis for subsequent targeted filtering.

[0104] It should also be noted that if the energy change rate of a certain resonant frequency band is less than the preset sensitivity threshold in some time windows and greater than or equal to the preset sensitivity threshold in others, it indicates that the frequency band may contain a mixture of line excitation and vehicle component vibration signals. In this case, the dominant component of the frequency band can be further determined by extending the time window and increasing the comparison period. If the energy change rate is less than the preset sensitivity threshold for more than 80% of the time window, it is still identified as the dominant frequency band of line excitation; if it is less than 80%, the frequency band is determined to contain valid signals and is not included in the range of the dominant frequency band of line excitation to avoid mistakenly filtering out valid signals.

[0105] S303. Use a band-stop filter to filter out the frequency components corresponding to the dominant frequency band of the line excitation, and determine the filtered signal as vehicle vibration response data.

[0106] Specifically, the purpose of this step is to perform directional filtering of the dominant frequency band of line excitation determined by S302, so as to achieve complete separation of interference signals and effective signals and obtain pure vehicle vibration response data.

[0107] Optionally, considering the signal characteristics of the dynamic response data of rail vehicles (mainly low-frequency interference and small signal amplitude fluctuations), an Infinite Impulse Response (IIR) band-stop filter can be selected. It has the advantages of high filtering accuracy, low phase distortion, and strong real-time performance, and can meet the real-time processing requirements of a large amount of dynamic response data from rail vehicles. If the real-time filtering requirements are lower, a Finite Impulse Response (FIR) band-stop filter can also be selected to further reduce filtering distortion.

[0108] It should be noted that the parameters of the band-stop filter must be precisely matched with the dominant frequency band of the line excitation determined by S302 to ensure that only interference frequency band components are filtered out without damaging the effective signals of vehicle component vibration. Specifically, the center frequency of the filter should be consistent with the center frequency of the dominant frequency band of the line excitation. For example, if the dominant frequency band of the line excitation is 19Hz~21Hz and the center frequency is 20Hz, then the center frequency of the filter should be set to 20Hz. The stopband bandwidth of the filter should be consistent with or slightly wider than the bandwidth of the dominant frequency band of the line excitation (0.1Hz~0.5Hz wider) to ensure that all interference components within the dominant frequency band of the line excitation can be completely filtered out, avoiding the mistaken filtering of adjacent effective signals due to an excessively wide stopband. The attenuation of the filter should be set to 40dB~60dB to ensure that the interference signal can be completely attenuated to meet the accuracy requirements of subsequent characteristic frequency extraction.

[0109] Finally, the preprocessed dynamic response data from S201 is input into a band-stop filter with pre-defined parameters. The filter selectively removes frequency components corresponding to the dominant frequency band of the line excitation, retaining signal components in other frequency ranges. After filtering, the output signal undergoes smoothing and noise reduction preprocessing to eliminate minor distortion signals generated during filtering. This output signal can then be identified as the vehicle vibration response data. This vehicle vibration response data has eliminated line excitation interference and contains only valid signals of vehicle body vibration and target component vibration, which can be used to extract the characteristic frequencies of the target components.

[0110] This application provides a method for determining the characteristic frequencies of dynamic response data of rail vehicles. By comparing the energy change rate of the resonant frequency band with a preset sensitivity threshold, the dominant frequency band of the line excitation is accurately identified. Then, a band-stop filter is used to selectively filter out its frequency components, effectively avoiding the limitations of fixed frequency band filtering. It can adapt to different line conditions, accurately separate line excitation interference, and retain the effective vibration information of the vehicle body to the greatest extent, thereby obtaining more accurate vehicle vibration response data.

[0111] As an optional implementation, based on any of the above embodiments, the characteristic frequencies of the target vehicle component are determined based on vehicle vibration response data, including the following steps:

[0112] First, Hilbert transform is performed on the vehicle vibration response data to extract the envelope signal of the vehicle vibration response data.

[0113] Specifically, the purpose of this step is to remove the useless carrier components from the vehicle vibration response data, extract the envelope signal that can reflect the fault characteristics of the target component, and highlight the weak vibration characteristics of early faults.

[0114] Optionally, the vehicle vibration response data obtained after the S203 separation is first acquired. Then, a Hilbert transform operation is performed on the vehicle vibration response data. The core of the Hilbert transform is to analyze the time-domain signal of the vibration response and generate a corresponding analytic signal. The amplitude of the analytic signal is the envelope signal of the vehicle vibration response data. Optionally, the formula for calculating the envelope signal can be:

[0115]

[0116] in, This is the remaining vehicle vibration response data after the previous stage of filtering. Let represent the Hilbert transform operator, j be the imaginary unit, and e(t) be the envelope of the vibration signal, which is the demodulated real function containing the modulation information of the periodic impact.

[0117] Optionally, the Hilbert transform operation can be implemented through signal processing algorithms (such as the fast Hilbert transform algorithm), and a reasonable transform window can be set (consistent with the time window for calculating the rate of change of energy in S203, 1s~5s) to avoid signal lag due to an excessively large window and envelope extraction distortion due to an excessively small window.

[0118] It should be noted that when early failures occur in target vehicle components (especially rolling bearings and wheelsets), their vibration signals are often modulated signals consisting of a combination of a high-frequency carrier and a low-frequency modulation. The high-frequency carrier is the inherent vibration frequency of the component during normal operation, while the low-frequency modulation signal is a weak vibration signal caused by the fault. Conventional frequency domain transformations (such as Fourier transforms) are difficult to distinguish between the carrier and modulation signals, which can easily lead to the masking of fault characteristics. However, the Hilbert transform can accurately extract the envelope of the modulation signal, remove the interference of the high-frequency carrier, and highlight the weak vibration characteristics corresponding to the fault, laying the foundation for subsequent spectral peak identification.

[0119] Optionally, the extracted envelope signal can be preprocessed (e.g., denoising and smoothing) to remove minor noise generated during the Hilbert transform process, ensuring that the envelope signal can accurately reflect the vibration characteristics of the target component and avoid noise interference with the accuracy of subsequent spectrum analysis.

[0120] Secondly, spectral analysis is performed on the envelope signal to obtain the envelope spectrum.

[0121] Specifically, the purpose of this step is to convert the envelope signal in the time domain into the envelope spectrum in the frequency domain, extract the fault characteristic frequency of the target component from the time domain signal, form intuitively identifiable spectral peaks, and provide a basis for subsequent spectral peak matching.

[0122] Optionally, considering the low-frequency characteristics of the envelope signal, Fast Fourier Transform (FFT) can be used for spectrum analysis. Fast Fourier Transform has the advantages of high computational efficiency and high spectral resolution, which can meet the real-time processing requirements of dynamic response data of rail vehicles. If there are non-stationary components in the envelope signal, wavelet packet transform can be used for spectrum analysis to further improve the accuracy of spectrum analysis.

[0123] Optionally, the formula for determining the envelope spectrum is: ;in, The envelope spectrum is a complex function in the frequency domain, and its amplitude represents the energy at the corresponding frequency. For frequency variables.

[0124] Optionally, the preprocessed envelope time-domain signal is input into the spectrum analysis module, and a reasonable spectrum resolution (e.g., 0.1Hz~1Hz) is set. Through spectrum analysis, the envelope time-domain signal is converted into a frequency domain distribution map, i.e., the envelope spectrum. The horizontal axis of the envelope spectrum is frequency, and the vertical axis is the amplitude of the spectral peaks. Among them, the spectral peaks with higher amplitudes are the spectral peaks that may correspond to the fault characteristics of the target component.

[0125] Optionally, after spectral analysis, the envelope spectrum can be normalized to map the amplitude of all spectral peaks to the range of 0 to 1, which facilitates the comparison between different spectral peaks, reduces the misjudgment of spectral peaks caused by excessive amplitude differences, and improves the accuracy of spectral peak identification.

[0126] Next, from the set of frequency peaks of the envelope spectrum, the target frequency peak with the smallest difference from the preset theoretical fault characteristic frequency is determined.

[0127] Specifically, the purpose of this step is to accurately select the spectral peaks corresponding to the fault characteristics of the target component from multiple spectral peaks in the envelope spectrum through theoretical frequency matching, thereby avoiding interference from irrelevant spectral peaks.

[0128] The preset theoretical fault characteristic frequency is the characteristic frequency that the target component should theoretically produce under fault conditions. The preset theoretical fault characteristic frequency can be derived from the structural parameters of the target component. For example, the theoretical fault characteristic frequency of an axle box bearing can be calculated using the formula: f=(n×d×ω) / (2×D), where n is the number of rolling elements, d is the diameter of the rolling elements, ω is the bearing speed, and D is the bearing pitch circle diameter. Optionally, the preset theoretical fault characteristic frequency can also be calibrated using a large amount of historical fault data, combined with fault monitoring data of similar components, to statistically obtain the characteristic frequency range of the target component under fault conditions, and the center value of the range is taken as the preset theoretical fault characteristic frequency. The preset theoretical fault characteristic frequency can be flexibly adjusted according to the model and specifications of the target component to adapt to different types of rail vehicle components.

[0129] Optionally, peak detection is performed on the envelope spectrum to filter out all peaks with amplitudes greater than a preset amplitude threshold (the preset amplitude threshold is set to 0.3~0.5 after normalization, and can be adjusted according to actual operating conditions). The frequencies corresponding to these peaks are extracted to form a set of peak frequencies. Peaks with amplitudes lower than the preset amplitude threshold are considered irrelevant noise peaks and are directly removed to reduce the computational load of subsequent matching.

[0130] Optionally, peak detection is performed on the envelope spectrum to filter out all peaks with amplitudes greater than a preset amplitude threshold and form a set of peak frequencies. The target peak is then determined from the set, and the actual characteristic frequency is identified. The quantification formula is:

[0131]

[0132] in, Indicates the frequency of all spectral peaks In the process, identify the frequency of fault characteristics that corresponds to the preset theoretical fault characteristics. The peak with the smallest difference, that is, the one closest to the theoretical value, It is the set of all significant spectral peak frequencies in the envelope spectrum. As a constraint, the spectrum is required. The amplitude of the spectral peak must be higher than the set value. It is used to filter out noise or insignificant peaks. A preset amplitude threshold is set based on the maximum peak value ratio, and the calculation formula is as follows: Where Amax is the highest spectral peak in the envelope spectrum, and ω is the relative significance coefficient, which is 0.1 to 0.2, meaning that only spectral peaks with amplitudes reaching 10% to 20% or more of the maximum peak value are retained.

[0133] Optionally, the absolute difference between each peak frequency in the set of peak frequencies and the preset theoretical fault characteristic frequency is calculated, and the peak with the smallest difference is determined as the target peak. Furthermore, a reasonable difference threshold can be set (the difference threshold is typically ±5% to ±8% of the preset theoretical fault characteristic frequency). If the minimum difference is greater than the difference threshold, it indicates that no fault characteristic peak of the target component is detected in the current envelope spectrum, and the target component can be determined to be in a normal state. If the minimum difference is less than or equal to the difference threshold, it indicates that the peak corresponds to the fault characteristic peak of the target component and is determined as the target peak.

[0134] It should be noted that if multiple spectral peaks have the same difference from the preset theoretical fault characteristic frequency and all of them are the minimum values, the amplitude stability of these spectral peaks can be further compared, and the spectral peak with the smallest amplitude fluctuation can be selected as the target spectral peak to ensure the accuracy of the target spectral peak.

[0135] Finally, the frequency corresponding to the target spectral peak is determined as the characteristic frequency of the vehicle target component.

[0136] This application provides a method for determining the characteristic frequencies of dynamic response data of rail vehicles. First, Hilbert transform is performed on the vehicle vibration response data to extract the envelope signal. Then, the envelope signal is analyzed to obtain the envelope spectrum. Finally, the target spectral peak with the smallest difference from the preset theoretical fault characteristic frequency is selected from the set of spectral peak frequencies to determine the component's characteristic frequency. This method can accurately capture the weak vibration characteristics generated by vehicle component faults, effectively eliminate other interferences, and improve the accuracy and reliability of characteristic frequency determination.

[0137] As an optional implementation, based on any of the above embodiments, after processing the dynamic response data generated by the vehicle on the operating route, the following steps are further included:

[0138] First, the dynamic response data is processed to remove the DC component in order to obtain the initial preprocessed data.

[0139] Specifically, the purpose of this step is to remove the superimposed DC component in the dynamic response data, eliminate the influence of static errors such as sensor zero drift, installation deviation, and ambient temperature drift, and ensure that the data can truly reflect the dynamic characteristics of vehicle vibration and line excitation.

[0140] The DC component in the dynamic response data mainly originates from three aspects: First, the zero-drift error of the vibration sensor itself (the small, fixed voltage / current signal output when the sensor has no vibration input); second, the static offset caused by sensor installation deviation (improper installation angle or tightness causes the sensor to be under slight stress for a long time, resulting in a fixed offset signal output); and third, the influence of environmental factors (such as temperature changes causing sensor sensitivity drift, generating a fixed offset signal). This type of DC component is static interference and does not contain any effective information about vehicle vibration or line excitation. If it is not removed, it will cause baseline shift in subsequent frequency domain analysis, interfering with the accurate positioning of the resonant frequency band.

[0141] Optionally, based on the time-domain characteristics of the dynamic response data (vibration acceleration signal, displacement vibration signal) of the rail vehicle, the mean method or the high-pass filtering method can be selected. The two methods can be flexibly selected according to the data acquisition scenario to ensure the DC removal effect without damaging the effective dynamic signal.

[0142] The mean method is suitable for scenarios where dynamic response data acquisition is stable and there are no obvious sudden interferences. Specifically, the raw dynamic response data (time-domain sequence) acquired by S201 is divided into data segments according to a fixed time window. The average value of all sampling points in each data segment is calculated, which is the estimated value of the DC component of that data segment. Then, the value of each sampling point in the data segment is subtracted from the corresponding average value to obtain the de-DC time-domain sequence. All data segments are traversed to complete the de-DC processing of all dynamic response data, and the initial preprocessed data is output. The mean method is simple to calculate, has strong real-time performance, does not require complex parameter settings, and can quickly achieve DC component removal.

[0143] The DC component is removed from the original dynamic response data using the mean method. The calculation formula is as follows:

[0144]

[0145] in, For sampling duration, This refers to the original vibration acceleration signal, i.e., the original dynamic response data. This represents the initial preprocessed data after removing the DC bias.

[0146] Among them, the high-pass filtering method is suitable for scenarios where the DC component fluctuates greatly and there is slight static drift in the dynamic response data. Specifically, a first-order or second-order RC high-pass filter is selected, and a reasonable cutoff frequency is set. The cutoff frequency is much lower than the line excitation frequency and the characteristic frequency of the vehicle components to ensure that the DC component and low-frequency static drift signal can be completely filtered out, while the effective dynamic signals of line excitation and vehicle component vibration are fully preserved. The original dynamic response data is input into the high-pass filter, and after filtering, the initial preprocessed data with the DC component removed is output.

[0147] Next, wavelet threshold denoising is used to denoise the initial preprocessed data to obtain clean dynamic response data.

[0148] Specifically, the purpose of this step is to remove random noise from the initial preprocessed data, such as electromagnetic interference, sensor noise, and external environmental noise. This type of noise is a high-frequency random interference with weak energy but wide distribution. If it is not removed, it will lead to false spectral peaks in the subsequent frequency domain analysis, interfering with the accuracy of resonance band positioning and characteristic frequency extraction.

[0149] Optionally, considering the non-stationary characteristics of the dynamic response data of rail vehicles (vibration signals fluctuate with changes in vehicle speed and operating conditions), a db4 wavelet basis or a db6 wavelet basis can be selected. These wavelet bases have advantages such as good compact support, high time-domain resolution, and strong anti-interference ability. They can accurately distinguish wavelet coefficients between signals and noise, and while eliminating noise, they can retain the weak vibration signals of early faults in vehicle components to the greatest extent possible, avoiding damage to effective signals during the noise elimination process.

[0150] Optionally, the number of wavelet decomposition layers is typically set to 3 to 5, depending on the sampling frequency of the dynamic response data (following the Nyquist sampling theorem). Specifically, the higher the sampling frequency (e.g., above 1000Hz), the more layers can be added (4 to 5 layers) to ensure that high-frequency noise can be adequately separated; the lower the sampling frequency (e.g., below 500Hz), the number of layers can be set to 3 to avoid excessive decomposition leading to increased computation and signal distortion. For example, when the sampling frequency is 1000Hz and the characteristic frequency of the target component is 10Hz to 500Hz, setting the number of wavelet decomposition layers to 4 layers can achieve accurate separation of high-frequency noise (above 500Hz) from the effective signal.

[0151] Among them, the wavelet threshold is the core parameter that distinguishes signal wavelet coefficients from noise wavelet coefficients. An adaptive threshold setting method can be used, which automatically calculates the threshold based on the noise intensity of the initial preprocessed data. The wavelet threshold formula is as follows: ,in, Let be the noise standard deviation (estimated by the first-level detail coefficients after wavelet decomposition), N be the number of sampling points, and ln be the natural logarithm function. This adaptive threshold can be dynamically adjusted according to the noise intensity in the data, adapting to the noise differences under different acquisition scenarios, and avoiding incomplete noise removal or damage to effective signals caused by a fixed threshold.

[0152] Optionally, the initial preprocessed data is input into the wavelet analysis module, and wavelet decomposition is performed according to the set wavelet basis and decomposition level to obtain the approximation coefficients (dominated by effective signal) and detail coefficients (dominated by noise) of each level; the detail coefficients of each level are thresholded using an improved soft threshold function to suppress the detail coefficients corresponding to noise; the processed detail coefficients and approximation coefficients are subjected to inverse wavelet transform to reconstruct the denoised signal, which is the clean dynamic response data.

[0153] Optionally, a modified soft thresholding function is used to threshold the detail coefficients of each layer to suppress the detail coefficients corresponding to noise. The formula for processing the detail coefficients is as follows:

[0154]

[0155] in, Represents the wavelet detail coefficients of the j-th layer. This indicates the result after thresholding. The absolute value represents the detail coefficients retained after noise reduction. The wavelet threshold is adaptively selected based on the noise estimation standard deviation and the signal length. for The sign function is defined as:

[0156]

[0157] Finally, the pure dynamic response data is normalized to obtain standardized dynamic response data.

[0158] Specifically, the purpose of this step is to unify the amplitude scale of the clean dynamic response data, eliminate the impact of signal amplitude differences under different acquisition scenarios (such as different vehicle speeds, different sensors, and different lines), and ensure the accuracy and consistency of subsequent resonant frequency band energy calculation and energy change rate comparison.

[0159] Optionally, based on the amplitude characteristics of the dynamic response data of the rail vehicle, the "linear normalization method" can be used for amplitude normalization. This method is simple to calculate, has no signal distortion, can map all data amplitudes to a fixed range, and preserves the relative amplitude relationship of the data (without changing the frequency domain characteristics of the signal), thus meeting the needs of subsequent energy calculation.

[0160] Optionally, all sampling points of the pure dynamic response data are traversed, and the normalization formula is used to calculate the standardized dynamic response data one by one. After normalization, the minimum and maximum amplitude values ​​of the pure dynamic response data need to be retained so that the signal amplitude can be inversely normalized after the feature frequency is extracted to restore the actual amplitude of the vehicle vibration.

[0161] This application provides a method for determining the characteristic frequencies of dynamic response data of rail vehicles. First, the DC component of the dynamic response data is removed to eliminate DC offset interference, resulting in initial preprocessed data. Then, wavelet threshold denoising is used to effectively remove noise, obtaining clean dynamic response data. Finally, amplitude normalization is performed. This improves data quality and reduces interference factors.

[0162] Figure 4 A flowchart illustrating a method for determining the characteristic frequencies of dynamic response data of a rail vehicle, provided in another embodiment of this application; as shown. Figure 4 As shown, as an optional implementation, based on any of the above embodiments, the following steps are also included:

[0163] S401. Calculate the dynamic response data using windowed Fourier transform to obtain the frequency domain energy corresponding to different times within the resonance band.

[0164] Specifically, the purpose of this step is to convert the dynamic response data in the time domain into a correspondence between "time and frequency domain energy", accurately capture the dynamic changes of energy in the resonant frequency band over time, and provide basic data for subsequent energy integration.

[0165] Optionally, based on the non-stationary characteristics of the dynamic response data of rail vehicles, the Hanning window can be selected as the windowing function. The Hanning window has good time-frequency focusing characteristics, which can effectively suppress spectral leakage, while taking into account both time domain resolution and frequency domain resolution. It can accurately capture the frequency domain energy changes at different times, and avoid excessive attenuation of the resonant frequency band energy by the window function.

[0166] Optionally, the window length of the windowed Fourier transform is set to match the sampling frequency and resonant frequency range of the dynamic response data. It is usually set to 1 / 10 to 1 / 5 of the sampling frequency. For example, when the sampling frequency is 1000Hz, the window length is set to 100 to 200 sampling points, corresponding to a time window of 0.1s to 0.2s. This ensures that the window length can cover the complete frequency components of the resonant frequency band and accurately reflect the instantaneous changes in energy, avoiding the energy change lag caused by an excessively large window length and the energy calculation distortion caused by an excessively small window length.

[0167] Optionally, to avoid abrupt energy changes between adjacent windows and reduce computational errors, the window overlap rate is set to 50%~75%, with 60% being preferred. This ensures reasonable overlap of sampling points between adjacent windows, making the frequency domain energy calculations at different times continuous and smooth. The number of transform points (FFT points) is set to an integer power of 2 greater than or equal to the window length. For example, when the window length is 100 sampling points, the number of transform points is set to 128 to ensure the computational efficiency and spectral resolution of the Fourier transform. The spectral resolution should be ≤0.5Hz to avoid loss of energy details within the resonant frequency band due to excessively low resolution.

[0168] Optionally, the preprocessed dynamic response data (time-domain sequence) is windowed according to a set window length and overlap rate. A Hanning window is applied to the time-domain signal within each window, and then a Fast Fourier Transform (FFT) is performed to convert the time-domain signal of each window into a frequency-domain signal (obtaining the frequency-domain distribution map of the corresponding time of the window). For each resonant frequency band obtained by S202 positioning, the frequency-domain energy values ​​corresponding to all frequency points within the resonant frequency band (e.g., 19Hz~21Hz) in the frequency-domain distribution map of each window are extracted. After summarizing, the frequency-domain energy corresponding to different times (each window corresponds to one time, and the time is the time of the center position of the window) within the resonant frequency band is obtained, forming a preliminary correspondence between "time and frequency-domain energy" to ensure that the frequency-domain energy of each time accurately corresponds to the energy contribution of the resonant frequency band.

[0169] Optionally, the total energy at different times within the resonant frequency band, i.e., the frequency band energy sequence, can be obtained by energy integration. The specific calculation formula is as follows:

[0170]

[0171] in, Indicates the signal in time Location, frequency The short-time Fourier frequency domain transform results are as follows: Indicates the first Multiplication analysis of frequency bands at time points The total energy in the frequency band energy sequence is called the energy value at a point in time, which will eventually form a curve showing how energy changes over time. This indicates that the energy at each frequency point is accumulated across the entire analysis frequency band. This is the center frequency of the frequency band, which is generally a harmonic of the known excitation frequency, for example... , The half-width of the frequency band is the range extending upwards and downwards from the center frequency. This refers to the frequency range of the entire analysis band, indicating that only this frequency band is considered when calculating energy. Represents frequency The power spectral density at a given frequency, i.e., the energy of that frequency component, is the square of the spectral amplitude.

[0172] S402. Based on the frequency range of the resonant frequency band, perform energy integration on the frequency domain energy corresponding to each moment to generate the frequency band energy sequence corresponding to the resonant frequency band.

[0173] Specifically, the purpose of this step is to convert the discrete frequency domain energy at each moment within the resonant frequency band into the total energy of the entire resonant frequency band at that moment, eliminate the error caused by energy fluctuations at discrete frequency points, and form a continuous and stable frequency band energy sequence.

[0174] Optionally, the frequency range of energy integration strictly corresponds to the resonant frequency band range determined by S202. If there are multiple resonant frequency bands, energy integration should be performed independently for each resonant frequency band to avoid mutual interference between the energies of different resonant frequency bands. The energy integration can adopt the "trapezoidal integration method", which is simple to calculate, highly accurate, and can adapt to the integration requirements of discrete frequency domain energy, avoiding the integration error caused by the rectangular integration method.

[0175] Optionally, for the discrete frequency domain energy of each time point and each resonant frequency band obtained in S401, the area between the curve formed by all discrete frequency points within the resonant frequency band and the horizontal axis is calculated using the trapezoidal integral method with frequency as the abscissa and frequency domain energy as the ordinate. This area is the total energy (band energy) of the resonant frequency band at that time point. By iterating through all times, the total energy of the resonant frequency band at each time point is calculated sequentially. The band energy of all times points is arranged in chronological order to generate the band energy sequence corresponding to the resonant frequency band. The length of the band energy sequence is consistent with the number of windows in the windowed Fourier transform. Each element corresponds to the total energy of the resonant frequency band at a time point, which can intuitively reflect the dynamic change law of the resonant frequency band energy over time.

[0176] Optionally, the generated frequency band energy sequence is smoothed using a moving average method. The moving window length is set to 3 to 5 time points. The frequency band energy sequence is traversed, and the average frequency band energy of each time point and its 2 to 4 adjacent time points is calculated. The original frequency band energy of that time point is replaced to eliminate the error caused by instantaneous energy fluctuations, ensure the continuity and stability of the frequency band energy sequence, and provide more accurate data support for the subsequent calculation of standard deviation and average value.

[0177] S403. Based on the frequency band energy sequence corresponding to the resonance frequency band, calculate the standard deviation and average value of the frequency band energy sequence respectively.

[0178] Specifically, the purpose of this step is to quantify the overall level and fluctuation of the frequency band energy by calculating the statistical parameters (standard deviation, mean) of the frequency band energy sequence.

[0179] The average value is the arithmetic mean of all elements in the frequency band energy sequence. It is used to characterize the average energy level of the resonant frequency band throughout the entire calculation period, reflecting the overall intensity of the line excitation. During the calculation, outliers in the frequency band energy sequence need to be removed, such as energy mutations caused by sudden line anomalies or instantaneous sensor errors. The criterion for removal is a value exceeding the average value by ±3 times the standard deviation, to avoid outliers affecting the accuracy of the average value calculation.

[0180] The standard deviation is used to characterize the degree of deviation of each element in the frequency band energy sequence from the average value, reflecting the fluctuation amplitude of the resonant frequency band energy. The smaller the standard deviation, the smaller the frequency band energy fluctuation and the more stable the line excitation; the larger the standard deviation, the greater the frequency band energy fluctuation, which may indicate the presence of vibration signals from vehicle components or abnormal line excitation.

[0181] It should be noted that if there are multiple resonant frequency bands, the standard deviation and average value of the energy sequence corresponding to each resonant frequency band need to be calculated separately to avoid interference between the statistical parameters of different resonant frequency bands and to ensure that the energy change rate calculation of each resonant frequency band is independent and accurate.

[0182] S404. The ratio of the standard deviation to the mean is determined as the energy change rate of the resonant frequency band.

[0183] Specifically, the purpose of this step is to complete the final calculation of the energy change rate of the resonant frequency band, combine the standard deviation (fluctuation amplitude) with the average value (overall level) to obtain a normalized parameter that can quantify the degree of energy fluctuation in the resonant frequency band, and provide a precise quantitative judgment basis for signal separation of S203.

[0184] Specifically, the energy change rate is the ratio of the standard deviation to the mean of the frequency band energy sequence. A normalized calculation method is used to eliminate the influence of differences in the absolute value of energy under different resonant frequency bands and different acquisition scenarios, so that the energy change rate can characterize the energy fluctuation characteristics of the resonant frequency band on a unified scale.

[0185] This application provides a method for determining the characteristic frequencies of dynamic response data of rail vehicles. It utilizes windowed Fourier transform to obtain the frequency domain energy at different times within the resonant frequency band, generates a frequency band energy sequence through energy integration, and then calculates its standard deviation and average value to obtain the ratio as the energy change rate. This method can accurately and quantitatively reflect the change of resonant frequency band energy over time, providing a basis for accurately identifying the dominant frequency band of line excitation and effectively separating line excitation interference, thereby improving the accuracy of determining the characteristic frequencies of target components of the vehicle.

[0186] Figure 5 This is a schematic diagram of the structure of a device for determining the characteristic frequency of dynamic response data of a rail vehicle according to an embodiment of this application, as shown below. Figure 5As shown, the characteristic frequency determination device for dynamic response data of a rail vehicle provided in this embodiment is located in an electronic device. The characteristic frequency determination device 50 for dynamic response data of a rail vehicle provided in this embodiment includes: an acquisition module 51, a resonance frequency band determination module 52, a separation module 53, and a characteristic frequency determination module 54.

[0187] Specifically, the acquisition module 51 is used to acquire dynamic response data generated by the vehicle on the operating route; the resonance frequency band determination module 52 is used to determine the resonance frequency band in the dynamic response that matches the known excitation frequency on the operating route; the separation module 53 is used to separate the dynamic response data according to the energy change rate of the resonance frequency band to obtain vehicle vibration response data; and the characteristic frequency determination module 54 is used to determine the characteristic frequency of the target component of the vehicle based on the vehicle vibration response data.

[0188] Optionally, the acquisition module 51, when acquiring the dynamic response data generated by the vehicle on the running line, is specifically used to: use a vibration acceleration sensor installed on the vehicle bogie to collect the original vibration signal generated by the target component of the vehicle during operation, and use the original vibration signal as dynamic response data.

[0189] Optionally, the resonance frequency band determination module 52, when determining the resonance frequency band in the dynamic response that matches the known excitation frequency based on the known excitation frequency on the operating line, is specifically used to: determine at least one analysis frequency band centered on an overtone of the known excitation frequency and having a preset bandwidth based on the known excitation frequency of the operating line, and determine the analysis frequency band as the resonance frequency band that matches the known excitation frequency.

[0190] Optionally, when the separation module 53 separates the dynamic response data according to the energy change rate of the resonant frequency band to obtain vehicle vibration response data, it is specifically used to: compare the energy change rate of each resonant frequency band with a preset sensitivity threshold; if the energy change rate of any resonant frequency band is less than the preset sensitivity threshold, determine the corresponding resonant frequency band as the dominant frequency band of the line excitation; use a band-stop filter to filter out the frequency components corresponding to the dominant frequency band of the line excitation, and determine the filtered signal as vehicle vibration response data.

[0191] Optionally, the characteristic frequency determination module 54, when determining the characteristic frequency of the target component of the vehicle based on the vehicle vibration response data, is specifically used for: performing a Hilbert transform on the vehicle vibration response data to extract the envelope signal of the vehicle vibration response data; performing spectral analysis on the envelope signal to obtain the envelope spectrum; determining the target spectral peak with the smallest difference from the preset theoretical fault characteristic frequency from the set of frequency peaks of the envelope spectrum; and determining the frequency corresponding to the target spectral peak as the characteristic frequency of the target component of the vehicle.

[0192] Optionally, the characteristic frequency determination device for dynamic response data of rail vehicles provided in this embodiment further includes a preprocessing module.

[0193] Optionally, the preprocessing module, after processing the dynamic response data generated by the vehicle on the operating route, performs the following steps: DC component removal processing on the dynamic response data to obtain initial preprocessed data; noise reduction processing on the initial preprocessed data using wavelet threshold denoising to obtain clean dynamic response data; and amplitude normalization processing on the clean dynamic response data to obtain standardized dynamic response data.

[0194] Optionally, the resonant frequency band determination module 52 is further configured to: calculate the dynamic response data using windowed Fourier transform to obtain the frequency domain energy corresponding to different times within the resonant frequency band; perform energy integration on the frequency domain energy corresponding to each time based on the frequency range of the resonant frequency band to generate the frequency band energy sequence corresponding to the resonant frequency band; calculate the standard deviation and average value of the frequency band energy sequence based on the frequency band energy sequence corresponding to the resonant frequency band; and determine the ratio of the standard deviation to the average value as the energy change rate of the resonant frequency band.

[0195] Figure 6 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application, as shown below. Figure 6 As shown, the electronic device 60 provided in this embodiment includes a processor 61 and a memory 62 communicatively connected to the processor 61.

[0196] The memory 62 stores computer execution instructions; the processor 61 executes the computer execution instructions stored in the memory 62 to implement the method provided in any of the above embodiments.

[0197] The program may include program code, which includes computer-executable instructions. Memory 62 may include high-speed RAM, and may also include non-volatile memory, such as at least one disk storage device.

[0198] In this embodiment, the memory 62 and the processor 61 are connected via a bus. The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. The bus can be divided into address bus, data bus, control bus, etc. For ease of representation, Figure 6The bus is represented by a single straight line, but this does not mean that there is only one bus or one type of bus.

[0199] This application also provides a computer-readable storage medium, including computer-executable instructions stored in the computer-readable storage medium, which, when executed by a processor, are used to implement the method provided in any of the above embodiments.

[0200] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the method provided in any of the above embodiments.

[0201] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to implement the solution of this embodiment according to actual needs.

[0202] Furthermore, the functional modules in the various embodiments of this application can be integrated into one processing unit, or each module can exist physically separately, or two or more modules can be integrated into one unit. The unit composed of the above modules can be implemented in hardware or in the form of hardware plus software functional units.

[0203] The integrated modules described above, implemented as software functional modules, can be stored in a computer-readable storage medium. These software functional modules, stored in a storage medium, include several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) or processor to execute some steps of the methods of the various embodiments of this application.

[0204] It should be understood that the aforementioned processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), etc. A general-purpose processor can be a microprocessor or any conventional processor. The steps of the method disclosed in this invention can be directly manifested as execution by a hardware processor, or execution by a combination of hardware and software modules within the processor.

[0205] The memory may include high-speed RAM, and may also include non-volatile storage (NVM), such as at least one disk storage device, and may also be a USB flash drive, external hard drive, read-only memory, disk or optical disc, etc.

[0206] The aforementioned storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk. The storage medium can be any available medium that can be accessed by a general-purpose or special-purpose computer.

[0207] An exemplary storage medium is coupled to a processor, enabling the processor to read information from and write information to the storage medium. Alternatively, the storage medium can be an integral part of the processor. The processor and storage medium can reside in an Application Specific Integrated Circuit (ASIC). Alternatively, the processor and storage medium can exist as discrete components in an electronic control unit or main control device.

[0208] Those skilled in the art will understand that all or part of the steps of the above-described method embodiments can be implemented by hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When executed, the program performs the steps of the above-described method embodiments; and the aforementioned storage medium includes various media capable of storing program code, such as ROM, RAM, magnetic disks, or optical disks.

[0209] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features therein. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of this application.

Claims

1. A method for determining the characteristic frequencies of dynamic response data of a rail vehicle, characterized in that, include: Acquire dynamic response data generated by the vehicle on the operating route; Based on the known excitation frequency on the operating line, the resonant frequency band in the dynamic response that matches the known excitation frequency is determined; Based on the energy change rate of the resonant frequency band, the dynamic response data is separated to obtain vehicle vibration response data; Based on the vehicle vibration response data, the characteristic frequencies of the target components of the vehicle are determined.

2. The method according to claim 1, characterized in that, The acquisition of dynamic response data generated by the vehicle on the operating route includes: Using a vibration acceleration sensor installed on the vehicle bogie, the original vibration signal generated by the target component of the vehicle during operation is collected, and the original vibration signal is used as the dynamic response data.

3. The method according to claim 1, characterized in that, The step of determining the resonant frequency band in the dynamic response that matches the known excitation frequency based on the known excitation frequency on the operating line includes: Based on the known excitation frequency of the operating line, at least one analysis frequency band centered on an overtone of the known excitation frequency and having a preset bandwidth is determined, and the analysis frequency band is determined as a resonant frequency band that matches the known excitation frequency.

4. The method according to claim 1, characterized in that, The step of separating and processing the dynamic response data according to the energy change rate of the resonant frequency band to obtain vehicle vibration response data includes: The energy change rate of each of the resonant frequency bands is compared with a preset sensitivity threshold; If the energy change rate of any of the resonant frequency bands is less than the preset sensitivity threshold, the corresponding resonant frequency band will be determined as the dominant frequency band for line excitation. A band-stop filter is used to filter out the frequency components corresponding to the dominant frequency band of the line excitation, and the filtered signal is determined as the vehicle vibration response data.

5. The method according to claim 1, characterized in that, The determination of the characteristic frequencies of the target vehicle component based on the vehicle vibration response data includes: Perform Hilbert transform on the vehicle vibration response data to extract the envelope signal of the vehicle vibration response data; The envelope signal is subjected to spectral analysis to obtain the envelope spectrum; From the set of frequency peaks of the envelope spectrum, determine the target frequency peak with the smallest difference from the preset theoretical fault characteristic frequency; The frequency corresponding to the target spectral peak is determined as the characteristic frequency of the target vehicle component.

6. The method according to any one of claims 1-5, characterized in that, Following the dynamic response data generated by the vehicle on the operating route, the system also includes: The dynamic response data is processed to remove the DC component in order to obtain initial preprocessed data; The initial preprocessed data is denoised using wavelet threshold denoising to obtain clean dynamic response data. The pure dynamic response data is subjected to amplitude normalization to obtain standardized dynamic response data.

7. The method according to any one of claims 1-5, characterized in that, Also includes: The dynamic response data is calculated using windowed Fourier transform to obtain the frequency domain energy corresponding to different times within the resonance band; Based on the frequency range of the resonant frequency band, the frequency domain energy corresponding to each moment is integrated to generate the frequency band energy sequence corresponding to the resonant frequency band. Based on the frequency band energy sequence corresponding to the resonant frequency band, calculate the standard deviation and average value of the frequency band energy sequence respectively; The ratio of the standard deviation to the average value is determined as the energy change rate of the resonant frequency band.

8. A device for determining the characteristic frequency of dynamic response data of a rail vehicle, characterized in that, include: The acquisition module is used to acquire dynamic response data generated by the vehicle on the operating route; The resonant frequency band determination module is used to determine the resonant frequency band in the dynamic response that matches the known excitation frequency based on the known excitation frequency on the operating line. The separation module is used to separate the dynamic response data according to the energy change rate of the resonant frequency band to obtain vehicle vibration response data. The characteristic frequency determination module is used to determine the characteristic frequencies of the target components of the vehicle based on the vehicle vibration response data.

9. An electronic device, characterized in that, include: A processor, and a memory communicatively connected to the processor; The memory stores computer-executed instructions; The processor executes computer execution instructions stored in the memory to implement the method as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer-executable instructions, which, when executed by a processor, are used to implement the method as described in any one of claims 1-7.

Images