Rolling bearing fault diagnosis method for trackside acoustic monitoring array

Abstract

The invention relates to the technical field of bearing fault monitoring, and discloses a rolling bearing fault diagnosis method for a trackside acoustic monitoring array. According to the fault diagnosis method disclosed by the invention, the sound signal of the rolling bearing can be taken as input, the energy difference value is taken as an index for screening the target parameter combination of the FMD algorithm, and meanwhile, the power spectrum kurtosis is taken as an index for screening the target modal component; and finally, the screened target modal component is demodulated through a frequency domain weighted energy operator to extract the fault information of the rolling bearing, so that the fault of the rolling bearing is diagnosed through the sound signal of the rolling bearing, and the accuracy and efficiency during fault diagnosis can be effectively improved.

Description

Technical Field

[0001] The present invention relates to the technical field of bearing fault monitoring, and more specifically, to a rolling bearing fault diagnosis method for a trackside acoustic monitoring array. Background Art

[0002] The content of this part only provides background information related to the present invention, which may not constitute prior art.

[0003] As a key component in the operation of high-speed trains, if a bogie fails, it will seriously threaten the safety of passengers and goods. To ensure the smooth operation of the train, it is necessary to ensure that the bogie is in good working condition. Among them, the rolling bearing, as a key component in the bogie, undertakes most of the functions of bearing and transmitting loads. To ensure the safety of the train during operation, it is very important to monitor the working state of key components such as rolling bearings in the bogie and achieve fault diagnosis.

[0004] At present, for the rolling bearings in the bogie, a trackside monitoring system is usually used to obtain the vibration signals of the rolling bearings, and the faults of the rolling bearings are diagnosed based on the vibration signals. However, in addition to obtaining the vibration signals of the rolling bearings, the trackside detection system can also obtain the sound signals of the rolling bearings. At present, there are few methods for diagnosing the faults of rolling bearings based on the sound signals of rolling bearings. Summary of the Invention

[0005] In view of this, the purpose of the present invention is to provide a rolling bearing fault diagnosis method for a trackside acoustic monitoring array, specifically a method for diagnosing the faults of rolling bearings through the sound signals of rolling bearings.

[0006] The purpose of the present invention is achieved through the following technical solutions:

[0007] The present invention discloses a rolling bearing fault diagnosis method for a trackside acoustic monitoring array, including the following steps:

[0008] Step S1. Obtain the sound signal of the rolling bearing;

[0009] Step S2. Based on the sound signal obtained in step S1, determine the target parameter combination of the FMD algorithm; wherein, the target parameter combination includes the target number of modes and the target filter length;

[0010] Step S3. Decompose the sound signal obtained in step S1 by using the FMD algorithm with the target parameter combination to obtain the target modal components;

[0011] Step S4. Perform frequency-domain weighted energy operator demodulation on the target modal components to extract the fault feature information of the rolling bearing.

[0012] Further, step S2 specifically includes:

[0013] Step S21. Load the sound signal obtained in step S1 and initialize the iteration times of the FMD algorithm;

[0014] Step S22. Set the range of the decomposition mode number, the range of the filter length, and the search step size, and construct a parameter combination composed of different decomposition mode numbers and filter lengths as the input parameters of the FMD algorithm;

[0015] Step S23. Use the FMD algorithm to iteratively calculate the total energy difference related to the modal components under each parameter combination;

[0016] Step S24. Select the parameter combination with the largest total energy difference as the target parameter combination.

[0017] Further, in step S23, the total energy difference under a single parameter combination is obtained through the following steps:

[0018] Step S231. Decompose the sound signal into the corresponding number of modal components according to the decomposition mode number under a single parameter combination;

[0019] Step S232. Calculate the energy difference between each adjacent two modal components;

[0020] Step S233. Sum up the energy differences between each adjacent two modal components to obtain the total energy difference related to the modal components under a single parameter combination.

[0021] Further, in step S22, the range of the decomposition mode number is 2 - 8, the filter length is 100 - 150, and the search step size is 2.

[0022] Further, the filter length is an even number.

[0023] Further, step S3 specifically includes:

[0024] Step S31. Input the sound signal obtained in step S1 into the FMD algorithm and set the input parameters of the FMD algorithm according to the target parameter combination;

[0025] Step S32. Use the FMD algorithm to decompose the sound signal to output modal components with the number matching the target decomposition mode number;

[0026] Step S33. Calculate the power spectral kurtosis of each modal component output in step S32, and use the modal component with the largest power spectral kurtosis as the target modal component.

[0027] Further, in step S33, the calculation formula for the power spectrum kurtosis of the modal component is:

[0028]

[0029] In the above formula, K(f) is the power spectrum kurtosis of the modal component; p(t,f) represents the complex envelope at frequency f.

[0030] The technical solution of the embodiment of the present invention has at least the following advantages and beneficial effects:

[0031] The fault diagnosis method disclosed in the present invention can take the sound signal of the rolling bearing as the input, use the energy difference as the index for screening the target parameter combination of the FMD algorithm, and use the power spectrum kurtosis as the index for screening the target modal component. Finally, the frequency-domain weighted energy operator is used to demodulate the screened target modal component to extract the fault information of the rolling bearing. It not only realizes the diagnosis of the fault of the rolling bearing through the sound signal of the rolling bearing, but also can effectively improve the accuracy and efficiency of fault diagnosis. Description of the Drawings

[0032] Figure 1 It is a flowchart of the rolling bearing fault diagnosis method for the trackside acoustic monitoring array provided by the embodiment of the present invention;

[0033] Figure 2 It is a flowchart of determining the target parameter combination of the FMD algorithm provided by the embodiment of the present invention;

[0034] Figure 3 It is a flowchart of determining the target modal component provided by the embodiment of the present invention;

[0035] Figure 4 It is the time-domain diagram of 5 modal components in the specific embodiment 1 of the present invention;

[0036] Figure 5 For Figure 4 in the time domain of 5 modal components Figure 1 The corresponding frequency-domain diagram;

[0037] Figure 6 It is the power spectrum kurtosis distribution diagram of 5 modal components in the specific embodiment 1 of the present invention;

[0038] Figure 7 It is the time-domain diagram of the target modal component in the specific embodiment 1 of the present invention;

[0039] Figure 8 For Figure 7 in the time domain of the target modal component corresponding to the time-domain diagram;

[0040] Figure 9This is the distribution diagram of the total energy difference related to the modal components under each parameter combination in Specific Embodiment 2 of the present invention;

[0041] Figure 10 This is the time-domain diagram of the four modal components in Specific Embodiment 2 of the present invention;

[0042] Figure 11 For Figure 10 the time domain of the four modal components in Figure 1 one-to-one corresponding frequency-domain diagram;

[0043] Figure 12 This is the power spectral kurtosis distribution diagram of the four modal components in Specific Embodiment 2 of the present invention;

[0044] Figure 13 This is the time-domain diagram of the target modal component in Specific Embodiment 2 of the present invention;

[0045] Figure 14 For Figure 13 the frequency-domain diagram corresponding to the time-domain diagram of the target modal component in Specific Embodiments

[0046] To make the objectives, technical solutions, and advantages of the present invention clearer, the technical solutions of the present invention will be clearly and completely described below in conjunction with specific embodiments. The same reference numerals in the drawings represent the same components. It should be noted that the described embodiments are part of the embodiments of the present invention, rather than all of the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the described embodiments of the present invention without creative efforts fall within the scope of protection of the present invention.

[0047] Compared with the embodiments shown in the drawings, the feasible embodiments within the scope of protection of the present invention may have fewer components, have other components not shown in the drawings, different components, differently arranged components, or differently connected components, etc. In addition, two or more components in the drawings may be implemented in a single component, or a single component shown in the drawings may be implemented as multiple separate components.

[0048] The embodiments of the present invention disclose a rolling bearing fault diagnosis method for a trackside acoustic monitoring array, aiming to diagnose the faults of rolling bearings based on the sound signals of rolling bearings and improve the diagnosis efficiency and the accuracy of diagnosis results.

[0049] Figure 1 The flowchart of the rolling bearing fault diagnosis method for a trackside acoustic monitoring array disclosed in this embodiment is shown. As Figure 1 shown, the fault diagnosis method disclosed in this embodiment may include the following steps:

[0050] Step S1. Obtain the sound signal of the rolling bearing.

[0051] In step S1, the sound signal of the rolling bearing may specifically be the sound signal of the faulty rolling bearing, and this sound signal can be collected by the trackside acoustic monitoring array (such as an acoustic sensor array) in the trackside monitoring system.

[0052] Step S2. Based on the sound signal obtained in step S1, determine the target parameter combination of the FMD algorithm. Among them, the target combination parameters include the target decomposition mode number and the target filter length.

[0053] It can be understood that the FMD algorithm described in this embodiment is the feature mode decomposition algorithm. This algorithm is a non-stationary signal processing method that designs an FIR filter bank and continuously updates the filter coefficients through iteration, and finally decomposes the signal into different mode components to complete the evaluation of the fault cycle and the selection of decomposition modes.

[0054] It is known that the decomposition effect of the FMD algorithm depends on the quality of the input parameters (i.e., the decomposition mode number and the filter length). However, in the known methods of using the FMD algorithm to decompose the vibration signal of the rolling bearing to diagnose the fault of the rolling bearing, the optimal input parameters of the FMD algorithm are usually set by the analyst's experience, and optimization algorithms such as PSO (particle swarm optimization algorithm) and SSA (sparrow algorithm) are used to evaluate the input parameters using information such as kurtosis and entropy. In this way, since the kurtosis index is not sensitive to periodic signals, and the entropy index is greatly affected by noise, and the optimization algorithm requires a corresponding objective function, the accuracy of the finally determined optimal input parameters may be poor, and the calculation period is long.

[0055] Therefore, this embodiment provides another method for determining the optimal input parameters of the FMD algorithm, that is, the target parameter combination, so as to improve the accuracy of the determined target parameter combination as much as possible and shorten the calculation period.

[0056] Among them, Figure 2 shows the flowchart of determining the target parameter combination of the FMD algorithm in this embodiment. As Figure 2 shown, step S2 specifically includes:

[0057] Step S21. Load the sound signal obtained in step S1, and initialize the iteration times of the FMD algorithm, i = 1, j = 1.

[0058] Step S22. Set the decomposition mode number range, filter length range, search step size, and construct a parameter combination composed of different decomposition mode numbers and filter lengths as the input parameters of the FMD algorithm.

[0059] It should be noted that when setting the range of the number of decomposition modes, the number of decomposition modes should not be too small or too large. Specifically, if the number of decomposition modes is too small, it may cause some modes to be unable to effectively represent the complete component information, resulting in the phenomenon of "under-decomposition"; correspondingly, if the number of decomposition modes is too large, since the sum of the energies of the modal components of each layer of decomposition is equal to the original signal, therefore, an excessive number of decomposition modes may lead to too high an energy sum, resulting in the phenomenon of "over-decomposition", and too many decomposition modes will also lead to too long a calculation time. Preferably, in this embodiment, it is recommended to set the range of the number of decomposition modes to 2-8.

[0060] At the same time, the filter length range can be 100-150, and the search step size can be 2. And the filter can select a filter with an even length, that is, the filter length can be even, so as to achieve complete filtering of the signal and avoid the Gibbs effect with the window edge, thereby affecting the filtering effect.

[0061] At the same time, the parameter combinations composed of different numbers of decomposition modes and filter lengths described above can be understood as follows: assuming that the range of the number of decomposition modes is 2-8, and there are n possible numbers of decomposition modes, the filter length is 100-150, and there are m possible filter lengths, then there can be n×m possible parameter combinations. For example, assuming that there are 7 possible numbers of decomposition modes: 2, 3, 4, 5, 6, 7, 8; and there are 6 possible filter lengths: 100, 110, 120, 130, 140, 150, then there are a total of 6×7 = 42 possible parameter combinations. In this case, the various parameter combinations are as follows: when the number of decomposition modes is 2, there are 6 parameter combinations, and the filter lengths in the 6 parameter combinations are respectively: 100, 110, 120, 130, 140; when the number of decomposition modes is 3, there are also 6 parameter combinations, and the filter lengths in the 6 parameter combinations are respectively: 100, 110, 120, 130, 140; and so on. When the number of decomposition modes is 4, 5, 6, 7, 8 respectively, there are 6 parameter combinations for each number of decomposition modes, and the filter lengths in the 6 parameter combinations are all respectively: 100, 110, 120, 130, 140.

[0062] It can be understood that although only the cases where the filter lengths are 100, 110, 120, 130, 140, 150 are shown above, in combination with the foregoing, the filter length can also be any other even number between 100-150, for example, 102, 112, 132, 142, etc.

[0063] Step S23. Use the FMD algorithm to iteratively calculate the total energy difference related to the modal components under each parameter combination.

[0064] Specifically, in step S23, the total energy difference under a single parameter combination can be obtained through the following steps:

[0065] Step S231. Decompose the sound signal into the corresponding number of modal components according to the number of decomposition modes under a single parameter combination;

[0066] Step S232. Calculate the energy difference between each adjacent two modal components;

[0067] Step S233. Sum up the energy differences between each adjacent two modal components to obtain the total energy difference related to the modal components under a single parameter combination.

[0068] For example, for the parameter combination consisting of a decomposition mode number of 3 and a filter length of 100, the process of obtaining the total energy difference related to the modal components under this parameter combination is as follows: First, decompose the sound signal into three modal components, namely the first modal component, the second modal component, and the third modal component; Subsequently, calculate the energy differences between the first modal component and the second modal component, and between the second modal component and the third modal component; Finally, sum up the energy difference between the first modal component and the second modal component and the energy difference between the second modal component and the third modal component to obtain the total energy difference related to the modal components under the parameter combination consisting of a decomposition mode number of 3 and a filter length of 100.

[0069] Step S24. Select the parameter combination with the largest total energy difference as the target parameter combination. For example, assume that after calculation, the total energy difference under the parameter combination consisting of a decomposition mode number of 5 and a filter length of 142 is the largest. Then, the parameter combination consisting of a decomposition mode number of 5 and a filter length of 142 is used as the target parameter combination, that is, the target decomposition mode number is determined to be 5, and the target filter length is determined to be 142.

[0070] It should be noted that in this embodiment, by using the energy difference as the screening index to determine the target parameter combination of the FMD algorithm, compared with the aforementioned known determination methods, since the kurtosis index and entropy index are no longer used, the accuracy of the determined target parameter combination can be effectively improved. At the same time, since no objective function is required throughout the process of determining the target parameter combination, it is beneficial to reduce the calculation period.

[0071] Step S3. On the basis of obtaining the target parameter combination of the FMD algorithm in step S2, decompose the sound signal obtained in step S1 by using the FMD algorithm with the target parameter combination to obtain the target modal component. Among them, the target modal component is also the modal component finally used for diagnosing the faults of rolling bearings.

[0072] Among them, Figure 3The flowchart for obtaining the target modal component in this embodiment is shown. As Figure 3 shown, step S3 specifically includes:

[0073] Step S31. Input the sound signal obtained in step S1 into the FMD algorithm, and set the input parameters of the FMD algorithm according to the target parameter combination, that is, set the decomposition modal number and the filter length of the FMD algorithm according to the target decomposition modal number and the target filter length respectively.

[0074] For example, if the target parameter combination of the FMD algorithm is a parameter combination consisting of a decomposition modal number of 5 and a filter length of 142, then set the decomposition modal number and the filter length of the FMD algorithm to 5 and 142 respectively.

[0075] Step S32. Use the FMD algorithm to decompose the sound signal to output modal components whose quantity matches the target decomposition modal number. For example, in the case where the target decomposition modal number is 5, finally 5 modal components will be output.

[0076] Among them, the operation of "using the FMD algorithm to process the signal to output several modal components" in step S32 is a conventional operation of the known FMD algorithm in the prior art, and will not be elaborated here too much.

[0077] Step S33. Calculate the power spectrum kurtosis of each modal component output in step S32, and use the modal component with the maximum power spectrum kurtosis as the target modal component.

[0078] In step S33, the calculation formula for the power spectrum kurtosis of the modal component is:

[0079]

[0080] In the above formula, K(f) is the power spectrum kurtosis of the modal component; p(t,f) represents the complex envelope at frequency f.

[0081] It should be noted that in the known rolling bearing fault diagnosis methods, the correlation coefficient between the modal component and the original signal is usually calculated by using the FMD algorithm, and the target modal component is screened according to indexes such as time-domain spectral kurtosis. However, since the sound signal of the faulty rolling bearing has stronger periodicity and smaller correlation with the original signal, the modal component screened by using the time-domain spectral kurtosis may not accurately reflect the spectral characteristics of the sound signal of the rolling bearing. Therefore, in this embodiment, the power spectral kurtosis (that is, the frequency-domain spectral kurtosis) is used as an index to screen the target modal component, making full use of the characteristic that the frequency-domain spectral kurtosis is more sensitive to the transient components in the component signal. The screened target modal component can more accurately reflect the spectral characteristics of the sound signal of the rolling bearing, which is more conducive to extracting the fault characteristic information of the rolling bearing to be described below.

[0082] Step S4. Perform frequency-domain weighted energy operator demodulation on the target modal component to extract the fault characteristic information of the rolling bearing, such as the fault frequency of the rolling bearing.

[0083] Among them, performing frequency-domain weighted energy operator demodulation on the target modal component can obtain a spectrogram, and the fault characteristic information of the rolling bearing, such as the fault frequency of the rolling bearing, can be extracted from the spectrogram.

[0084] In this embodiment, the frequency-domain weighted energy operator is defined as follows:

[0085]

[0086] In the above formula, H[x(n)] is the Hilbert transform of the discrete signal x(n). And for the discrete signal x(n), three-point symmetric difference can be adopted, that is:

[0087]

[0088] At the same time, define H[x(n)] = h(n), and substitute the above formula (3) into formula (2), then for the discrete signal x(n), its frequency-domain weighted energy operator expression is:

[0089]

[0090] It should be noted that the method disclosed in the present invention performs demodulation on the target modal component by using the frequency-domain weighted energy operator. Compared with the known Hilbert envelope demodulation method in the prior art, the frequency-domain weighted energy operator is not easily interfered by other frequencies; at the same time, when calculating the instantaneous energy, the frequency-domain weighted energy operator only retains the instantaneous amplitude and ignores the influence of the instantaneous frequency, which helps to improve the accuracy of the extracted fault characteristic information of the rolling bearing and is beneficial to improving the calculation efficiency.

[0091] To more clearly and intuitively see the effectiveness of the fault diagnosis method disclosed in the above embodiments of the present invention, the fault diagnosis method disclosed in the above embodiments of the present invention will be further described below by taking the simulated sound signal and the measured sound signal of a rolling bearing with a fault as examples respectively.

[0092] Specific implementation case 1

[0093] In this specific implementation case, the simulated sound signal of a rolling bearing with a fault is used as experimental data. Among them, the fault of the rolling bearing is: damage to the outer ring of the bearing. It is known that the sampling frequency of the simulated sound signal is 12000 Hz, the rotational speed of the rolling bearing is 1800 r / min, the fault frequency fi of the outer ring of the bearing is 120 Hz, the number of sampling points is 18000, and the signal-to-noise ratio is -13 dB.

[0094] Among them, the parameters of the simulated sound signal are shown in Table 1 below.

[0095] Table 1 Parameters of the simulated sound signal

[0096]

[0097] On this basis, the fault diagnosis method disclosed in the foregoing embodiments of the present invention is used to analyze the simulated sound signal.

[0098] First, according to the set range of the number of decomposition modes and the range of filter lengths, a parameter combination composed of different numbers of decomposition modes and filter lengths is constructed. Based on the simulated sound signal of the rolling bearing, the total energy difference related to the modal components under each parameter combination is calculated by iterative calculation of the FMD algorithm, and the parameter combination with the largest total energy difference is selected as the target parameter combination of the FMD algorithm. Specifically, after calculation, the target parameter combination of this implementation case is: the target number of decomposition modes is 5, and the target filter length is 142.

[0099] Secondly, the target parameter combination is used as the input parameter of the FMD algorithm, that is, the number of decomposition modes of the FMD algorithm is set to 5, and the filter length is set to 142, so as to decompose the original simulated sound signal through the FMD algorithm to obtain 5 modal components.

[0100] Among them, Figure 4 and Figure 5 respectively show the time-domain diagram and the frequency-domain diagram of the 5 modal components. Combining the frequency-domain diagram, it can be seen that there is no obvious fault frequency in the frequency-domain diagram at this time, and there are many harmonic and noise components. Therefore, the power spectrum kurtosis of the 5 modal components is calculated respectively, and the calculation results are as shown in Figure 6 shown.

[0101] From Figure 6It can be seen that among the five modal components, the power spectrum kurtosis of the fourth modal component is the largest. Therefore, the fourth modal component is used as the target modal component for final fault diagnosis.

[0102] Finally, the fourth modal component, which is the target modal component, is demodulated using the frequency-domain weighted energy operator. The demodulated time-domain graph and its corresponding frequency-domain graph are respectively as Figure 7 and Figure 8 shown. Combining Figure 8 It can be seen that by using the modal component with the largest power spectrum kurtosis as the target modal component and demodulating it using the frequency-domain weighted energy operator, the noise part in the signal is well suppressed, and the fault frequency of the rolling bearing and its multiple harmonics can be intuitively obtained. Among them, Figure 8 the fault frequency in the shown frequency-domain graph is 120.01 Hz, with a very small error from the previously mentioned bearing outer ring fault frequency of 120 Hz, and the rotation frequency of 30 Hz can be identified.

[0103] Specific implementation case 2

[0104] In this specific implementation case, the measured sound signal data of the rolling bearing provided by Case Western Reserve University in the United States is used as the experimental data. These data have been widely used in the research of the rolling bearing fault diagnosis field, and their reliability and representativeness have been widely recognized.

[0105] Specifically, the rolling bearing used is the drive-end bearing (SKF6205-2RS). The specific parameters of this bearing are shown in Table 2 below:

[0106] Table 2 Structural parameters of the rolling bearing

[0107]

[0108] Calculation shows that the bearing outer ring fault frequency is 107.31 Hz, the rotational speed is 1797 r / min, the sampling frequency is 12000 Hz, the sampling duration is 1.5 s, and the signal-to-noise ratio of the noisy signal SNR = -21 dB.

[0109] On this basis, the fault diagnosis method disclosed in the foregoing embodiments of the present invention is used to analyze this simulated sound signal.

[0110] First, according to the set decomposition modal number range and filter length range, a parameter combination consisting of different decomposition modal numbers and filter lengths is constructed. Based on the measured sound signal of the rolling bearing, the total energy difference related to the modal components under each parameter combination is iteratively calculated through the FMD algorithm, and the parameter combination with the largest total energy difference is selected as the target parameter combination of the FMD algorithm. Specifically, combining Figure 9As shown, after calculation, when the number of decomposition modes is 4 and the filter length is 100, the total energy difference is the largest. Therefore, the target parameter combination of this embodiment is: the target number of decomposition modes is 4, and the target filter length is 100.

[0111] Secondly, use the target parameter combination as the input parameter of the FMD algorithm. That is, set the number of decomposition modes of the FMD algorithm to 4 and the filter length to 100, so as to decompose the original measured sound signal through the FMD algorithm to obtain 4 modal components.

[0112] Among them, Figure 10 and Figure 11 respectively show the time-domain diagram and frequency-domain diagram of the 4 modal components. Combining the frequency-domain diagram, it can be seen that there is no obvious fault frequency in the frequency-domain diagram at this time, and there are many harmonic and noise components. Therefore, calculate the power spectrum kurtosis of the 4 modal components respectively, and the calculation results are as shown in Figure 12 shown.

[0113] From Figure 12 it can be seen that among the 4 modal components, the power spectrum kurtosis of the third modal component is the largest. Therefore, the third modal component is used as the target modal component finally used for fault diagnosis.

[0114] Finally, use the frequency-domain weighted energy operator to demodulate the third modal component as the target modal component. The time-domain diagram after demodulation and its corresponding frequency-domain diagram are respectively as shown in Figure 13 and Figure 14 shown. Combining Figure 14 it can be seen that by using the modal component with the largest power spectrum kurtosis as the target modal component and demodulating it with the frequency-domain weighted energy operator, the noise part in the signal is well suppressed, and the fault frequency of the rolling bearing and its multiple harmonics can be intuitively obtained. Among them, Figure 14 the fault frequency in the frequency-domain diagram shown is 107.5Hz, with a very small error from the bearing outer ring fault frequency of 107.31Hz mentioned above, and the rotation frequency of 29.9Hz can be identified.

[0115] In summary, the fault diagnosis method disclosed in the embodiments of the present invention can take the sound signal of the rolling bearing as the input, use the energy difference as the index for screening the target parameter combination of the FMD algorithm, and at the same time use the power spectrum kurtosis as the index for screening the target modal component. Finally, use the frequency-domain weighted energy operator to demodulate the selected target modal component to extract the fault information of the rolling bearing, which not only realizes the diagnosis of the fault of the rolling bearing through the sound signal of the rolling bearing, but also can effectively improve the accuracy and efficiency of fault diagnosis.

[0116] The above are only the preferred embodiments of the present invention and are not intended to limit the present invention. For those skilled in the art, the present invention can have various modifications and changes. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. A rolling bearing fault diagnosis method for a trackside acoustic monitoring array, characterized in that: The steps include: Step S1. Acquire the sound signal of the rolling bearing; Step S2. Based on the sound signal obtained in step S1, determine the target parameter combination of the FMD algorithm; wherein the target parameter combination includes a target mode number and a target filter length; Step S3. Decomposing the sound signal obtained in step S1 using the FMD algorithm with the target parameter combination to obtain the target modal component; Step S4: Perform frequency domain weighted energy operator demodulation on the target modal component to extract the fault feature information of the rolling bearing.

2. The rolling bearing fault diagnosis method for trackside acoustic monitoring array according to claim 1 is characterized in that: Step S2 specifically includes: Step S21. Load the sound signal obtained in step S1 and initialize the number of iterations of the FMD algorithm; Step S22. Setting the range of decomposition mode numbers, filter length range, and search step length, and constructing parameter combinations consisting of different decomposition mode numbers and filter lengths as input parameters of the FMD algorithm; Step S23. Using the FMD algorithm to iteratively calculate the total energy difference associated with the modal component under each parameter combination; Step S24: Select the parameter combination with the largest total energy difference and use it as the target parameter combination.

3. The rolling bearing fault diagnosis method for trackside acoustic monitoring array according to claim 2 is characterized in that: In step S23, the total energy difference under a single parameter combination is obtained by the following steps: Step S231. Decompose the sound signal into a corresponding number of modal components according to the number of decomposition modes under a single parameter combination; Step S232. Calculate the energy difference between each two adjacent modal components; Step S233: summing up the energy differences between each two adjacent modal components to obtain a total energy difference associated with the modal component under a single parameter combination.

4. The rolling bearing fault diagnosis method for trackside acoustic monitoring array according to claim 2 is characterized in that: In step S22, the number of decomposed modes ranges from 2 to 8, the filter length is 100 to 150, and the search step is 2.

5. The rolling bearing fault diagnosis method for trackside acoustic monitoring array according to claim 2, characterized in that: The filter length is an even number.

6. The rolling bearing fault diagnosis method for trackside acoustic monitoring array according to claim 1, characterized in that: Step S3 specifically includes: Step S31. Input the sound signal obtained in step S1 into the FMD algorithm, and set the input parameters of the FMD algorithm according to the target parameter combination; Step S32. Decompose the sound signal using the FMD algorithm to output modal components whose number matches the target decomposition modal number; Step S33. Calculate the power spectrum kurtosis of each modal component output in step S32, so as to take the modal component with the largest power spectrum kurtosis as the target modal component.

7. The rolling bearing fault diagnosis method for trackside acoustic monitoring array according to claim 6 is characterized in that: In step S33, the calculation formula of the power spectrum kurtosis of the modal component is: In the above formula, K(f) is the power spectrum kurtosis of the modal component; p(t,f) represents the complex envelope at frequency f.

Images