Variable speed gearbox fault diagnosis method based on multi-harmonic time-frequency ridge line enhancement extraction
By using the multi-harmonic time-frequency ridge enhancement extraction method, the problem of insufficient accuracy in instantaneous angular velocity estimation in fault diagnosis of variable speed gearboxes is solved, achieving high-precision and efficient fault diagnosis, which is applicable to variable speed gearboxes and other rotating machinery.
Patent Information
- Application Number
- CN202211678230.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2022-12-26
- Publication Date
- 2026-02-06
- Estimated Expiration
- 2042-12-26
AI Technical Summary
Existing instantaneous angular velocity estimation methods lack accuracy and computational speed in fault diagnosis of variable speed gearboxes, making it difficult to meet the high-precision and high-speed diagnostic requirements of industrial sites. Traditional signal processing methods are also ineffective in diagnosing gearbox faults under variable speed conditions.
A multi-harmonic time-frequency ridge enhancement extraction method is adopted. Through steps such as downsampling, short-time Fourier transform, multi-harmonic time-frequency ridge enhancement, Vold-Kalman filtering and Hilbert transform, the instantaneous angular velocity and phase of the gearbox are calculated. After equal-angle resampling, Fourier transform and order spectrum analysis are performed to achieve fault diagnosis.
It improves calculation speed and accuracy, has high robustness and low noise sensitivity, is suitable for gearbox fault diagnosis under conditions of large speed fluctuations, expands the application range, and is applicable to other rotating machinery.
Smart Images
Figure CN116380451B_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The present application relates to the field of gear box fault, in particular to a variable speed gear box fault diagnosis method based on multi-harmonic time-frequency ridge line enhancement extraction. BACKGROUND
[0002] Gear box is a very important transmission and speed change device in mechanical equipment, which is widely used in wind power generation, aerospace, rail transportation and other fields. The fault rate of gear box is very high during transmission, and weak fault will seriously affect the stability and reliability of the whole gear transmission system. After the gear box fails, the downtime is long, the maintenance is difficult and the cost is high, which seriously affects the safety of life and property. Therefore, it is of great significance to diagnose the fault of gear box, to early detect weak fault and timely repair to ensure the safe operation of the whole machine.
[0003] The working state of gear box can be divided into constant speed and variable speed. In actual engineering application, due to the influence of load or speed-up and down, gear box is mostly operated under variable speed condition. Since the speed and load are time-varying, the vibration signal has non-stationary characteristics such as amplitude modulation, frequency modulation and phase modulation, which leads to "spectrum blur" phenomenon, making it difficult for traditional signal processing methods to diagnose the fault. The keyphasor order tracking based on equal-angle resampling has become one of the most effective variable speed gear box fault diagnosis methods.
[0004] For variable speed gear box fault diagnosis based on keyphasor order tracking, accurate estimation of instantaneous angular velocity is the key factor to determine whether it can accurately diagnose the fault of gear box. However, the estimation accuracy and calculation rate of existing instantaneous angular velocity estimation methods need to be further improved to meet the high-precision and high-speed diagnosis requirements of industrial field. SUMMARY
[0005] The purpose of the present application is to provide a variable speed gear box fault diagnosis method based on multi-harmonic time-frequency ridge line enhancement extraction, which comprises the following steps:
[0006] 1) Collecting the original vibration signal of variable speed rotating machinery;
[0007] 2) Down-sampling the original vibration signal, and then performing short-time Fourier transform on the down-sampled signal to obtain the time-frequency representation of the vibration signal;
[0008] 3) Processing the time-frequency representation of the vibration signal by using the multi-harmonic time-frequency ridge line enhancement extraction method, and calculating the instantaneous angular velocity of the reference shaft of the gear box;
[0009] 4) Using the instantaneous angular velocity of the reference shaft to perform Vold-Kalman filtering on the original vibration signal to extract the target harmonic component, and then performing Hilbert transform on the target harmonic component to obtain the instantaneous phase;
[0010] 5) Equal-angle resample the original vibration signal with instantaneous phase, convert the original time-domain non-stationary signal into angle-domain stationary signal;
[0011] 6) Fourier transform the resampled angle-domain stationary signal into order domain, conduct order spectrum analysis, and thus realize fault diagnosis.
[0012] Further, in step 2), the step of obtaining the time-frequency feature of the vibration signal comprises:
[0013] 2.1) resample the original vibration signal at a down-sampling frequency f sd to generate a down-sampled signal x(t);
[0014] 2.2) conduct short-time Fourier transform on the down-sampled signal to obtain the time-frequency representation TFR(t, f) of the vibration signal, i.e.
[0015]
[0016] wherein h(τ-t) is a window function; f is frequency; τ and t are time; x(τ) is the down-sampled signal;
[0017] 2.3) discretize the time-frequency representation TFR(t, f) to obtain the discrete time-frequency feature TFR(t n ,f m ) of the vibration signal, time t n =nΔt d , frequency f m =mΔf d ; n and m are time discrete points and frequency discrete points; n=1, 2,..., N, m=1, 2,..., M; N and M are the number of time discrete points and frequency discrete points, respectively; Δt d and Δf d are the time and frequency resolution of the time-frequency representation, respectively.
[0018] Further, in step 3), the step of calculating the instantaneous angular velocity of the reference shaft of the gearbox comprises:
[0019] 3.1) initialize parameters; set the initial search frequency center point f t(0) , the search frequency band bandwidth f w , and the harmonic order H i involved in ridge line extraction, i is the harmonic number;
[0020] 3.2) construct the cost function for multi-harmonic ridge line enhancement extraction, i.e.
[0021]
[0022] wherein fm ∈[f t(n-1) -f w / 2,f t(n-1) +f w / 2] as the search frequency; f t(n-1) is the frequency at the n-1 time point in the time-frequency representation matrix, which is taken as the center of the search band during the search process;
[0023] where the weight factor is as follows:
[0024]
[0025] 3.3) Minimizing the cost function to obtain the frequency f t(n) at the n time point in the time-frequency feature matrix, i.e.
[0026]
[0027] 3.4) Based on the frequency f t(n) at the n time point, the instantaneous angular velocity set V of the reference axis is established, i.e.
[0028] V = [f t(1) ,f t(2) ,...,f t(N) ] (5)
[0029] where f t(N) is the frequency at the N time point.
[0030] Further, in step 4), the step of obtaining the instantaneous phase includes:
[0031] 4.1) Calculating the instantaneous angular velocity V k of the target harmonic component, i.e.
[0032] V k = k · V = k · [f t(1) ,f t(2) ,...,f t(N) ] (6)
[0033] where k is the order of the target harmonic component;
[0034] 4.2) Using the instantaneous angular velocity V k of the target harmonic component to perform Vold-Kalman filtering on the original vibration signal to extract the target harmonic component, denoted as x k (t);
[0035] 4.3) Hilbert transforming the filtered single harmonic component x k (t) to calculate the instantaneous phase φ(t) of the gearbox reference axis, i.e.
[0036]
[0037] where unwrap[·] is an unwrap operation;
[0038] where x k the Hilbert transform of (t) as follows:
[0039]
[0040] Further, in step 5), the step of using the instantaneous phase to equally angle resample the original vibration signal comprises:
[0041] 5.1) determining the size of the equally angle sampling interval Δφ, i.e.
[0042]
[0043] where min{f(n)} is the minimum rotation frequency of the reference shaft of the gearbox, f s is the sampling frequency of the gearbox vibration signal;
[0044] 5.2) according to the instantaneous phase of the reference shaft of the gearbox, the vibration signal is resampled with an equally angle increment Δφ, and the time domain non-stationary signal is converted into an angle domain stationary signal x(φ).
[0045] Further, in step 6), the step of realizing fault diagnosis comprises:
[0046] 6.1) performing Fourier transform on the resampled angle domain stationary signal x(φ) to obtain an order spectrum x(O);
[0047] 6.2) identifying the side frequency characteristics in the order spectrum x(O) of the gearbox, identifying the fault source and the fault degree of the gearbox according to the interval and amplitude of the side frequency band, so as to realize the fault diagnosis of the gearbox.
[0048] The technical effects of the present application are self-evident, and the beneficial effects of the present application are as follows:
[0049] (1) The present application adopts a downsampling data preprocessing method, which improves the calculation rate under the premise of ensuring the accuracy.
[0050] (2) The present application extracts the time-frequency ridge line by using the method of sampling multi-harmonic time-frequency ridge line enhancement extraction, so that it has high robustness, high precision and low sensitivity to background noise.
[0051] (3) The present application is suitable for gearbox fault diagnosis under the condition of large rotation speed fluctuation, and has a wider application range, and has a good application prospect for other rotating machines in actual industrial fields. BRIEF DESCRIPTION OF DRAWINGS
[0052] Figure 1 Flowchart of the variable speed gearbox fault diagnosis method based on multi-harmonic time-frequency ridge line extraction of the application;
[0053] Figure 2 Time-domain graph of the wind turbine gearbox vibration signal in the embodiment of the application;
[0054] Figure 3 Time-domain graph of the down-sampled vibration signal in the embodiment of the application;
[0055] Figure 4 Time-frequency domain graph of the down-sampled vibration signal in the embodiment of the application;
[0056] Figure 5 Instantaneous angular velocity graph of the reference shaft in the embodiment of the application;
[0057] Figure 6 Instantaneous phase graph of the reference shaft in the embodiment of the application;
[0058] Figure 7 Time-domain graph of the equal-angle resampled signal in the embodiment of the application;
[0059] Figure 8 Order spectrum graph of the equal-angle resampled signal in the embodiment of the application. DETAILED DESCRIPTION
[0060] The application will be further described below in conjunction with the embodiments, but should not be understood as limiting the above-mentioned subject matter of the application to the following embodiments. According to the ordinary technical knowledge and common practice in the art, various substitutions and modifications can be made without departing from the technical idea of the application, and all of them should be included in the protection scope of the application.
[0061] Embodiment 1:
[0062] Referring to Figures 1 to 8 , the variable speed gearbox fault diagnosis method based on multi-harmonic time-frequency ridge line extraction, comprising the following steps:
[0063] 1) Collecting the original vibration signal of the variable speed rotating machinery;
[0064] 2) Down-sampling the original vibration signal, and then performing short-time Fourier transform on the down-sampled signal to obtain the time-frequency representation of the vibration signal;
[0065] 3) Processing the time-frequency representation of the vibration signal by using the multi-harmonic time-frequency ridge line extraction method, and calculating the instantaneous angular velocity of the reference shaft of the gearbox;
[0066] 4) Vold-Kalman filtering of the original vibration signal with the instantaneous angular velocity of the reference shaft to extract the target harmonic component, and then Hilbert transform of the target harmonic component to obtain the instantaneous phase;
[0067] 5) Equiangular resampling of the original vibration signal with the instantaneous phase to convert the original time-domain non-stationary signal into an angular-domain stationary signal;
[0068] 6) Fourier transform of the resampled angular-domain stationary signal into the order domain for order spectrum analysis, thereby realizing fault diagnosis.
[0069] In step 2), the step of obtaining the time-frequency feature of the vibration signal includes:
[0070] 2.1) Down-sampling the original vibration signal at a down-sampling frequency f sd Down-sampling the original vibration signal to generate a down-sampled signal x(t);
[0071] 2.2) Short-time Fourier transform of the down-sampled signal to obtain the time-frequency feature TFR(t, f) of the vibration signal, i.e.,
[0072]
[0073] where h(τ-t) is a window function; f is a frequency; τ and t are both time, and τ≠t; x(τ) is the down-sampled signal;
[0074] 2.3) Discretization of the time-frequency feature TFR(t, f) to obtain the discrete time-frequency feature TFR(t n ,f m ) of the vibration signal, where time t n =nΔt d , frequency f m =mΔf d ; n and m are time discrete points and frequency discrete points; n=1, 2,..., N, m=1, 2,..., M; N and M are the number of time discrete points and frequency discrete points, respectively; Δt d , and Δf d are the time and frequency resolutions of the time-frequency representation, respectively.
[0075] In step 3), the step of calculating the instantaneous angular velocity of the reference shaft of the gearbox includes:
[0076] 3.1) Initialization of parameters; setting the initial search frequency center point f t(0) , the search band bandwidth fw, and the harmonic order H i involved in ridge line extraction, i is the harmonic number;
[0077] 3.2) Construction of the cost function for multi-harmonic ridge line enhancement extraction That is,
[0078]
[0079] where f m ∈ [f t(n-1) -f w / 2,f t(n-1) +f w / 2] is the search frequency; f t(n-1) is the frequency at the n-1 time point in the time-frequency representation matrix, which is used as the center of the search band in the search process;
[0080] where the weight factor is as follows:
[0081]
[0082] 3.3) Minimizing the cost function to obtain the frequency f t(n) at the n time point in the time-frequency feature matrix, that is,
[0083]
[0084] 3.4) Based on the frequency f t(n) at the n time point, the instantaneous angular velocity set V of the reference axis is established, that is,
[0085] V = [f t(1) ,f t(2) ,...,f t(N) ] (5)
[0086] where f t(N) is the frequency at the N time point.
[0087] In step 4), the step of obtaining the instantaneous phase includes:
[0088] 4.1) Calculating the instantaneous angular velocity V k of the target harmonic component, that is,
[0089] V k = k · V = k · [f t(1) ,f t(2) ,...,f t(N) ] (6)
[0090] where k is the order of the target harmonic component;
[0091] 4.2) Using the instantaneous angular velocity V k of the target harmonic component to perform Vold-Kalman filtering on the original vibration signal to extract the target harmonic component, denoted as x k (t);
[0092] 4.3) the filtered single harmonic component x k (t) the Hilbert transform calculates the instantaneous phase φ(t) of the gearbox reference shaft, i.e.
[0093]
[0094] where unwrap[·] is the unwrapping operation;
[0095] where x k (t) is the Hilbert transform of the filtered single harmonic component x is given by:
[0096]
[0097] where τ represents a temporary symbol in the integral, which disappears after the integral is completed.
[0098] In step 5), the step of using the instantaneous phase to equally angle resample the original vibration signal includes:
[0099] 5.1) determining the size of the equally angle sampling interval Δφ, i.e.
[0100]
[0101] where min{f(n)} is the minimum rotational frequency of the gearbox reference shaft, f s is the sampling frequency of the gearbox vibration signal;
[0102] 5.2) resampling the vibration signal with the equally angle increment Δφ according to the instantaneous phase of the gearbox reference shaft, converting the time domain non-stationary signal into an angle domain stationary signal x(φ).
[0103] In step 6), the step of realizing fault diagnosis includes:
[0104] 6.1) performing Fourier transform on the resampled angle domain stationary signal x(φ) to obtain an order spectrum x(O);
[0105] 6.2) identifying the sideband features in the order spectrum x(O) of the gearbox, identifying the fault source and fault degree of the gearbox according to the interval and amplitude of the sideband, thereby realizing fault diagnosis of the gearbox.
[0106] Specifically, the interval and amplitude of the sideband are compared with the interval and amplitude of the inherent sideband of the gearbox to determine the fault source and fault degree of the gearbox.
[0107] Embodiment 2:
[0108] Referring to Figures 1 to 8 , the variable speed gearbox fault diagnosis method based on multi-harmonic time-frequency ridge line enhancement extraction includes the following steps:
[0109] 1) Collecting the vibration signal of the rotating machinery with variable rotating speed.
[0110] 2) Downsampling the original vibration signal, then performing short-time Fourier transform on the downsampled signal to obtain the time-frequency representation of the vibration signal.
[0111] The step of obtaining the time-frequency representation of the downsampled signal comprises:
[0112] 2.1) Downsampling the original vibration signal at a downsampled frequency f sd Downsampling the original vibration signal to generate a downsampled signal x(t);
[0113] 2.2) Performing short-time Fourier transform on the downsampled signal to obtain:
[0114]
[0115] where h(t) is a window function, thus the discrete time-frequency representation of the vibration signal can be expressed as TFR(t n ,f m ), t n =nΔt d , f m =mΔf d ; n and m are discrete points, n=1, 2,..., N, m=1, 2,..., M; Δt d , Δf d are the time and frequency resolutions of the time-frequency representation, respectively.
[0116] 3) Calculating the instantaneous angular velocity of the reference shaft of the gearbox using the multi-harmonic time-frequency ridge enhancement extraction method.
[0117] The step of calculating the instantaneous angular velocity of the reference shaft of the gearbox using the multi-harmonic time-frequency ridge enhancement extraction method comprises:
[0118] 3.1) Initializing parameters; setting the initial search frequency center point f t(0) , the search frequency band bandwidth f w , and the harmonic order H i involved in ridge extraction, i is the harmonic number.
[0119] 3.2) Constructing the cost function for multi-harmonic ridge enhancement extraction, which is expressed as:
[0120]
[0121] where f m ∈[f t(n-1) -f w / 2, f t(n-1) +f w / 2] is the search frequency, ft(n-1) is the frequency at the n-1 time point in the time-frequency representation matrix, which is used as the center of the search band in the search process, is a weight factor, whose expression is:
[0122]
[0123] 3.3) Minimizing the cost function to obtain the frequency f at the n time point in the time-frequency representation matrix t(n) , that is:
[0124]
[0125] The instantaneous angular velocity of the reference axis is:
[0126] V = [f t(1) ,f t(2) ,...,f t(N) ].
[0127] 4) Using the instantaneous angular velocity of the reference axis to perform Vold-Kalman filtering on the original vibration signal to extract the target harmonic component, and then performing Hilbert transform on the harmonic component to obtain the instantaneous phase.
[0128] The step of performing Hilbert transform on the harmonic component to obtain the instantaneous phase includes:
[0129] 4.1) Calculating the instantaneous angular velocity V k of the target harmonic component, whose expression is: V k = k·V = k·[f t(1) ,f t(2) ,...,f t(N) ]
[0130] In the formula, k is the order of the target harmonic component;
[0131] 4.2) Using the instantaneous angular velocity V k of the target harmonic component to perform Vold-Kalman filtering on the original vibration signal to extract the target harmonic component, denoted as x k (t);
[0132] 4.3) Hilbert transforming the extracted single harmonic component x k (t) to calculate the instantaneous phase φ(t) of the gearbox reference axis, whose expression is:
[0133]
[0134] In the formula, unwrap[·] is an unwrapping operation, is the Hilbert transform of the single harmonic component x k (t), whose calculation formula is:
[0135]
[0136] 5) Re-sampling the original vibration signal with equal angle increment by using the instantaneous phase, converting the original time-domain non-stationary signal into angle-domain stationary signal;
[0137] The step of re-sampling the original vibration signal with equal angle increment includes:
[0138] 5.1) Determining the equal angle sampling interval Δφ size:
[0139]
[0140] In the formula, min{f(n)} is the minimum rotation frequency of the reference shaft of the gearbox, f s is the sampling frequency of the gearbox vibration signal;
[0141] 5.2) Re-sampling the vibration signal with equal angle increment Δφ according to the instantaneous phase of the reference shaft of the gearbox, converting the time-domain non-stationary signal into angle-domain stationary signal x(φ).
[0142] 6) Fourier transform the re-sampled angle-domain stationary signal into order domain, and perform order spectrum analysis to realize fault diagnosis.
[0143] The step of performing order spectrum analysis to realize fault diagnosis includes:
[0144] 6.1) Fourier transform the re-sampled angle-domain stationary signal x(φ) to obtain order spectrum x(O);
[0145] 6.2) Identify the side frequency characteristics in the order spectrum of the gearbox, identify the fault source and fault degree of the gearbox according to the interval and amplitude of the side frequency band, and thus realize the fault diagnosis of the gearbox.
[0146] Embodiment 3:
[0147] This embodiment specifically illustrates the variable speed gearbox fault diagnosis method based on multi-harmonic time-frequency ridge line enhancement extraction of the present application through a wind power gearbox diagnosis experiment.
[0148] The meshing order of the experimental wind power gearbox relative to the reference shaft is shown in Table 1.
[0149] Table 1 Meshing frequency of the gearbox relative to the reference shaft
[0150]
[0151] The specific implementation steps are as follows:
[0152] Step 1: install vibration acceleration sensors at appropriate positions of the wind turbine gearbox, the sampling frequency of the acceleration sensors is 5 kHz, and the sampling duration is 550 s. The vibration signal x(t) of the variable speed gearbox is obtained, and the time-domain waveform of the original vibration is shown in Figure 2 .
[0153] Step 2: downsample the collected original vibration signal at a sampling frequency of 200 Hz, and the time-domain waveform of the downsampled signal is shown in Figure 3 . Then, short-time Fourier transform is performed on the downsampled signal to obtain the time-frequency representation of the downsampled signal, as shown in Figure 4 .
[0154] Step 3: estimate the instantaneous angular velocity of the reference axis from the three harmonic time-frequency ridges in the time-frequency representation of the downsampled signal using the multi-harmonic ridge enhancement extraction method, and the calculation result of the instantaneous angular velocity of the reference axis of the gearbox is shown in Figure 5 .
[0155] Step 4: calculate the instantaneous angular velocity of the target harmonic component to be separated according to the instantaneous angular velocity of the reference axis, and then perform Vold-Kalman filtering on the original vibration signal according to the instantaneous angular velocity of the target harmonic component to extract the target harmonic component. Then, phase demodulation is performed on the target harmonic component, i.e., the instantaneous phase of the harmonic component is calculated using Hilbert transform, and then the instantaneous phase of the reference axis is calculated according to the instantaneous phase of the target harmonic component, as shown in Figure 6 .
[0156] Step 5: resample the vibration signal with equal angle increments according to the instantaneous phase of the reference axis of the gearbox, convert the time-domain non-stationary signal to the angular-domain stationary signal, and the time-domain waveform of the resampled signal is shown in Figure 7 .
[0157] Step 6: perform Fourier transform on the angular-domain stationary signal to the order domain, perform order spectrum analysis to identify the sideband features, identify the fault source and fault degree of the gearbox according to the interval and amplitude of the sideband, and realize the fault diagnosis of the variable speed gearbox, and the order spectrum is shown in Figure 8 .
[0158] Experimental verification shows that the variable speed gearbox fault diagnosis method based on multi-harmonic time-frequency ridge extraction can effectively extract the fault features of the gearbox and can be applied to the fault diagnosis of the variable speed gearbox.
[0159] Example 4
[0160] The variable speed gearbox fault diagnosis method based on multi-harmonic time-frequency ridge enhancement extraction includes the following steps:
[0161] 1) collect the original vibration signal of the variable speed rotating machinery.
[0162] 2) down-sampling the original vibration signal, then short-time Fourier transform the down-sampled signal to obtain the time-frequency representation of the vibration signal;
[0163] 3) processing the time-frequency representation of the vibration signal by using the multi-harmonic time-frequency ridge enhancement extraction method to calculate the instantaneous angular velocity of the reference shaft of the gearbox;
[0164] 4) using the instantaneous angular velocity of the reference shaft to perform Vold-Kalman filtering on the original vibration signal to extract the target harmonic component, then performing Hilbert transform on the target harmonic component to obtain the instantaneous phase;
[0165] 5) using the instantaneous phase to perform equal-angle resampling on the original vibration signal to convert the original time-domain non-stationary signal into an angular-domain stationary signal;
[0166] 6) performing Fourier transform on the resampled angular-domain stationary signal to convert it into order-domain, performing order spectrum analysis to realize fault diagnosis.
[0167] Embodiment 5:
[0168] The multi-harmonic time-frequency ridge enhancement extraction-based variable-speed gearbox fault diagnosis method, the main content of which is shown in Embodiment 4, wherein in step 2), the step of obtaining the time-frequency representation of the vibration signal comprises:
[0169] 1) down-sampling the original vibration signal at a down-sampling frequency f sd down-sampling the original vibration signal to generate a down-sampled signal x(t);
[0170] 2) performing short-time Fourier transform on the down-sampled signal to obtain the time-frequency representation TFR(t, f) of the vibration signal, i.e.
[0171]
[0172] wherein h(τ-t) is a window function; f is a frequency; τ and t are times; x(τ) is a down-sampled signal;
[0173] 3) discretizing the time-frequency representation TFR(t, f) to obtain the discrete time-frequency representation TFR(t n ,f m ) of the vibration signal; time t n =nΔt d , frequency f m =mΔf d ; n and m are time discrete points and frequency discrete points; n=1, 2,..., N, m=1, 2,..., M; N and M are the number of time discrete points and frequency discrete points respectively; Δt d , Δf dTime and frequency resolution of the time-frequency representation, respectively.
[0174] Embodiment 6:
[0175] The variable speed gearbox fault diagnosis method based on multi-harmonic time-frequency ridge enhancement extraction, the main content of which is seen in embodiment 4, wherein in step 3), the step of calculating the instantaneous angular velocity of the reference shaft of the gearbox comprises:
[0176] 1) Initialize parameters; set the initial search frequency center point f t(0) , search band bandwidth f w , and harmonic order H i involved in ridge extraction, i is the harmonic number;
[0177] 2) Construct the cost function of multi-harmonic ridge enhancement extraction , that is:
[0178]
[0179] In the formula, f m ∈[f t(n-1) -f w / 2,f t(n-1) +f w / 2] is the search frequency; f t(n-1) is the frequency of the n-1 time point in the time-frequency representation matrix, which is used as the search band center in the search process;
[0180] Wherein, the weight factor is as follows:
[0181]
[0182] 3) Minimize the cost function to obtain the frequency f t(n) of the n time point in the time-frequency feature matrix, that is:
[0183]
[0184] 4) Based on the frequency f t(n) of the n time point, the instantaneous angular velocity set V of the reference shaft is established, that is:
[0185] V=[f t(1) ,f t(2) ,...,f t(N) ](4)
[0186] In the formula, f t(N) is the frequency of the N time point.
[0187] Embodiment 7:
[0188] The variable speed gearbox fault diagnosis method based on multi-harmonic time-frequency ridge line enhancement extraction, the main content of which is seen in embodiment 4, wherein in step 4), the step of obtaining the instantaneous phase includes:
[0189] 1) Calculate the instantaneous angular velocity V of the target harmonic component k , that is:
[0190] V k = k·V = k·[f t(1) ,f t(2) ,...,f t(N) ] (1)
[0191] In the formula, k is the order of the target harmonic component;
[0192] 2) Use the instantaneous angular velocity V of the target harmonic component k to perform Vold-Kalman filtering on the original vibration signal to extract the target harmonic component, denoted as x k (t);
[0193] 3) Calculate the instantaneous phase φ(t) of the gearbox reference shaft by Hilbert transforming the filtered single harmonic component x k (t), that is:
[0194]
[0195] In the formula, unwrap[·] is an unwrapping operation;
[0196] Wherein, the Hilbert transform of the single harmonic component x k (t) is as follows:
[0197]
[0198] In the formula, τ, t are time.
[0199] Embodiment 8:
[0200] The variable speed gearbox fault diagnosis method based on multi-harmonic time-frequency ridge line enhancement extraction, the main content of which is seen in embodiment 4, wherein in step 5), the step of using the instantaneous phase to perform equi-angle resampling on the original vibration signal includes:
[0201] 1) Determine the equi-angle sampling interval Δφ size, that is:
[0202]
[0203] In the formula, min{f(n)} is the minimum rotational frequency of the gearbox reference shaft, f s is the sampling frequency of the gearbox vibration signal;
[0204] 2) Resample the vibration signal with equal angle increment Δφ according to the instantaneous phase of the gearbox reference shaft, convert the time-domain non-stationary signal into angular-domain stationary signal x(φ).
[0205] Embodiment 9:
[0206] The variable speed gearbox fault diagnosis method based on multi-harmonic time-frequency ridge line enhancement extraction, the main content is seen in embodiment 4, wherein, in step 6), the steps of realizing fault diagnosis include:
[0207] 1) Fourier transform the resampled angular-domain stationary signal x(φ) to obtain order spectrum x(O);
[0208] 2) Identify the sideband features in the order spectrum x(O) of the gearbox, identify the fault source and fault degree of the gearbox according to the interval and amplitude of the sideband, so as to realize the fault diagnosis of the gearbox.
Claims
1. A fault diagnosis method for variable speed gearboxes based on multi-harmonic time-frequency ridge enhancement extraction, characterized in that, Includes the following steps: 1) Collect raw vibration signals from rotating machinery with varying rotational speeds; 2) Downsample the original vibration signal, and then perform a short-time Fourier transform on the downsampled signal to obtain the time-frequency representation of the vibration signal; 3) The time-frequency representation of the vibration signal is processed using the multi-harmonic time-frequency ridge enhancement extraction method to calculate the instantaneous angular velocity of the gearbox reference shaft; 4) Use the instantaneous angular velocity of the reference axis to perform Volk-Kalman filtering on the original vibration signal to extract the target harmonic components, and then perform Hilbert transform on the target harmonic components to obtain the instantaneous phase. 5) The original vibration signal is resampled at equal angles using the instantaneous phase to convert the original time-domain non-stationary signal into an angular-domain stationary signal; 6) Perform Fourier transform on the resampled angular domain stationary signal to the order domain, and perform order spectrum analysis to achieve fault diagnosis; Step 3), the steps for calculating the instantaneous angular velocity of the gearbox reference shaft include: 3.1) Initialize parameters; set the initial search frequency center point f t(0) Search bandwidth f w And the harmonic order H involved in ridge line extraction i , where i is the harmonic number; 3.2) Constructing the cost function for multiharmonic ridge enhancement extraction Right now: In the formula, f m ∈[f t(n-1) -f w / 2,f t(n-1) +f w [ / 2] represents the search frequency; f t(n-1) The frequency at time n-1 in the time-frequency representation matrix serves as the center of the search band during the search process; Among them, weighting factors As shown below: 3.3) Minimize the cost function to obtain the frequency f at time point n in the time-frequency feature matrix. t(n) ,Right now: 3.4) Frequency f based on time point n t(n) Establish the set of instantaneous angular velocities V along the reference axis, that is: V=[f t(1) ,f t(2) ,...,f t(N) ] (5) In the formula, f t(N) Let N be the frequency at time N.
2. The method for fault diagnosis of variable speed gearboxes based on multi-harmonic time-frequency ridge enhancement extraction according to claim 1, characterized in that, Step 2), the steps for obtaining the time-frequency representation of the vibration signal include: 1) Reduce the sampling frequency f sd The original vibration signal is downsampled to generate a downsampled signal x(t); 2) Perform a short-time Fourier transform on the downsampled signal to obtain the time-frequency representation of the vibration signal, TFR(t, f), that is: In the formula, h(τ-t) is the window function; f is the frequency; τ and t are time; and x(τ) is the downsampled signal. 3) Discretize the time-frequency representation TFR(t, f) to obtain the discrete time-frequency representation TFR(t) of the vibration signal. n ,f m ), time t n =nΔt d Frequency f m =mΔf d n and m are the time discrete points and frequency discrete points, respectively; n = 1, 2, ..., N, m = 1, 2, ..., M; N and M are the number of time discrete points and frequency discrete points, respectively; Δt d , Δf d These represent the time and frequency resolutions of the time-frequency representation, respectively.
3. The method for fault diagnosis of variable speed gearboxes based on multi-harmonic time-frequency ridge enhancement extraction according to claim 1, characterized in that, Step 4) involves obtaining the instantaneous phase, including: 1) Calculate the instantaneous angular velocity V of the target harmonic component. k ,Right now: V k =k·V=k·[f t(1) ,f t(2) ,...,f t(N) ](6) In the formula, k is the order of the target harmonic component; 2) Utilizing the instantaneous angular velocity V of the target harmonic component k The original vibration signal is subjected to Volk-Kalman filtering to extract the target harmonic component, denoted as x. k (t); 3) For the filtered single harmonic component x k The Hilbert transform calculates the instantaneous phase φ(t) of the gearbox reference shaft, i.e.: In the formula, unwrap[·] represents the dewrap operation; Among them, the single harmonic component x k Hilbert transform of (t) As shown below: In the formula, τ and t represent time.
4. The method for fault diagnosis of variable speed gearboxes based on multi-harmonic time-frequency ridge enhancement extraction according to claim 1, characterized in that, Step 5), the step of resampling the original vibration signal at equal angles using the instantaneous phase, includes: 1) Determine the magnitude of the equal-angle sampling interval Δφ, i.e.: In the formula, min{f(n)} is the minimum rotational frequency of the gearbox reference shaft, f s The sampling frequency of the gearbox vibration signal; 2) Based on the instantaneous phase of the gearbox reference shaft, the vibration signal is resampled with equal angular increments Δφ to convert the time-domain non-stationary signal into an angular-domain stationary signal x(φ).
5. The method for fault diagnosis of variable speed gearboxes based on multi-harmonic time-frequency ridge enhancement extraction according to claim 4, characterized in that, Step 6) includes the following steps for fault diagnosis: 1) Perform Fourier transform on the resampled angular domain stationary signal x(φ) to obtain the order spectrum x(O); 2) Identify the sideband characteristics in the order spectrum x(O) of the gearbox, and identify the fault source and fault degree of the gearbox based on the spacing and amplitude of the sidebands, thereby realizing the fault diagnosis of the gearbox.