Method for gearbox fault diagnosis of a drive train based on an encoder transfer error signal

Abstract

The present disclosure discloses a kind of transmission system gearbox fault diagnosis methods based on encoder transmission error signal, comprising: transmission system planetary gearbox input shaft encoder signal and output shaft encoder signal are collected;According to the encoder signal obtained by collection, calculate original transmission error signal TE (t), and the signal is de-counting trend and zero mean, obtain signal TE1 (t);Variational modal decomposition is carried out to signal TE1 (t), and the low-frequency intrinsic modal component representing the speed fluctuation trend term is extracted, using TE1 (t) to subtract the speed fluctuation trend term, obtain the de-trend transmission error signal TE2 (t) after decoupling speed fluctuation;According to time mark signal, time domain synchronous average analysis is carried out to de-trend transmission error signal TE2 (t), obtain time domain synchronous average signal, then it is carried out FFT spectrum analysis, when the amplitude ratio of meshing frequency and side frequency of fault state is greater than that of normal state, it can be judged that there is fault.

Description

Technical Field

[0001] This disclosure belongs to the field of planetary gearbox fault diagnosis technology, specifically relating to a fault diagnosis method for a transmission system gearbox based on encoder-transmitted error signals. Background Technology

[0002] Existing technologies commonly use vibration sensors to collect vibration signals from planetary gearboxes, and many effective fault diagnosis methods have been developed. However, due to the extremely complex structure of transmission systems, numerous working parts, and extremely limited remaining space, it is difficult to install vibration sensors to measure vibration signals. Furthermore, installing vibration sensors increases the complexity of the transmission system, and the vibration sensors themselves may also be a source of failure. Therefore, directly using built-in control signals to diagnose faults in bevel gearboxes is a feasible solution. Encoder signals are one of the most commonly used built-in control signals, and using encoder signals for fault diagnosis has many advantages. First, encoders are motion measurement devices, usually directly connected to the monitored critical components, resulting in shorter transmission paths and relatively higher signal fidelity. Second, unlike vibration signals, encoder signals primarily reflect the torsional vibration behavior of rotating machinery, making them more sensitive to stiffness losses caused by early faults such as gear cracks. Finally, since the encoder is already installed in the transmission system, its dynamic signals can be easily accessed from the control system, making it a zero-cost health assessment solution. However, existing vibration signal spectrum analysis techniques suffer from insufficient accuracy, large errors, and complex and unstable analysis.

[0003] The information disclosed in the background section is only intended to enhance the understanding of the background of the present invention, and therefore may contain information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0004] To address the shortcomings of existing technologies, the purpose of this disclosure is to provide a fault diagnosis method for gearboxes in transmission systems based on encoder transmission error signals. The transmission error signal, as a built-in sensing signal, has a shorter transmission path, better signal fidelity, no time-varying transmission path, and can directly reflect the torsional vibration state of the system. Therefore, using the transmission error signal for fault diagnosis yields a cleaner spectrum, simplifies fault identification, overcomes the shortcomings of traditional vibration signal spectrum analysis methods, and enables fault diagnosis directly using built-in control signals.

[0005] To achieve the above objectives, this disclosure provides the following technical solutions:

[0006] A method for fault diagnosis of a transmission system gearbox based on encoder-transmitted error signals includes the following steps:

[0007] S100: Acquires the input shaft encoder signal E of the planetary gearbox in the transmission system. in (t) and the output shaft encoder signal E out (t);

[0008] S200: Based on the input shaft encoder signal E in (t) and the output shaft encoder signal E out (t) Calculate the original transmission error signal TE(t), and perform decounting and zero-meaning on the original transmission error signal TE(t) to obtain the signal TE1(t);

[0009] S300: Perform variational mode decomposition on the signal TE1(t) to extract the low-frequency intrinsic mode component U, which represents the speed fluctuation trend. LF (t), using the signal TE1(t) minus the low-frequency intrinsic mode component U LF (t), to obtain the detrended propagation error signal TE2(t) after decoupling speed fluctuation;

[0010] S400: Calculates the rotational frequency, meshing frequency, and fault characteristic frequencies of the planetary gearbox, so that the amplitude can be compared based on the fault characteristic frequencies later.

[0011] S500: The input shaft time stamp signal Pulse(t) is obtained based on the input encoder signal. Then, the detrending propagation error signal TE2(t) is analyzed in the time domain based on the input shaft time stamp signal Pulse(t) to obtain the time domain synchronous average signal. Then, FFT spectrum analysis is performed on it to observe the meshing frequency harmonic components and their sideband amplitudes. When the meshing frequency and sideband amplitudes in the fault state are larger than those in the normal state, a fault is determined to exist.

[0012] In the method described, in step S100, the encoder is an absolute or incremental encoder with more than 100 lines.

[0013] In the method described above, in step S200, the method for calculating the original transmission error signal TE(t) is as follows:

[0014] TE(t)=E in (t)-R×E out (t),

[0015] Where R is the reduction ratio from the input shaft to the output shaft.

[0016] In the method described, in step S200, the de-counting trending is to remove the linear trend term of the original transmission error signal TE(t), and the zero-meaning is to remove the DC component of the original transmission error signal TE(t), as expressed by the following formula:

[0017]

[0018] Where Detrend(·) means removing the best-fit line from the original propagation error signal TE(t), and Mean(·) means calculating the mean of the original propagation error signal TE(t).

[0019] In the method described, step S300 involves the variational mode decomposition decomposing the signal into a series of eigenmode components by constructing and solving a variational problem. The steps are as follows:

[0020] S301: The intrinsic mode function is an amplitude-frequency modulated signal with a finite bandwidth, written as:

[0021] s k (t)=a k (t)cos(φ k (t))'

[0022] Among them, s k (t) is the signal to be solved, a k (t) is the amplitude function, φ k (t) is the phase function;

[0023] S302: The constructed variational problem is expressed as:

[0024]

[0025]

[0026] Among them, u k ={u1, u2, ..., u k} represents the individual intrinsic mode functions obtained from the decomposition, f represents the original signal to be decomposed, and ω represents the eigenmode functions. k ={ω1, ω2, ... ω k} represents the center frequency of each mode, δ(·) is the Dirac Delta function; the symbol * indicates convolution operation; K is the number of target intrinsic mode functions extracted; .j is the imaginary unit; π is pi; e is the natural constant;

[0027] S303: By using the augmented Lagrangian function, the constrained optimization problem of the constructed variational problem is equivalent to an unconstrained optimization problem. The augmented Lagrangian function can be expressed as:

[0028]

[0029] Where λ(t) is the Lagrange multiplier, α is the balance parameter of the data authenticity constraint; <·, ·> denote the inner product; the constrained optimization problem is solved by the Alternating Direction Method of Multipliers (ADMM);

[0030] S304: Solving for u k With ω k Then, find the minimum modal center frequency ω. kmin The corresponding intrinsic mode function u kminω Its low-frequency IMF component U represents the trend term of rotational speed fluctuation. LF (t).

[0031] In the method described, in step S300, the detrending propagation error signal TE2(t) after decoupling the speed fluctuation is obtained by the following formula:

[0032] TE2(t)=TE l (t)-U LF (t),

[0033] In the method described, in step S400, the component fault characteristic frequencies include the local fault characteristic frequencies of the sun gear and the local fault characteristic frequencies of the planet gears. The rotational frequency, meshing frequency, and component fault characteristic frequencies are calculated using the following formulas. The rotational frequency can be calculated using the following formula:

[0034]

[0035] Among them, f in n in These refer to the input shaft frequency and speed, respectively, f out n out These refer to the output shaft rotational frequency and rotational speed, respectively. The meshing frequency of the planetary gearbox can be calculated using the following formula:

[0036] f m =(f in -f out )×Z s

[0037] Among them, f m Planetary gearbox meshing frequency, Z s The sun gear has a number of teeth; the characteristic frequency of local faults in the sun gear can be calculated using the following formula:

[0038] f s =N*f m / Z s '

[0039] Among them, f sZ represents the characteristic frequency of partial failure of the sun gear, N is the number of planet gears, and Z is the frequency of partial failure of the sun gear. s The sun gear has the number of teeth; the characteristic frequency of local faults in the planetary gears can be calculated using the following formula:

[0040] f p1 =f m / Z p ;f p2 =2*f m / Z p ,

[0041] In the formula, Z p The number of teeth on the planetary gear indicates that a local fault in the planetary gear has two characteristic frequencies. The first characteristic frequency, f, occurs when the impact amplitude of the faulty tooth meshing with the sun gear differs from the impact amplitude of the tooth meshing with the ring gear. p1 When the difference between impact amplitudes is ignored, the second characteristic frequency f is... p2 .

[0042] In the method described above, in step S500, the input shaft time stamp signal Pulse(t) is obtained based on the input encoder signal. The time point or angle point corresponding to each revolution of the input shaft is found directly based on the equal time sequence signal or equal angle sequence signal recorded by the original encoder, and the corresponding input shaft time stamp signal Pulse(t) is obtained based on the sampling frequency.

[0043] In the method described above, step S500, the time-domain synchronous averaging and Fast Fourier Transform (FFT) of the detrending propagation error signal, is expressed as the following two equations:

[0044] TSAsig=TSA(TE2(t),Pulse(t),fs,f s f t f m (M, segN);

[0045] f(ω)=FFT(TSAsig,RSf)

[0046] Where TSAsig refers to the signal after time-domain synchronous averaging, fs refers to the original sampling frequency of the detrending propagation error signal, and f s The frequency of the solar wheel rotation, f t The frequency of the local fault characteristic of interest, f m The frequency of engagement is denoted by M, the number of average segments is denoted by segN, the number of characteristic periods in the obtained TSAsig signal is denoted by RSf, and f(ω) is the spectrum obtained after performing FFT on the time-domain synchronous average signal.

[0047] Compared with existing technologies, the beneficial effects of this disclosure are as follows: This disclosure enables fault diagnosis of planetary gearboxes in transmission systems by directly using the encoder's built-in sensing signal, i.e., the transmission error signal. Compared with traditional fault diagnosis using vibration signals, it has the following advantages: First, it is a built-in sensing signal, eliminating the need for additional vibration sensors, and can be considered a zero-cost fault diagnosis sensing solution. Second, the transmission error signal directly reflects the torsional vibration behavior of the gearbox, with a shorter transmission path, making it more sensitive to early faults such as gear cracks. Without complex time-varying transmission paths, the signal fidelity is relatively higher, and its spectrum is relatively simpler and cleaner, thus allowing for better extraction of fault characteristics. Attached Figure Description

[0048] Figure 1 This disclosure provides a flowchart of a method for diagnosing gearbox faults in a transmission system based on encoder-transmitted error signals, according to one embodiment.

[0049] Figure 2 This is a schematic diagram of the structure and sensor arrangement of a planetary gear test bench according to an embodiment of this disclosure;

[0050] Figure 3 This is a schematic diagram of a gear failure mode provided in one embodiment of the present disclosure;

[0051] Figure 4 This is a schematic diagram of a comparison of the original encoder signal in normal state and the original time-domain transmission error curves in different states calculated according to the transmission error calculation formula, provided in an embodiment of this disclosure.

[0052] Figures 5(a) to 5(c) This is a schematic diagram of the time-domain propagation error signal TE1(t) and its spectrum after the original time-domain TE signal has been decounted and zero-meaned, according to an embodiment of this disclosure;

[0053] Figures 6(a) and 6(b) show the low-frequency component U provided in one embodiment of this disclosure. LF A schematic diagram of the error signal TE2(t) and the speed trend transmission error signal;

[0054] Figures 7(a) and 7(b) are schematic diagrams of the time and frequency domains of normal and stripped signals after TSA when the frequency of interest is the planetary gear frequency, according to an embodiment of this disclosure.

[0055] Figures 8(a) and 8(b) are schematic diagrams comparing the biting frequency and its first-order sideband amplitude in the normal state and the peeling fault state provided in an embodiment of this disclosure. Detailed Implementation

[0056] The following will refer to the appendix. Figures 1 to 8(b)Specific embodiments of this disclosure are described in detail. While specific embodiments of this disclosure are shown in the accompanying drawings, it should be understood that this disclosure may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of this disclosure to those skilled in the art.

[0057] It should be noted that certain terms are used in the specification and claims to refer to specific components. Those skilled in the art will understand that different terms may be used to refer to the same component. This specification and claims do not distinguish components based on differences in terminology, but rather on differences in function. The terms "comprising" or "including" used throughout the specification and claims are open-ended and should be interpreted as "comprising but not limited to." The following descriptions are preferred embodiments for carrying out the invention; however, these descriptions are for the purpose of understanding the general principles of the specification and are not intended to limit the scope of the invention. The scope of protection of this disclosure is determined by the appended claims.

[0058] To facilitate understanding of the embodiments of this disclosure, further explanations and descriptions will be provided below with reference to the accompanying drawings and specific embodiments. The accompanying drawings do not constitute a limitation on the embodiments of this disclosure.

[0059] In one embodiment, such as Figure 1 As shown, a method for fault diagnosis of a transmission system gearbox based on encoder-transmitted error signals includes the following steps:

[0060] S100: Acquires encoder signal E from the input shaft of the planetary gearbox in the transmission system. in (t) and the output shaft encoder signal E out (t);

[0061] S200: Calculate the original transmission error signal TE(t) based on the acquired encoder signal, and perform de-counting trending and zero-meaning on the signal to obtain signal TE1(t);

[0062] S300: For signal TE l (t) Perform variational mode decomposition to extract the U-values ​​of the low-frequency intrinsic mode (IMF) components representing the rotational speed fluctuation trend. LF (t), use TE1(t) to subtract the speed fluctuation trend term U LF (t), to obtain the detrended propagation error signal TE2(t) after decoupling speed fluctuation;

[0063] S400: Calculates the rotational frequency, meshing frequency, and fault characteristic frequencies of the planetary gearbox, so that the amplitude can be compared based on the fault characteristic frequencies later.

[0064] S500: The input shaft time stamp signal Pulse(t) is obtained based on the input encoder signal. Then, the detrending propagation error signal TE2(t) is analyzed in the time domain based on the input shaft time stamp signal Pulse(t) to obtain the time domain synchronous average signal. Then, FFT spectrum analysis is performed on it to observe the meshing frequency harmonic components and their sideband amplitudes. When the amplitudes of the meshing frequency and sidebands in the fault state are larger than those in the normal state, it can be determined that a fault exists.

[0065] The above embodiments constitute the complete technical solution of this disclosure. Unlike existing technologies, the above embodiments use encoder-transmitted error signals for fault diagnosis, which has many advantages over using vibration signals. Firstly, as a built-in signal, it eliminates the need for additional vibration sensors. Secondly, the transmission path is short, making it more sensitive to early faults, and it lacks complex time-varying transmission path effects, resulting in a cleaner spectrum and easier extraction of fault features. Furthermore, variational mode decomposition can effectively extract the low-frequency intrinsic mode components representing the rotational speed fluctuation trend, thereby decoupling the rotational frequency and meshing frequency components of the signal. The presence of the encoder signal also eliminates the need for additional key phase signal sensors to obtain the time-domain synchronous averaging method required for fault diagnosis. Moreover, the time-domain averaging method of this disclosure is more accurate than fitting using key phase signals. Time-domain synchronous averaging of the detrended transmission error signal can also effectively reduce noise and highlight fault features.

[0066] In another embodiment, step S100 is characterized in that the encoder signal is acquired by an absolute or incremental encoder signal, and the encoder line count is preferably more than 100 lines.

[0067] In another embodiment, in step S200, the method for calculating the original transmission error signal is as follows:

[0068] TE(t)=E in (t)-R×E out (t)

[0069] Where R is the reduction ratio from the input shaft to the output shaft.

[0070] In another embodiment, in step S200, the de-counting trending refers to removing the linear trend term of the TE signal, and zero-meaning refers to removing the DC component of the TE signal, which can be expressed by the following formula:

[0071] TE(t) = Detrend(TE(t));

[0072] TE1(t) = TE(t) - Mean(TE(t))

[0073] In this context, Detrend(·) represents removing the best-fit line from the TE data, and Mean(·) represents calculating the mean of the TE signal.

[0074] In another embodiment, in step S300, the variational mode decomposition decomposes the signal into a series of intrinsic mode components (IMF components) by constructing and solving a variational problem. The specific steps are as follows:

[0075] S301: Assuming the intrinsic mode functions are finite bandwidth amplitude-frequency modulated signals, they can be written as:

[0076] s k (t)=a k (t)cos(φ k (t))

[0077] Among them, s k (t) is the signal to be solved, a k (t) is the amplitude function, φ k (t) is the phase function.

[0078] S302: The constructed variational problem is expressed as:

[0079]

[0080]

[0081] Among them, u k ={u1, u2, ..., u k} represents the individual eigenmode functions obtained from the decomposition, f is the original signal to be decomposed, and ω k ={ω1, ω2, ... ω k} represents the center frequency of each mode, δ(·) is the Dirac Delta function; the symbol * indicates the convolution operation; K is the number of target intrinsic mode functions extracted; j is the imaginary unit; π is pi; e is the natural constant;

[0082] S303: By using the augmented Lagrangian function, the constrained optimization problem of the constructed variational problem is equivalent to an unconstrained optimization problem. The augmented Lagrangian function can be expressed as:

[0083]

[0084] Where λ(t) is the Lagrange multiplier, α is the balance parameter of the data authenticity constraint; <·, ·> denote the inner product; the constrained optimization problem is solved by the Alternating Direction Method of Multipliers (ADMM);

[0085] S304: Solving for u k With ω k Then, find the minimum modal center frequency ω. kmin The corresponding intrinsic mode function u kminω u kminω The low-frequency IMF component representing the trend of rotational speed fluctuation is denoted as U below. LF (t).

[0086] In another embodiment, in step S300, the detrending propagation error signal TE2(t) after decoupling the speed fluctuation can be obtained by the following formula:

[0087] TE2(t)=TE1(t)-U LF (t)

[0088] In another embodiment, in step S400, the component failure characteristic frequencies include the local failure characteristic frequencies of the sun gear and the local failure characteristic frequencies of the planet gears. The rotational frequency, meshing frequency, and component failure characteristic frequencies are calculated using the following formula. The rotational frequency can be calculated using the following formula:

[0089]

[0090] Among them, f in n in These refer to the input shaft frequency and speed, respectively, f out n out These refer to the output shaft rotational frequency and rotational speed, respectively. The meshing frequency of the planetary gearbox can be calculated using the following formula:

[0091] f m =(f in -f out )×Z s

[0092] Among them, f m Planetary gearbox meshing frequency, Z s The sun gear has a number of teeth; the characteristic frequency of local faults in the sun gear can be calculated using the following formula:

[0093] f s =N*f m / Z s '

[0094] Among them, f s Z represents the characteristic frequency of partial failure of the sun gear, N is the number of planet gears, and Z is the frequency of partial failure of the sun gear. s The sun gear has the number of teeth; the characteristic frequency of local faults in the planetary gears can be calculated using the following formula:

[0095] f p1 =f m / Z p ;fp2 =2*f m / Z p ,

[0096] In the formula, Z p The number of teeth on the planetary gear indicates that a local fault in the planetary gear has two characteristic frequencies. The first characteristic frequency, f, occurs when the impact amplitude of the faulty tooth meshing with the sun gear differs from the impact amplitude of the tooth meshing with the ring gear. p1 When the difference between impact amplitudes is ignored, the second characteristic frequency f is... p2 .

[0097] In another embodiment, in step S500, obtaining the input shaft time stamp signal based on the input encoder signal means directly finding the time point or angle point corresponding to each revolution of the input shaft based on the equal time sequence signal or equal angle sequence signal recorded by the original encoder, and obtaining the corresponding input shaft time stamp signal Pulse(t) based on the sampling frequency.

[0098] In another embodiment, step S500, where the time-domain synchronous averaging and Fast Fourier Transform (FFT) of the detrending propagation error signal is expressed as the following two equations:

[0099] TSAsig=TSA(TE2(t),Pulse(t),fs,f s f t f m (M, segN);

[0100] f(ω)=FFT(TSAsig,RSf)

[0101] Where TSAsig refers to the signal after time-domain synchronous averaging, fs refers to the original sampling frequency of the detrending propagation error signal, and f s The frequency of the solar wheel rotation, f t The frequency of the local fault characteristic of interest, f m The frequency of engagement is denoted by M, the number of average segments is denoted by segN, the number of characteristic periods in the obtained TSAsig signal is denoted by RSf, and f(ω) is the spectrum obtained after performing FFT on the time-domain synchronous average signal.

[0102] In this embodiment, if the amplitude of the biting frequency in the spectrum of the fault signal after TSA at the frequency of interest is significantly different from that of the normal signal and there is a sideband with the frequency of interest as the sideband, it indicates that there is an obvious fault; otherwise, it indicates that there is no obvious fault.

[0103] The following is combined with Figures 2 to 8(b) The technical solution disclosed herein will be further described.

[0104] In a specific embodiment, such as Figure 2 As shown, a fault simulation experiment was conducted on a planetary gear fault simulation test bench. The gear fault modes are as follows: Figure 3 As shown, the specific diagnostic process is as follows:

[0105] 1. In the planetary gear test bench, encoders (Kistler sensors) are installed at both ends of the planetary gearbox to measure the encoder signal E on the input shaft of the planetary gearbox. in (t) and E out (t), such as Figure 2 As shown in the figure. This embodiment simulates a planetary gear peeling failure state. The planetary gearbox has an input frequency of 15Hz, a reduction ratio of 4.95, and an output frequency of 3.03Hz.

[0106] 2. Figure 4 The graph compares the original encoder signal under normal conditions with the original time-domain transmission error curves for different fault states calculated using the transmission error calculation formula. As can be seen from the graph, the original signal is a counting increment signal with a strong linear trend, which is one of the characteristics of encoder signals. The transmission error signal obtained after calculating the transmission error signal of the dual-ended encoder signal using the transmission error formula is shown below. Figure 4 As shown in the figure, the TE signal is still heavily influenced by the original linear trend. Although different faults can be classified based on this linear trend, it may not be universally applicable because it mainly relies on the influence of the linear components on the signal. Furthermore, judging the fault type solely from the linear trend coefficient is insufficient and lacks interpretability. Therefore, it is necessary to de-linearize the signal. Additionally, zero-mean normalization of the signal is also required.

[0107] 3. Figures 5(a) to 5(c) To analyze the original time-domain TE signal after removing the counting trend and zero-meaning, the time-domain propagation error signal TE1(t) and its spectrum are presented. The time-domain signal graph shows that even after removing the linear trend, the TE signal still exhibits abrupt changes with a frequency conversion period, as well as slowly varying periodic terms. The spectrum indicates that low-frequency signals still dominate, suggesting that fault information is still submerged in low-order harmonics such as the frequency conversion. Therefore, to extract fault information, it is necessary to fully demodulate the single-component signals of low-order harmonics such as the frequency conversion and obtain fault characteristic information from the high-frequency signal.

[0108] 4. Figure 6(a) shows the low-frequency component U after 10-level variational mode decomposition of TE1(t). LFFigure 6(t) shows a schematic diagram with 10 modes, a penalty coefficient of 2000, and a convergence tolerance of 1e-6. It can be seen that the low-frequency components effectively reflect the low-frequency speed fluctuation trend term in TE1(t). Figure 6(b) shows the speed trend propagation error signal TE2(t) obtained after subtraction. It can be directly seen from the figure that the remaining part consists of the meshing frequency sideband signal components. From the amplitude of the time-domain signal, it can be determined that the torsional vibration degree caused by different faults is inconsistent.

[0109] 5. The calculated rotational frequency, meshing frequency, and fault characteristic frequencies of each component of this planetary gearbox are shown in Table 1.

[0110] Table 1. Experimental Fault Characteristic Frequencies of Planetary Gearboxes

[0111] normal 235 0 Planetary gear peeling 235 8.10；16.20

[0112] 6. Figures 7(a) and 7(b) are the time-domain and frequency-domain plots of the normal TE signal and the stripped TE signal after time-domain synchronous averaging based on the planetary gear fault characteristic frequency of 8.10Hz. Figure 7(a) shows that the amplitude of the decoupled TE stripped signal after TSA differs significantly from the normal signal, exhibiting a clear fault impact. Both the stripped and normal signals show this impact, but the impact of the stripped signal is obviously greater. The presence of impact in the normal signal may be due to rotor imbalance and misalignment in the planetary gear system. Figure 7(b) shows that the grit frequency amplitude of the fault signal is much larger than that of the normal signal, especially noticeable in the higher harmonics of the grit frequency. Furthermore, there are sidebands around the grit frequency and its harmonics with the planetary gear fault characteristic frequency as the sideband, which are almost absent in the normal signal. Therefore, this indicates a localized fault in the planetary gear.

[0113] 7. Figures 8(a) and 8(b) are comparison diagrams of the first three meshing frequencies and their first-order fault sideband amplitudes obtained further from the spectra in Figures 7(a) and 7(b). Figure 8(a) compares the amplitudes of the meshing frequencies in the normal state and the spalling fault state. It can be seen from the figure that the amplitudes of the first three meshing frequencies in the spalling fault state are much larger than those in the normal state, especially the amplitude of the double meshing frequency. Figure 8(b) compares the amplitudes of the first-order fault sideband of the first three meshing frequencies in the normal state and the spalling fault state. It can be seen from the figure that the amplitude variation of the meshing frequency sideband is generally larger in the spalling fault state than in the normal state, while the variation is relatively gradual in the normal state. The amplitude of the first-order fault sideband of the double meshing frequency in the spalling fault state is close to three times that of the normal state. Therefore, based on the above analysis, it can be determined that the planetary gear has a local fault.

[0114] The transmission error signal (TE) can be calculated from the input and output shaft encoder signals of the monitored gearbox, and it can be used to describe transmission instability. Furthermore, it directly reflects the torsional vibration state of the gearbox, unlike traditional methods that rely solely on the output encoder signal to obtain the instantaneous angular velocity (IAS) and instantaneous angular acceleration (IAA) signals. The differential process is essentially a high-pass filtering process, which amplifies noise components in the instantaneous angular velocity and acceleration signals. Since fault information is often concentrated in the low-frequency range below 1kHz, using IAS and IAA increases the difficulty of fault feature extraction. Because the TE signal directly reflects the difference in torsional position fluctuations between the input and output shafts, it is more accurate and retains all the advantages of the encoder signal. Analyzing the TE signal can better identify whether a gearbox fault has occurred.

[0115] The above are merely preferred embodiments of this disclosure and are not intended to limit the implementation methods and protection scope of this disclosure. Those skilled in the art should recognize that any equivalent substitutions and obvious changes made using the content of this disclosure should be included within the protection scope of this disclosure.

Claims

1. A method for fault diagnosis of a gearbox in a transmission system based on encoder-transmitted error signals, characterized in that, Includes the following steps: S100: Acquires the input shaft encoder signal E of the planetary gearbox in the transmission system. in (t) and the output shaft encoder signal E out (t); S200: Based on the input shaft encoder signal E in (t) and the output shaft encoder signal E out (t) Calculate the original transmission error signal TE(t), and perform decounting and zero-meaning on the original transmission error signal TE(t) to obtain the signal TE1(t); S300: Perform variational mode decomposition on the signal TE1(t) to extract the low-frequency intrinsic mode component U, which represents the speed fluctuation trend. LF (t), using the signal TE1(t) minus the low-frequency intrinsic mode component U LF (t), to obtain the detrended propagation error signal TE2(t) after decoupling speed fluctuation; S4 00: Calculate the rotational frequency, meshing frequency, and fault characteristic frequencies of each component of the planetary gearbox so that the amplitude can be compared based on the fault characteristic frequencies later. S500: The input shaft time stamp signal Pulse(t) is obtained based on the input encoder signal. Then, the detrending propagation error signal TE2(t) is analyzed in the time domain based on the input shaft time stamp signal Pulse(t) to obtain the time domain synchronous average signal. Then, FFT spectrum analysis is performed on it to observe the meshing frequency harmonic components and their sideband amplitudes. When the meshing frequency and sideband amplitudes in the fault state are larger than those in the normal state, a fault is determined to exist. In step S200, the de-counting trend and zero-meaning refer to removing the linear trend term of the original transmission error signal TE(t), and the zero-meaning refer to removing the DC component of the original transmission error signal TE(t), as expressed by the following formula: ， Where Detrend(·) means removing the best-fit line from the original propagation error signal TE(t) data, and Mean(·) means calculating the mean of the original propagation error signal TE(t); In step S500, the time-domain synchronous averaging and Fast Fourier Transform (FFT) of the detrending propagation error signal are expressed as the following two equations: ， Where TSAsig refers to the signal after time-domain synchronous averaging, fs refers to the original sampling frequency of the detrending propagation error signal, and f s The frequency of the solar rotation, f t The frequency of the local fault characteristic of interest, f m The term refers to the meshing frequency, M refers to the average number of segments, segN refers to the number of characteristic periods in the obtained TSAsig signal, and RSf is the sampling frequency of TSAsig. This is the spectrum obtained after performing an FFT on the time-domain synchronous average signal.

2. The method according to claim 1, characterized in that, In step S100, the encoder is an absolute or incremental encoder with more than 100 lines.

3. The method according to claim 1, characterized in that, In step S200, the method for calculating the original transmission error signal TE(t) is as follows: ， Where R is the reduction ratio from the input shaft to the output shaft.

4. The method according to claim 1, characterized in that, In step S300, the variational mode decomposition decomposes the signal into a series of eigenmode components by constructing and solving a variational problem. The steps are as follows: S301: The intrinsic mode function is a finite bandwidth amplitude-frequency modulated signal, denoted as: ， in, The signal to be solved, It is an amplitude function. It is a phase function; S302: The constructed variational problem is expressed as: ， in, Let f be the original signal to be decomposed, and let f be the eigenmode functions obtained from the decomposition. For the center frequencies of each mode, For Dirac Delta functions; the symbol * indicates convolution operation; K is the number of target intrinsic mode functions extracted; j is the imaginary unit; Pi is the mathematical constant; e is the natural constant. S303: By using the augmented Lagrangian function, the constrained optimization problem of the constructed variational problem is equivalent to an unconstrained optimization problem. The augmented Lagrangian function is expressed as: ， in, For Lagrange multipliers, The balancing parameter for data authenticity constraints; The inner product is represented by the alternating direction multiplier method; the constrained optimization problem is solved by this method. S304: Solved to obtain and Then, find the minimum modal center frequency. The corresponding intrinsic mode function Its low-frequency IMF component U represents the trend term of rotational speed fluctuation. LF (t).

5. The method according to claim 4, characterized in that, In step S300, the detrending propagation error signal TE2(t) after decoupling the speed fluctuation is obtained by the following formula: TE2(t)= TE1(t)-U LF (t)。 6. The method according to claim 1, characterized in that, In step S400, the component fault characteristic frequencies include the local fault characteristic frequencies of the sun gear and the local fault characteristic frequencies of the planet gears. The rotational frequency, meshing frequency, and component fault characteristic frequencies are calculated using the following formulas, with the rotational frequency calculated using the following formula: ， in, , These refer to the input shaft frequency and rotational speed, respectively. , These refer to the output shaft rotational frequency and rotational speed, respectively. The meshing frequency of the planetary gearbox is calculated using the following formula: ， in, Planetary gearbox meshing frequency, The number of teeth on the sun gear; the characteristic frequency of partial failure of the sun gear is calculated by the following formula: ， in, Let N be the characteristic frequency of a partial failure of the sun gear, and N be the number of planet gears. The sun gear has the number of teeth; the characteristic frequency of local faults in the planetary gears is calculated using the following formula: ， In the formula, The number of teeth on a planetary gear indicates that a local fault in a planetary gear has two characteristic frequencies. The first characteristic frequency is defined as the difference between the impact amplitude of the faulty tooth meshing with the sun gear and the impact amplitude of the tooth meshing with the ring gear. When the difference between impact amplitudes is ignored, it is the second characteristic frequency. .

7. The method according to claim 1, characterized in that, In step S500, the input shaft time stamp signal Pulse(t) is obtained based on the input encoder signal. The time point or angle point corresponding to each revolution of the input shaft is found directly based on the equal time sequence signal or equal angle sequence signal recorded by the original encoder, and the corresponding input shaft time stamp signal Pulse(t) is obtained based on the sampling frequency.

Images