Planetary gear transmission fault detection method based on frequency domain characteristics

Abstract

The invention relates to the technical field of planetary gear fault detection, in particular to a planetary gear transmission fault detection method based on frequency domain characteristics. The method comprises the following steps: acquiring a vibration signal of the planetary gear transmission system; decomposing the vibration signal by using an improved variational mode decomposition algorithm to obtain a plurality of IMF components; performing envelope spectrum analysis on the IMF component to detect whether the planetary gear transmission system has a fault or not; wherein the improved variational mode decomposition algorithm comprises a self-adaptive mode decomposition number and a self-adaptive secondary penalty factor. According to the scheme of the invention, the transmission fault of the planetary gear can be accurately detected.

Description

Technical Field

[0001] This invention relates to the field of planetary gear fault detection technology. More specifically, this invention relates to a planetary gear transmission fault detection method based on frequency domain characteristics. Background Technology

[0002] Planetary gear transmission systems are widely used in key equipment such as industrial robots and planetary reducers due to their advantages of compact structure, high load-bearing capacity, and high transmission efficiency.

[0003] In the field of industrial robots, planetary reducers are the core components of joint drives. They are responsible for converting the high speed and low torque of the motor into the low speed and high torque required by the robot end effector, ensuring precise and smooth movement.

[0004] However, the unique planetary carrier rotation and multi-gear meshing mechanism of planetary gears cause their vibration signals to exhibit complex non-stationary modulation characteristics, containing a large number of sideband signals. In order to effectively extract fault features, variational mode decomposition (VMD), as a signal decomposition technique with a solid theoretical foundation, is widely used.

[0005] VMD can decompose complex signals into several intrinsic mode functions (IMFs) with specific center frequencies, effectively avoiding the endpoint effects of traditional empirical mode decomposition. However, existing VMD methods have significant limitations in application, as follows: The core parameters of Variational Mode Decomposition (VMD) include the number of mode decompositions and the second-order penalty factor. The number of mode decompositions and the second-order penalty factor are typically preset and kept fixed based on human experience. However, planetary gear transmission systems usually operate under non-stationary conditions with fluctuating speeds or varying loads, and their signal spectral structure (the number and bandwidth of frequency components) changes dynamically over time.

[0006] Therefore, the fixed number of modal decompositions and the quadratic penalty factor cannot adapt to such dynamic changes, causing the algorithm to frequently suffer from "modal aliasing" (such as the mixing of fault features with normal meshing frequencies) or "signal loss" (such as misjudging weak fault impacts as noise filtering) when processing signals under varying operating conditions, which seriously affects the detection accuracy of early weak faults.

[0007] Therefore, it is particularly important to detect planetary gear transmission faults more accurately. Summary of the Invention

[0008] The purpose of this invention is to propose a fault detection method for planetary gear transmission based on frequency domain characteristics, so as to solve the problem that the existing technology cannot accurately detect faults in planetary gear transmission; to this end, the present invention provides a solution in one aspect.

[0009] The planetary gear transmission fault detection method based on frequency domain features provided by this invention includes: Acquire vibration signals from the planetary gear transmission system; The vibration signal is decomposed into multiple IMF components using an improved variational mode decomposition algorithm. Envelope spectrum analysis is performed on the IMF components to detect whether there is a fault in the planetary gear transmission system; The improved variational mode decomposition algorithm includes an adaptive number of mode decompositions and an adaptive quadratic penalty factor. The adaptive number of mode decompositions is positively correlated with both the logarithm of the frequency domain discretization index and the significance factor. The adaptive quadratic penalty factor is positively correlated with the significance factor and negatively correlated with the frequency domain discretization index. The frequency domain discretization index is used to characterize the sparsity of the energy distribution of the vibration signal's spectrum. The significance factor is used to characterize the strength of the modulation sideband structure in the vibration signal's spectrum.

[0010] The above-mentioned method, by introducing frequency domain discrete index and significance factor to adaptively determine the number of mode decompositions and the second-order penalty factor of variational mode decomposition, solves the problem that fixed variational mode decomposition parameters in the prior art easily lead to mode aliasing or signal loss under planetary gear variable operating conditions (speed fluctuation, load change). It can adjust the decomposition strategy according to the dynamic changes of the signal spectrum structure, thereby achieving accurate extraction of weak fault features of planetary gear transmission system, and significantly improving the accuracy and robustness of fault detection.

[0011] Optionally, the frequency domain discrete index for: ; In the formula, The amplitude of the spectrum, To set the upper frequency limit of the frequency band, log() is a logarithmic function. The gradient of the spectral amplitude. It is an absolute value function.

[0012] The aforementioned frequency domain discreteness index can effectively reflect the intensity of oscillations and the amount of non-stationary frequency components in the spectrum curve, providing an accurate quantitative basis for subsequent judgment of signal complexity and determination of the number of mode decompositions, and ensuring the rationality of parameter adaptive adjustment.

[0013] Optionally, the significance factor for: ; In the formula, For frequency domain discrete index, The number of main peaks in the vibration spectrum. and The first The amplitude and frequency of each main peak. The average of all main peak amplitudes. The variance of the amplitude of all major peaks, This is the theoretical gear meshing frequency. The specified bandwidth range is given by exp(), which is an exponential function. It is an absolute value function.

[0014] The above scheme constructs a significance factor by comprehensively considering the number of main peaks, amplitude variance, and distance from the theoretical meshing frequency. This factor can effectively distinguish between broadband noise and the unique modulation sidebands caused by gear faults, thereby accurately characterizing the existence intensity of the modulation sideband structure. This provides a key basis for the selection of the secondary penalty factor and enhances the algorithm's sensitivity to specific fault characteristics.

[0015] Optionally, the main peak is located by performing a peak search algorithm on the spectrum to locate local maxima points where the amplitude is greater than a preset peak threshold.

[0016] The above scheme, by setting a peak threshold to execute the search algorithm, can quickly and accurately locate local maxima in the amplitude spectrum, effectively filter out low-amplitude background noise interference, ensure that the peak selected when calculating the significance factor is representative, and further improve the reliability of feature extraction.

[0017] Optionally, the number of adaptive mode decompositions for: ; In the formula, It is a frequency domain energy dispersion index. As the significance factor of the modulation sideband, and These are the first adjustment coefficient and the second adjustment coefficient, respectively. This indicates rounding up to the nearest integer.

[0018] The above scheme establishes a nonlinear mapping relationship between the number of modal decompositions and frequency domain characteristics. When the signal frequency domain structure is complex (high dispersion) or rich in sidebands (strong significance), the number of decompositions is automatically increased, which effectively prevents under-decomposition and ensures that the main fault components are effectively separated rather than omitted or mixed, thus realizing intelligent control of the decomposition granularity.

[0019] Optionally, the adaptive quadratic penalty factor for: ; In the formula, It is a frequency domain energy dispersion index. As the significance factor of the modulation sideband, This is the third adjustment coefficient. The sampling frequency of the vibration signal is [missing information]. To prevent tiny positive numbers with a denominator of zero.

[0020] The above method utilizes a signal-to-noise ratio (SNR) countermeasure mechanism to dynamically adjust the secondary penalty factor. When the modulation characteristics are clear, the penalty factor is increased to narrow the bandwidth and focus on the fault. When the broadband noise is significant, the penalty factor is decreased to broaden the frequency band and capture the impact. This optimizes the bandwidth setting of the mode filter and preserves the fault impact information to the greatest extent while suppressing noise.

[0021] Optionally, the envelope spectrum analysis of the IMF components includes: Calculate the spectral kurtosis of each IMF component; The IMF component with the largest spectral kurtosis value is selected as the principal component. The principal components are subjected to Hilbert transform to obtain the envelope signal, and the envelope spectrum of the envelope signal is obtained.

[0022] Optionally, the detection of whether a fault exists in the planetary gear transmission system includes: The amplitude of the characteristic fault frequency in the envelope spectrum is compared with a preset safety threshold; if the amplitude is greater than the preset safety threshold, a fault is determined to exist and an alarm signal is output.

[0023] Optionally, before acquiring the vibration signal, the method further includes: performing detrending processing on the acquired raw vibration signal to obtain the vibration signal.

[0024] Optionally, the original vibration signal is acquired by an accelerometer mounted on the surface of the planetary gearbox.

[0025] The beneficial effects of this invention are as follows: This invention proposes an adaptive variational mode decomposition (VMD) method based on frequency domain features for planetary gear signals under varying operating conditions. By constructing a frequency domain discretization index characterizing energy sparsity and a saliency factor characterizing sideband strength, an adaptive mapping model of VMD parameters (number of modes and penalty factor) is established. This method abandons the manual parameter setting approach, solves the problems of mode aliasing and signal loss in non-stationary signal processing, and achieves accurate extraction and detection of weak fault features in planetary gears. Attached Figure Description

[0026] Figure 1 The flowchart illustrating the steps of the planetary gear transmission fault detection method based on frequency domain features in this embodiment is shown in the schematic diagram. Figure 2 The diagram illustrates the spectrum of the vibration signal after frequency domain transformation in this embodiment. Detailed Implementation

[0027] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention.

[0028] This invention proposes a fault detection method for planetary gear transmissions based on frequency domain features. This method addresses the dynamic changes in the spectral structure of vibration signals from planetary gears under varying operating conditions (speed fluctuations, load changes). It abandons the fixed parameter model of traditional variational mode decomposition (VMD) and adaptively calculates the number of mode decompositions and the quadratic penalty factor using the topological characteristics of frequency domain energy, thereby achieving accurate extraction of subtle fault features.

[0029] Specifically, as shown in the figure As shown, the planetary gear transmission fault detection method based on frequency domain features in this embodiment includes the following steps: Step S1: Obtain the vibration signal of the planetary gear transmission system.

[0030] In this embodiment, the raw vibration signal is first acquired in real time by an accelerometer mounted on the surface of the planetary gearbox.

[0031] Considering the complex operating conditions of planetary gear transmission systems in actual industrial settings (such as industrial robot joints), a sampling frequency is set to ensure coverage of the meshing frequency of the planetary gears and its main harmonic components, and to prevent aliasing. Preferred to Between. For example, in this embodiment, the sampling frequency is set to. .

[0032] In this embodiment, the acquired original vibration signal is denoted as . Since the original vibration signal often contains trend terms caused by equipment drift or low-frequency interference, these trend terms can affect the accuracy of subsequent spectrum analysis. Therefore, this embodiment uses a polynomial fitting method to remove the trend terms from the original vibration signal to obtain the final vibration signal.

[0033] Step S2: The vibration signal is decomposed into multiple IMF components using an improved variational mode decomposition algorithm.

[0034] Variational Mode Decomposition (VMD) is a signal decomposition and estimation method. This method determines the frequency center and bandwidth of each component by iteratively searching for the optimal solution of the variational model during the acquisition of decomposed components, thereby adaptively achieving frequency domain partitioning and effective separation of each component. Its specific implementation process includes: First, initializing the mode functions and their corresponding center frequencies, typically using random initialization or pre-estimation based on signal characteristics. Second, the iterative optimization process alternately updates the frequency and time domains using the Alternating Directional Multiplier Method (ADMM): In the frequency domain, a Hilbert transform is performed on each mode component to obtain a one-sided spectrum, which is then shifted to the baseband to estimate the bandwidth; in the time domain, L2 regularization is used to constrain the gradient smoothness of the mode components, thereby updating the mode components. Then, the iteration is repeated until convergence conditions are met (e.g., the change in mode components is less than a preset tolerance or the maximum number of iterations is reached).

[0035] In this embodiment, the improved variational mode decomposition algorithm includes an adaptive number of mode decompositions and an adaptive quadratic penalty factor.

[0036] Specifically, the process of obtaining the number of adaptive mode decompositions and the adaptive quadratic penalty factor is as follows: First, a frequency domain discrete index is constructed to quantify the sparsity of the spectral energy distribution.

[0037] The core of this embodiment lies in solving the problem of adaptive parameter selection in the variational mode decomposition (VMD) algorithm. First, it is necessary to evaluate the complexity and energy dispersion of the signal spectrum. A frequency domain discrete index is constructed, which can reflect the sparsity of the energy distribution in the vibration signal spectrum.

[0038] Before constructing the frequency domain discrete index, a Fast Fourier Transform (FFT) is performed on the vibration signal to convert the time-domain signal into a frequency-domain signal and obtain the spectrum. See details Figure 2 As shown.

[0039] Among them, frequency domain discrete index The calculation formula is as follows: ; in, The amplitude of the spectrum, To set the upper limit frequency of the frequency band, a value is typically taken as 1. ; The gradient represents the amplitude of the spectrum; Represents the spectral energy, and log() is the logarithmic function. It is an absolute value function.

[0040] The logarithmic function mentioned above is a function with base 2. To find the derivative, This represents the integral.

[0041] It is understandable that the gradient term in the numerator of the above formula... This reflects the severity of oscillations in the spectral curve. When the planetary gear system is in a faulty state or there is a large amount of broadband noise interference, numerous non-stationary frequency components and spikes will appear in the spectrum, leading to a significant increase in the amplitude gradient and an increase in the integral result. The denominator is a total energy normalization term, ensuring that the index is not affected by the overall signal strength.

[0042] In the above formula, when the frequency domain discrete index As the numerical value increases, the frequency domain structure of the signal becomes more complex, and the number of independent frequency components increases. This provides a direct basis for subsequently determining the number of mode decompositions, i.e., the frequency domain discretization index. The larger the value, the more modal decompositions are needed to effectively separate these mixed components; otherwise, under-decomposition will occur.

[0043] Secondly, a significance factor is constructed to characterize the presence and strength of the modulation sideband structure.

[0044] To distinguish broadband noise in the spectrum from the unique modulation sidebands caused by gear faults, and to provide bandwidth control parameters The selection of significance factors is based on this, and this embodiment further constructs significance factors.

[0045] Specifically, firstly in the spectrum The peak search algorithm is executed to locate local maxima points whose amplitudes exceed a preset peak threshold, and the point with the largest amplitude is selected from these local maxima. A local peak.

[0046] The peak search algorithms mentioned above are a class of computational methods used to locate peak elements in a data sequence. They mainly include linear traversal, binary search, and AMPD algorithm.

[0047] In this embodiment, The preferred value is One, to cover the main meshing frequencies and harmonic components.

[0048] The significance factor The calculation formula is as follows: ; in, For frequency domain discrete index, The number of main peaks in the vibration spectrum. and The first The amplitude and frequency of each main peak. The average of all main peak amplitudes, i.e. ; The variance of the amplitude of all main peak values; This is the theoretical gear meshing frequency. It is an absolute value function; The size of the set bandwidth range is preferably set to The planetary carrier rotation frequency is times that of the planetary carrier.

[0049] The above formula introduces an exponential decay term. and variance term When the main peak value is identified Closely related to the theoretical gear meshing frequency Surrounding (i.e.) When the value is small, the exponential term approaches the value of . The highest weight is given to the peaks; at the same time, if the amplitude differences of these peaks are large (i.e., the modulation sidebands are significant), the weight is increased by the peaks. (reflection), then significance factor The value will increase significantly.

[0050] Conversely, if it is random broadband noise, its peak frequency distribution is disordered and far from the target frequency. The exponential term approaches ,lead to The significance factor is relatively small. A larger value indicates the presence of significant narrowband modulation characteristics in the signal, in which case the penalty factor should be increased. The value is used to limit the modal bandwidth and focus on fault characteristics.

[0051] Then, the number of adaptive mode decompositions and the adaptive quadratic penalty factor are calculated.

[0052] Based on the frequency domain discrete indexes and significance factors obtained above, a nonlinear mapping relationship is established to generate the optimal parameter combination for variational mode decomposition that is adapted to the current vibration signal.

[0053] Among them, the number of adaptive mode decompositions for: ; In the formula, This indicates the rounding up operation; and Let be the first adjustment coefficient and the second adjustment coefficient, respectively, and ln() be the logarithmic function.

[0054] In this embodiment, the first adjustment coefficient The preferred range is Second adjustment coefficient The preferred range is Of course, other implementation methods can be determined based on the actual situation.

[0055] As can be seen from the formula, the number of adaptive mode decompositions With frequency domain discrete index Logarithm and significance factor Both are positively correlated. When the signal complexity When increasing, the number of adaptive mode decompositions This is subsequently increased to cover more frequency components; simultaneously, the significance factor... It will also affect the number of adaptive mode decompositions. Perform positive corrections to prevent the loss of key sidebands.

[0056] For example, when the frequency domain discrete index from Increase to When the logarithmic term increases significantly, the number of adaptive mode decompositions automatically increases, thus avoiding under-decomposition.

[0057] Among them, the adaptive quadratic penalty factor for: ; In the formula, For frequency domain discrete index, As a significant factor, This is the third adjustment coefficient. The sampling frequency of the vibration signal is [missing information]. To prevent tiny positive numbers with a denominator of zero (e.g.) ).

[0058] In this embodiment, The preferred value range is .

[0059] The above formula reflects the signal-to-noise ratio (SNR) countermeasure mechanism. When the significance factor... Larger (clear modulation) and frequency domain discrete index When the value is relatively small (low noise), the ratio increases, and the calculated adaptive quadratic penalty factor... Larger values ​​(e.g., reaching) ).

[0060] In the VMD algorithm, a larger adaptive quadratic penalty factor implies a narrower bandwidth for the mode filter, which is beneficial for accurately extracting narrowband fault features concentrated at the center frequency. Conversely, when the frequency domain discrete index... When the value is large (significant broadband noise or impact), the ratio decreases, and the adaptive quadratic penalty factor... The value decreased (e.g., dropped to) A smaller adaptive quadratic penalty factor allows the mode to have a wider bandwidth, thus enabling the capture of impulse signals with wideband characteristics and avoiding the omission of fault information.

[0061] In this embodiment, the vibration signal is decomposed into multiple IMF components using an improved variational mode decomposition algorithm.

[0062] Specifically, the vibration signal is subjected to variational mode decomposition using an improved variational mode decomposition algorithm, and the decomposition yields... There are 1 intrinsic mode functions (IMFs), denoted as . .

[0063] To select the principal component containing fault information from multiple IMF components, the spectral kurtosis of each IMF component is calculated, and the IMF component with the largest spectral kurtosis value is selected as the optimal component. .

[0064] Since the calculation of spectral kurtosis is a current technique, it will not be elaborated here.

[0065] The aforementioned optimal component is considered to contain the richest non-Gaussian impact components, i.e., fault characteristics.

[0066] Using the aforementioned spectral kurtosis as a screening criterion, the principal component containing the richest non-Gaussian impulse components can be automatically identified from the multiple decomposed modal components. Envelope demodulation is then performed using Hilbert transform, effectively transforming the complex modulation signal into an easily analyzable envelope spectrum, and intuitively revealing the fault characteristic frequencies hidden in the sidebands.

[0067] Step S3: Perform envelope spectrum analysis on the IMF components to detect whether there is a fault in the planetary gear transmission system.

[0068] For the selected optimal component Perform a Hilbert transform, calculate the magnitude of the analytic signal, and obtain the envelope signal. Specifically: ; in This represents the Hilbert transform.

[0069] The Hilbert transform described above is a linear operator that produces functions with the same domain as the function, and it can derive the optimal components. The analytical representation of Hilbert transform is omitted here as it is a current technique.

[0070] Next, the envelope signal The envelope spectrum is obtained by performing a Fourier transform. Since the Fourier transform is a current technique, it will not be described in detail here.

[0071] Search the envelope spectrum for the presence of fault frequencies characteristic of planetary gears. If the amplitude of a fault frequency exceeds the safety threshold, the planetary gear transmission system is deemed to be faulty, and an alarm signal is output.

[0072] Specifically, the process for obtaining the amplitude of the fault frequency is as follows: First, obtain the fault characteristic frequencies of the planetary gears.

[0073] In this embodiment, twice the rotation frequency of the planetary gear relative to the planet carrier is used as the fault characteristic frequency.

[0074] Secondly, set the frequency tolerance, determine the frequency search range based on the fault characteristic frequency and the frequency tolerance, and find the maximum amplitude of the fault frequency in the frequency search range.

[0075] The frequency tolerance mentioned above is generally 1%-2% of the fault characteristic frequency. Of course, as other implementation methods, it can be set according to the actual situation.

[0076] The frequency search range mentioned above is [fault characteristic frequency - frequency tolerance, fault characteristic frequency + frequency tolerance].

[0077] Then, the maximum amplitude value is searched in the search range of higher harmonics such as twice the theoretical frequency and three times the fault frequency.

[0078] The above safety threshold is taken as the fundamental frequency amplitude under normal conditions. times.

[0079] By setting a safety threshold based on normal conditions, the above-mentioned automatic judgment of the health status of the planetary gear transmission system is realized. It can promptly detect abnormal increases in characteristic frequency amplitude and issue alarms, providing timely decision support for preventive maintenance of the equipment.

[0080] The present invention constructs a frequency domain discrete index and a significance factor to adaptively adjust the number of modes and the penalty factor of the VMD algorithm, enabling it to adaptively optimize and adapt to planetary gears under varying operating conditions (speed fluctuations, load changes). This effectively solves the problems of mode aliasing and signal loss caused by fixed parameters under varying operating conditions, significantly improving the diagnostic accuracy of early faults in planetary gears. In the description of this specification, "multiple" means at least two, such as two, three, or more, unless otherwise explicitly specified.

[0081] While various embodiments of the invention have been shown and described in this specification, it will be apparent to those skilled in the art that such embodiments are provided by way of example only. Many modifications, alterations, and alternatives will occur to those skilled in the art without departing from the spirit and essence of the invention.

Claims

1. A fault detection method for planetary gear transmissions based on frequency domain characteristics, characterized in that, include: Acquire vibration signals from the planetary gear transmission system; The vibration signal is decomposed into multiple IMF components using an improved variational mode decomposition algorithm. Envelope spectrum analysis is performed on the IMF components to detect whether there is a fault in the planetary gear transmission system; The improved variational mode decomposition algorithm includes an adaptive number of mode decompositions and an adaptive quadratic penalty factor. The adaptive number of mode decompositions is positively correlated with both the logarithm of the frequency domain discretization index and the significance factor. The adaptive quadratic penalty factor is positively correlated with the significance factor and negatively correlated with the frequency domain discretization index. The frequency domain discretization index is used to characterize the sparsity of the energy distribution of the vibration signal's spectrum. The significance factor is used to characterize the strength of the modulation sideband structure in the vibration signal's spectrum.

2. The planetary gear transmission fault detection method based on frequency domain features according to claim 1, characterized in that, The frequency domain discrete index for: ； In the formula, The amplitude of the spectrum, To set the upper frequency limit of the frequency band, log() is a logarithmic function. The gradient of the spectral amplitude. It is an absolute value function.

3. The planetary gear transmission fault detection method based on frequency domain features according to claim 1, characterized in that, The significance factor for: ； In the formula, For frequency domain discrete index, The number of main peaks in the vibration spectrum. and The first The amplitude and frequency of each main peak. The average of all main peak amplitudes. The variance of the amplitude of all major peaks. This is the theoretical gear meshing frequency. The specified bandwidth range is given by exp(), which is an exponential function. It is an absolute value function.

4. The planetary gear transmission fault detection method based on frequency domain characteristics according to claim 3, characterized in that, The main peak is determined by performing a peak search algorithm on the spectrum to locate local maxima points where the amplitude is greater than a preset peak threshold.

5. The planetary gear transmission fault detection method based on frequency domain features according to claim 1, characterized in that, The number of adaptive mode decompositions for: ； In the formula, It is a frequency domain energy dispersion index. As the modulation sideband significance factor, and These are the first adjustment coefficient and the second adjustment coefficient, respectively. This indicates rounding up to the nearest integer.

6. The planetary gear transmission fault detection method based on frequency domain features according to claim 1, characterized in that, The adaptive quadratic penalty factor for: ； In the formula, It is a frequency domain energy dispersion index. As the modulation sideband significance factor, This is the third adjustment coefficient. The sampling frequency of the vibration signal is [missing information]. To prevent tiny positive numbers with a denominator of zero.

7. The planetary gear transmission fault detection method based on frequency domain features according to claim 1, characterized in that, The envelope spectrum analysis of the IMF components includes: Calculate the spectral kurtosis of each IMF component; The IMF component with the largest spectral kurtosis value is selected as the principal component. The principal components are subjected to Hilbert transform to obtain the envelope signal, and the envelope spectrum of the envelope signal is obtained.

8. The planetary gear transmission fault detection method based on frequency domain features according to claim 7, characterized in that, The detection of faults in the planetary gear transmission system includes: The amplitude of the characteristic fault frequency in the envelope spectrum is compared with a preset safety threshold; if the amplitude is greater than the preset safety threshold, a fault is determined to exist and an alarm signal is output.

9. The planetary gear transmission fault detection method based on frequency domain features according to claim 1, characterized in that, Before acquiring the vibration signal, the method further includes: performing detrending processing on the acquired raw vibration signal to obtain the vibration signal.

10. The planetary gear transmission fault detection method based on frequency domain features according to claim 1, characterized in that, The original vibration signal was acquired by an accelerometer mounted on the surface of the planetary gearbox.

Images