Planetary gearbox fault real-time detection method based on multiple filtering and synchronous diversity entropy

Abstract

The invention provides a planetary gearbox fault real-time detection method based on multiple filtering and synchronous diversity entropy, which comprises the following steps: selecting a sensor and a mounting position thereof, and setting parameters; collecting vibration data X, and re-sampling the vibration data X to obtain synchronous vibration data Y; performing first filtering processing on the synchronous vibration data Y by adopting an angular domain synchronous average filtering processing method to obtain a harmonic component Z; performing secondary filtering processing on the harmonic component Z by adopting a symbol dynamic filtering processing method to obtain a symbol sequence S; performing phase space reconstruction on the symbol sequence S to obtain a phase space P; performing third filtering processing on each vector in the phase space P by adopting a mean filtering processing method to obtain a phase space Q; calculating the cosine similarity between every two adjacent vectors in the phase space Q; calculating synchronous diversity entropy (SDE); and judging whether a fault occurs based on the synchronous diversity entropy SDE and a threshold value T. According to the method, the influence of disturbance noise is effectively suppressed by designing a multiple filtering strategy, and the planetary gearbox fault can be rapidly and reliably detected.

Description

Technical Field

[0001] The present invention belongs to the technical field of rotating machinery fault diagnosis, and in particular relates to a real-time detection method for planetary gearbox faults based on multiple filtering and synchronous diversity entropy. Background Art

[0002] In modern industrial systems, gearboxes are widely used in mechanical transmission systems. Under high-intensity operating conditions, gears within gearboxes are prone to various types of failure. These failures not only impair mechanical system performance but can also cause serious safety incidents, posing threats to personnel and property. With increasing awareness of workplace safety and property protection, gearbox fault detection is gaining increasing attention. Promptly detecting abnormal gearbox conditions can largely prevent damage, downtime, and even accidents. Therefore, real-time gearbox fault detection has significant economic benefits.

[0003] Entropy-based fault detection methods are widely adopted due to their ability to efficiently capture fault information. Common entropy methods include sample entropy, fuzzy entropy, permutation entropy, and their improved versions. Although these entropy methods have demonstrated good performance in rotating machinery fault detection, they still have limitations, such as long fault detection delays and low reliability.

[0004] Long fault detection delays severely limit the application of fault detection methods. Mechanical transmission systems are closely interconnected, and a failure in one component is likely to cause failures in other components, potentially rendering the entire mechanical system useless. Therefore, real-time detection methods must be able to rapidly detect faults and complete gearbox fault detection within a relatively short timeframe to minimize the potential damage.

[0005] The reliability of fault detection seriously affects the performance of fault detection methods. In common large-scale mechanical transmission systems, there are generally complex power transmission processes, and unplanned shutdowns will result in a large amount of kinetic energy waste. Therefore, the reliability of real-time fault detection is particularly important. In the planetary gearbox fault detection process, the fault information generated by the gear fault needs to go through a long transmission route before it can be acquired by the sensor. The acquired fault signal inevitably contains interference noise. In order to improve the reliability of fault detection, the influence of interference noise on fault detection should be suppressed. In addition, the working environment of the planetary gearbox may change, thereby reducing the reliability of fault detection. Existing entropy-based fault detection methods often lack effective filtering measures, which makes it difficult to separate weak fault information from strong noise, seriously limiting the reliability of fault detection.

[0006] In order to complete the real-time fault detection of planetary gearboxes more quickly and reliably, it is necessary to develop new fault detection algorithms and technologies, which can not only complete the gearbox fault detection work in a very short time, but also improve the reliability of fault detection by better identifying the signs of gearbox faults through the use of multiple filtering technologies. Summary of the Invention

[0007] Purpose of the invention: The present invention provides a real-time detection method for planetary gearbox faults based on multiple filtering and synchronous diversity entropy. By adopting multiple filtering technology to suppress interference noise in the signal, the planetary gearbox faults can be detected quickly and reliably.

[0008] Technical solution: The present invention proposes a real-time detection method for planetary gearbox faults based on multiple filtering and synchronous diversity entropy, which specifically includes the following steps:

[0009] (1) Select the sensor and its installation location, and set the parameters;

[0010] (2) collecting vibration data X and resampling it to obtain synchronized vibration data Y;

[0011] (3) Using the angular domain synchronous average filtering method to perform the first filtering on the synchronous vibration data Y to obtain the harmonic component Z;

[0012] (4) Using the symbol dynamic filtering method to perform a second filtering process on the harmonic component Z to obtain the symbol sequence S;

[0013] (5) Reconstruct the phase space of the symbol sequence S to obtain the phase space P;

[0014] (6) Using the mean filtering method to perform a third filtering process on each vector in the phase space P, the phase space Q is obtained;

[0015] (7) Calculate the cosine similarity between each adjacent vector in the phase space Q;

[0016] (8) Calculate the synchronous diversity entropy SDE;

[0017] (9) Determine whether a fault occurs based on the synchronization diversity entropy SDE and the threshold T.

[0018] Furthermore, the sensors are vibration sensors and speed sensors; the vibration sensor is installed on the planetary gearbox, and the speed sensor is installed near the sun gear main shaft; the setting parameters include: sampling frequency, number of sampling points M per sun gear rotation, number of sun gear teeth c, detection time window length, embedding dimension m, time delay τ, and number of symbols ε; the number of sampling points M per sun gear rotation must be an integer multiple of the number of sun gear teeth c; the detection time window length is the length of sampling data within k rotations of the sun gear; the time delay τ is the meshing period of the sun gear and the planetary gear, that is, the sun gear rotation period divided by the number of sun gear teeth; the embedding dimension m and the number of symbols ε are determined according to the gear fault characteristics.

[0019] Furthermore, the implementation process of step (2) is as follows:

[0020] A speed sensor is used to measure the sun gear speed signal, and this speed signal is used as a reference signal. A vibration sensor is used to obtain the vibration data X during the operation of the planetary gearbox. The vibration data X can be expressed as:

[0021] X={x(1),x(2),…,x(i),…,x(N)};

[0022] According to the set M and speed signal, the vibration data X is resampled to obtain the synchronous vibration data Y. The synchronous vibration data Y can be expressed as follows:

[0023] Y={y(1),y(2),…,y(i),…,y(kM)},

[0024] Wherein, k is the total number of revolutions of the sun gear within a single detection time window.

[0025] Furthermore, the implementation process of step (3) is as follows:

[0026] Taking the number of sampling points M per sun gear revolution as a reference, the angular domain synchronous averaging algorithm is used to perform the first filtering process on the synchronous vibration data Y to obtain the harmonic component Z. The angular domain synchronous averaging process is as follows:

[0027]

[0028] Among them, y(i+jM) is the vibration signal collected at equal angles; through angular domain synchronous averaging processing, the harmonic components of the synchronous vibration data Z = {z(1),z(2),…,z(i),…,z(kM)} are obtained.

[0029] Furthermore, the implementation process of step (4) is as follows:

[0030] Arrange the harmonic components Z in ascending order, and then divide the sorted sequence evenly according to the number of symbols ε to form ε intervals, each of which has The symbol value of each interval is the mean value of all the data in the interval; the harmonic component Z is subjected to a second filtering process using the symbol dynamic filtering method to obtain the symbol sequence S = {s(1), s(2),…, s(i),…, s(kM)} of the harmonic component.

[0031] Furthermore, the implementation process of step (5) is as follows:

[0032] According to the preset time delay τ and embedding dimension m, the symbol sequence S is reconstructed into the phase space P:

[0033]

[0034] Among them, the time delay That is, the number of sampling points within the gear meshing cycle.

[0035] Furthermore, the implementation process of step (6) is as follows:

[0036] The mean filtering method is used to perform a third filtering process on each m-dimensional vector in the phase space P to obtain the filtered phase space Q. The mean filtering process is as follows:

[0037]

[0038] Furthermore, the implementation process of step (7) is as follows:

[0039] Calculate the cosine similarity of each adjacent vector in the phase space Q:

[0040]

[0041] Among them, p i (m) and p i+1 (m) represent the i-th and i+1-th m-dimensional vectors in the phase space Q, respectively, ||p i (m)|| and||p i+1 (m)|| are vectors p i (m) and p i+1 (m) modulus, ||p iThe calculation process of (m)|| is as follows:

[0042]

[0043] Furthermore, the implementation process of step (8) is as follows:

[0044] Divide the cosine similarity value range [-1,1] into ε intervals evenly, divide the kM-(m-1)τ-1 cosine similarities CS into ε intervals according to their values, and count the number of cosine similarities in each interval I i , calculate the probability of cosine similarity distribution in each interval, using G1, G2, ..., G ε To express it, the probability calculation process is as follows:

[0045]

[0046] Calculate the synchronous diversity entropy value based on the obtained probability:

[0047]

[0048] Furthermore, the implementation process of step (9) is as follows:

[0049] The calculated synchronization diversity entropy value SDE is used as a fault indicator and combined with the fault threshold T to perform real-time fault detection. If the synchronization diversity entropy value SDE is greater than or equal to the threshold T, the gearbox is judged to have a fault; otherwise, the gearbox is judged to be in a healthy working state.

[0050] Beneficial effects

[0051] Compared with the prior art, the present invention mainly has the following beneficial effects:

[0052] (1) The present invention discloses a real-time fault detection method for planetary gearboxes based on multiple filtering and synchronous diversity entropy. By designing a multiple filtering scheme, the method filters out and suppresses the influence of interference noise in the signal on fault detection, thereby achieving higher fault detection reliability.

[0053] (2) The real-time planetary gearbox fault detection method based on multiple filtering and synchronous diversity entropy disclosed in the present invention has low computational cost, short fault detection time delay, and can detect planetary gearbox faults in real time;

[0054] (3) The real-time detection method for planetary gearbox faults based on multiple filtering and synchronous diversity entropy disclosed in the patent of this invention is not affected by the change of the planetary gearbox speed. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] Figure 1This is a flow chart of the real-time detection method for planetary gearbox faults based on multiple filtering and synchronous diversity entropy disclosed by the present invention;

[0056] Figure 2 This is a photo of the WT-planetary gearbox test bench in an embodiment of the present invention;

[0057] Figure 3 The sun wheel in five different health states;

[0058] Figure 4 is the synchronous diversity entropy of the sun gear under five health states obtained by the method of the present invention;

[0059] Figure 5 is the fuzzy entropy of the sun gear under five health states obtained by the traditional fuzzy entropy method; DETAILED DESCRIPTION

[0060] The present invention will be described in further detail below with reference to the accompanying drawings. Figure 1 The present invention discloses a real-time planetary gearbox fault detection method based on multiple filtering and synchronous diversity entropy, which specifically includes the following steps:

[0061] Step (1): Select the sensor and its installation location, and set the parameters.

[0062] The sensors selected are vibration sensors and speed sensors; the vibration sensor is installed on the planetary gearbox, while the speed sensor is installed near the sun gear main shaft; the set parameters include: sampling frequency, number of sampling points M per sun gear revolution, number of sun gear teeth c, embedding dimension m, time delay τ, and number of symbols ε; the number of sampling points M per sun gear revolution must be an integer multiple of the number of sun gear teeth c.

[0063] The gearbox fault dataset used in this embodiment is the WT-planetary gearbox dataset published in the literature by Liu Dongdong et al. from Beijing University of Technology.

[0064] Figure 2 This is a photo of a WT planetary gearbox test bench. The test bench consists of a motor, planetary gearbox, fixed-axis gearbox, and loading device. In a planetary gearbox, four planetary gears rotate around a sun gear.

[0065] The test bench simulates the working process of the sun gear under five health conditions. The health conditions of the sun gear include gear health, gear damage, tooth root fracture, tooth surface wear and gear tooth missing. Figure 3 During the experiment, the sampling frequency was set to 48KHz and the sun gear speed was set to 20Hz. A total of 100 sets of sample data were collected, and the length of each set of sample data was the length of the data within five rotations of the sun gear.

[0066] In this embodiment, the number of sun gear teeth c=28. To ensure that the number of sampling points M per sun gear rotation is an integer multiple of the number of sun gear teeth c, we set the number of sampling points M per sun gear rotation to 2800, and the time delay is set to the number of sampling points in the sun gear meshing period, that is, The embedding dimension m is set to 3 and the number of symbols ε is set to 20.

[0067] Step (2): Collect vibration data X and resample it to obtain synchronous vibration data Y.

[0068] A speed sensor is used to measure the sun gear speed signal, and this speed signal is used as a reference signal. A vibration sensor is used to obtain the vibration data X during the operation of the planetary gearbox. The vibration data X can be expressed as:

[0069] X={x(1),x(2),…,x(i),…,x(N)}.

[0070] According to the set M, the vibration data X is resampled according to the speed signal to obtain the synchronous vibration data Y. The synchronous vibration data can be expressed as follows:

[0071] Y={y(1),y(2),…,y(i),…,y(kM)}.

[0072] Since the sun gear rotates at 20 Hz, the data length of each time window is the data length within five rotations of the sun gear, that is, the number of rotations k = 5. Therefore, the synchronous vibration data Y can be expressed as follows:

[0073] Y={y(1),y(2),…,y(i),…,y(14000)}.

[0074] Step (3): Use the angular domain synchronous average filtering method to perform the first filtering process on the synchronous vibration data Y to obtain the harmonic component Z.

[0075] Taking the number of sampling points M per sun gear revolution as a reference, the angular domain synchronous averaging algorithm is used to perform the first filtering process on the synchronous vibration data Y to obtain the harmonic component Z. The angular domain synchronous averaging process is as follows:

[0076]

[0077] Among them, y(i+jM) is the vibration signal collected at equal angles; through angular domain synchronous averaging processing, the harmonic components of the synchronous vibration data Z = {z(1),z(2),…,z(i),…,z(kM)} are obtained.

[0078] Step (4): Use the symbol dynamic filtering method to perform a second filtering process on the harmonic component Z to obtain a symbol sequence S.

[0079] Arrange the harmonic components Z in ascending order, and then divide the sorted sequence evenly according to the number of symbols ε to form ε intervals, each of which has The symbol value of each interval is the mean value of all the data in the interval; the harmonic component Z is subjected to a second filtering process using the symbol dynamic filtering method to obtain the symbol sequence S = {s(1), s(2),…, s(i),…, s(kM)} of the harmonic component.

[0080] Step (5): Reconstruct the phase space of the symbol sequence S to obtain the phase space P.

[0081] According to the preset time delay τ and embedding dimension m, the symbol sequence S is reconstructed into the phase space P:

[0082]

[0083] Among them, the time delay That is, the number of sampling points within the gear meshing cycle.

[0084] Step (6): Use the mean filtering method to perform a third filtering process on each vector in the phase space P to obtain the phase space Q.

[0085] The mean filtering method is used to perform a third filtering process on each m-dimensional vector in the phase space P to obtain the filtered phase space Q. The mean filtering process is as follows:

[0086]

[0087] Step (7): Calculate the cosine similarity between each adjacent vector in the phase space Q.

[0088] Calculate the cosine similarity of each adjacent vector in the phase space Q:

[0089]

[0090] Among them, p i (m) and p i+1 (m) represent the i-th and i+1-th m-dimensional vectors of the phase space Q, respectively, ||p i (m)|| and||p i+1 (m)|| are vectors p i (m) and p i+1 (m) modulus, ||p i The calculation process of (m)|| is as follows:

[0091]

[0092] Step (8): Calculate the synchronized diversity entropy SDE.

[0093] Divide the cosine similarity value range [-1,1] into ε intervals evenly, divide the kM-(m-1)τ-1 cosine similarities CS into ε intervals according to their values, and count the number of cosine similarities in each interval I i , calculate the probability of cosine similarity distribution in each interval, using G1, G2, ..., G ε To express it, the probability calculation process is as follows:

[0094]

[0095] Calculate the synchronous diversity entropy value based on the obtained probability:

[0096]

[0097] Step (9): Determine whether a fault occurs based on the synchronization diversity entropy SDE and the threshold T.

[0098] Synchronous diversity entropy under the five health states of the sun wheel is as follows Figure 4 As shown in the figure, the healthy working state of the gear is well distinguished and has a clear boundary with various fault states. Therefore, the patented method has a high reliability in gearbox fault detection. In this embodiment, we set the fault threshold T to 0.08, as shown in Figure 4 During the fault detection process, if the synchronous diversity entropy is less than 0.08, the gearbox is judged to be in normal working condition; otherwise, it is judged that the gearbox has a fault.

[0099] In order to further illustrate the advantages of the invented method in fault detection, we also calculated the traditional fuzzy entropy of the same sample data. The calculation results are as follows: Figure 5 As shown in the figure, it can be seen that the traditional fuzzy entropy cannot reliably identify the gearbox fault.

[0100] The present invention proposes a real-time planetary gearbox fault detection method based on multiple filtering and synchronous diversity entropy. While having high fault detection reliability, it also has low computational costs and can meet the needs of real-time fault detection. In this embodiment, we used a desktop computer with an Intel Core i9-9900 3.4GHz CPU to test the computational cost of the method of the present invention. The test showed that the calculation time of the method of the present invention is approximately 0.03 seconds, while the calculation time of the traditional fuzzy entropy is approximately 1.59 seconds. The computational time cost of the method of the present invention is only 1.9% of that of the traditional fuzzy entropy method, with a shorter fault detection time delay, and can achieve real-time fault detection.

[0101] In summary, the method of the present invention has higher reliability and lower computational cost compared with the traditional gearbox fault diagnosis method based on fuzzy entropy.

Claims

1. A real-time detection method for planetary gearbox faults based on multiple filtering and synchronous diversity entropy, characterized in that: The following steps are involved: (1) Select the sensor and its installation location, and set the parameters; (2) collecting vibration data X and resampling it to obtain synchronized vibration data Y; (3) Using the angular domain synchronous average filtering method to perform the first filtering on the synchronous vibration data Y to obtain the harmonic component Z; (4) Using the symbol dynamic filtering method to perform a second filtering process on the harmonic component Z to obtain the symbol sequence S; (5) Reconstruct the phase space of the symbol sequence S to obtain the phase space P; (6) Using the mean filtering method to perform a third filtering process on each vector in the phase space P, the phase space Q is obtained; (7) Calculate the cosine similarity between each adjacent vector in the phase space Q; (8) Calculate the synchronous diversity entropy SDE; (9) Determine whether a fault occurs based on the synchronization diversity entropy SDE and the threshold T.

2. The real-time detection method for planetary gearbox faults based on multiple filtering and synchronous diversity entropy according to claim 1 is characterized in that: The sensors described in step (1) are a vibration sensor and a speed sensor; the vibration sensor is installed on the planetary gearbox, and the speed sensor is installed near the main shaft of the sun gear; the setting parameters include: sampling frequency, number of sampling points M per rotation of the sun gear, number of sun gear teeth c, detection time window length, embedding dimension m, time delay τ, and number of symbols ε; the number of sampling points M per rotation of the sun gear must be an integer multiple of the number of sun gear teeth c; the detection time window length is the length of the sampling data within k rotations of the sun gear; the time delay τ is the meshing period of the sun gear and the planetary gear, that is, the sun gear rotation period divided by the number of sun gear teeth; the embedding dimension m and the number of symbols ε are determined according to the gear fault characteristics.

3. The real-time detection method for planetary gearbox faults based on multiple filtering and synchronous diversity entropy according to claim 1, characterized in that: The implementation process of step (2) is as follows: using a speed sensor to measure the sun gear speed signal, and using the speed signal as a reference signal; using a vibration sensor to obtain vibration data X during the operation of the planetary gearbox, the vibration data X can be expressed as: X={x(1),x(2),…,x(i),…,x(N)}; According to the set M and speed signal, the vibration data X is resampled to obtain the synchronous vibration data Y. The synchronous vibration data Y can be expressed as follows: Y={y(1),y(2),…,y(i),…,y(kM)}, Wherein, k is the total number of revolutions of the sun gear within a single detection time window.

4. The real-time detection method for planetary gearbox faults based on multiple filtering and synchronous diversity entropy according to claim 1, characterized in that: The implementation process of step (3) is as follows: Taking the number of sampling points M per sun gear revolution as a reference, the angular domain synchronous averaging algorithm is used to perform the first filtering process on the synchronous vibration data Y to obtain the harmonic component Z. The angular domain synchronous averaging process is as follows: Among them, y(i+jM) is the vibration signal collected at equal angles; through angular domain synchronous averaging processing, the harmonic components of the synchronous vibration data Z = {z(1),z(2),…,z(i),…,z(kM)} are obtained.

5. The real-time detection method for planetary gearbox faults based on multiple filtering and synchronous diversity entropy according to claim 1, characterized in that: The implementation process of step (4) is as follows: Arrange the harmonic components Z in ascending order, and then divide the sorted sequence evenly according to the number of symbols ε to form ε intervals, each of which has The symbol value of each interval is the mean value of all the data in the interval; the harmonic component Z is subjected to a second filtering process using the symbol dynamic filtering method to obtain the symbol sequence S = {s(1), s(2),…, s(i),…, s(kM)} of the harmonic component.

6. The real-time detection method for planetary gearbox faults based on multiple filtering and synchronous diversity entropy according to claim 1, characterized in that: The implementation process of step (5) is as follows: According to the preset time delay τ and embedding dimension m, the symbol sequence S is reconstructed into the phase space P: Among them, the time delay That is, the number of sampling points within the gear meshing cycle.

7. The real-time detection method for planetary gearbox faults based on multiple filtering and synchronous diversity entropy according to claim 1, characterized in that: The implementation process of step (6) is as follows: The mean filtering method is used to perform a third filtering process on each m-dimensional vector in the phase space P to obtain the filtered phase space Q. The mean filtering process is as follows:

8. The real-time planetary gearbox fault detection method based on multiple filtering and synchronous diversity entropy according to claim 1 is characterized in that: The implementation process of step (7) is as follows: Calculate the cosine similarity of each adjacent vector in the phase space Q: Among them, p i (m) and p i+1 (m) represent the i-th and i+1-th m-dimensional vectors in the phase space Q, respectively, ||p i (m)|| and||p i+1 (m)|| are respectively vector p i (m) and p i+1 (m) modulus, ||p i The calculation process of (m)|| is as follows:

9. The real-time planetary gearbox fault detection method based on multiple filtering and synchronous diversity entropy according to claim 1, characterized in that: The implementation process of step (8) is as follows: Divide the cosine similarity value range [-1,1] into ε intervals evenly, divide the kM-(m-1)τ-1 cosine similarities CS into ε intervals according to their values, and count the number of cosine similarities in each interval I i , calculate the probability of cosine similarity distribution in each interval, using G1, G2, ..., G ε To express it, the probability calculation process is as follows: Calculate the synchronous diversity entropy value based on the obtained probability:

10. The real-time detection method for planetary gearbox faults based on multiple filtering and synchronous diversity entropy according to claim 1, characterized in that: The implementation process of step (9) is as follows: The calculated synchronization diversity entropy value SDE is used as a fault indicator and combined with the fault threshold T to perform real-time fault detection. If the synchronization diversity entropy value SDE is greater than or equal to the threshold T, it is determined that the gearbox has a fault. Otherwise, it is determined that the gearbox is in a healthy working condition.

Images