A method for separating vibration signals of planetary gearbox based on multipath dictionary

Abstract

The present invention discloses a planetary gearbox vibration signal separation method based on a multi-path dictionary, comprising: establishing a planetary gear train dynamics model, deriving the relative motion equations of coupled components based on the mutual motion relationship of transmission components, and establishing a planetary gearbox dynamics concentrated mass model; obtaining the meshing vibration signal characteristics under normal and fault conditions; establishing an energy transfer loss model, combining the power flow method to construct a transfer path contribution evaluation index, identifying the main transfer path and non-correlated transfer path of the planetary gearbox vibration; filtering the transfer signal of the non-correlated path by the ANC non-correlated path method; analyzing the meshing vibration signal characteristics under normal and fault conditions to determine the most relevant transfer path; the vibration signal separation method based on the multi-path dictionary separates the most relevant transfer path signal from the measuring point signal to achieve vibration signal separation. The method can accurately and effectively determine whether a fault occurs through feature classification.

Description

Technical Field

[0001] The present invention belongs to the technical field of vibration signals, and in particular relates to a method for separating vibration signals of a planetary gearbox based on a multi-path dictionary. Background Art

[0002] The study of the vibration characteristics of planetary gearboxes faces the technical difficulty of decomposing vibration signals due to the presence of multiple transmission paths. Planetary gearboxes, as key components in mechanical transmission systems, are widely used in wind power generation, aerospace, ship propulsion and other fields. However, due to the complexity of the structure of planetary gearboxes and the variability of the working environment, their vibration signals usually contain multiple frequency components and noise interference, making the extraction and identification of signal vibration characteristics, especially gear crack fault signals, particularly difficult. In order to achieve accurate separation and analysis of planetary gearbox vibration signals, researchers have proposed a variety of methods.

[0003] At present, common planetary gearbox vibration signal separation methods mainly include time-frequency analysis, empirical mode decomposition (EMD), variational mode decomposition (VMD), etc. The time-frequency analysis method can intuitively display the changes of each frequency component in the signal over time by mapping the signal to the time-frequency plane, but for non-stationary signals, the resolution of time-frequency analysis is limited. The EMD method can better adapt to the separation of nonlinear and non-stationary signals by decomposing the signal into a series of intrinsic mode functions (IMFs), but it is easily affected by modal aliasing. As an adaptive signal decomposition technology, the VMD method can avoid the problem of modal aliasing and can better separate different frequency components, but its computational complexity is high.

[0004] Although these methods have achieved certain results in separating vibration signals of planetary gearboxes, there is still a problem of insufficient accuracy in signal feature extraction. Due to the complex structure of planetary gearboxes, signals often contain multiple modulation frequencies, and existing methods are difficult to accurately separate the characteristic frequencies in the correct state and gear crack fault state.

[0005] In summary, although the existing planetary gearbox vibration signal separation methods have their own strengths, they still face challenges in practical applications. Therefore, it is urgent to develop a more effective signal separation method to improve the reliability of planetary gearbox signal separation. Summary of the invention

[0006] In view of the deficiencies of the above-mentioned prior art, the present invention provides a planetary gearbox vibration signal separation method based on a multi-path dictionary, which identifies the most relevant transmission path through a planetary gearbox dynamic model and an ANC method of filtering out non-relevant path signals, combines a multi-path dictionary and realizes signal separation of the most relevant path through an FBS algorithm.

[0007] In order to achieve the above technical objectives, the present invention provides the following technical solutions:

[0008] A method for separating vibration signals of a planetary gearbox based on a multipath dictionary, comprising:

[0009] S1. Establish a planetary gear train dynamics model based on the concentrated mass method; derive the relative motion equations of the coupled components based on the mutual motion relationship of the transmission components, and establish a planetary gearbox dynamics concentrated mass model; obtain the vibration signal characteristics of the planetary gearbox under normal conditions and crack fault conditions based on the established concentrated mass model;

[0010] S2. Based on the nonlinear interface energy equivalent loss model, a multi-interface energy transfer loss model from the crack fault source to the target point is established. The transfer path contribution evaluation index is constructed in combination with the power flow method to obtain the contribution of each transfer path and identify the main transfer path and non-correlated transfer path of the planetary gearbox vibration signal.

[0011] S3, filtering out the transmission signal on the non-correlated path through the ANC (Adaptive Noise Cancellation) non-correlated path method;

[0012] S4, determining the most relevant transfer path by analyzing the vibration signal characteristics under normal conditions and crack fault conditions obtained in step S1 and the transfer path contribution obtained in step S2;

[0013] S5. The vibration signal separation method based on the multi-path dictionary separates the transmission signal on the most relevant transmission path from the measurement point signal to achieve the separation of the vibration signal.

[0014] Furthermore, in step S1, the planetary gear train dynamics model is established as follows: based on the concentrated mass method, the sun gear, planetary gears, inner ring gear, planet carrier, etc. are regarded as concentrated mass points, and the gear meshing is equivalent to a spring with stiffness and damping, so as to establish the planetary gear train dynamics model.

[0015] Furthermore, in step S1, the vibration signal characteristics obtained under normal conditions and crack fault conditions are divided into internal meshing vibration signal characteristics and external meshing vibration signal characteristics; the planetary gear-sun gear meshing vibration is internal meshing, and the inner ring gear-planetary gear meshing vibration is external meshing.

[0016] Further, in step S2, the transmission path includes:

[0017] Path 1: planetary gear → inner ring → housing → sensor;

[0018] Path 2: planetary gear → planetary gear center axis → planetary carrier → box → sensor;

[0019] Path 3: planetary gear → sun gear → sun gear center axis → housing → sensor;

[0020] Path 4: sun gear → planetary gear → planetary gear center → planetary carrier → housing → sensor;

[0021] Path 5: sun gear → planetary gear → inner ring gear → housing → sensor;

[0022] Path 6: sun gear → sun gear center axis → housing → sensor;

[0023] Among them, the distance between the meshing point and the sensor in paths 1 and 5 is periodic, which is a time-varying transmission path; the distance between the meshing point and the sensor in paths 2, 3, 4, and 6 is fixed, which is a time-invariant transmission path.

[0024] Furthermore, step S3 specifically includes the following sub-steps:

[0025] S31, determining that the main input signal is a coupling signal of the vibration source signal on the main transmission path of the planetary gearbox and the vibration source signal on the non-correlated transmission path to the measuring point via different transmission paths; determining that the reference signal is the vibration source signal on the non-correlated transmission path; the reference signal here is determined by the dynamic analysis and the transmission path influence law in step S1 and step S2;

[0026] S32, the adaptive filter is designed using the least mean square algorithm (LMS). By adjusting the adaptive filter parameters for multiple iterations, the residual signal after the main input signal is subtracted from the reference signal passing through the adaptive filter is ε The correlation evaluation index is minimized, realizing the separation of the non-correlated transfer path signal components and the correlated transfer path signals in the main input signal.

[0027] Furthermore, step S4 specifically analyzes the time domain curves and spectrum diagrams of the vibration signal characteristics under normal conditions and crack fault conditions obtained in step S1 to identify the spectrum characteristics; and combines the contribution of each transmission path obtained in step S2 to determine the most relevant transmission path.

[0028] Furthermore, step S5 specifically includes the following sub-steps:

[0029] S51, establish a multi-path dictionary; from the perspective of the transmission path of the planetary gearbox vibration mechanism, under normal conditions, the gear is constructed by a pulse response function modulated by the same transmission path; under the condition of a gear crack fault, the amplitude and frequency of the meshing vibration will change, thereby generating harmonic components, and then an amplitude modulation and frequency modulation function item is added on the basis of the dictionary of normal gear conditions;

[0030] S52, perform signal separation; on the basis of filtering out the transmission signal on the non-correlated path by the ANC non-correlated path method in step S3, establish a mathematical model of the vibration signal on the surface of the cabinet, further establish a convolution sparse model, solve the model by FBS (front-back segmentation algorithm), and achieve the final signal separation.

[0031] Through the above technical solution, the present invention has the following beneficial effects:

[0032] The method proposed in the present invention is aimed at gearboxes with complex transmission structures. The gear meshing vibration signal after passing through the most relevant transmission path is separated from the vibration signal collected from the surface of the box body, and Fourier transform and envelope spectrum analysis are performed. By analyzing the vibration signal characteristics on the most relevant transmission path, the gear meshing vibration signal can be separated more accurately and effectively, and feature classification can be used to further identify whether a gear crack fault has occurred. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 The overall flow chart of the method proposed by the present invention;

[0034] Figure 2 A flowchart of the ANC non-correlated path method in the method proposed by the present invention;

[0035] Figure 3 This is a schematic diagram of the vibration transmission path of the planetary gearbox;

[0036] Figure 4 Flowchart of the multi-path dictionary convolution sparse vibration signal separation method. DETAILED DESCRIPTION

[0037] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0038] Although the steps in the present invention are arranged with numbers, they are not used to limit the order of the steps. Unless the order of the steps is clearly stated or the execution of a certain step requires other steps as a basis, the relative order of the steps can be adjusted. It is understood that the term "and / or" used in this article involves and covers any and all possible combinations of one or more of the associated listed items.

[0039] like Figures 1 to 4 , a specific embodiment of the present invention is given:

[0040] like Figure 1 As shown, a planetary gearbox vibration signal separation method based on a multi-path dictionary proposed by the present invention is provided, which specifically includes the following steps:

[0041] S1. Establish a planetary gear train dynamics model based on the concentrated mass method; derive the relative motion equations of the coupled components based on the mutual motion relationship of the transmission components, and establish a planetary gearbox dynamics concentrated mass model; obtain the vibration signal characteristics of the planetary gearbox under normal conditions and crack fault conditions based on the established concentrated mass model;

[0042] As a preferred implementation of step S1, step S1 specifically includes:

[0043] The specific steps of establishing the dynamic model of the planetary gear train are as follows: based on the concentrated mass method, the sun gear, planetary gear, inner ring gear, planet carrier, etc. are regarded as concentrated mass points, and the gear meshing is equivalent to a spring with stiffness and damping, and the dynamic model of the planetary gear train is established.

[0044] The vibration signal characteristics obtained under normal conditions and crack fault conditions are divided into internal meshing vibration signal characteristics and external meshing vibration signal characteristics; the planetary gear-sun gear meshing vibration is internal meshing, and the inner ring gear-planetary gear meshing vibration is external meshing.

[0045] S2. Based on the nonlinear interface energy equivalent loss model, a multi-interface energy transfer loss model from the crack fault source to the target point is established. The transfer path contribution evaluation index is constructed in combination with the power flow method to obtain the contribution of each transfer path and identify the main transfer path and non-correlated transfer path of the planetary gearbox vibration signal.

[0046] More specifically, in step S2, based on the nonlinear interface energy equivalent loss model, the stiffness and damping equivalent dynamic parameters of the bearing interface and the housing joint surface at the key parts of the planetary gearbox are characterized; a multi-interface energy transfer loss model from the crack fault source to the target point is established, and a transfer path contribution evaluation index is constructed in combination with the power flow method, that is, the vibration energy reaching the housing surface sensor from different transfer paths is quantitatively compared, and the greater the energy, the greater the contribution, thereby identifying the main vibration transfer path and non-related transfer path of the planetary gearbox;

[0047] In addition, in this embodiment, if Figure 3 As shown, the transmission path in step S2 specifically includes:

[0048] Path 1: planetary gear → inner ring → housing → sensor;

[0049] Path 2: planetary gear → planetary gear center axis → planetary carrier → box → sensor;

[0050] Path 3: planetary gear → sun gear → sun gear center axis → housing → sensor;

[0051] Path 4: sun gear → planetary gear → planetary gear center → planetary carrier → housing → sensor;

[0052] Path 5: sun gear → planetary gear → inner ring gear → housing → sensor;

[0053] Path 6: sun gear → sun gear center axis → housing → sensor;

[0054] Among them, the distance between the meshing point and the sensor in paths 1 and 5 is periodic, which is a time-varying transmission path; the distance between the meshing point and the sensor in paths 2, 3, 4, and 6 is fixed, which is a time-invariant transmission path.

[0055] In this embodiment, in the next step S3, it is necessary to use ANC to filter out the non-correlated transfer path signal. Therefore, it is necessary to first identify the non-correlated transfer path, then determine the signal of the non-correlated transfer path (that is, the reference signal), and finally filter this signal through ANC. This is mainly achieved by analyzing the influence of the dynamic model and the transfer path in step S2, that is, the influence of the transfer path on the vibration signal received by the vibration sensor on the surface of the box, to determine the reference signal.

[0056] S3, filtering out the transmission signal on the non-correlated path through the ANC (Adaptive Noise Cancellation) non-correlated path method;

[0057] As a preferred implementation of step S3, Figure 2 As shown, specifically including:

[0058] S31, determining that the main input signal is a coupling signal of the vibration source signal on the main transmission path of the planetary gearbox and the vibration source signal on the non-correlated transmission path to the measuring point via different transmission paths; determining that the reference signal is the vibration source signal on the non-correlated transmission path; the reference signal here is determined by the dynamic analysis and the transmission path influence law in step S1 and step S2;

[0059] S32. The adaptive filter is designed using the Least Mean Square (LMS) algorithm. By adjusting the adaptive filter parameters for multiple iterations, the correlation evaluation index of the residual signal ε after the main input signal is subtracted from the reference signal passing through the adaptive filter is minimized, thereby achieving separation of the non-correlated transmission path signal components and the correlated transmission path signals in the main input signal.

[0060] S4, determining the most relevant transfer path by analyzing the vibration signal characteristics under normal conditions and crack fault conditions obtained in step S1 and the transfer path contribution obtained in step S2;

[0061] As a preferred implementation of step S4, it specifically includes:

[0062] S41, analyzing the time domain curves and spectrum diagrams of the inner and outer meshing vibration signal characteristics under normal conditions and crack fault conditions obtained in step S1 to identify spectrum characteristics;

[0063] S42. Determine the most relevant transmission path by combining the contribution of each transmission path obtained in step S2.

[0064] In this embodiment, under normal conditions, the vibration signals of the internal meshing and external meshing will not cause frequency modulation under the time-invariant transmission path (i.e., paths 2, 3, 4, and 6), but under the gear crack fault, the time-invariant of the internal meshing will have the characteristic of sideband components. Therefore, it is necessary to regard the internal meshing time-invariant path as the most relevant transmission path, in order to verify whether the separation of the gear meshing vibration signal under the fault crack condition is successful through the separated signal, and further identify the gear state; at the same time, the time-varying paths of the internal and external meshing are the main paths for transmitting vibration information, among which the path with the largest contribution contains the most complete vibration signal, and therefore it is also regarded as the most relevant transmission path; therefore, the meshing mode is finally determined to be internal meshing, and then combined with the contribution of each transmission path, the most relevant transmission paths are obtained: internal meshing time-varying path 5 (transmission path inside the box), internal meshing time-invariant path 6 (transmission path outside the box).

[0065] S5. The vibration signal separation method based on the multi-path dictionary separates the transmission signal on the most relevant transmission path from the measurement point signal to achieve the separation of the vibration signal.

[0066] As a preferred implementation of step S5, it specifically includes:

[0067] S51, establish a multi-path dictionary; from the perspective of the transmission path of the planetary gearbox vibration mechanism, under normal conditions, the gear is constructed by a pulse response function modulated by the same transmission path; under the condition of a gear crack fault, the amplitude and frequency of the meshing vibration will change, thereby generating harmonic components, and then an amplitude modulation and frequency modulation function item is added on the basis of the dictionary of normal gear conditions;

[0068] S52, perform signal separation; on the basis of filtering out the transmission signal on the non-correlated path by the ANC non-correlated path method in step S3, establish a mathematical model of the vibration signal on the surface of the cabinet, further establish a convolution sparse model, solve the model by FBS (front-back segmentation algorithm), and achieve the final signal separation.

[0069] In this embodiment, Figure 4 As shown:

[0070] The method for establishing the multi-path dictionary in step S51 is as follows: based on the most relevant transmission path determined in step S4, from the perspective of the transmission path of the planetary gearbox vibration mechanism, a multi-path dictionary is established for the normal gear case and the gear crack fault case (the dictionary of the transmission path in the gearbox is recorded as , the dictionary of the transmission path outside the box is recorded as ), under normal circumstances, the dictionary is constructed by the impulse response function modulated by the same transmission path; the formula of the impulse response function is expressed as:

[0071] ;

[0072] in represents a single pulse;

[0073] Get the dictionary of the transfer path in the box at the current time t under normal circumstances for:

[0074] ;

[0075] Under normal circumstances, the current time t is the external transfer path dictionary for:

[0076] ;

[0077] In the case of crack fault, the amplitude and frequency of meshing vibration will change, thus generating harmonic components. It is necessary to use the amplitude modulation function based on the normal gear dictionary. , frequency modulation function Adjust to get:

[0078] In the case of crack failure, the transfer path dictionary inside the box at the current time t for:

[0079] ;

[0080] In the case of crack failure, the transfer path dictionary outside the box at the current time t for:

[0081] ;

[0082] Among the above four formulas, Represents the amplitude intensity. is the natural vibration of the gearbox, is the load distribution ratio, Represents the number of planetary gears; is the gear meshing impact vibration represented by the impulse response function, is the number of single pulses; is the relative position between the first planetary gear and the remaining planetary gears; is the error of each planetary gear deviating from the nominal position along the tangential direction (circumferential direction), is the phase difference between the nth planetary gear meshing with the sun gear and the ring gear, Planet carrier rotation frequency, Sun wheel rotation frequency, is the meshing frequency; and The influence of the time-invariant transfer path and the time-varying transfer path on the internal meshing (sun gear-planet gear) vibration, is the vibration of the sun gear and the i-th planet gear.

[0083] In this embodiment, the signal separation in step S52 specifically includes the following steps:

[0084] S521. Establishing a mathematical model of the vibration signal on the box surface:

[0085] ;

[0086] in, is the observed signal, It is the weighted sum of the vibration signals of the external meshing after being modulated by transmission paths 1, 2, and 3. It is the signal modulated by the transmission path in the box. It is the signal modulated by the transmission path outside the box. Represents the harmonic interference component, Represents the noise component. The problem of using the sparse convolution model to separate the most relevant transmission path for the vibration signal of the measuring point is transformed into Separating the vibration signal on the most relevant transfer path obtained in step S4;

[0087] S522. Establish a sparse convolution model:

[0088] ;

[0089] in, is the observed signal, It is the weighted sum of the vibration signals of the external meshing after being modulated by transmission paths 1, 2, and 3. is the GMC regularization term. This non-convex penalty function avoids the problem of underestimation of the true value. A sparse dictionary representing the modulation coupling between the transfer path inside the box and the transfer path outside the box; represents the sparse coefficient of the corresponding dictionary, is the regularization term adjustment parameter, which is used to adjust the data sparsity in the regularization term function.

[0090] S523: According to the above steps S521 and S522, a dictionary highly similar to the vibration component of the gear transmission path is constructed, and the generalized minimum sag penalty is used as a sparse regularization term to suppress the transmission path modulation, and the optimal signal sparse coefficient is found through the FBS (front-back segmentation algorithm) iterative algorithm. , then the signal modulated by the transmission path in the box can be expressed as: ; The signal modulated by the transmission path outside the box is expressed as: ; Finally, the vibration signal separation on multiple transmission paths is achieved. According to the characteristics of the separated vibration signal, it can also be determined whether the gear has crack fault.

[0091] In summary, the method proposed in the present invention identifies the most relevant transmission path through the planetary gearbox dynamics model and the ANC filtering method for non-relevant path signals, combines the multi-path dictionary and realizes the signal separation of the most relevant path through the FBS algorithm. Compared with the traditional method, by analyzing the vibration signal characteristics on the relevant transmission path, it can more accurately and effectively determine whether a crack fault has occurred on the gear on the transmission path, and can further identify whether a fault has occurred through feature classification.

[0092] It will be apparent to those skilled in the art that the invention is not limited to the details of the exemplary embodiments described above and that the invention can be implemented in other specific forms without departing from the spirit or essential features of the invention. Therefore, the embodiments should be considered exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description, and it is intended that all variations falling within the meaning and scope of the equivalent elements of the claims be included in the invention. Any reference numeral in a claim should not be considered as limiting the claim to which it relates.

[0093] In addition, it should be understood that although the present specification is described according to implementation modes, not every implementation mode contains only one independent technical solution. This description of the specification is only for the sake of clarity. Those skilled in the art should regard the specification as a whole. The technical solutions in each embodiment may also be appropriately combined to form other implementation modes that can be understood by those skilled in the art.

Claims

1. A method for separating vibration signals of a planetary gearbox based on a multipath dictionary, characterized in that: The specific steps include: S1. Establish a planetary gear train dynamics model based on the concentrated mass method; derive the relative motion equations of the coupled components based on the mutual motion relationship of the transmission components, and establish a planetary gearbox dynamics concentrated mass model; obtain the vibration signal characteristics of the planetary gearbox under normal conditions and crack fault conditions based on the established concentrated mass model; S2. Based on the nonlinear interface energy equivalent loss model, a multi-interface energy transfer loss model from the crack fault source to the target point is established. The transfer path contribution evaluation index is constructed in combination with the power flow method to obtain the contribution of each transfer path and identify the main transfer path and non-correlated transfer path of the planetary gearbox vibration signal. S3, filtering out the transmission signal on the non-correlated path by using an ANC adaptive noise elimination non-correlated path method; S4, determining the most relevant transfer path by analyzing the vibration signal characteristics under normal conditions and crack fault conditions obtained in step S1 and the transfer path contribution obtained in step S2; S5. The vibration signal separation method based on the multipath dictionary separates the transmission signal on the most relevant transmission path from the measurement point signal to achieve the separation of the vibration signal, specifically including: S51, establish a multi-path dictionary; from the perspective of the transmission path of the planetary gearbox vibration mechanism, under normal conditions, the gear is constructed by a pulse response function modulated by the same transmission path; under the condition of a gear crack fault, the amplitude and frequency of the meshing vibration will change, thereby generating harmonic components, and then an amplitude modulation and frequency modulation function item is added on the basis of the dictionary of normal gear conditions; S52, perform signal separation; on the basis of filtering out the transmission signal on the non-correlated path by the ANC adaptive noise elimination non-correlated path method in step S3, establish a mathematical model of the vibration signal on the surface of the box, further establish a convolution sparse model, solve the model by the FBS front and back segmentation algorithm, and realize the final signal separation.

2. A planetary gearbox vibration signal separation method based on a multipath dictionary according to claim 1, characterized in that: The specific steps of establishing the planetary gear train dynamics model in step S1 are as follows: Based on the concentrated mass method, the sun gear, planetary gears, inner ring gear and planet carrier are regarded as concentrated mass points, and the gear meshing is equivalent to a spring with stiffness and damping, and the dynamic model of the planetary gear train is established.

3. The method for separating vibration signals of a planetary gearbox based on a multipath dictionary according to claim 1, characterized in that: In step S1, the vibration signal characteristics obtained under normal conditions and crack fault conditions are divided into internal meshing vibration signal characteristics and external meshing vibration signal characteristics; the planetary gear-sun gear meshing vibration is internal meshing, and the inner ring gear-planetary gear meshing vibration is external meshing.

4. The method for separating vibration signals of a planetary gearbox based on a multipath dictionary according to claim 1, characterized in that: The transfer path in step S2 includes: Path 1: planetary gear → inner ring gear → housing → sensor; Path 2: planetary gear → planetary gear center axis → planetary carrier → box → sensor; Path 3: planetary gear → sun gear → sun gear center axis → housing → sensor; Path 4: sun gear → planetary gear → planetary gear center → planetary carrier → housing → sensor; Path 5: sun gear → planetary gear → inner ring gear → housing → sensor; Path 6: sun gear → sun gear center axis → housing → sensor; Among them, the distance between the meshing point and the sensor in paths 1 and 5 is periodic, which is a time-varying transmission path; the distance between the meshing point and the sensor in paths 2, 3, 4, and 6 is fixed, which is a time-invariant transmission path.

5. The method for separating vibration signals of a planetary gearbox based on a multipath dictionary according to claim 1, characterized in that: Step S3 specifically includes: S31, determining that the main input signal is a coupling signal of the vibration source signal on the main transmission path of the planetary gearbox and the vibration source signal on the non-correlated transmission path to the measuring point via different transmission paths; determining that the reference signal is the vibration source signal on the non-correlated transmission path; the reference signal here is determined by the dynamic analysis and the transmission path influence law in step S1 and step S2; S32, the adaptive filter is designed by the least mean square algorithm LMS, and the residual signal after the main input signal minus the reference signal after passing through the adaptive filter is adjusted by multiple iterations. ε The correlation evaluation index is minimized, realizing the separation of the non-correlated transfer path signal components and the correlated transfer path signals in the main input signal.

6. The method for separating vibration signals of a planetary gearbox based on a multipath dictionary according to claim 1, characterized in that: Step S4 is specifically as follows: Analyze the time domain curves and spectrum diagrams of the vibration signal characteristics under normal conditions and crack fault conditions obtained in step S1 to identify spectrum characteristics; The contribution of each transfer path obtained in step S2 is combined to determine the most relevant transfer path.

Images