Fault feature identification method and system for rotating part in tachometer-free scene

Abstract

The invention belongs to the field of aero-engine test and fault diagnosis, relates to an engine rotating part fault positioning technology, and provides a rotating part fault feature recognition method and system in a tachometer-free scene, and the method comprises the steps: obtaining a vibration signal and a theoretical fault feature order of a rotating part; obtaining an instantaneous rotating frequency signal of the rotating part through the constructed non-convex linear transformation regular term sparse representation model; carrying out iterative solution through the constructed sparse representation model to obtain a fault characteristic signal of the rotating part; obtaining an order envelope spectrum according to the instantaneous frequency conversion signal and the fault characteristic signal; and performing health assessment and fault identification on the rotating part according to the order envelope spectrum and the theoretical fault feature order. The method can solve the problem that the variable-speed fault of the rotating shaft part of the aero-engine cannot be recognized in a non-rotating-speed scene. Through the implementation of the method, the fault of the rotating shaft part of the aero-engine is effectively identified, and technical support is provided for troubleshooting in an aero-engine test.

Description

TECHNICAL FIELD

[0001] The application belongs to the field of aero-engine test and fault diagnosis, and relates to a rotating component fault positioning technology, in particular to a rotating component fault feature recognition method and system in a tachometer-free scene. BACKGROUND

[0002] Under harsh service environments such as high temperature, high pressure and heavy load, rotating components such as rotors, bearings and accessory gears of an aero-engine are prone to local damage, and the damage will continue to expand. If not detected in time, it may lead to serious economic damage or even safety risks. Therefore, early fault feature extraction of rotating components on an aero-engine is of great significance for aero-engine operation state perception and maintenance decision.

[0003] Compared with general rotating mechanical systems, the aero-engine operating condition has the following differences: 1. Variable speed operating condition: The flight mission profile of the aero-engine such as climbing, accelerating, turning and descending is a typical variable speed operating condition; 2. There are tachometer-free application scenarios: Due to the limitation of installation space or harsh service environment, there are application scenarios where tachometers cannot be installed; 3. Weak fault features due to strong noise interference environment: The aero-engine is affected by strong aerodynamic, vibration and combustion noise during operation, and the fault features of the rotating components are submerged in strong background noise, making it difficult to extract weak fault features. The current variable speed operating condition rotating component fault feature representation methods mainly include order tracking, envelope order spectrum and order spectrum correlation. These methods are easily affected by noise signals, which weakens their fault feature extraction effect under early damage or strong background noise environment.

[0004] In recent years, a rich sparse representation model of rotating component fault features has been constructed for constant speed operating conditions. However, the fault features under variable speed operating conditions are not the same as those of rotating components under constant speed, making it difficult to directly apply these methods to variable speed operating conditions. A few sparse representation models require the instantaneous rotational frequency signal of the rotor and a large number of parameters to be optimized and calculated, which is low in calculation efficiency and difficult to apply to tachometer-free application scenarios.

[0005] Therefore, a variable speed fault feature recognition method that is easy to implement, consumes less resources and is not limited by noise is designed, which has very important engineering significance for aero-engine shaft fault diagnosis. SUMMARY

[0006] In order to solve the limitation of the prior art that the rotating mechanical component variable speed fault of the aero-engine under the non-tachometer scene cannot be identified, the present application discloses a rotating component fault feature identification method under a non-tachometer scene, which comprises the following steps: S1, obtaining the vibration signal of the rotating component in the operation process of the aero-engine and the theoretical fault feature order; S2, constructing a non-convex linear transformation regular term sparse representation model by using the vibration signal, and obtaining the instantaneous rotational frequency signal of the rotating component through the non-convex linear transformation regular term sparse representation model; S3, constructing a Morlet envelope enhanced sparse representation model by using the vibration signal, and obtaining the fault feature signal of the rotating component by iteratively solving the sparse representation model; S4, obtaining the feature signal by using the interpolation method according to the instantaneous rotational frequency signal and the fault feature signal, and obtaining the order envelope spectrum by performing envelope calculation and Fourier transform on the feature signal; S5, performing health assessment and fault identification on the rotating component according to the order envelope spectrum and the theoretical fault feature order.

[0007] Further, in the above step S1, the vibration signal of the rotating component in the operation process of the aero-engine and the theoretical fault feature order are obtained, which comprises: S11, installing a vibration acceleration sensor on the outer or inner casing of the support point of the rotating component, and collecting the vibration signal of the rotating component through the vibration acceleration sensor; S12, calculating the theoretical fault feature order of the rotating component according to the type and geometric parameters of the rotating component.

[0008] Further, in the above step S12, when the type of the rotating component is a bearing, the calculation formula of the theoretical fault feature order comprises: , ; , ; wherein, f o 、f i 、f b 、f c are the theoretical fault feature orders of the outer ring, inner ring, rolling body and retainer of the bearing, is the number of rolling bodies, is the diameter of the rolling body, is the pitch diameter of the rolling bearing, is the contact angle.

[0009] Further, in the step S12, when the rotating component is a gear, the formula for calculating the theoretical fault feature order includes: , ; wherein, given is the driving gear of the meshing gear pair, is the driven gear of the meshing gear pair, f g1 , f g2 are the theoretical fault feature orders of the driving gear and the driven gear respectively, is the number of teeth of the driving gear, is the number of teeth of the driven gear.

[0010] Further, in the step S2, the non-convex linear transformation regular term sparse representation model is constructed using the vibration signal, and the instantaneous rotating frequency signal of the rotating component is obtained through the non-convex linear transformation regular term sparse representation model, comprising: S21, Fourier transform is performed on the vibration signal to obtain a Fourier spectrum, frequency spectrum line coefficients in a set frequency band range of the Fourier spectrum are extracted, and inverse Fourier transform is performed on the frequency spectrum line coefficients to obtain a filtered vibration signal; S22, the non-convex linear transformation regular term sparse representation model is constructed using the filtered vibration signal; S23, the rotating frequency information related signal is obtained by iteratively solving the non-convex linear transformation regular term sparse representation model through a linear transformation regular term denoising algorithm; S24, short-time Fourier transform is performed on the rotating frequency information related signal to obtain a time-frequency spectrum, and synchronous compression transform is performed on the time-frequency spectrum to obtain a time-frequency distribution; S25, the ridge line of the time-frequency distribution is extracted using a ridge line extraction algorithm to obtain the instantaneous rotating frequency signal of the rotating component.

[0011] Further, in the step S2, the expression of the non-convex linear transformation regular term sparse representation model is: ; wherein, is the rotating frequency information related signal, is an intermediate variable, is a discrete cosine inverse transform, and are regular term parameters respectively, , ; when , the regular term is a non-convex function of , is a convex function of , is the filtered vibration signal.

[0012] Further, in the step S3, the expression of the Morlet envelope enhanced sparse representation model is: wherein, is a global optimization objective function, is a fault feature signal, is a first-order difference matrix, is a group size; and are regularization term parameters, satisfying and , is an intermediate variable; x n is a vibration signal; for any signal sequence , is the signal data point sequence number, represents a regularization term function for the signal sequence , is an integer, represents the calculation of the sum of squares of the data points in the signal sequence from to , and are regularization term functions corresponding to the signal sequences and .

[0013] Further, in the step S4, according to the instantaneous rotation frequency signal and the fault feature signal, a feature signal is obtained by interpolation, envelope calculation and Fourier transform are performed on the feature signal to obtain an order envelope spectrum, including: S41, rotation angle integration is performed on the instantaneous rotation frequency signal to obtain a rotation angle signal, and equal-angle sampling points signal is obtained by rotation angle sampling on the rotation angle signal; S42, according to the mapping relationship between the rotation angle signal and the fault feature signal, the equal-angle sampling point signal corresponding feature signal is obtained by spline interpolation or polynomial interpolation; S43, envelope calculation and Fourier transform are performed on the feature signal to obtain an order envelope spectrum.

[0014] Further, in the step S5, according to the order envelope spectrum and the theoretical fault feature order, the health assessment and fault identification of the rotating component are performed, including: S51, when the deviation between the order envelope spectrum and the theoretical fault feature order is less than a threshold value, or the order envelope spectrum contains an integral multiple of the theoretical fault feature order, it is judged that the rotating component is unhealthy and has a fault; S52, when the deviation between the order envelope spectrum and the theoretical fault characteristic order is greater than or equal to a threshold value, and the order envelope spectrum does not contain the integral multiple of the theoretical fault characteristic order, judging that the rotating component health has no fault.

[0015] The embodiment of the present application also provides a rotating component fault feature identification system in a tachometer-free scene, comprising a vibration signal acquisition module, a first calculation module, a model construction module, a second calculation module, an order envelope spectrum acquisition module and a health evaluation and fault identification module.

[0016] The vibration signal acquisition module is used to acquire the vibration signal of the rotating component in the operation process of the aero-engine. The first calculation module is used to acquire the theoretical fault characteristic order of the rotating component in the operation process of the aero-engine. The model construction module is used to construct the Morlet envelope enhanced sparse representation model through the vibration signal. The second calculation module is used to iteratively solve the sparse representation model to obtain the fault feature signal of the rotating component. The order envelope spectrum acquisition module is used to acquire the characteristic signal by using the interpolation method according to the instantaneous rotational frequency signal and the fault feature signal, and perform envelope calculation and Fourier transform on the characteristic signal to obtain the order envelope spectrum. The health evaluation and fault identification module is used to perform health evaluation and fault identification on the rotating component according to the order envelope spectrum and the theoretical fault characteristic order.

[0017] Compared with the prior art, the fault feature identification method has at least the following beneficial effects: 1) By extracting the instantaneous rotational frequency information and the fault feature signal of the rotating component from the aero-engine casing vibration monitoring signal, the rotating component variable speed fault feature extraction and characterization can be applied to the tachometer-free scene.

[0018] 2) The constructed Morlet envelope enhanced sparse representation model is based on the time domain group sparse characteristics of the non-periodic fault feature signal, and uses a first-order difference matrix for sparse matching. Compared with the traditional sparse representation model, the super parameter of the constructed sparse model is greatly reduced, the model super parameter optimization process is saved, and the efficiency of fault diagnosis is improved. BRIEF DESCRIPTION OF DRAWINGS

[0019] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following will briefly introduce the drawings needed to be used in the embodiments. Obviously, the drawings described in the following are only some of the embodiments of the present application, and not all the embodiments. The present application can also be implemented or applied through other different specific embodiments, and the details in the specification can be modified or changed based on different views and applications without departing from the spirit of the present application. It should be noted that the following embodiments and features of the embodiments can be combined with each other without conflict. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without making creative efforts are within the scope of protection of the present application.

[0020] Figure 1 Flow chart of the fault feature recognition method of the rotating component in the non-tachometer scene of the present application; Figure 2 Vibration signal of the rolling bearing on the test bearing of the double rotor of the certain type of engine; Figure 3 Rotation frequency information related signal of the test bearing; Figure 4 Time-frequency distribution of the rolling bearing on the test bearing; Figure 5 Instantaneous rotation frequency signal of the rolling bearing on the test bearing; Figure 6 Rolling bearing fault feature signal of the rolling bearing on the test bearing; Figure 7 Envelope order spectrum of the rolling bearing on the test bearing; Figure 8 Architecture diagram of the fault feature recognition system of the rotating component in the non-tachometer scene of the present application; Among them, 801, vibration signal acquisition module; 802, first calculation module; 803, model construction module; 804, second calculation module; 805, order envelope spectrum acquisition module; 806, health assessment and fault recognition module. DETAILED DESCRIPTION

[0021] The embodiments of the present application will be described in detail below with reference to the drawings.

[0022] The embodiments of the present application will be described in detail below with reference to the drawings.

[0023] The embodiment of the application discloses a rotating component fault feature identification method in a tachometer-free scene, referring to Figure 1 The method comprises the following steps: S1, acquiring vibration signals of a rotating component in an aero-engine running process and theoretical fault feature orders; S2, constructing a non-convex linear transformation regular term sparse representation model by using the vibration signals, and acquiring an instantaneous rotating frequency signal of the rotating component through the non-convex linear transformation regular term sparse representation model; S3, constructing a Morlet envelope enhanced sparse representation model by using the vibration signals, and iteratively solving the sparse representation model to obtain a fault feature signal of the rotating component; S4, acquiring a feature signal by using an interpolation method according to the instantaneous rotating frequency signal and the fault feature signal, and performing envelope calculation and Fourier transform on the feature signal to obtain an order envelope spectrum; S5, performing health assessment and fault identification on the rotating component according to the order envelope spectrum and the theoretical fault feature orders.

[0024] Further, in the step S1, the vibration signals of the rotating component in the aero-engine running process and the theoretical fault feature orders are acquired, comprising: S11, installing a vibration acceleration sensor on an outer or inner casing of a support point of the rotating component, and acquiring vibration signals of the rotating component through the vibration acceleration sensor, wherein the vibration signals can be expressed as wherein is a sampling sequence number of the vibration signals, is the total length of the vibration signals, and the sampling frequency of the vibration signals is .

[0025] S12, calculating the theoretical fault feature orders of the rotating component according to the type and geometric parameters of the rotating component.

[0026] Further, the type of the rotating component usually includes bearings, gears and the like, when the type of the rotating component is a bearing, the calculation formula of the theoretical fault feature orders comprises: , ; , ; wherein, f o 、f i 、f b 、f c are the theoretical fault feature orders of the outer ring, the inner ring, the rolling body and the retainer of the bearing, respectively is the number of rolling elements, is the diameter of rolling elements, is the pitch diameter of rolling elements, is the contact angle.

[0027] When the type of the rotating component is a gear, the calculation formula of the theoretical fault characteristic order comprises: , ; wherein, given is the driving gear of the meshing gear pair, is the driven gear of the meshing gear pair, f g1 , f g2 are the theoretical fault characteristic orders of the driving gear and the driven gear respectively, is the number of teeth of the driving gear, is the number of teeth of the driven gear.

[0028] Further, in the step S2, the non-convex linear transformation regular term sparse representation model is constructed by using the vibration signal, and the instantaneous rotating frequency signal of the rotating component is obtained through the non-convex linear transformation regular term sparse representation model, comprising: S21, Fourier transform is performed on the vibration signal to obtain a Fourier spectrum, frequency spectrum line coefficients in a set frequency band range of the Fourier spectrum are extracted, and inverse Fourier transform is performed on the frequency spectrum line coefficients to obtain a filtered vibration signal.

[0029] In implementation, the frequency band range can be set as , and the filtered vibration signal can be expressed as ; wherein is the highest rotating frequency designed or regularly operated by the aero-engine, and the frequency spectrum line coefficients outside the frequency band range are set to 0.

[0030] S22, the non-convex linear transformation regular term sparse representation model is constructed by using the filtered vibration signal, and the expression of the model is: ; wherein, is the instantaneous rotating frequency signal, is an intermediate variable, is a discrete cosine inverse transform, and are regular term parameters respectively, , ; when , the regular term is a non-convex function of , is a convex function of .

[0031] S23, the non-convex linear transformation regular term sparse representation model is solved by the linear transformation regular term denoising algorithm to obtain the rotation frequency information related signal; in the iterative solving, the transformation rate of the rotation frequency information related signals obtained by two continuous iterations is calculated, and when the maximum transformation rate is less than a given threshold or the maximum iteration number is reached, the iteration calculation is terminated, that is, it can be expressed as: or ; wherein, , is the rotation frequency information related signal obtained by the first and the first iteration process, is the iteration round; is a signal point division operator with the same length; is a maximum value operator; is a given threshold, generally selected as ; is the maximum iteration number, generally selected as . When the iteration calculation termination condition is reached, the of the last iteration calculation is taken as the final rotation frequency information related signal, and the acquisition process of the rotation frequency information related signal is shown in the following algorithm 1: Algorithm 1: Linear transformation regular term denoising algorithm Input: low-pass filtered vibration signal , , , , ; Initialization: let , ; Loop : 1. ; 2. ; 3. Determine the termination condition: or ; End: output Y0.

[0032] S24, the time-frequency spectrum is obtained by performing short-time Fourier transform on the rotation frequency information related signal, and the time-frequency distribution is obtained by performing synchronous compression transform on the time-frequency spectrum, which can be expressed as: ; wherein, is the time-frequency spectrum, is the Dirac function, is the time-frequency distribution, is a time series, is the frequency, is the sign substitution of frequency in the integration operation, j is an imaginary number , is the partial derivative of the time spectrum with respect to time.

[0033] S25. Use a ridge extraction algorithm to extract the ridge of the time-frequency distribution to obtain the instantaneous rotation frequency signal of the rotating component. The instantaneous rotation frequency signal can be expressed as , instantaneous frequency conversion signal Vibration monitoring signal The same length.

[0034] Furthermore, in the above step S3, a sparse representation model enhanced by Morrow envelope is constructed by using the vibration signal, and the sparse representation model is iteratively solved to obtain the fault characteristic signal of the rotating component, including: S31, through vibration signal Construct a sparse representation model enhanced by Morrow envelope, expressed as: ,in, is the global optimization objective function, is the fault characteristic signal, is the first-order difference matrix, is the group size; and are regularization parameters respectively, satisfying and , is the intermediate variable; x n is a vibration signal; for any signal sequence , is the sequence number of the signal data point, Signal sequence The regularization function of is an integer, Indicates the signal sequence from to Calculate the sum of squares of the data points, and Corresponding to the signal sequence and The regularization function of .

[0035] Among them, the first-order difference matrix D can be expressed as: ;matrix The size of .

[0036] S32. Global optimization objective function in the above sparse representation model is a convex function, and the fault feature signal of the rotating component is obtained by iteratively solving the group sparse total variation denoising algorithm . .

[0037] In implementation, the process of iteratively solving the fault feature signal by the group sparse total variation denoising algorithm is shown in the following algorithm 2, wherein, is a transpose matrix of, and the termination condition in iterative calculation is the maximum transformation rate or the maximum iteration number of the sequence obtained by two consecutive iterative calculations, which is represented as: or

[0038] wherein, , are the signal sequences obtained by the first and the second iterative processes, respectively; is a two-same-length signal point division operator; is a maximum value taking operator; is a maximum change rate threshold, and is generally selected as ; is a maximum iteration number, and is generally selected as . When the iterative calculation termination condition is reached, the sequence in the last round of iterative calculation is the extracted frequency conversion information related signal.

[0039] Algorithm: group sparse total variation denoising algorithm 2: constructor: ; ; ; ; input: vibration signal , , , ; initialization: let , ; loop : 1. ; 2. ; 3. judge the termination condition: or .

[0040] Further, in step S4, the characteristic signal is obtained by interpolation according to the instantaneous rotation frequency signal and the fault characteristic signal, envelope calculation and Fourier transform are performed on the characteristic signal to obtain the order envelope spectrum, including: S41, the rotation angle signal is obtained by rotation frequency integration on the instantaneous rotation frequency signal, and the equi-angle sampling point signal is obtained by rotation angle sampling on the rotation angle signal.

[0041] In implementation, the instantaneous rotation frequency signal of the rotating component and the fault characteristic signal are both equal-length and equal-time interval sampling signals, and the sampling frequency of the signal is At the same time, and the signal data points are the same, and the corresponding rotation angle signal is calculated by rotation frequency integration calculation The rotation angle signal can be expressed as: is the i-th data point in the instantaneous rotation frequency signal, i is a natural number.

[0042] The rotation angle signal is sampled in the rotation angle range according to the rotation angle sampling interval, and the equi-angle sampling point signal can be obtained. Since the rotation angle signal is consistent with the length of the sequence , in the rotation angle range , the equi-angle sampling point signal can be obtained by setting the rotation angle sampling point with as the angle interval, where is the rotation angle sampling frequency, and the equi-angle sampling point signal can be expressed as: wherein, is the length of the equi-angle sampling point sequence .

[0043] S42, according to the mapping relationship between the rotation angle signal and the fault characteristic signal, the characteristic signal corresponding to the equi-angle sampling point signal is obtained by spline interpolation or polynomial interpolation. In implementation, according to the mapping relationship between , the characteristic signal corresponding to the equi-angle sampling point signal is calculated by spline interpolation or polynomial interpolation method.

[0044] S43, envelope calculation and Fourier transform are performed on the characteristic signal to obtain the order envelope spectrum, which can be expressed as: wherein,​ is a Hilbert transform operator; is a modulus operator; is a Fourier transform.

[0045] Further, in the step S5, the rotating component is health evaluated and fault identified according to the order envelope spectrum and the theoretical fault characteristic order, comprising: S51, when the deviation between the order envelope spectrum and the theoretical fault characteristic order is less than a threshold value, or the order envelope spectrum contains the integral multiple of the theoretical fault characteristic order, judging that the rotating component is unhealthy and has a fault; S52, when the deviation between the order envelope spectrum and the theoretical fault characteristic order is greater than or equal to a threshold value, and the order envelope spectrum does not contain the integral multiple of the theoretical fault characteristic order, judging that the rotating component is healthy and has no fault, wherein the threshold value can be set to 1%, and the multiple relationship can be set to 2-4 order components.

[0046] The present application takes a certain type of engine double rotor test bearing as an example to illustrate the above method, and a crack fault is prefabricated in the inner ring of the test bearing before the experiment. According to the geometric dimensions of the test bearing, the calculated theoretical fault characteristic orders of each component of the test rolling bearing are shown in Table 1 as follows: Table 1: Geometric parameters and theoretical fault characteristic orders of test bearing

[0047] In the experiment, the acceleration sensor installed on the bearing seat collects the vibration signal of the rolling bearing on the test bearing, and the sampling frequency in the experiment is 200 kHz. The collected vibration signal is as shown in Figure 2 . The data signal length is 2000000 data points, and the low-pass filter passband range is set to . The vibration test signal is low-pass filtered. On this basis, a non-convex linear transformation regularization term sparse representation model is constructed, wherein the regularization term parameter is set to , The rotating frequency information related signal obtained by solving the sparse representation model is as shown in Figure 3 , which is mainly a frequency modulation harmonic signal. The rotating frequency information related signal is subjected to synchronous compression transformation based on short-time Fourier transform, and the obtained time-frequency distribution is as shown in Figure 4 . Further, the time-frequency distribution is analyzed by a time-frequency face ridge line extraction algorithm, and the extracted instantaneous rotating frequency signal is as shown in Figure 5 .

[0048] By constructing a Morlet envelope enhanced sparse representation model, the parameters in the model are set to , The rolling bearing fault characteristic signal obtained by solving the sparse representation model is as follows: Figure 6 As shown, it can be seen that there is a significant impact component in the extracted fault feature signal.

[0049] On this basis, the characteristic signal can be calculated by extracting the instantaneous frequency conversion signal and the fault characteristic signal. Figure 7 The envelope order spectrum shown is Figure 7 Where FCO is the theoretical fault characteristic order of a component calculated on the bearing. The main components of the envelope order spectrum are , this component is consistent with the theoretical fault characteristic order of the inner ring of the test bearing The relative deviation between the two is 0.07%. At the same time, there is The presence of integer multiples of the order of 2, 3, and so on indicates that the test bearing is faulty and located on the inner race. This data demonstrates that the method of the present invention can accurately extract the instantaneous rotational frequency and non-periodic fault impact characteristic signals of aircraft engine rotating components under variable speed conditions, and characterize and identify rotating component faults.

[0050] Based on the same inventive concept, an embodiment of the present invention further provides a system for identifying fault characteristics of rotating parts in a tachometer-free scenario, as described in the following embodiments. Since the principle of solving the problem by the system for identifying fault characteristics of rotating parts in a tachometer-free scenario is similar to the method for identifying fault characteristics of rotating parts in a tachometer-free scenario disclosed in the above embodiments, the implementation of the system for identifying fault characteristics of rotating parts in a tachometer-free scenario can refer to the implementation of the method for identifying fault characteristics of rotating parts in a tachometer-free scenario, and the repeated parts will not be repeated. As used below, the terms "unit" or "module" can be a combination of software and / or hardware that implements a predetermined function. Although the apparatus described in the following embodiments is preferably implemented in software, implementation in hardware, or a combination of software and hardware, is also possible and conceived.

[0051] Figure 8 This is a structural block diagram of a fault feature identification system for rotating components in a tachometer-free scenario disclosed in an embodiment of the present invention. Figure 8 As shown, the system includes a vibration signal acquisition module 801, a first calculation module 802, a model construction module 803, a second calculation module 804, an order envelope spectrum acquisition module 805 and a health assessment and fault identification module 806. The structure is described below.

[0052] The vibration signal acquisition module 801 is used to acquire vibration signals of rotating parts during the operation of the aircraft engine; The first calculation module 802 is used to obtain the theoretical fault characteristic order of the rotating components during the operation of the aircraft engine; The model construction module 803 is configured to construct a Morlet envelope enhanced sparse representation model from the vibration signal, and construct the Morlet envelope enhanced sparse representation model from the vibration signal. The second calculation module 804 is configured to iteratively solve the sparse representation model to obtain the fault characteristic signal of the rotating component, and iteratively solve the sparse representation model to obtain the fault characteristic signal of the rotating component. The order envelope spectrum acquisition module 805 is configured to acquire a characteristic signal from the instantaneous rotational frequency signal and the fault characteristic signal by using an interpolation method, and perform envelope calculation and Fourier transform on the characteristic signal to obtain an order envelope spectrum. The health assessment and fault identification module 806 is configured to perform health assessment and fault identification on the rotating component according to the order envelope spectrum and a theoretical fault characteristic order.

[0053] Compared with the prior art, the fault characteristic identification method has at least the following beneficial effects: 1) The instantaneous rotational frequency information and the fault characteristic signal of the rotating component are extracted from the aero-engine casing vibration monitoring signal, so that the fault characteristic of the rotating component in the non-tachometer scene can be extracted and characterized.

[0054] 2) The constructed Morlet envelope enhanced sparse representation model is based on the time domain group sparse characteristics of the non-periodic fault characteristic signal, and a first-order difference matrix is used for sparse matching, so that, compared with the traditional sparse representation model, the super parameter of the constructed sparse model is greatly reduced, the model super parameter optimization process is saved, and the efficiency of fault diagnosis is improved.

[0055] In this embodiment, a computer device is provided, which includes a memory, a processor, and a computer program stored in the memory and executable on the processor, and the processor implements the above-mentioned any fault characteristic identification method of the rotating component in the non-tachometer scene.

[0056] Specifically, the computer device can be a computer terminal, a server, or a similar computing device.

[0057] In this embodiment, a computer readable storage medium is provided, which stores a computer program for executing the above-mentioned any fault characteristic identification method of the rotating component in the non-tachometer scene.

[0058] In particular, computer readable storage media tangibly embody the processing circuit. In fact, a computer readable storage medium does not reflect a transitory propagating signal per se. The computer readable storage medium can include without limitation volatile and non-volatile, removable and non-removable media implemented in any method or technology for storage of information such as computer readable instructions, data structures, program modules or other data. Examples of computer readable storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read only memory (ROM), electrically erasable programmable read only memory (EEPROM), flash memory or other memory technology, compact disc read-only memory (CD-ROM), digital versatile disks (DVD) or other optical storage, magnetic cassettes, magnetic tape, magnetic disk storage or other magnetic storage devices or any other non-transmission medium that can be used to store information for access by a computing device. In accordance with the teachings provided herein, computer readable storage media does not include a transitory medium, such as modulated data signal and carrier wave.

[0059] It is apparent that software modules or steps can be released separately from each other, and thus can be stored and transmitted in the form of codes. It is also apparent that the modules or steps can be implemented in the form of codes, codes being stored in a storage medium and being executed by a processor. Although the preferred embodiments of the present application have been disclosed with reference to the drawings, various changes and modifications can be suggested to one skilled in the art, and it is intended that the present application encompass such changes and modifications as fall within the scope of the appended claims.

[0060] The preferred embodiments of the present application have been disclosed with reference to the drawings, and various changes and modifications can be suggested to one skilled in the art and it is intended that the present application encompass such changes and modifications as fall within the scope of the appended claims.

Claims

1. A method for identifying fault characteristics of rotating components in a tachometer-free scenario, characterized in that: include: Obtain vibration signals and theoretical fault characteristic orders of rotating parts during aircraft engine operation; A non-convex linear transformation regularization term sparse representation model is constructed using the vibration signal, and an instantaneous rotation frequency signal of the rotating component is obtained through the non-convex linear transformation regularization term sparse representation model; Constructing a Morrow envelope enhanced sparse representation model based on the vibration signal, and iteratively solving the sparse representation model to obtain a fault characteristic signal of the rotating component; According to the instantaneous frequency conversion signal and the fault characteristic signal, an interpolation method is used to obtain a characteristic signal, and envelope calculation and Fourier transform are performed on the characteristic signal to obtain an order envelope spectrum; Health assessment and fault identification are performed on the rotating component according to the order envelope spectrum and the theoretical fault characteristic order.

2. The method for identifying fault characteristics of rotating components in a tachometer-free scenario according to claim 1, characterized in that: Obtain vibration signals and theoretical fault characteristic orders of rotating components during aircraft engine operation, including: Installing a vibration acceleration sensor on the outer or inner casing of the support point on the rotating component, and collecting the vibration signal of the rotating component through the vibration acceleration sensor; According to the type and geometric parameters of the rotating component, a theoretical fault characteristic order of the rotating component is calculated.

3. The method for identifying fault characteristics of rotating components in a tachometer-free scenario according to claim 2, characterized in that: When the type of the rotating component is a bearing, the calculation formula of the theoretical fault characteristic order includes: ， ； ， ； in, f o 、f i 、f b 、f c The theoretical fault characteristic orders of the bearing's outer ring, inner ring, rolling element and cage are as follows: is the number of rolling elements, is the rolling element diameter, is the pitch diameter of the rolling bearing, is the contact angle.

4. The method for identifying fault characteristics of rotating components in a tachometer-free scenario according to claim 2, characterized in that: When the type of the rotating component is a gear, the calculation formula of the theoretical fault characteristic order includes: , ; where given It is the driving wheel of the meshing gear pair. is the driven gear of the meshing gear pair, f g1 、f g2 are the theoretical fault characteristic orders of the driving wheel and the driven wheel respectively, is the number of teeth on the driving gear, is the number of driven gear teeth.

5. The method for identifying fault characteristics of rotating components in a tachometer-free scenario according to claim 1, characterized in that: The vibration signal is used to construct a non-convex linear transformation regularization term sparse representation model, and the instantaneous rotation frequency signal of the rotating component is obtained through the non-convex linear transformation regularization term sparse representation model, including: Performing a Fourier transform on the vibration signal to obtain a Fourier spectrum, extracting frequency spectrum line coefficients within a set frequency band range within the Fourier spectrum, and performing an inverse Fourier transform on the frequency spectrum line coefficients to obtain a filtered vibration signal; Using the filtered vibration signal to construct a non-convex linear transformation regularization term sparse representation model; Iteratively solving the non-convex linear transformation regularization term sparse representation model by a linear transformation regularization term denoising algorithm to obtain a frequency conversion information related signal; Performing short-time Fourier transform on the frequency conversion information related signal to obtain a time-frequency spectrum, and performing synchronous compression transform on the time-frequency spectrum to obtain a time-frequency distribution; A ridge extraction algorithm is used to extract the ridge of the time-frequency distribution to obtain the instantaneous rotation frequency signal of the rotating component.

6. The method for identifying fault characteristics of rotating components in a tachometer-free scenario according to claim 5, characterized in that: The expression of the non-convex linear transformation regularization term sparse representation model is: ;in, is the frequency conversion information related signal, is an intermediate variable, is the inverse discrete cosine transform, and are the regularization parameters, 、 ;when When for A non-convex function of for The convex function of is the vibration signal after filtering.

7. The method for identifying fault characteristics of rotating components in a tachometer-free scenario according to claim 1, characterized in that: The expression of the Morrow envelope enhanced sparse representation model is: ,in, is the global optimization objective function, is the fault characteristic signal, is the first-order difference matrix, is the group size; and are regularization parameters respectively, satisfying and , is the intermediate variable; x n is a vibration signal; for any signal sequence , is the sequence number of the signal data point, Signal sequence The regularization function of is an integer, Indicates the signal sequence from to Calculate the sum of squares of the data points, and Corresponding to the signal sequence and The regularization function of .

8. The method for identifying fault characteristics of rotating components in a tachometer-free scenario according to claim 1, characterized in that: According to the instantaneous frequency conversion signal and the fault characteristic signal, an interpolation method is used to obtain a characteristic signal, and envelope calculation and Fourier transform are performed on the characteristic signal to obtain an order envelope spectrum, including: Performing frequency integration on the instantaneous frequency signal to obtain an angle signal, and performing angle sampling on the angle signal to obtain an equal-angle sampling point signal; According to the mapping relationship between the rotation angle signal and the fault characteristic signal, the characteristic signal corresponding to the equal-angle sampling point signal is obtained by spline interpolation or polynomial interpolation; Envelope calculation and Fourier transform are performed on the characteristic signal to obtain an order envelope spectrum.

9. The method for identifying fault characteristics of rotating components in a tachometer-free scenario according to claim 1, characterized in that: Performing health assessment and fault identification on the rotating component according to the order envelope spectrum and the theoretical fault characteristic order, including: When the deviation between the order envelope spectrum and the theoretical fault characteristic order is less than a threshold, or the order envelope spectrum contains an integer multiple of the theoretical fault characteristic order, it is determined that the rotating component is unhealthy and has a fault; When the deviation between the order envelope spectrum and the theoretical fault characteristic order is greater than or equal to a threshold, and the order envelope spectrum does not contain an integer multiple of the theoretical fault characteristic order, it is determined that the rotating component is healthy and has no faults.

10. A fault feature identification system for rotating parts in a tachometer-free scenario, characterized in that: include: A vibration signal acquisition module, which is used to acquire vibration signals of rotating parts during the operation of the aircraft engine; a first calculation module, configured to obtain a theoretical fault characteristic order of a rotating component during operation of an aircraft engine; a model construction module, the model construction module being configured to construct a Morrow envelope enhanced sparse representation model based on the vibration signal; a second computing module, configured to iteratively solve the sparse representation model to obtain a fault characteristic signal of the rotating component; An order envelope spectrum acquisition module is used to obtain a characteristic signal using an interpolation method based on the instantaneous frequency conversion signal and the fault characteristic signal, and to perform envelope calculation and Fourier transform on the characteristic signal to obtain an order envelope spectrum; A health assessment and fault identification module is used to perform health assessment and fault identification on the rotating component according to the order envelope spectrum and the theoretical fault characteristic order.

Images