Liquid rocket engine turbine pump bearing fault diagnosis method based on dual adaptive filtering and related equipment

Abstract

The invention discloses a fault diagnosis method for a turbine pump bearing of a liquid rocket engine based on dual adaptive filtering and related equipment, and the method comprises the steps: processing an original vibration signal through an adaptive line spectrum intensifier, and obtaining a vibration signal after the interference of a first-order cyclostationary signal is eliminated; blind cyclic frequency detection is carried out on the vibration signals after the first-order cyclostationary signal interference is eliminated, and a cyclic frequency set is obtained; taking the vibration signal after eliminating the interference of the first-order cyclostationary signal as the input of a multi-channel adaptive FRESH filter, taking the frequency shift component used by each channel as the frequency in a cyclic frequency set, and outputting the filtered vibration signal; and analyzing the filtered vibration signal by using cyclostationary analysis to obtain a diagnosis result. According to the method, a pure second-order cyclostationary signal related to a fault can be accurately extracted from a fault liquid rocket engine turbine pump bearing vibration signal, the filtering process can adapt to the change of a system state, and a reliable pretreatment result can be provided for a bearing fault diagnosis method based on cyclostationary analysis; and finally, accurate diagnosis of the turbine pump bearing fault of the liquid-propellant rocket engine is realized.

Description

Technical Field

[0001] The present invention belongs to the field of liquid rocket engine turbine pump bearing fault diagnosis, and in particular relates to a liquid rocket engine turbine pump bearing fault diagnosis method based on dual adaptive filtering and related equipment. Background Art

[0002] Bearings are an important component of various rotating mechanical transmission systems. They are designed to carry shaft system components to improve the mechanical efficiency and reliability of the system. They are widely used in high-speed rotating occasions such as liquid rocket engine turbopumps. If they fail, the performance of the liquid rocket engine system will be affected. In severe cases, they may even cause catastrophic accidents and bring huge economic losses.

[0003] The vibration signal of a faulty bearing often exhibits second-order cyclostationary characteristics, so cyclostationary analysis is a powerful tool for bearing fault diagnosis. In order to improve the diagnostic accuracy of cyclostationary analysis for bearing faults, the collected vibration signal needs to be filtered. In fact, there are often first-order cyclostationary components in the vibration signal collected, which will produce a large amount of harmonic interference. In the existing liquid rocket engine turbine pump bearing fault diagnosis process, this interference is often ignored, resulting in deviations in the diagnosis results; in addition, in the existing liquid rocket engine turbine pump bearing fault diagnosis process, the adaptive filter needs to accurately know the prior knowledge about the fault characteristics, that is, the accurate fault characteristic reference signal, so that the filter can have better extraction performance, but the actual situation is complicated. In most cases, the fault extraction characteristic frequency cannot be predicted in advance, that is, the actual cycle frequency set is unknown. Summary of the invention

[0004] In response to the problems existing in the prior art, the present invention provides a liquid rocket engine turbine pump bearing fault diagnosis method and related equipment based on dual adaptive filtering, which can accurately extract a pure second-order cyclostationary signal related to the fault from the vibration signal of the faulty liquid rocket engine turbine pump bearing. The filtering process can adapt to the changes in the system state and can provide reliable pre-processing results for the bearing fault diagnosis method based on cyclostationary analysis, ultimately achieving accurate diagnosis of liquid rocket engine turbine pump bearing faults.

[0005] In order to solve the above technical problems, the present invention is implemented by the following technical solutions:

[0006] According to a first aspect of the present invention, a liquid rocket engine turbopump bearing fault diagnosis method based on dual adaptive filtering is provided, comprising:

[0007] Obtaining the original vibration signal of the bearing to be diagnosed;

[0008] Processing the original vibration signal through an adaptive line spectrum enhancer to obtain a vibration signal after eliminating first-order cyclostationary signal interference;

[0009] Performing blind cyclic frequency detection on the vibration signal after eliminating the interference of the first-order cyclostationary signal to obtain a cyclic frequency set;

[0010] The vibration signal after eliminating the interference of the first-order cyclostationary signal is used as the input of the multi-channel adaptive FRESH filter, and the frequency shift component used by each channel is the frequency in the cyclic frequency set, and the filtered vibration signal is output;

[0011] The filtered vibration signal is analyzed using cyclostationary analysis to obtain a diagnosis result.

[0012] In a possible implementation manner of the first aspect, the original vibration signal is processed by an adaptive line spectrum enhancer to obtain a vibration signal after first-order cyclostationary signal interference is eliminated, specifically:

[0013] Using the original noisy vibration signal as a reference signal for the adaptive line spectrum enhancer;

[0014] Time-shift the original vibration signal backward by a number of samples to serve as an input signal of the adaptive line spectrum enhancer;

[0015] Inputting the input signal into a finite length unit impulse response filter, adaptively adjusting the coefficient of the finite length unit impulse response filter according to the state of the liquid rocket engine turbo pump by a recursive least square method, and obtaining a first-order cyclostationary signal to be eliminated, which is recorded as an output signal;

[0016] The reference signal is subtracted from the output signal to obtain a vibration signal after first-order cyclostationary signal interference is eliminated.

[0017] In a possible implementation manner of the first aspect, performing blind cyclic frequency detection on the vibration signal after eliminating the interference of the first-order cyclostationary signal to obtain a cyclic frequency set is specifically:

[0018] Using the vibration signal after eliminating the interference of the first-order cyclostationary signal as the input of the blind cyclic frequency detector SSCA;

[0019] Performing sliding windowing processing on the vibration signal after eliminating the interference of the first-order cyclostationary signal, wherein the length of the vibration signal after eliminating the interference of the first-order cyclostationary signal is N, the sliding window length is N′, the overlapping area between the sliding windows is N′-1, N is a natural number not equal to 0, and N′ is less than N;

[0020] The signal of each sliding window is converted into a column vector, and the column vectors are arranged backward in sequence as the sliding window moves until the sliding window is translated to the tail of the vibration signal after eliminating the interference of the first-order cyclostationary signal, so as to obtain an N′×P-dimensional channelized input signal matrix;

[0021] Performing a fast Fourier transform on each column of the N′×P-dimensional channelized input signal matrix to obtain an N′-point fast Fourier transform matrix;

[0022] Repeatedly expanding the N′-point fast Fourier transform matrix by columns so that the number of columns of the N′-point fast Fourier transform matrix is ​​equal to the length N of the vibration signal after eliminating the interference of the first-order cyclostationary signal, to obtain an N′×N dimensional matrix;

[0023] Multiplying each row in the N′×N dimensional matrix with the conjugate of the vibration signal after eliminating the interference of the first-order cyclostationary signal to obtain a conjugate multiplied N′×N dimensional matrix;

[0024] Performing a fast Fourier transform on each row of the conjugate multiplied N′×N dimensional matrix to obtain an N′×N dimensional spectral correlation estimation matrix;

[0025] Taking the square of each element in the N′×N dimensional spectrum correlation estimation matrix, and summing the N′×N dimensional spectrum correlation estimation matrix row by row to obtain an N dimensional row vector;

[0026] The frequencies corresponding to the first three largest elements in the N-dimensional row vector and their integer multiples are selected to form a set to obtain a cyclic frequency set.

[0027] In a possible implementation manner of the first aspect, the vibration signal after eliminating the interference of the first-order cyclostationary signal is used as the input of the multi-channel adaptive FRESH filter, and the frequency shift component used by each channel is the frequency in the cyclic frequency set, and the filtered vibration signal is output, specifically:

[0028] Frequency shifting the vibration signal after eliminating the interference of the first-order cyclostationary signal according to the frequencies in the cyclic frequency set to obtain frequency shifted signals of each channel of the multi-channel adaptive FRESH filter;

[0029] Input the frequency shift signal of each channel into its own finite length unit impulse response filter to obtain the output signal of each channel;

[0030] The output signals of each channel are summed to obtain the output signal of the multi-channel adaptive FRESH filter, which is the filtered vibration signal.

[0031] In a possible implementation manner of the first aspect, the filtered vibration signal is analyzed by using cyclostationary analysis to obtain a diagnosis result, specifically:

[0032] Calculating a spectral correlation or spectral coherence function of the filtered vibration signal;

[0033] Fault diagnosis is achieved based on the spectral correlation or spectral coherence function.

[0034] According to a second aspect of the present invention, there is provided a liquid rocket engine turbo pump bearing fault diagnosis device based on dual adaptive filtering, comprising:

[0035] An acquisition module, used for acquiring an original vibration signal of a bearing to be diagnosed;

[0036] An elimination module, used for processing the original vibration signal through an adaptive line spectrum enhancer to obtain a vibration signal after eliminating interference from a first-order cyclostationary signal;

[0037] A detection module, used for performing blind cyclic frequency detection on the vibration signal after eliminating the interference of the first-order cyclostationary signal, to obtain a cyclic frequency set;

[0038] An output module is used to use the vibration signal after eliminating the interference of the first-order cyclostationary signal as the input of a multi-channel adaptive FRESH filter, and the frequency shift component used by each channel is the frequency in the cyclic frequency set, and output the filtered vibration signal;

[0039] The diagnosis module is used to analyze the filtered vibration signal by using cyclostationary analysis to obtain a diagnosis result.

[0040] In a possible implementation manner of the second aspect, the elimination module is specifically configured to:

[0041] Using the original noisy vibration signal as a reference signal for the adaptive line spectrum enhancer;

[0042] Time-shift the original vibration signal backward by a number of samples to serve as an input signal of the adaptive line spectrum enhancer;

[0043] Inputting the input signal into a finite length unit impulse response filter, adaptively adjusting the coefficient of the finite length unit impulse response filter according to the state of the liquid rocket engine turbo pump by a recursive least square method, and obtaining a first-order cyclostationary signal to be eliminated, which is recorded as an output signal;

[0044] The reference signal is subtracted from the output signal to obtain a vibration signal after first-order cyclostationary signal interference is eliminated.

[0045] In a possible implementation manner of the second aspect, the detection module is specifically configured to:

[0046] Using the vibration signal after eliminating the interference of the first-order cyclostationary signal as the input of the blind cyclic frequency detector SSCA;

[0047] Performing sliding windowing processing on the vibration signal after eliminating the interference of the first-order cyclostationary signal, wherein the length of the vibration signal after eliminating the interference of the first-order cyclostationary signal is N, the sliding window length is N′, the overlapping area between the sliding windows is N′-1, N is a natural number not equal to 0, and N′ is less than N;

[0048] The signal of each sliding window is converted into a column vector, and the column vectors are arranged backward in sequence as the sliding window moves until the sliding window is translated to the tail of the vibration signal after eliminating the interference of the first-order cyclostationary signal, so as to obtain an N′×P-dimensional channelized input signal matrix;

[0049] Performing a fast Fourier transform on each column of the N′×P-dimensional channelized input signal matrix to obtain an N′-point fast Fourier transform matrix;

[0050] Repeatedly expanding the N′-point fast Fourier transform matrix by columns so that the number of columns of the N′-point fast Fourier transform matrix is ​​equal to the length N of the vibration signal after eliminating the interference of the first-order cyclostationary signal, to obtain an N′×N dimensional matrix;

[0051] Multiplying each row in the N′×N dimensional matrix with the conjugate of the vibration signal after eliminating the interference of the first-order cyclostationary signal to obtain a conjugate multiplied N′×N dimensional matrix;

[0052] Performing a fast Fourier transform on each row of the conjugate multiplied N′×N dimensional matrix to obtain an N′×N dimensional spectral correlation estimation matrix;

[0053] Taking the modulus and squaring each element in the N′×N dimensional spectral correlation estimation matrix, and summing the N′×N dimensional spectral correlation estimation matrix row by row to obtain an N dimensional row vector;

[0054] The frequencies corresponding to the first three largest elements in the N-dimensional row vector and their integer multiples are selected to form a set to obtain a cyclic frequency set.

[0055] According to a third aspect of the present invention, there is provided a device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the steps of the method for diagnosing liquid rocket engine turbine pump bearing faults based on dual adaptive filtering are implemented.

[0056] According to a fourth aspect of the present invention, there is provided a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps of a liquid rocket engine turbine pump bearing fault diagnosis method based on dual adaptive filtering.

[0057] Compared with the prior art, the present invention has at least the following beneficial effects:

[0058] The present invention provides a liquid rocket engine turbo pump bearing fault diagnosis method based on dual adaptive filtering. First, an adaptive line spectrum enhancer is used to effectively separate the first-order cyclostationary signal. The interference from the first-order cyclostationary signal in the remaining signal is eliminated, allowing the detection of cyclic frequency information; then, blind cyclic frequency detection is performed on the vibration signal after eliminating the interference of the first-order cyclostationary signal, that is, blind estimation of the cyclic frequency set from the vibration signal; after obtaining the cyclic frequency set, a multi-channel adaptive FRESH filter is used to further extract the second-order cyclostationary signal from the vibration signal. The present invention can separate two types of cyclostationary features without prior information, and analyze the second-order cyclostationary component containing the bearing fault feature separately, so as to obtain a more accurate diagnosis conclusion. The present invention shows stronger noise robustness, and the adaptive strategy adopted basically meets the needs of real-time processing; in addition, the present invention is suitable for real-time fault diagnosis scenarios, wherein the adaptive filtering strategy allows the system state to change, that is, it can adaptively track the statistical characteristics of the signal to be analyzed, and by further calculating the spectral correlation or spectral coherence of the extracted fault feature signal, it can effectively detect weaker bearing fault features, which is conducive to early fault warning.

[0059] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, preferred embodiments are given below and described in detail with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] In order to more clearly illustrate the technical solutions in the specific implementation modes of the present invention, the drawings required for use in the description of the specific implementation modes will be briefly introduced below. Obviously, the drawings described below are some implementation modes of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0061] Figure 1 The present invention is a flowchart of a method for diagnosing a liquid rocket engine turbopump bearing fault based on dual adaptive filtering according to an embodiment of the present invention.

[0062] Figure 2It is a part of the simulation signal used in the embodiment of the present invention; wherein, a is the second-order cyclostationary component in the simulation signal, b is the first-order cyclostationary component in the simulation signal, c is Gaussian white noise obeying the standard normal distribution, and d is a simulation signal combining a, b and c.

[0063] Figure 3 In FIG. 1 , a is a first-order cyclostationary signal extracted after the simulation signal passes through the ALE structure in an embodiment of the present invention; b is a vibration signal after isolating the first-order cyclostationary signal.

[0064] Figure 4 It is the spectral coherence result calculated after the vibration signal undergoes SSCA in the embodiment of the present invention; the cyclic frequency corresponding to the dotted line in the figure represents the result of blind cyclic frequency estimation, that is, as a cyclic frequency set.

[0065] Figure 5 In the figure, the solid line represents the second-order cyclostationary component of the simulation signal after double adaptive filtering, and the dotted line represents the second-order cyclostationary component in the original simulation signal.

[0066] Figure 6 It represents the spectral coherence estimation result of the signal after double adaptive filtering, which is used to diagnose the bearing fault exhibiting second-order cyclostationary characteristics. DETAILED DESCRIPTION

[0067] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0068] like Figure 1 As shown, as a specific embodiment of the present invention, a liquid rocket engine turbopump bearing fault diagnosis method based on dual adaptive filtering is intended to improve the accuracy and robustness of fault diagnosis, and specifically includes the following steps:

[0069] Step 1: Obtain the original vibration signal of the bearing to be diagnosed.

[0070] Specifically, a high-precision vibration sensor is used to collect the original vibration signal of the bearing when the liquid rocket engine turbopump is in operation.

[0071] Step 2: Process the original vibration signal through an adaptive line spectrum enhancer to obtain a vibration signal after eliminating the interference of the first-order cyclostationary signal.

[0072] That is to say, the collected original vibration signal is input into the adaptive line spectrum enhancer (Adaptive Line Enhancer, ALE). ALE is designed to identify and suppress the first-order cyclostationary components in the signal, that is, harmonic interference with obvious periodicity but not conforming to the second-order cyclostationary characteristics. Through ALE processing, the impact of these interference signals on subsequent analysis can be effectively reduced, and a purer vibration signal can be obtained.

[0073] Step 3, performing blind cyclic frequency detection on the vibration signal after eliminating the interference of the first-order cyclostationary signal to obtain a cyclic frequency set.

[0074] Specifically, blind cyclic frequency detection is performed on the vibration signal after ALE processing. Blind cyclic frequency detection can automatically detect and extract the cyclostationary characteristic frequencies in the signal without relying on any prior knowledge to form a cyclic frequency set. The cyclic frequency is a key parameter in the subsequent multi-channel adaptive FRESH filter design.

[0075] Step 4: Use the vibration signal after eliminating the interference of the first-order cyclostationary signal as the input of the multi-channel adaptive FRESH filter, and the frequency shift component used by each channel is the frequency in the cyclic frequency set, and output the filtered vibration signal.

[0076] Specifically, the cyclic frequency set detected by the blind cyclic frequency is used as the input parameter of the multi-channel adaptive FRESH filter. The multi-channel adaptive FRESH filter is a high-order spectrum estimation method for processing cyclostationary signals. By applying different frequency shift components (i.e., frequencies in the cyclic frequency set) on different channels, the fault features in the signal can be effectively extracted. The output of each channel is the filtering result of the original signal at a specific cyclic frequency, and these results together constitute a more refined and accurate representation of the vibration signal features.

[0077] Step 5: Analyze the filtered vibration signal using cyclostationary analysis to obtain a diagnosis result.

[0078] Specifically, the filtered vibration signal output by the multi-channel adaptive FRESH filter is further analyzed using cyclostationary analysis. The cyclostationary analysis method can make full use of the cyclostationary characteristics in the signal and identify the bearing fault type, fault degree, fault location and other information through the periodic changes of statistical characteristics. For example, a diagnostic report is obtained based on the analysis results to provide a basis for subsequent maintenance decisions.

[0079] Through the above-mentioned specific implementation method, the first-order cyclostationary component interference in the original vibration signal can be effectively filtered out, and the fault characteristic signal can be adaptively extracted, so that accurate fault diagnosis can be achieved even when the cyclic frequency set is unknown, thereby improving the accuracy and reliability of liquid rocket engine turbine pump bearing fault diagnosis.

[0080] In one possible implementation, the original vibration signal is processed by an adaptive line spectrum enhancer to obtain a vibration signal after eliminating the interference of the first-order cyclostationary signal, as follows:

[0081] a. The original noisy vibration signal is directly used as the reference signal of the adaptive line spectrum enhancer. The reference signal contains all the vibration information that needs to be retained and the first-order cyclostationary signal interference that needs to be suppressed.

[0082] b. Time-shift the original vibration signal backward by a number of samples, and use the time-shifted signal as the input signal of the adaptive line spectrum enhancer.

[0083] It should be noted that the time shift amount can be adjusted according to actual conditions and experimental results, usually to introduce a certain time delay between the input signal and the reference signal so that ALE can better distinguish and suppress periodic interference.

[0084] c. Input the input signal into the finite impulse response (FIR) filter. The FIR filter is the core part of the adaptive line spectrum enhancer, and its coefficients are adaptively adjusted by recursive least squares (RLS). The signal processed by the FIR filter is the first-order cyclostationary signal to be eliminated, which is recorded as the output signal.

[0085] More specifically, the RLS algorithm dynamically adjusts the coefficients of the FIR filter according to the actual working state of the liquid rocket engine turbo pump (such as speed, load, etc.) and the characteristics of the input signal to minimize the error between the reference signal and the output signal. In this process, the FIR filter will gradually learn the characteristics of the first-order cyclostationary signal interference.

[0086] d. Perform a difference operation between the reference signal (i.e., the original noisy vibration signal) and the output signal. The difference is the vibration signal after eliminating the interference of the first-order cyclostationary signal.

[0087] In one possible implementation, blind cyclic frequency detection is performed on the vibration signal after eliminating the interference of the first-order cyclostationary signal to obtain a cyclic frequency set. The specific process is as follows:

[0088] ① The vibration signal processed by the adaptive line spectrum enhancer (i.e., the vibration signal after eliminating the interference of the first-order cyclostationary signal) is used as the input of the blind cyclic frequency detector SSCA.

[0089] ② Perform sliding windowing on the input signal. Assume that the length of the input signal is N, the sliding window length is N′ (usually the minimum length that can cover the main features of the output signal is selected, such as N′=3N / 4), and the length of the overlapping area between sliding windows is N′-1.

[0090] ③ As the sliding window moves, each sliding windowed signal segment is converted into a column vector, and the column vectors are arranged backward in sequence as the sliding window moves until the sliding window is translated to the tail of the input signal, obtaining an N′×P-dimensional channelized input signal matrix.

[0091] ④ Perform fast Fourier transform (FFT) on each column of the N′×P dimensional channelized input signal matrix to obtain an N′-point fast Fourier transform matrix, where each column of the N′-point fast Fourier transform matrix represents the representation of a signal segment in the frequency domain.

[0092] ⑤ Repeatedly expand the N′-point fast Fourier transform matrix by column until the number of columns of the N′-point fast Fourier transform matrix is ​​equal to the length N of the vibration signal after eliminating the interference of the first-order cyclostationary signal, and obtain an N′×N dimensional matrix.

[0093] ⑥ Multiply each row of the expanded N′×N dimensional matrix by the conjugate of the vibration signal after eliminating the interference of the first-order cyclostationary signal, and obtain a conjugate multiplied N′×N dimensional matrix. This operation utilizes the cyclostationary characteristics of the signal and enhances the cyclic frequency component through conjugate multiplication.

[0094] ⑦ Perform fast Fourier transform on each row of the conjugate multiplication N′×N matrix to obtain an N′×N dimensional spectral correlation estimation matrix, each element of which contains information about the cyclic frequency of the input signal.

[0095] ⑧ Take the modulus and square each element in the N′×N dimensional spectral correlation estimation matrix, and sum the N′×N dimensional spectral correlation estimation matrix row by row to obtain an N dimensional row vector, which reflects the cyclostationarity strength of different frequency components.

[0096] ⑨ Select the frequencies corresponding to the first three largest elements in the N-dimensional row vector and their integer multiples to form a set to obtain a cyclic frequency set.

[0097] Through the above steps, the blind cyclic frequency detector SSCA can effectively extract the cyclic frequency set from the vibration signal after eliminating the interference of the first-order cyclostationary signal, providing a basis for subsequent multi-channel adaptive FRESH filtering and fault diagnosis.

[0098] In one possible implementation, the vibration signal after eliminating the interference of the first-order cyclostationary signal is used as the input of the multi-channel adaptive FRESH filter, and the frequency shift component used by each channel is the frequency in the cyclic frequency set, and the filtered vibration signal is output. The specific process is:

[0099] a. The vibration signal after eliminating the interference of the first-order cyclostationary signal is frequency-shifted according to the frequencies in the cyclic frequency set to obtain the frequency-shifted signals of each channel of the multi-channel adaptive FRESH filter.

[0100] That is to say, according to the cyclic frequency set previously obtained by blind cyclic frequency detection SSCA, the vibration signal after eliminating the interference of the first-order cyclostationary signal is frequency shifted.

[0101] b. Input the frequency-shifted signal of each channel into its own finite-length unit impulse response filter to obtain the output signal of each channel.

[0102] Specifically, the frequency shift signal of each channel is input into the corresponding finite-length unit impulse response (FIR) filter to obtain the output signal of each channel.

[0103] c. Sum the output signals of each channel to obtain the output signal of the multi-channel adaptive FRESH filter, which is the filtered vibration signal.

[0104] That is to say, the output signals of all channels after being processed by the FIR filter are summed up in order to merge the characteristic signals related to the bearing fault that may exist in each channel to form the final output signal of the multi-channel adaptive FRESH filter, which is the filtered vibration signal. The filtered vibration signal not only eliminates the first-order cyclostationary signal interference in the original vibration signal, but also further enhances the characteristic signals related to the bearing fault through frequency shift and filtering processing, providing more reliable and effective data support for subsequent fault diagnosis.

[0105] In one possible implementation, the filtered vibration signal is analyzed using cyclostationary analysis to obtain a diagnostic result, which is as follows:

[0106] First, the spectral correlation or spectral coherence function of the filtered vibration signal is calculated. It should be understood that the spectral correlation function or spectral coherence function is a key tool in cyclostationary analysis, which is used to describe the correlation of signals at different frequencies and cyclic frequencies. These functions can reveal the periodic components in the signal and their interactions.

[0107] Finally, fault diagnosis is achieved based on the calculated spectral correlation or spectral coherence function.

[0108] A more detailed explanation is given below in conjunction with specific embodiments.

[0109] A liquid rocket engine turbo pump bearing fault diagnosis method based on dual adaptive filtering comprises the following steps:

[0110] S1. Simulate and obtain the original vibration signal of the bearing to be diagnosed.

[0111] In this embodiment, the original vibration signal is recorded as x(n), where x(n) consists of three parts, as follows:

[0112]

[0113] In the formula, x cs1 (n) represents the first-order cyclostationary component; f m represents the modulation frequency; f c Indicates the carrier frequency; x cs2 (n) represents the second-order cyclostationary component; f i and represents the frequency and phase of the amplitude modulated random signal; n(n) represents white noise that obeys the standard Gaussian distribution; s(n) is narrowband white noise that obeys the standard Gaussian distribution.

[0114] like Figure 2 The following are some simulation signals used in this embodiment; where a is the second-order cyclostationary component in the simulation signal, that is, x cs2 (n); b is the first-order cyclostationary component in the simulation signal, that is, x cs1 (n); c is Gaussian white noise that obeys standard normal distribution, that is, n(n); d is the simulated signal that combines a, b and c, that is, x(n).

[0115] S2. Pass the original vibration signal through the ALE structure to obtain the vibration signal after eliminating the first-order cyclostationary signal.

[0116] The specific implementation process uses the following steps:

[0117] S21. Use the original noisy vibration signal as a reference signal of the ALE structure.

[0118] In this embodiment, the reference signal of the ALE structure is denoted as d(n), then d(n)=x(n).

[0119] S22. Time-shift the original vibration signal backward by Δ samples to obtain an input signal of the ALE structure.

[0120] In this embodiment, the input signal of the time-shifted ALE structure is x(n-Δ).

[0121] S23, passing the input signal obtained in step S22 through an FIR filter to obtain a first-order cyclostationary signal to be eliminated as an output signal.

[0122] In this embodiment, the first-order cyclostationary signal to be eliminated is y1(n), and the first-order cyclostationary signal to be eliminated output by the FIR filter is like Figure 3 As shown in a.

[0123] S24. Subtract the reference signal d(n) in step S21 from the output signal y1(n) in step S23 to obtain the vibration signal e(n) after eliminating the first-order cyclostationary signal and use it as the output signal of the ALE structure.

[0124] In this embodiment, the vibration signal after eliminating the first-order cyclostationary signal is e(n)=x(n)-y1(n), as Figure 3 As shown in b.

[0125] S25. Adaptively adjust the coefficients of the FIR filter in step S23 according to the system status through the RLS algorithm.

[0126] In this embodiment, the coefficients of the FIR filter are continuously adjusted in a recursive manner in the sense of least squares, so that the mean square error between the filter output signal and the reference signal is minimized.

[0127] S3, performing blind cyclic frequency detection on the vibration signal obtained after eliminating the interference of the first-order cyclostationary signal to obtain a cyclic frequency set, which can be specifically carried out by the following steps:

[0128] S31, using the vibration output signal after eliminating the first-order cyclostationary signal in step S24 as the input signal of the blind cyclic frequency detector SSCA.

[0129] In this embodiment, the input signal of SSCA is

[0130] S32, performing sliding windowing processing on the input signal in step S31.

[0131] In this embodiment, the length of the input signal is N, the length of the sliding window is N′, and the overlapping area between the sliding windows is N′-1.

[0132] S33, converting the windowed signal in step S32 into a column vector, and sequentially arranging the column vectors backwards as the sliding window moves until the sliding window is translated to the tail of the input signal.

[0133] In this embodiment, after step S33, an N′×P-dimensional channelized input signal matrix is ​​obtained.

[0134] S34, performing FFT on each column of the N′×P dimensional channelized input signal matrix obtained in step S33.

[0135] In this embodiment, the result obtained in step S34 is recorded as the complex demodulation X T (n,f), as follows:

[0136]

[0137] Where, T s Indicates the sampling period.

[0138] S35. Repeatedly expand the matrix obtained in step S34 by columns.

[0139] In this embodiment, after the matrix is ​​repeatedly expanded column by column in step S35, the number of columns is equal to the length N of the input signal, and an N′×N dimensional matrix is ​​obtained.

[0140] S36. Multiply each row of the N′×N dimensional matrix obtained in step S35 by the conjugate of the input signal in step S31.

[0141] In this embodiment, after the matrix and the input signal are multiplied row-wise by conjugate in step S36, an N′×N dimensional matrix is ​​obtained.

[0142] S37. Perform FFT on each row of the N′×N dimensional matrix obtained in step S36 to obtain an N′×N dimensional spectral correlation estimation matrix.

[0143] In this embodiment, after step S36, the spectral correlation function is obtained, which is recorded as The details are as follows:

[0144]

[0145] Where g(n) represents the rectangular window function, N′ corresponds to the spectral frequency, and N corresponds to the cyclic frequency.

[0146] Figure 4 The figure shows the spectral coherence result calculated after SSCA of the vibration signal after first-order cyclostationary signal interference in this embodiment.

[0147] S38. Square each element in the N′×N dimensional spectral correlation estimation matrix obtained in step S37, and sum the matrix row by row.

[0148] In this embodiment, after the calculation in step S38, the N-dimensional row vector is obtained.

[0149] S39. Select the frequencies corresponding to the first three largest elements in the N-dimensional row vector obtained in step S38 and their integer multiples to form a set, and obtain a cyclic frequency set.

[0150] In this embodiment, the cyclic frequency set is recorded as A = {α1, α2, ..., α M}.

[0151] S4. Select the vibration signal obtained in step S2 after eliminating the interference of the first-order cyclostationary signal as the input of the multi-channel adaptive FRESH filter, wherein the frequency shift component used by each channel is selected as the frequency in the cyclic frequency set obtained in step S3.

[0152] The following steps can be used in the specific implementation process:

[0153] S41. Frequency shift the vibration signal obtained in step S24 after eliminating the first-order cyclostationary signal according to the cyclic frequency set obtained in step S39 to obtain input signals of multiple channels of the FRESH filter.

[0154] In this embodiment, the input signal of each channel after frequency shift is x i (n), then

[0155] S42, inputting the frequency shift signal of each channel obtained in step S41 into the respective FIR filter to obtain the output signal of each channel.

[0156] In this embodiment, the output of the input signal of each channel after passing through the respective FIR filters in step S41 is recorded as but

[0157] S43, summing the output signals of the channels obtained in step S42 to obtain the output signal of the multi-channel adaptive FRESH filter, which is the fault characteristic signal after double adaptive filtering.

[0158] In this embodiment, the output signal of the multi-channel adaptive FRESH filter is recorded as y2(n), as Figure 5 As shown, Figure 5 In the figure, the solid line represents the second-order cyclostationary component of the simulation signal after double adaptive filtering, and the dotted line represents the second-order cyclostationary component in the original simulation signal.

[0159] S44. Adaptively adjust the coefficients of the FIR filter in step S42 according to the system status through the RLS algorithm.

[0160] In this embodiment, the coefficients of the FIR filter are continuously adjusted in a recursive manner in the sense of least squares, so that the mean square error between the filter output signal and the reference signal is minimized.

[0161] It should be noted that the reference signal in the multi-channel adaptive FRESH filter is still the vibration signal e(n) obtained in step S24 after eliminating the first-order cyclostationary signal.

[0162] S5. Analyze the vibration signal after passing through the dual adaptive filter using cyclostationary analysis to obtain a diagnosis result.

[0163] The following steps can be used in the specific implementation process:

[0164] S51, calculating the spectral correlation or spectral coherence function of the vibration signal after the dual adaptive filter.

[0165] In this embodiment, the spectral coherence result of the vibration signal after passing through the dual adaptive filter is like Figure 6 shown.

[0166] S52: Implement fault diagnosis according to the spectral correlation or spectral coherence result obtained in step S51.

[0167] In this embodiment, after the simulation signal is subjected to double adaptive filtering, the cyclostationary analysis is applied to successfully detect the fault characteristic frequency simulated by the simulation signal and its integer multiple harmonic frequencies, that is, Figure 6 The corresponding fault characteristic frequency components can be clearly observed in the image, thus realizing the bearing fault diagnosis.

[0168] In yet another embodiment of the present invention, a liquid rocket engine turbo pump bearing fault diagnosis device based on dual adaptive filtering is provided, comprising:

[0169] The acquisition module is used to acquire the original vibration signal of the bearing to be diagnosed.

[0170] The elimination module is used to process the original vibration signal through an adaptive line spectrum enhancer to obtain a vibration signal after eliminating the interference of the first-order cyclostationary signal.

[0171] The detection module is used to perform blind cyclic frequency detection on the vibration signal after eliminating the interference of the first-order cyclostationary signal to obtain a cyclic frequency set.

[0172] The output module is used to use the vibration signal after eliminating the interference of the first-order cyclostationary signal as the input of the multi-channel adaptive FRESH filter, and the frequency shift component used by each channel is the frequency in the cyclic frequency set, and output the filtered vibration signal.

[0173] The diagnosis module is used to analyze the filtered vibration signal by using cyclostationary analysis to obtain a diagnosis result.

[0174] All relevant contents of each step involved in the aforementioned embodiment of a liquid rocket engine turbine pump bearing fault diagnosis method based on dual adaptive filtering can be referred to the functional description of the functional module corresponding to the liquid rocket engine turbine pump bearing fault diagnosis device based on dual adaptive filtering in the embodiment of the present invention, and will not be repeated here. The division of modules in the embodiment of the present invention is schematic, which is only a logical function division. There may be other division methods in actual implementation. In addition, each functional module in each embodiment of the present invention can be integrated into a processor, or it can exist physically separately, or two or more modules can be integrated into one module. The above-mentioned integrated module can be implemented in the form of hardware or in the form of software functional modules.

[0175] In another embodiment of the present invention, a computer device is provided, the computer device comprising a processor and a memory, the memory being used to store a computer program, the computer program comprising program instructions, and the processor being used to execute the program instructions stored in the computer storage medium. The processor may be a central processing unit (CPU), or may be other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, etc., which are the computing core and control core of the terminal, and are suitable for implementing one or more instructions, and are specifically suitable for loading and executing one or more instructions in a computer storage medium to implement a corresponding method flow or corresponding function; the processor described in the embodiment of the present invention can be used for the operation of a liquid rocket engine turbine pump bearing fault diagnosis method based on dual adaptive filtering.

[0176] In another embodiment of the present invention, the present invention also provides a storage medium, specifically a computer-readable storage medium (Memory), which is a memory device in a computer device for storing programs and data. It is understandable that the computer-readable storage medium here can include both built-in storage media in the computer device and, of course, extended storage media supported by the computer device. The computer-readable storage medium provides a storage space, which stores the operating system of the terminal. In addition, one or more instructions suitable for being loaded and executed by the processor are also stored in the storage space, and these instructions can be one or more computer programs (including program codes). It should be noted that the computer-readable storage medium here can be a high-speed RAM memory or a non-volatile memory, such as at least one disk memory. The processor can load and execute one or more instructions stored in the computer-readable storage medium to implement the corresponding steps of the above-mentioned embodiment of a liquid rocket engine turbine pump bearing fault diagnosis method based on dual adaptive filtering.

[0177] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0178] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0179] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0180] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process in the computer or other programmable device. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0181] In the present invention, the terms "one embodiment", "some embodiments", "examples", "specific examples", or "some examples" etc. mean that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the described specific features, structures, materials or characteristics may be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art may combine and combine the different embodiments or examples described in this specification and the features of the different embodiments or examples, without contradiction.

[0182] Finally, it should be noted that the above-described embodiments are only specific implementations of the present invention, which are used to illustrate the technical solutions of the present invention, rather than to limit them. The protection scope of the present invention is not limited thereto. Although the present invention is described in detail with reference to the above-described embodiments, ordinary technicians in the field should understand that any technician familiar with the technical field can still modify the technical solutions recorded in the above-described embodiments within the technical scope disclosed by the present invention, or can easily think of changes, or make equivalent replacements for some of the technical features therein; and these modifications, changes or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention shall be based on the protection scope of the claims.

Claims

1. A liquid rocket engine turbopump bearing fault diagnosis method based on dual adaptive filtering, characterized in that: include: Obtaining the original vibration signal of the bearing to be diagnosed; Processing the original vibration signal through an adaptive line spectrum enhancer to obtain a vibration signal after eliminating first-order cyclostationary signal interference; Performing blind cyclic frequency detection on the vibration signal after eliminating the interference of the first-order cyclostationary signal to obtain a cyclic frequency set; The vibration signal after eliminating the interference of the first-order cyclostationary signal is used as the input of the multi-channel adaptive FRESH filter, and the frequency shift component used by each channel is the frequency in the cyclic frequency set, and the filtered vibration signal is output; The filtered vibration signal is analyzed using cyclostationary analysis to obtain a diagnosis result.

2. A liquid rocket engine turbo pump bearing fault diagnosis method based on dual adaptive filtering according to claim 1, characterized in that: The original vibration signal is processed by an adaptive line spectrum enhancer to obtain a vibration signal after eliminating the interference of the first-order cyclostationary signal, specifically: Using the original noisy vibration signal as a reference signal for the adaptive line spectrum enhancer; Time-shift the original vibration signal backward by a number of samples to serve as an input signal of the adaptive line spectrum enhancer; Inputting the input signal into a finite length unit impulse response filter, adaptively adjusting the coefficient of the finite length unit impulse response filter according to the state of the liquid rocket engine turbo pump by a recursive least square method, and obtaining a first-order cyclostationary signal to be eliminated, which is recorded as an output signal; The reference signal is subtracted from the output signal to obtain a vibration signal after first-order cyclostationary signal interference is eliminated.

3. The liquid rocket engine turbo pump bearing fault diagnosis method based on dual adaptive filtering according to claim 1 is characterized in that: The blind cyclic frequency detection is performed on the vibration signal after eliminating the interference of the first-order cyclostationary signal to obtain a cyclic frequency set, which is specifically: Using the vibration signal after eliminating the interference of the first-order cyclostationary signal as the input of the blind cyclic frequency detector SSCA; Performing sliding windowing processing on the vibration signal after eliminating the interference of the first-order cyclostationary signal, wherein the length of the vibration signal after eliminating the interference of the first-order cyclostationary signal is N, the sliding window length is N′, the overlapping area between the sliding windows is N′-1, N is a natural number not equal to 0, and N′ is less than N; The signal of each sliding window is converted into a column vector, and the column vectors are arranged backward in sequence as the sliding window moves until the sliding window is translated to the tail of the vibration signal after eliminating the interference of the first-order cyclostationary signal, so as to obtain an N′×P-dimensional channelized input signal matrix; Performing a fast Fourier transform on each column of the N′×P-dimensional channelized input signal matrix to obtain an N′-point fast Fourier transform matrix; Repeatedly expanding the N′-point fast Fourier transform matrix by columns so that the number of columns of the N′-point fast Fourier transform matrix is ​​equal to the length N of the vibration signal after eliminating the interference of the first-order cyclostationary signal, to obtain an N′×N dimensional matrix; Multiplying each row in the N′×N dimensional matrix with the conjugate of the vibration signal after eliminating the interference of the first-order cyclostationary signal to obtain a conjugate multiplied N′×N dimensional matrix; Performing a fast Fourier transform on each row of the conjugate multiplied N′×N dimensional matrix to obtain an N′×N dimensional spectral correlation estimation matrix; Taking the modulus and squaring each element in the N′×N dimensional spectral correlation estimation matrix, and summing the N′×N dimensional spectral correlation estimation matrix row by row to obtain an N dimensional row vector; The frequencies corresponding to the first three largest elements in the N-dimensional row vector and their integer multiples are selected to form a set to obtain a cyclic frequency set.

4. The liquid rocket engine turbo pump bearing fault diagnosis method based on dual adaptive filtering according to claim 1 is characterized in that: The vibration signal after eliminating the interference of the first-order cyclostationary signal is used as the input of the multi-channel adaptive FRESH filter, and the frequency shift component used by each channel is the frequency in the cyclic frequency set, and the filtered vibration signal is output, specifically: Frequency shifting the vibration signal after eliminating the interference of the first-order cyclostationary signal according to the frequencies in the cyclic frequency set to obtain frequency shifted signals of each channel of the multi-channel adaptive FRESH filter; Input the frequency shift signal of each channel into its own finite length unit impulse response filter to obtain the output signal of each channel; The output signals of each channel are summed to obtain the output signal of the multi-channel adaptive FRESH filter, which is the filtered vibration signal.

5. The liquid rocket engine turbo pump bearing fault diagnosis method based on dual adaptive filtering according to claim 1, characterized in that: The filtered vibration signal is analyzed by cyclostationary analysis to obtain a diagnosis result, which is specifically: Calculating a spectral correlation or spectral coherence function of the filtered vibration signal; Fault diagnosis is achieved based on the spectral correlation or spectral coherence function.

6. A liquid rocket engine turbo pump bearing fault diagnosis device based on dual adaptive filtering, characterized in that: include: An acquisition module, used for acquiring an original vibration signal of a bearing to be diagnosed; An elimination module, used for processing the original vibration signal through an adaptive line spectrum enhancer to obtain a vibration signal after eliminating interference from a first-order cyclostationary signal; A detection module, used for performing blind cyclic frequency detection on the vibration signal after eliminating the interference of the first-order cyclostationary signal, to obtain a cyclic frequency set; An output module is used to use the vibration signal after eliminating the interference of the first-order cyclostationary signal as the input of a multi-channel adaptive FRESH filter, and the frequency shift component used by each channel is the frequency in the cyclic frequency set, and output the filtered vibration signal; The diagnosis module is used to analyze the filtered vibration signal by using cyclostationary analysis to obtain a diagnosis result.

7. A liquid rocket engine turbo pump bearing fault diagnosis device based on dual adaptive filtering according to claim 6, characterized in that: The elimination module is specifically used for: Using the original noisy vibration signal as a reference signal for the adaptive line spectrum enhancer; Time-shift the original vibration signal backward by a number of samples to serve as an input signal of the adaptive line spectrum enhancer; Inputting the input signal into a finite length unit impulse response filter, adaptively adjusting the coefficient of the finite length unit impulse response filter according to the state of the liquid rocket engine turbo pump by a recursive least square method, and obtaining a first-order cyclostationary signal to be eliminated, which is recorded as an output signal; The reference signal is subtracted from the output signal to obtain a vibration signal after first-order cyclostationary signal interference is eliminated.

8. The liquid rocket engine turbo pump bearing fault diagnosis device based on dual adaptive filtering according to claim 6, characterized in that: The detection module is specifically used for: Using the vibration signal after eliminating the interference of the first-order cyclostationary signal as the input of the blind cyclic frequency detector SSCA; Performing sliding windowing processing on the vibration signal after eliminating the interference of the first-order cyclostationary signal, wherein the length of the vibration signal after eliminating the interference of the first-order cyclostationary signal is N, the sliding window length is N′, the overlapping area between the sliding windows is N′-1, N is a natural number not equal to 0, and N′ is less than N; The signal of each sliding window is converted into a column vector, and the column vectors are arranged backward in sequence as the sliding window moves until the sliding window is translated to the tail of the vibration signal after eliminating the interference of the first-order cyclostationary signal, so as to obtain an N′×P-dimensional channelized input signal matrix; Performing a fast Fourier transform on each column of the N′×P-dimensional channelized input signal matrix to obtain an N′-point fast Fourier transform matrix; Repeatedly expanding the N′-point fast Fourier transform matrix by columns so that the number of columns of the N′-point fast Fourier transform matrix is ​​equal to the length N of the vibration signal after eliminating the interference of the first-order cyclostationary signal, to obtain an N′×N dimensional matrix; Multiplying each row in the N′×N dimensional matrix with the conjugate of the vibration signal after eliminating the interference of the first-order cyclostationary signal to obtain a conjugate multiplied N′×N dimensional matrix; Performing a fast Fourier transform on each row of the conjugate multiplied N′×N dimensional matrix to obtain an N′×N dimensional spectral correlation estimation matrix; Taking the modulus and squaring each element in the N′×N dimensional spectral correlation estimation matrix, and summing the N′×N dimensional spectral correlation estimation matrix row by row to obtain an N dimensional row vector; The frequencies corresponding to the first three largest elements in the N-dimensional row vector and their integer multiples are selected to form a set to obtain a cyclic frequency set.

9. A device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of a liquid rocket engine turbine pump bearing fault diagnosis method based on dual adaptive filtering as described in any one of claims 1 to 5 are implemented.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of a liquid rocket engine turbine pump bearing fault diagnosis method based on dual adaptive filtering as described in any one of claims 1 to 5 are implemented.

Images