Aerial engine compressor rotation stall voiceprint monitoring method, system, medium and equipment

Abstract

The invention discloses a voiceprint monitoring method, system, medium and equipment for rotation stall of an aero-engine compressor, and the method comprises the steps: determining the layout of a microphone array through the model parameters of the compressor of the aero-engine, and carrying out the synchronous collection of sound pressure time domain signals of multiple channels of the compressor of the aero-engine through the microphone array; arranging the multi-channel sound pressure time-domain signals in sequence to construct a time-domain signal matrix; performing spectral analysis on the sound pressure time-domain signals by adopting a fast Fourier transform algorithm to obtain a spectrogram of the sound pressure time-domain signals of each channel, monitoring whether an abnormal single-tone frequency peak value lower than one time of rotation frequency appears in the spectrogram or not, obtaining an acoustic mode spectrogram of the abnormal single-tone frequency peak value by adopting a single-frequency acoustic mode decomposition method, and obtaining an acoustic mode spectrum of the abnormal single-tone frequency peak value; and whether the rotating stall characteristic of the gas compressor occurs in the acoustic mode spectrogram is monitored to judge whether the rotating stall characteristic of the gas compressor occurs or not.

Description

Technical Field

[0001] The present invention relates to the technical field of aeroengine noise testing, and particularly relates to a method, a system, a medium and a device for monitoring the rotating stall sound pattern of an aeroengine compressor. Background Art

[0002] In order to improve the working performance, the single-stage pressure ratio of a compressor is often increased in a large aeroengine, while the number of compressor stages is reduced. The increase in the single-stage pressure ratio and the reduction in the number of stages necessarily require an increase in the blade loading of the compressor, which causes flow separation and generates flow instability problems inside the compressor such as rotating stall and surge. The occurrence of gas path faults such as rotating stall and surge seriously affects the stability and safety of the operation of the aeroengine, and often brings devastating disasters to the aeroengine.

[0003] Rotating stall can generally be divided into two categories: progressive stall and abrupt stall. When progressive stall occurs in a compressor, the pressure ratio gradually decreases but does not drop suddenly; when abrupt stall occurs in a compressor, the pressure in the sound field drops suddenly, reflecting the discontinuous characteristics on the pressure ratio curve. When the tip flow of the compressor blade peels off and a stall air mass is formed at the tip, rotating stall will occur and gradually develop circumferentially and radially. The outlet pressure of the compressor decreases and a periodic oscillation fluctuation with a relatively high frequency and a relatively low amplitude is formed. When the fluctuation develops to the entire compressor, a surge fault will be triggered.

[0004] Rotating stall not only deteriorates the performance of the aeroengine, limits the working range of the aeroengine, but also once rotating stall occurs, the engine is very likely to stall immediately, damaging the key components of the engine. Therefore, accurately identifying and analyzing its fault characteristics is crucial for the maintenance and repair of the aeroengine. Traditional fault detection methods mainly rely on means such as visual inspection, non-destructive testing techniques, and vibration analysis. However, these methods often have limitations when facing the rotating stall fault of the compressor.

[0005] The above information disclosed in the background art section is only used to enhance the understanding of the background of the present invention, and therefore may include information that does not constitute the prior art known to those of ordinary skill in the art. Summary of the Invention

[0006] The present invention provides a method, a system, a medium and a device for monitoring the rotating stall sound pattern of an aeroengine compressor. An annular acoustic array is used to measure acoustic signals. Through a spectrum analysis module and an acoustic mode decomposition module, the transformation of the acoustic signal from the time domain to the wavenumber domain is realized, and a more effective rotating stall detection index, the sound pattern diagram, is obtained, realizing the conversion of the sound field signal in the time domain - frequency domain - wavenumber domain, and fully extracting the fault characteristics of the acoustic signal when the rotating stall fault occurs for identification.

[0007] A method for monitoring the rotating stall sound pattern of an aeroengine compressor includes:

[0008] In the first step, the layout of the microphone array is determined based on the compressor model parameters of the aero-engine. According to the number of rotor blades of the compressor and the number of stator blades the rotational-stator interference modal order of the compressor single-tone noise is calculated . Based on the maximum modal order of the rotational-stator interference modal order the acoustic modal monitoring range, the number of microphones and the circumferential installation angle of the microphones are determined ;

[0009] In the second step, the microphone array is constructed by the number of microphones and the installation angle of the microphones. The multi-channel acoustic pressure time-domain signals of the aero-engine compressor are synchronously collected by the microphone array; the time-domain signal matrix is constructed by arranging the multi-channel acoustic pressure time-domain signals in sequence; In the third step, the fast Fourier transform algorithm is used to perform spectral analysis on the acoustic pressure time-domain signals to obtain the spectrograms of the acoustic pressure time-domain signals of each channel. By monitoring the spectrograms to check if there are abnormal single-tone frequency peaks lower than 1 times the rotational frequency, if not, it indicates that the compressor is in normal operating condition, and if so, it enters the fourth step;

[0010] In the fourth step, the acoustic modal spectrogram of the abnormal single-tone frequency peak is obtained by the single-frequency acoustic modal decomposition method. By monitoring whether the compressor rotating stall characteristics appear in the acoustic modal spectrogram to judge whether the compressor has rotating stall. When the compressor has rotating stall characteristics under supersonic working conditions, in the wavenumber domain, there will be a rotating stall characteristic frequency with a frequency component lower than 1 times the rotational frequency and an acoustic modal order of 1.

[0011] In the method for monitoring the rotating stall acoustic signature of an aero-engine compressor described above, it further includes

[0012] In the fifth step, the wave spectrogram of the acoustic pressure time-domain signals is obtained by continuous broadband acoustic modal decomposition. The abscissa of the wave spectrogram represents the frequency analysis range, the ordinate represents the acoustic modal monitoring range, and the depth of the color represents the amplitude size, the darker the color, the larger the amplitude. By monitoring whether the compressor rotating stall characteristics appear in the wave spectrogram to further illustrate whether the compressor has rotating stall. When the compressor has rotating stall under supersonic working conditions, there will be a rotating stall abnormal sound source and its harmonics in the wave spectrogram that are all non-integer multiples of the rotational frequency, and at the same time, there will be an acoustic modal oblique bright band component starting from the modal order of 1, corresponding to the rotating stall characteristic frequency and its harmonics.

[0013]

[0014] In the method for monitoring the rotating stall acoustic signature of an aero-engine compressor described above, in the first step, the rotational-stator interference modal order of the compressor​ is , where represents the order of pressure pulsation caused by the unsteady aerodynamic force resulting from the compressor stator-rotor interaction, represents an integer that determines the number of microphones in the microphone array and the microphone installation angles. Among them,

[0015] When adopting the uniform acoustic array layout scheme, calculate the number of sensors required for modal detection based on the Nyquist sampling law , which has the following relationship with the stator-rotor interaction modal order : ; When adopting the non-uniform acoustic array layout scheme with fewer measurement points, randomly select positions from the positions of the virtual uniform layout to install sensors, ;

[0016] When adopting the uniform acoustic array layout scheme, sensors form a circular acoustic array, and the spacing between sensors is , and the installation angle of the microphone is , where , , , and so on; When adopting the non-uniform acoustic array layout scheme with fewer measurement points, the microphone installation angle is randomly selected, .

[0017] In the described method for monitoring the rotating stall sound pattern of an aeroengine compressor, the second step includes the following steps:

[0018] S201. Use the circular acoustic array to measure the sound pressure signal of the aeroengine compressor. The measured sound pressure time-domain signal is , where the length of the time-domain signal sequence measured by a single microphone is , and the subscripts are the microphones corresponding to the installation angle positions respectively;

[0019] S202. According to the sound pressure time-domain signals measured by the microphones at different installation angle positions, form a time-domain signal matrix , where the element represents the signal measured by the microphone at the corresponding installation angle position at the th time,

[0020] , when adopting the uniform acoustic array layout, the size of the time-domain signal matrix is ; When using a non-uniform acoustic array layout with few measurement points, the time domain signal matrix The size is .

[0021] In the method for monitoring rotating stall soundprint of an aero-engine compressor, the third step includes:

[0022] S301, measuring the time domain signal matrix of the annular acoustic array Perform Fourier transform on each column to obtain the frequency domain matrix , where the elements express The signal measured by the microphone at the corresponding installation angle position is The amplitude at the frequency, The length is , according to Nyquist sampling theorem, ,

[0023] , in a uniform acoustic array layout, the frequency domain matrix The size is ; Frequency domain matrix in non-uniform acoustic array layout The size is ,

[0024] S302, according to the frequency domain matrix Draw the spectrum of the signal measured by the sensor at different installation angles to observe whether there is any difference in the frequency of the blade passing through. The abnormal single-tone frequency peak value outside the rotation frequency is the rotating stall characteristic frequency and its double and triple frequencies, which have a frequency lower than the rotation frequency and an amplitude more than twice the amplitude of the current blade passing frequency.

[0025] In the method for monitoring rotating stall soundprint of an aero-engine compressor, the fourth step includes:

[0026] S401, when using uniform acoustic array layout, subscript The microphone at the corresponding installation angle position is at the predetermined frequency The frequency domain signal at It can be regarded as the linear superposition of different circumferential acoustic modes, that is, , construct the transformation matrix The form is as follows:

[0027] ,

[0028] When using a non-uniform acoustic array layout with few measurement points, the observation matrix is constructed based on the randomly selected sensor installation angles. , the observation matrix Size ,

[0029] ,

[0030] S402. When adopting a uniform acoustic array layout, perform a spatial Fourier transform on the frequency domain matrix to obtain the wavenumber domain matrix , , where is the transpose matrix of the frequency domain matrix , , where represents the pseudo-inverse of the transformation matrix ;

[0031] When adopting a non-uniform acoustic array layout with fewer measurement points, the compressive sensing model is , and the sparse dictionary is composed of orthogonal Fourier transform bases, is the sensing matrix, and sparsely reconstruct the wavenumber domain matrix based on the compressive sensing model;

[0032] S403. According to the obtained wavenumber domain matrix , plot the acoustic mode spectrogram; under supersonic conditions, observe whether the rotating stall characteristic frequency and its harmonics appear. The rotating stall characteristic frequency is lower than 1 times the rotational frequency and the mode order is 1.

[0033] In the method for monitoring the rotating stall soundprint of an aero-engine compressor described above, the fifth step includes,

[0034] S501. Expand the wavenumber domain matrix to the full frequency domain. The size of the wavenumber domain matrix is expanded to , where the element represents the amplitude of the th order acoustic mode at the frequency ,

[0035] ,

[0036] S502. According to the wavenumber domain matrix , plot the wave spectrum. Under supersonic conditions, observe whether there are abnormal rotating stall sound sources and their harmonics that are not integer multiples of the rotational frequency, and the acoustic mode oblique bright band components starting from the mode order of 1 corresponding to the rotating stall characteristic frequency and its harmonics. The mathematical expression of the acoustic mode oblique bright band component is

[0037]

[0038] In the formula, is the rotating stall characteristic frequency, represents the harmonic of the characteristic frequency, is a non - negative integer, represents the circumferential acoustic mode order. The characteristic frequencies of rotating stall appearing in the spectrum, the corresponding mode orders at the characteristic frequencies in the acoustic mode spectrogram, the abnormal sound sources in the wavenumber spectrogram, and the inclined bright band components of the acoustic mode are defined as the acoustic fingerprint characteristics for monitoring the rotating stall of the compressor. Based on this, it is judged whether the compressor has rotating stall.

[0039] A monitoring system for implementing the described method includes:

[0040] An acoustic field measurement module, which includes an acoustic array measurement sub - module and a data acquisition sub - module, and is used to measure the acoustic field information propagated to the acoustic array installation position in the pipeline when the compressor is operating.

[0041] A spectrum analysis module, which is used to transform the time - domain acoustic field signal at the acoustic array position to the frequency domain and detect whether there are abnormal frequencies other than the blade passing frequency and the rotating frequency.

[0042] An acoustic mode decomposition module, which is used to perform single - tone acoustic mode decomposition and continuous broadband acoustic mode decomposition, transform the acoustic field information from the frequency domain to the wavenumber domain, and detect whether the characteristics of compressor rotating stall appear in the acoustic mode spectrogram and the wavenumber spectrogram.

[0043] A computer storage medium, the storage medium includes computer instructions, when it runs on a computer, it enables the computer to execute the described method.

[0044] An electronic device, the electronic device includes:

[0045] A memory, a processor, and a computer program stored on the memory and executable on the processor, wherein,

[0046] When the processor executes the program, it implements the described method.

[0047] Compared with the prior art, the present invention has the following advantages: The present invention uses acoustic information to monitor the rotating stall fault of the compressor, which has higher sensitivity and stronger adaptability compared with the traditional method using vibration signals and inter - stage pressure signals; The present invention obtains the rotating stall characteristics of the compressor in the frequency domain of the acoustic signal through fast Fourier transform, with obvious characteristics and a simple process; Obtain the rotating stall characteristics of the compressor in the wavenumber domain of the acoustic signal through spatial Fourier transform, with high sensitivity and adaptability to different rotational speeds. In the case of limited test resources, a non - uniform layout scheme can also be used to extract the acoustic fingerprint characteristics of the rotating stall of the aero - engine compressor. Defining the abnormal components in the spectrum, acoustic mode spectrogram, and wavenumber spectrogram of the acoustic signal as the acoustic fingerprint characteristics of the compressor rotating stall improves the monitoring accuracy. Description of the Drawings

[0048] By reading the detailed description in the following preferred specific embodiments, various other advantages and benefits of the present invention will become clear to those of ordinary skill in the art. The accompanying drawings of the specification are only for the purpose of showing the preferred embodiments and are not considered as limiting the present invention. Obviously, the following described drawings are only some embodiments of the present invention, and those of ordinary skill in the art can obtain other drawings based on these drawings without creative efforts. Moreover, throughout the drawings, the same reference numerals are used to represent the same components.

[0049] In the drawings:

[0050] Figure 1 is a flowchart of the present disclosure;

[0051] Figure 2 is a schematic diagram of a detection device for a compressor rotating stall monitoring method based on acoustic signature provided by an embodiment of the present disclosure;

[0052] Figures 3(a) to 3(c) is a spectrogram of no rotating stall and spectrograms of rotating stall at different speeds under the supersonic condition of the compressor provided by an embodiment of the present disclosure; wherein, Fig. 3(a) is the spectrogram of no rotating stall, Fig. 3(b) is the spectrogram of rotating stall at a higher speed of the compressor, and Fig. 3(c) is the spectrogram of rotating stall at a lower speed of the compressor;

[0053] Figures 4(a) to 4(b) is an acoustic modal spectrogram at the compressor rotating stall characteristic frequency and at the blade passing frequency provided by an embodiment of the present disclosure; wherein, Fig. 4(a) is the acoustic modal spectrogram at the compressor rotating stall characteristic frequency, and Fig. 4(b) is the acoustic modal spectrogram at the blade passing frequency;

[0054] Figures 5(a) to 5(c) is a spectrogram of no fault and spectrograms of rotating stall at different speeds under the supersonic condition of the compressor provided by an embodiment of the present disclosure; wherein, Fig. 5(a) is the spectrogram of no fault, Fig. 5(b) is the spectrogram of rotating stall at a higher speed of the compressor, and Fig. 5(c) is the spectrogram of rotating stall at a lower speed of the compressor.

[0055] The present invention will be further explained below with reference to the accompanying drawings and embodiments. Detailed Embodiments

[0056] The specific embodiments of the present invention will be described in more detail below with reference to the accompanying drawings. Although the specific embodiments of the present invention are shown in the drawings, it should be understood that the present invention can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided so that the present invention can be more thoroughly understood and the scope of the present invention can be completely conveyed to those skilled in the art.

[0057] It should be noted that certain terms are used in the specification and claims to refer to specific components. Those skilled in the art should understand that technicians may use different terms to refer to the same component. The specification and claims do not distinguish components by the difference in terms, but by the difference in the functions of the components. For example, the terms "comprising" or "including" mentioned throughout the specification and claims are open-ended terms, so they should be interpreted as "including but not limited to". The subsequent description in the specification is the preferred embodiment for implementing the present invention, but the description is for the purpose of the general principles of the specification and is not used to limit the scope of the present invention. The scope of protection of the present invention shall be subject to what is defined by the appended claims.

[0058] For ease of understanding the embodiments of the present invention, the following will further explain with specific embodiments in conjunction with the drawings, and the various drawings do not constitute a limitation to the embodiments of the present invention.

[0059] As Figures 1 to 5(c) shown, the method for monitoring the rotating stall acoustic signature of an aero-engine compressor includes the following steps:

[0060] In the first step S1, the layout of the microphone array is determined based on the compressor model parameters of the aero-engine. According to the number of rotor blades and the number of stator blades of the compressor, calculate the rotor-stator interference modal order of the compressor's tonal noise. Based on the maximum modal order of the rotor-stator interference modal order , determine the acoustic modal monitoring range, the number of microphones and the circumferential installation angle of the microphones. ;

[0061] In the second step S2, construct a microphone array based on the number of microphones and the installation angle of the microphones. Synchronously collect the multi-channel acoustic pressure time-domain signals of the aero-engine compressor through the microphone array. Arrange the multi-channel acoustic pressure time-domain signals in sequence to construct a time-domain signal matrix. ;

[0062] In the third step S3, use the fast Fourier transform algorithm to perform spectral analysis on the acoustic pressure time-domain signal to obtain the spectrogram of the acoustic pressure time-domain signal of each channel. Monitor whether there is an abnormal tonal frequency peak below 1 times the rotational frequency in the spectrogram. If not, it indicates that the compressor is in a normal working state. If so, enter the fourth step S4.

[0063] In the fourth step S4, an acoustic mode spectrogram of abnormal single-tone frequency peaks is obtained through the single-frequency acoustic mode decomposition method. By monitoring whether the characteristics of compressor rotating stall appear in the acoustic mode spectrogram, it is determined whether the compressor has rotating stall. When the characteristics of compressor rotating stall appear under supersonic conditions, in the wavenumber domain, the frequency component of this frequency is lower than the 1x rotating frequency and the acoustic mode order is 1 for the rotating stall characteristic frequency.

[0064] In a preferred embodiment of the described method for monitoring the acoustic signature of aero-engine compressor rotating stall, it further includes,

[0065] In the fifth step S5, a spectrogram of the acoustic pressure time-domain signal is obtained through continuous broadband acoustic mode decomposition. The abscissa of the spectrogram represents the frequency analysis range, the ordinate represents the acoustic mode monitoring range, and the depth of the color represents the amplitude size. The darker the color, the larger the amplitude. By monitoring whether the characteristics of compressor rotating stall appear in the spectrogram, it is further explained whether the compressor has rotating stall. When the compressor has rotating stall under supersonic conditions, the abnormal sound source of rotating stall and its harmonics appear in the spectrogram and have a non-integer multiple relationship with the rotating frequency. At the same time, the acoustic mode oblique bright band components starting from the acoustic mode order of 1 appear, corresponding to the rotating stall characteristic frequency and its harmonics.

[0066] In a preferred embodiment of the described method for monitoring the acoustic signature of aero-engine compressor rotating stall, in the first step S1, the rotor-stator interference mode order of the compressor is, , where, represents the pressure pulsation order caused by the unsteady aerodynamic force caused by the rotor-stator interference of the compressor, represents an integer, determining the number of microphones in the microphone array and the microphone installation angle,

[0067] When adopting a uniform acoustic array layout scheme, calculate the number of sensors required for modal detection based on the Nyquist sampling law , and its relationship with the rotor-stator interference mode order is: ; When adopting a non-uniform acoustic array layout scheme with fewer measurement points, randomly select positions from the positions in the virtual uniform layout to install sensors, ;

[0068] When adopting a uniform acoustic array layout scheme, sensors form a circular acoustic array, and the spacing between sensors is , the installation angle of the microphone is , where , , , and so on; when adopting the non-uniform acoustic array layout scheme with fewer measurement points, the installation angle of the microphone is randomly selected, .

[0069] In the preferred implementation manner of the aero-engine compressor rotating stall acoustic fingerprint monitoring method described above, the second step S2 includes the following steps:

[0070] S201. Use the annular acoustic array to measure the acoustic pressure signal of the aero-engine compressor, and the measured acoustic pressure time-domain signal is , where the length of the time-domain signal sequence measured by a single microphone is , and the subscript are the microphones at the corresponding installation angle positions respectively;

[0071] S202. According to the acoustic pressure time-domain signals measured by the microphones at different installation angle positions, form a time-domain signal matrix , where the element represents the signal measured by the microphone at the corresponding installation angle position at the th,

[0072] . When adopting the uniform acoustic array layout, the size of the time-domain signal matrix is ; when adopting the non-uniform acoustic array layout scheme with fewer measurement points, the size of the time-domain signal matrix is .

[0073] In the preferred implementation manner of the aero-engine compressor rotating stall acoustic fingerprint monitoring method described above, the third step S3 includes

[0074] S301. Perform Fourier transform on each column of the time-domain signal matrix measured by the annular acoustic array to obtain a frequency-domain matrix , where the element represents the amplitude of the signal measured by the microphone at the corresponding installation angle position at the th frequency, the length of is , according to the Nyquist sampling theorem,

[0075] . In the uniform acoustic array layout, the size of the frequency-domain matrix is ; in the non-uniform acoustic array layout, the size of the frequency-domain matrix is ,

[0076] S302. According to the frequency domain matrix Draw the spectrograms of the signals measured by the position sensors at different installation angles, and observe whether there are abnormal single-tone frequency peaks other than the blade passing frequency and the rotational frequency. The abnormal single-tone frequency peak is the rotating stall characteristic frequency whose frequency is lower than the rotational frequency and whose amplitude is more than twice the amplitude of the current blade passing frequency, as well as its second and third harmonics.

[0077] In the preferred implementation of the method for monitoring the rotating stall soundprint of an aero-engine compressor, the fourth step S4 includes

[0078] S401. When using a uniform acoustic array layout, the subscript The frequency domain signal of the microphone corresponding to the installation angle at a predetermined frequency is regarded as a linear superposition of different circumferential acoustic modes, that is , and a transformation matrix is constructed in the form of as follows:

[0079] ,

[0080] When using a non-uniform acoustic array layout with fewer measurement points, an observation matrix is constructed according to the randomly selected sensor installation angles , and the observation matrix is of size ,

[0081] ,

[0082] S402. When using a uniform acoustic array layout, perform a spatial Fourier transform on the frequency domain matrix to obtain a wavenumber domain matrix , , where is the transpose matrix of the frequency domain matrix , , where represents the pseudo-inverse of the transformation matrix ;

[0083] When using a non-uniform acoustic array layout with fewer measurement points, the compressive sensing model is , the sparse dictionary is composed of orthogonal Fourier transform bases, is the sensing matrix, and the wavenumber domain matrix is sparsely reconstructed based on the compressive sensing model;

[0084] S403. According to the obtained wavenumber domain matrix , draw the acoustic mode spectrogram; under supersonic conditions, observe whether the characteristic frequency of rotating stall and its multiple frequencies appear. The characteristic frequency of rotating stall is lower than the first harmonic of the rotational frequency and the mode order is 1.

[0085] In a preferred embodiment of the method for monitoring the rotating stall acoustic signature of an aero-engine compressor, the fifth step S5 includes

[0086] S501. Expand the wavenumber domain matrix to the full frequency domain. The size of the wavenumber domain matrix is expanded to , where the element represents the amplitude of the -th order acoustic mode at frequency .

[0087] ,

[0088] S502. According to the wavenumber domain matrix , draw the wave spectrogram. Under supersonic conditions, observe whether there are abnormal sound sources of rotating stall and their multiple frequencies that are not in an integer multiple relationship with the rotational frequency, as well as the acoustic mode oblique bright band components starting from the mode order of 1 corresponding to the characteristic frequency of rotating stall and its multiple frequencies. The mathematical expression of the acoustic mode oblique bright band components is

[0089]

[0090] In the formula, is the characteristic frequency of rotating stall, represents the multiple frequency of the characteristic frequency, is a non-negative integer, represents the circumferential acoustic mode order. Define the characteristic frequency of rotating stall in the spectrum, the corresponding mode order at the characteristic frequency in the acoustic mode spectrogram, the abnormal sound source and the acoustic mode oblique bright band components in the wave spectrogram as the acoustic signature for monitoring the rotating stall of the compressor, and judge whether the compressor has a rotating stall accordingly.

[0091] A monitoring system for implementing the above method includes:

[0092] An acoustic field measurement module, which includes an acoustic array measurement sub-module and a data acquisition sub-module, and is used to measure the acoustic field information propagated to the acoustic array installation position in the pipeline during the operation of the compressor;

[0093] A spectrum analysis module, which is used to transform the time-domain acoustic field signal at the acoustic array position to the frequency domain and detect whether there are abnormal frequencies other than the blade passing frequency and the rotational frequency;

[0094] A sound mode decomposition module, which is used to perform single - tone sound mode decomposition and continuous broadband sound mode decomposition, transform the sound field information from the frequency domain to the wavenumber domain, and detect whether the compressor rotating stall characteristics appear in the sound mode spectrogram and the wave spectrogram.

[0095] A computer storage medium, the storage medium includes computer instructions, when it runs on a computer, it causes the computer to execute the described method.

[0096] An electronic device, the electronic device includes:

[0097] A memory, a processor, and a computer program stored on the memory and executable on the processor, wherein,

[0098] When the processor executes the program, it implements the described method.

[0099] In one embodiment, the method for monitoring the rotating stall sound signature of an aero - engine compressor includes the following steps:

[0100] In the first step S1, according to the number of rotor blades and the number of stator blades of the aero - engine compressor, calculate the rotor - stator interference mode order of the compressor single - tone noise. , in the selected implementation case, a uniform sound array layout is adopted. Based on the maximum mode order of the sound mode order, determine the sound mode monitoring range. According to the Nyquist sampling theorem: determine the number of microphones and the installation angles of the microphones ;

[0101] In the second step S2, through the number of microphones and the installation angles of the microphones construct a microphone array; synchronously collect the multi - channel sound pressure of the aero - engine compressor through the microphone array; arrange the multi - channel sound pressure time - domain signals in sequence to construct a time - domain signal matrix;

[0102] In the third step S3, use the fast Fourier transform algorithm to perform spectral analysis on the sound array signal to obtain the spectrograms of the sound signals of each channel. When the compressor is working normally, only the blade passing frequency (and the rotation frequency) exists in the spectrum. Monitor the signal spectrogram to check whether there is an abnormal single - tone frequency peak lower than 1 times the rotation frequency. If not, it means the compressor is in a normal working state. If so, further analysis is required;

[0103] In the fourth step S4, based on the constructed transformation matrix, the acoustic mode spectrogram of the abnormal single-tone frequency peak is obtained by the single-frequency acoustic mode decomposition method; whether the compressor has rotating stall is judged by monitoring whether the characteristics of compressor rotating stall appear in the acoustic mode spectrogram. When the compressor has the characteristics of rotating stall under supersonic conditions, the main components in the frequency spectrum are the characteristic frequencies of rotating stall below 1 times the rotational frequency, and in the wavenumber domain, the characteristic frequencies of rotating stall with frequencies below 1 times the rotational frequency and a modal order of 1 appear. To more fully illustrate that the compressor has a rotating stall fault at this time, the acoustic array signal is further analyzed;

[0104] In the fifth step S5, the wavenumber spectrogram of the array signal is obtained by continuous broadband acoustic mode decomposition; whether the compressor has rotating stall is further illustrated by monitoring whether the characteristics of compressor rotating stall appear in the wavenumber spectrogram. When the compressor has rotating stall under supersonic conditions, the abnormal sound source of rotating stall and its multiple frequencies appear in the wavenumber spectrogram, and they are all non-integer multiple relationships with the rotational frequency; at the same time, the acoustic mode oblique bright band components starting from the modal order of 1 appear, corresponding to the characteristic frequencies of rotating stall and their multiple frequencies.

[0105] In a preferred embodiment, in the first step S1, according to the number of rotor blades of the aeroengine compressor and the number of stator blades calculate the acoustic mode order of the compressor single-tone noise : , where represents the pressure pulsation order caused by the unsteady aerodynamic force caused by the rotor-stator interference of the compressor, represents an integer. Determining the number and installation angles of the sensors in the microphone array includes the following steps:

[0106] S101. Calculate the number of sensors required for modal detection based on the Nyquist sampling law , and its relationship with the acoustic mode order is: ;

[0107] S102. sensors form a circular acoustic array in a uniform layout, the spacing between the sensors is , and the installation angle of the microphone is , where , , , and so on;

[0108] In a preferred embodiment, in the second step S2, conduct a stall test on the aeroengine compressor, and use the microphone array to obtain the sound field information in the compressor pipeline. Establishing the construction of the time-domain signal matrix includes the following steps:

[0109] S201. Measure the sound pressure signal of the aero-engine compressor using a circular acoustic array. The measured sound pressure signal is , indicating that the length of the time-domain signal sequence measured by a single microphone is , and the subscripts are the microphones at the corresponding angular positions respectively;

[0110] S202. According to the time-domain signals measured by the microphones at different angular positions, construct a time-domain signal matrix , where the element represents the signal measured by the microphone at the corresponding angular position at the th.

[0111]

[0112] In a preferred embodiment, in the third step S3, obtain a spectrogram through fast Fourier transform, and observe whether there are significantly energetic abnormal single-tone components in the spectrum and whether there is no integer relationship with the blade rotation frequency.

[0113] S301. Perform Fourier transform on each column of the time-domain signal matrix measured by the circular acoustic array to obtain a frequency-domain matrix , where the element represents the amplitude of the signal measured by the microphone at the corresponding angular position at the th frequency. has a length of . According to the Nyquist sampling theorem, .

[0114]

[0115] S302. According to the obtained frequency-domain matrix , plot the spectrograms of the signals measured by the sensors at different angular positions, and observe whether there are abnormal single-tone frequency peaks other than the blade passing frequency and the rotation frequency. Specifically, it shows large-amplitude rotating stall characteristic frequencies below 1 times the rotation frequency and their harmonics.

[0116] In a preferred embodiment, in the fourth step S4, perform single-frequency acoustic mode decomposition to obtain an acoustic mode spectrogram, and observe whether there are compressor rotating stall characteristics:

[0117] S401. The acoustic signal of the microphone at the corresponding angular position at a specific frequency can be regarded as a linear superposition of different circumferential acoustic modes, that is, , so a transformation matrix can be constructed , thus , and its specific form is as follows:

[0118]

[0119] S402. Perform a spatial Fourier transform on the frequency-domain matrix to obtain a wavenumber-domain matrix . According to S401, it can be known that , where is the transpose matrix of the frequency-domain matrix . Therefore, there is , where represents the pseudo-inverse of the transformation matrix .

[0120] S403. According to the obtained wavenumber-domain matrix , plot the acoustic mode spectrogram; under supersonic conditions, observe whether there are obvious rotating stall characteristic frequencies and their harmonics. The rotating stall characteristic frequency is lower than 1 times the rotational frequency, and the mode order is 1.

[0121] In a preferred embodiment, in the fifth step S5, continuous broadband acoustic mode decomposition is performed to obtain a spectrogram, and the presence of compressor rotating stall characteristics is observed:

[0122] S501. Perform continuous broadband acoustic mode decomposition on the full frequency domain, then the size of the wavenumber-domain matrix is expanded to , where the element represents the amplitude of the -th order acoustic mode at the frequency .

[0123]

[0124] S502. According to the obtained wavenumber-domain matrix , plot the spectrogram; under supersonic conditions, observe whether there are abnormal rotating stall sound sources and their harmonics that are not in an integer multiple relationship with the rotational frequency, as well as the acoustic mode oblique bright band components starting from the mode order of 1 corresponding to the rotating stall characteristic frequency and its harmonics. The mathematical expression of this oblique bright band is

[0125]

[0126] Define the above characteristics as the acoustic fingerprint characteristics for monitoring compressor rotating stall, and accordingly judge whether the compressor has rotating stall.

[0127] In a preferred embodiment, the sound field information is measured by a microphone acoustic array to obtain a time-domain signal matrix , where the element represents the measurement of the Signal

[0128] Figure 1 is the flow chart of the aero-engine compressor rotating stall monitoring method based on voiceprint features completed by the present invention. This method calculates the order of the rotor-stator interference mode in the compressor single tone noise through the compressor model of the aero-engine, and determines the acoustic mode monitoring range, the number of microphones, the axial installation position and the circumferential installation angle according to the maximum mode order concerned; performs a fast Fourier transform on the acoustic array signal to obtain the frequency domain signal matrix and outputs the spectrogram to observe whether there are abnormal peaks other than the blade passing frequency and the rotating frequency; based on the constructed transformation matrix, obtains the acoustic mode spectrogram of the abnormal single tone frequency peak through the single frequency acoustic mode decomposition method, and observes whether there is a rotating stall characteristic frequency with a frequency component lower than 1 times the rotating frequency and a mode order of 1; obtains the wave spectrogram of the array signal through continuous broadband acoustic mode decomposition, and observes whether there are rotating stall abnormal sound sources and their harmonics with a non-integer multiple relationship with the rotating frequency, as well as the acoustic mode oblique bright band components starting from the mode order of 1 corresponding to the rotating stall characteristic frequency and its harmonics. The specific steps are as follows:

[0129] 1) Assume the number of rotor blades and the number of stator blades of the aero-engine compressor. According to the acoustic mode order calculation formula of the compressor single tone noise , usually take to represent that the pressure pulsation order caused by the unsteady aerodynamic force caused by the rotor-stator interference of the compressor is 1. At this time, take to represent the first-order acoustic mode, and calculate to obtain , take to represent the second-order acoustic mode, and calculate to obtain . According to the Nyquist sampling theorem: , it is required that the number of sensors , so the number of sensors selected is ;

[0130] 2) A total of [[number]] sensors form a circular acoustic array in a uniform layout. The sampling frequency is set to 20000 hz, the spacing between sensors is , and the installation angle of the microphone is , where , , , and so on;

[0131] 3) Use the circular acoustic array to measure the sound pressure signal of the aero-engine compressor. Each channel measures data points, then the size of the constructed time domain signal matrix is Measure the time-domain signal matrix for the circular sound array Perform Fourier transform on each column of the matrix to transform the signal from the time domain to the frequency domain, obtaining the frequency-domain matrix The size of the frequency-domain matrix is .

[0132] 4) According to the obtained frequency-domain matrix , plot the spectrograms of the signals measured by sensors at different angular positions. Taking the microphone signal with the installation position as an example, when there is no rotating stall in Fig. 3(a), the main components in the spectrum are the rotation frequency and its harmonics. When rotating stall occurs in Fig. 3(b) and Fig. 3(c), in addition to the blade passing frequency and the rotation frequency, abnormal peaks appear at less than one times the rotation frequency.

[0133] 5) The element in the frequency-domain matrix represents the acoustic signal of the microphone at the corresponding angular position with the subscript at a specific frequency . It can be regarded as a linear superposition of different circumferential acoustic modes, that is . Therefore, the constructed transformation matrix has a size of .

[0134] 6) Perform single-frequency acoustic mode decomposition at the abnormal peak frequency to obtain a wavenumber-domain matrix with a size of and plot the acoustic mode spectrogram. When rotating stall occurs, in the acoustic mode spectrogram at the rotating stall characteristic frequency in Fig. 4(a), its acoustic mode is observed to be . In the acoustic mode spectrogram at the blade passing frequency in Fig. 4(b), its acoustic mode is observed to be and .

[0135] 7) Perform continuous broadband acoustic mode decomposition on the frequency-domain matrix to obtain a wavenumber-domain matrix with a size of . The element in the wavenumber-domain matrix represents the amplitude of the th order acoustic mode at the frequency , a spectrogram is plotted. Under supersonic conditions, when there is no rotating stall in Fig. 5(a), the main sound source in the spectrogram is the blade passing frequency. When rotating stall occurs in Fig. 5(b) and Fig. 5(c), abnormal sound sources of rotating stall and their harmonics appear in the spectrogram, all of which have a non-integer multiple relationship with the rotation frequency; at the same time, the sound mode oblique bright band components starting from the modal order of 1 appear, corresponding to the rotating stall characteristic frequency and its harmonics.

[0136] Although the embodiments of the present invention have been described above in conjunction with the accompanying drawings, the present invention is not limited to the above specific embodiments and application fields. The above specific embodiments are merely illustrative and guiding, rather than restrictive. Those of ordinary skill in the art can also make many forms under the inspiration of this specification and without departing from the scope protected by the claims of the present invention, and all of these fall within the scope of protection of the present invention.

Claims

1. A method for monitoring the rotating stall sound pattern of an aero-engine compressor, characterized in that It includes the following steps: In the first step (S1), the layout of the microphone array is determined based on the compressor model parameters of the aeroengine, and according to the number of rotor blades of the compressor and the number of stator blades the order of the rotor-stator interference mode of the compressor single tone noise is calculated , and based on the maximum mode order of the rotor-stator interference mode the acoustic mode monitoring range, the number of microphones and the circumferential installation angle of the microphones are determined ; In the second step (S2), through the number of microphones and the installation angles of the microphones a microphone array is constructed; the multi-channel sound pressure time-domain signals of the aero-engine compressor are synchronously collected through the microphone array; a time-domain signal matrix is constructed by arranging the multi-channel sound pressure time-domain signals in sequence; In the third step (S3), the fast Fourier transform algorithm is used to perform spectral analysis on the sound pressure time-domain signal to obtain the spectrogram of the sound pressure time-domain signal of each channel. By monitoring the spectrogram to check if there is an abnormal single-tone frequency peak below 1 times the rotation frequency, if not, it indicates that the compressor is in a normal working state; if so, it enters the fourth step (S4). In the fourth step (S4), the acoustic mode spectrogram of the abnormal single-tone frequency peak is obtained through the single-frequency acoustic mode decomposition method. By monitoring whether the compressor stall characteristics appear in the acoustic mode spectrogram to determine whether the compressor has a rotating stall. When the compressor shows rotating stall characteristics under supersonic conditions, in the wavenumber domain, there will be a rotating stall characteristic frequency with a frequency component lower than 1 times the rotation frequency and an acoustic mode order of 1.

2. The method for monitoring the rotating stall sound pattern of an aero-engine compressor according to claim 1, characterized in that Preferably, it further includes: In the fifth step (S5), the wavenumber spectrogram of the sound pressure time-domain signal is obtained through continuous broadband acoustic mode decomposition. The abscissa of the wavenumber spectrogram represents the frequency analysis range, the ordinate represents the acoustic mode monitoring range, and the depth of the color represents the amplitude size. The darker the color, the larger the amplitude. By monitoring whether the compressor stall characteristics appear in the wavenumber spectrogram, it further indicates whether the compressor has a rotating stall. When the compressor has a rotating stall under supersonic conditions, there will be a rotating stall abnormal sound source and its harmonics in the wavenumber spectrogram that are not in an integer multiple relationship with the rotation frequency, and at the same time, there will be an acoustic mode oblique bright band component starting from the acoustic mode order of 1, corresponding to the rotating stall characteristic frequency and its harmonics.

3. The method for monitoring the rotating stall sound pattern of an aero-engine compressor according to claim 2, wherein In the first step (S1), the order of the rotor-stator interference mode of the compressor is , where represents the order of the pressure pulsation caused by the unsteady aerodynamic force caused by the rotor-stator interference of the compressor, represents an integer, and determines the number of microphones in the microphone array and the microphone installation angle When adopting the uniform acoustic array layout scheme, calculate the number of sensors required for modal detection based on the Nyquist sampling law , and its relationship with the rotor-stator interference modal order is as follows: ; When adopting the non-uniform acoustic array layout scheme with fewer measurement points, randomly select positions from the positions of the virtual uniform layout to install sensors, ; When adopting the uniform acoustic array layout scheme, sensors form a circular acoustic array, and the spacing between sensors is , and the installation angle of the microphone is , where , , , and so on; when adopting the non-uniform acoustic array layout scheme with fewer measurement points, the installation angle of the microphone is randomly selected, .

4. A method for monitoring the rotating stall sound pattern of an aero-engine compressor according to claim 1, characterized in that, The second step (S2) includes the following steps: S201. Measure the sound pressure signal of the aero-engine compressor using a circular microphone array, and the measured sound pressure time-domain signal is , where the length of the time-domain signal sequence measured by a single microphone is , and the subscripts are the microphones at the corresponding installation angle positions respectively; S202. Based on the time-domain sound pressure signals measured by microphones at different installation angle positions, a time-domain signal matrix is formed , where the element represents the signal measured by the microphone at the corresponding installation angle position , when adopting a uniform acoustic array layout, the size of the time-domain signal matrix is ; when adopting a non-uniform acoustic array layout scheme with fewer measurement points, the size of the time-domain signal matrix is .

5. The method for monitoring the rotating stall sound pattern of an aeroengine compressor according to claim 4, characterized in that The third step (S3) includes: S301. Perform Fourier transform on each column of the time-domain signal matrix measured by the circular acoustic array to obtain the frequency-domain matrix , where the element represents the amplitude of the signal measured by the microphone at the corresponding installation angle position at frequency. The length of is . According to the Nyquist sampling theorem, . , in the uniform acoustic array layout, the size of the frequency domain matrix is ; in the non-uniform acoustic array layout, the size of the frequency domain matrix is , S302. According to the frequency domain matrix Draw the spectrogram of the signals measured by the position sensors at different installation angles, and observe whether there are abnormal single-tone frequency peaks other than the blade passing frequency and the rotating frequency. The abnormal single-tone frequency peak is the rotating stall characteristic frequency whose frequency is lower than the rotating frequency and whose amplitude is more than twice the amplitude of the current blade passing frequency, as well as its second and third harmonics.

6. The method for monitoring the rotating stall sound pattern of an aero-engine compressor according to claim 5, characterized in that, The fourth step (S4) includes: S401. When adopting a uniform acoustic array layout, the subscript corresponding to the microphone at the installation angle position is the frequency-domain signal at a predetermined frequency regarded as a linear superposition of different circumferential acoustic modes, that is , construct a transformation matrix in the following form: ， When adopting a non-uniform acoustic array layout with fewer measurement points, an observation matrix is constructed according to the randomly selected sensor installation angles , the observation matrix has a size of , ， S402. When adopting a uniform acoustic array layout, perform a spatial Fourier transform on the frequency-domain matrix to obtain the wavenumber-domain matrix , , where is the transpose matrix of the frequency-domain matrix , , where represents the pseudo-inverse of the transformation matrix . When adopting a non-uniform acoustic array layout with fewer measurement points, the compressive sensing model is , and the sparse dictionary is composed of orthogonal Fourier transform bases, is the sensing matrix, and the wavenumber domain matrix is sparsely reconstructed based on the compressive sensing model; S403. According to the obtained wavenumber domain matrix , plot the acoustic mode spectrogram; under supersonic conditions, observe whether the rotating stall characteristic frequency and its multiple frequencies appear. The rotating stall characteristic frequency is lower than the first-order rotating frequency and the mode order is 1.

7. A method for monitoring the rotating stall sound pattern of an aeroengine compressor according to claim 6, characterized in that The fifth step (S5) includes: S501. Expand the wavenumber domain matrix to the full frequency domain. The size of the wavenumber domain matrix is expanded to . Among them, the element represents the amplitude of the -th order acoustic mode at the frequency . ， S502. According to the wavenumber domain matrix , draw a spectrogram. Under supersonic conditions, observe whether there are rotating stall abnormal sound sources and their harmonics that are not integer multiples of the rotation frequency, as well as the acoustic mode oblique bright band components starting from mode order 1 corresponding to the rotating stall characteristic frequency and its harmonics. The mathematical expression of the acoustic mode oblique bright band components is ， In the formula, is the characteristic frequency of rotating stall, represents the multiple frequency of the characteristic frequency, is a non-negative integer, represents the circumferential acoustic mode order. The characteristic frequency of rotating stall appearing in the spectrum, the corresponding mode order appearing at the characteristic frequency in the acoustic mode spectrogram, the abnormal sound source in the wavenumber spectrogram, and the inclined bright band component of the acoustic mode are defined as the acoustic fingerprint characteristics for monitoring the rotating stall of the compressor. Based on this, it is judged whether the compressor has rotating stall.

8. A monitoring system for implementing the method according to any one of claims 1-7, characterized in that, It includes: An acoustic field measurement module, which includes an acoustic array measurement sub-module and a data acquisition sub-module, and is used to measure the acoustic field information propagated to the acoustic array installation position in the pipeline during the operation of the compressor. A spectral analysis module, which is used to transform the time-domain acoustic field signal at the acoustic array position to the frequency domain and detect whether there is an abnormal frequency outside the blade passing frequency and the rotation frequency. An acoustic mode decomposition module, which is used to perform single-tone acoustic mode decomposition and continuous broadband acoustic mode decomposition, transform the acoustic field information from the frequency domain to the wavenumber domain, and detect whether the compressor stall characteristics appear in the acoustic mode spectrogram and the wavenumber spectrogram.

9. A computer storage medium, characterized in that, The storage medium includes computer instructions, which when running on a computer, cause the computer to execute the method according to any one of claims 1-7.

10. An electronic device, characterized in that, The electronic device includes: A memory, a processor, and a computer program stored on the memory and executable on the processor, where When the processor executes the program, it implements the method according to any one of claims 1-7.

Images