CN-120489327-B - Method, system, medium and equipment for continuously monitoring near stall state voiceprint of air compressor

CN120489327BCN 120489327 BCN120489327 BCN 120489327BCN-120489327-B

Abstract

A method for continuously monitoring the sound wave of near stall state of air compressor includes such steps as randomly choosing the microphone measuring points with predefined number at entrance to form microphone array, synchronously collecting the sound pressure time-domain signals of multiple channels, adding rectangular window to the sound pressure time-domain signals of multiple channels, creating the sound time-domain signal segments, fast Fourier transform to the sound time-domain signal segments in the current time-domain window to form multiple sound pressure frequency spectrums, judging if there are multiple non-synchronous single-tone features with non-integer multiple current frequencies, creating the multi-channel microphone continuous frequency domain observation matrix with non-synchronous single-tone features, creating orthogonal Fourier transmission matrix based on the installation azimuth angle of microphone sensor and the highest order of mode, creating the near stall feature mode reconstruction model of air compressor, and solving the amplitude based on the block orthogonal matching tracking algorithm.

Inventors

Li Zepi
WEN BI
QIAO BAIJIE
WANG YANAN
CHEN XUEFENG
TIAN JINHU

Assignees

太行国家实验室
西安交通大学

Dates

Publication Date: 20260512
Application Date: 20250331

Claims (6)

1. A method for continuously monitoring the voiceprint of a near stall state of a compressor is characterized by comprising the following steps: In a first step (S1), according to the number of rotor blades of the compressor Number of stator blades Calculating the highest circumferential mode order of the compressor And the number of microphones Randomly selecting a preset number of microphone measuring points at the inlet to form a microphone array, and carrying out synchronous acquisition on sound pressure time domain signals of multiple channels; in the second step (S2), a rectangular window is added to the multi-channel sound pressure time domain signal, an acoustic time domain signal segment is constructed, fast Fourier transformation is carried out on the multi-channel acoustic time domain signal segment in the current time domain window to form a plurality of sound pressure spectrums, whether a plurality of asynchronous single-tone characteristic frequencies with current conversion frequency being non-integral multiple exist in the sound pressure spectrums or not is judged, if not, the compressor is in a normal working state, the second step (S2) is repeated, and if so, the near stall working condition monitoring is further carried out; In the third step (S3), a multi-channel microphone continuous frequency domain observation matrix of unsynchronized single-tone characteristic frequency is constructed, an orthogonal Fourier transfer matrix is constructed based on a microphone sensor installation azimuth angle and a mode highest order, and a near stall characteristic mode reconstruction model of the compressor under non-uniform few-measuring-point is established; In the fourth step (S4), the amplitude of a near stall characteristic modal reconstruction model of the compressor is solved based on a block orthogonal matching pursuit algorithm, so that the near stall working condition monitoring of the compressor based on voiceprint characteristics is realized; Wherein, the The first step (S1) comprises, Step S101, calculating the highest circumferential mode order, wherein the mode order of Zhou Xiangsheng modes is , Represents the pressure pulsation order caused by unsteady aerodynamic forces caused by fan-to-static interference, Representing non-negative integers, determining the highest circumferential modal order Determining a measurable modal range as ; Step S102, setting the number of virtual annular microphones And the installation position, by Nyquist-shannon sampling theory, the number of microphones And the highest modal order The relation of (2) is that The virtual microphone arrays are uniformly arranged along the circumferential direction, and the distance between each virtual grid point is ; Step S103, randomly installing the number of sensors K on the wall surface of the casing along the circumferential direction, wherein K is the number of microphones under Nyquist sampling 50% Of the installation angle ; The second step (S2) comprises the steps of: Step S201, determining the window length of an acoustic time domain signal segment by sampling frequency, adding a rectangular window to the sound pressure time domain signal of the multiple channels, and increasing the window length And is also provided with , Is an arbitrary integer number of the whole, Is the sampling frequency of the microphone sensor, and every other in the continuous sampling process Windowing the sound pressure time domain signal of the multi-channel signal, overlapping length of each window ; Step S202, current acoustic time domain signal segment Performing a fast fourier transform on the K sound pressure time domain signals of (a): , wherein, A current acoustic time domain signal segment matrix representing sound pressure time domain signals measured by the non-uniform microphone array, the current time domain signal segment matrix having dimensions of , Representing the discrete fourier transform of the signal, In the first multi-channel array A frequency domain matrix of acoustic time domain signal segments having dimensions of , The expression is as follows: , Wherein the method comprises the steps of , Step S203, according to the obtained frequency domain matrix Drawing sound pressure frequency spectrograms measured by sensors at different positions, and observing whether a plurality of divided integer multiple frequency conversion exists External non-synchronous tone characteristic frequency, i.e. presence If the stall condition exists, further carrying out the monitoring on the stall condition; the third step (S3) comprises, Step S301, establishing asynchronous tone characteristic frequency Multi-channel microphone continuous frequency domain observation matrix : , Wherein, the Representing acoustic time domain signal segments Asynchronous single tone characteristic frequency of kth measuring point of non-uniform array The amplitude value of the amplitude value, Step S302, constructing an orthogonal Fourier transfer matrix , , Wherein the method comprises the steps of , Step S303, defining an acoustic modal matrix Establishing a near stall characteristic mode reconstruction model of the compressor, , Wherein, the Is the first of the matrix The number of row vectors, For the number of dominant modes of interest, Is a sparse norm of structure and has: , Returning 1 when the condition is true as the indication function; the fourth step (S4) comprises, Step S401, setting residual error matrix Index sequence number set Structural sparsity For the dominant number of modes of interest, the number of iterations ; Step S402, iteration number update ; Step S403, calculating a support set index ; Step S404, updating index sequence number set ; Step S405, calculating the acoustic mode amplitude vector of the corresponding support set ; Step S406, updating residual error matrix ; Step S407, checking the iteration number Whether or not to achieve structural sparsity And outputting the amplitude monitoring result of the near stall characteristic mode reconstruction model of the compressor if the amplitude monitoring result is reached, otherwise, returning to the step S402.
2. The method for continuously monitoring the sound print of a near stall condition of a gas compressor according to claim 1, wherein the gas compressor is a gas compressor of an aeroengine.
3. The method for continuously monitoring the voiceprint of the near stall condition of the compressor according to claim 1, the compressor is characterized by comprising a fan structure.
4. A fan blade synchronous vibration identification system implementing the method of any of claims 1-3, comprising: A sound field measuring module for measuring the number of rotor blades of the compressor Number of stator blades Calculating the highest circumferential mode order of the compressor And the number of microphones Randomly selecting a preset number of microphone measuring points at the inlet to form a microphone array, and carrying out synchronous acquisition on sound pressure time domain signals of multiple channels; the frequency spectrum analysis module is used for adding a rectangular window to the multi-channel sound pressure time domain signal, constructing an acoustic time domain signal segment, performing fast Fourier transform on the multi-channel acoustic time domain signal segment in the current time domain window to form a plurality of sound pressure frequency spectrums, and judging whether a plurality of current frequency conversion non-integer times of asynchronous single-tone characteristic frequencies exist in the sound pressure frequency spectrums; The construction module is used for constructing a multi-channel microphone continuous frequency domain observation matrix of unsynchronized single-tone characteristic frequency, constructing an orthogonal Fourier transfer matrix based on a microphone sensor installation azimuth angle and a mode highest order, and constructing a near stall characteristic mode reconstruction model of the compressor under non-uniform few-measuring points; and the monitoring module is used for solving the amplitude of the near-stall characteristic modal reconstruction model of the compressor based on the block orthogonal matching pursuit algorithm and realizing the near-stall working condition monitoring of the compressor based on voiceprint characteristics.
5. A computer storage medium comprising computer instructions which, when run on a computer, cause the computer to perform the method of any of claims 1-3.
6. An electronic device, the electronic device comprising: a memory, a processor, and a computer program stored on the memory and executable on the processor, wherein, The processor, when executing the program, implements the method of any one of claims 1-3.

Description

Method, system, medium and equipment for continuously monitoring near stall state voiceprint of air compressor Technical Field The invention relates to the technical field of compressor state monitoring, in particular to a method, a system, a medium and equipment for continuously monitoring the near stall state voiceprint of a compressor. Background As aircraft have increased in flying height, speed and maneuver performance requirements, the compressor operating environment has become more severe. When the engine enters a pneumatic unstable working state, the performance of the engine is reduced, the vibration stress of the rotor blade of the compression part is increased, the thermal load and the thermal stress of the turbine are increased, the stable working range of the combustion chamber is narrowed, and even the structural integrity of the engine can be damaged, so that the flight safety is seriously threatened. By monitoring the unsteady pressure pulsation of the compressor in the near stall working condition, the pre-existing abnormal characteristics shown in the near stall working condition are monitored and diagnosed, and the method has important significance in the design of available stability margin and pneumatic stability control of an aeroengine compression system. At present, a strain gauge, a pressure probe and other sensors are usually arranged on a rotating blade to monitor the state of the air compressor, so that the cost is high, and an invasive measurement mode can interfere the flow field of the air compressor. The acoustic sensor has the characteristics of high sensitivity, short transmission path and the like, and is non-invasive measurement. The compressor typically has strong single-tone noise generation at near stall conditions, and acoustic signals are one of the important response forms for faults. However, voiceprint characteristics of the compressor under the near-stall working condition are less researched, and a compressor near-stall monitoring method based on the voiceprint characteristics is not established. The above information disclosed in the background section is only for enhancement of understanding of the background of the invention and therefore may contain information that does not form the prior art that is already known to a person of ordinary skill in the art. Disclosure of Invention The invention provides a method, a system, a medium and equipment for continuously monitoring the sound print of a near stall state of a gas compressor, which are characterized in that a small quantity of non-uniform annular microphone arrays are arranged at the gas inlet of the gas compressor, the non-synchronous characteristic single-tone frequency of the near stall working condition of the gas compressor is judged by a frequency spectrum analysis module of a sound array test signal, and constructing a multi-channel microphone continuous frequency domain observation matrix of unsynchronized single-tone characteristic frequency, constructing a near stall characteristic mode reconstruction model of the compressor under non-uniform few-measuring-point conditions, solving the near stall characteristic acoustic mode amplitude of the compressor based on a block orthogonal matching tracking algorithm, and realizing the near stall working condition monitoring of the compressor based on acoustic print characteristics. The method realizes the real-time monitoring of the near stall characteristic modal amplitude based on a small number of acoustic measuring points for the first time, and provides an effective tool for engine control in a near stall running state. Meanwhile, the method uses an acoustic monitoring mode which is more sensitive in response, shorter in transmission path and non-invasive, and solves the dilemma that dynamic sensitive feedback is difficult to realize when the traditional static pressure sensor monitors the working condition near the stall point. A method for continuously monitoring the voiceprint of a near stall state of a compressor comprises the following steps: in a first step, according to the number of rotor blades of the compressor Number of stator bladesCalculating the highest circumferential mode order of the compressorAnd the number of microphonesRandomly selecting a preset number of microphone measuring points at the inlet to form a microphone array, and carrying out synchronous acquisition on sound pressure time domain signals of multiple channels; In the second step, a rectangular window is added to the multi-channel sound pressure time domain signal, an acoustic time domain signal segment is constructed, fast Fourier transform is carried out on the multi-channel acoustic time domain signal segment in the current time domain window to form a plurality of sound pressure spectrums, whether a plurality of asynchronous single-tone characteristic frequencies with current conversion frequency being non-integral multiple exist in the sound pressure spectrums