Method and system for intelligently diagnosing and analyzing faults of spacecraft vibration test products

Abstract

The invention provides an intelligent fault diagnosis and analysis method and system for a spacecraft vibration test product. The system comprises a microphone array, voiceprint signal processing software and a vibration response acquisition device. A microphone array is reasonably arranged, voiceprint signals generated when a spacecraft is subjected to a vibration test on a vibration table are collected through microphones, weighting, time delay and summation processing are carried out on the received signals, spatial directivity is formed, and therefore a fault occurrence area is rapidly positioned. And then time domain response data collected by measuring points near the fault area is analyzed, and the fault type is accurately positioned by comparing the voiceprint time-frequency signal with the acceleration time-frequency signal. And in combination with a multi-scale convolutional memory neural network, fault features are extracted, different fault models are trained, and intelligent diagnosis of faults is realized. According to the method, various faults in the spacecraft vibration test process can be accurately identified, troubleshooting and maintenance of spacecraft parts can be rapidly guided, and technical guarantee is provided for smooth implementation of the spacecraft vibration test.

Description

Technical Field

[0001] The present invention relates to the field of fault diagnosis and analysis of spacecraft vibration test products, and in particular to a method and system for intelligent fault diagnosis and analysis of spacecraft vibration test products. Background Art

[0002] In the aerospace field, with the rapid advancement of technology, new-generation spacecraft have seen significant improvements in functionality and performance. On the one hand, their capabilities are becoming increasingly diverse, their performance is constantly improving, their payloads are significantly increased, and their number of moving and deployable parts is increasing. These changes have led to a more complex overall configuration of spacecraft, and the connections between components are no longer simple, resulting in a more diverse and complex design.

[0003] On the other hand, to meet the unique demands of space missions, most spacecraft are increasingly incorporating new materials and processes. Design concepts are gradually shifting toward lighter weight and greater flexibility. However, these innovations and improvements also present new challenges. To ensure the quality and safety of spacecraft, ground vibration testing is required to verify whether critical payloads and components are free of design or process flaws, and to assess their ability to withstand the dynamic environment.

[0004] A key challenge facing current spacecraft vibration testing is the difficulty in quickly and accurately locating and identifying fault damage generated during testing. Currently, when unusual noises occur during spacecraft vibration testing, the human ear is primarily used to determine their location and, in turn, infer the likely fault location. However, this traditional approach has significant limitations. First, due to the inherent human factor, individual testers may make errors in fault location determination due to differences in personal experience and auditory sensitivity. Such errors not only hinder accurate fault identification but can also delay subsequent testing, increasing testing costs and time. Second, during vibration testing, many unusual noises are transient and non-repeatable. In such cases, relying solely on human hearing to accurately determine the location of the noise is virtually impossible, making it difficult to quickly locate and effectively address the fault.

[0005] A search of patent documents revealed an invention patent with publication number CN115931318A, which discloses an intelligent fault diagnosis method, including: collecting vibration signals from the device under test and extracting vibration characteristic information from the vibration signals using a data processing algorithm; processing the vibration characteristic information using a pre-trained machine learning classification model to determine whether a rotor shafting fault has occurred; if so, outputting the cause of the rotor shafting fault; if not, processing the vibration signals using a pre-trained deep learning classification model to determine whether a bearing fault has occurred; if so, outputting the cause of the bearing fault; if not, the fault diagnosis result is no fault. This patent focuses on fault diagnosis of rotor shafting and bearings, and has a narrow scope of application. The diagnostic method is limited to vibration signals and lacks fault location.

[0006] In summary, in response to the above-mentioned problems of the existing technology, researching an intelligent diagnosis and analysis method and system for spacecraft vibration test product faults has become a key task that needs to be solved urgently. Summary of the Invention

[0007] In view of the defects in the prior art, the purpose of the present invention is to provide a method and system for intelligent diagnosis and analysis of spacecraft vibration test product faults.

[0008] According to the present invention, a method for intelligent diagnosis and analysis of faults in a spacecraft vibration test product is provided, comprising the following steps:

[0009] Step S1: hoist the spacecraft onto a vibration table, deploy a vibration response acquisition device, and reasonably arrange a microphone array around the spacecraft;

[0010] Step S2: starting the spacecraft vibration test, using the microphone array to collect the voiceprint signal generated during the spacecraft vibration test, and storing the voiceprint signal in the device; at the same time, using the vibration response acquisition device to collect time domain data;

[0011] Step S3, processing the voiceprint signal, extracting voiceprint features, and generating a spectrogram;

[0012] Step S4, determining the fault occurrence area based on the sound spectrogram;

[0013] Step S5: selecting a measurement point in the fault occurrence area and performing time-frequency analysis on the time domain data of the measurement point to obtain frequency domain features;

[0014] Step S6, constructing a fault image library based on the spectrogram and frequency domain features;

[0015] Step S7: input the fault image library into the convolution layer to extract high-level feature signals;

[0016] Step S8, processing the high-level feature signal through the long short-term memory network to further extract the fault features;

[0017] Step S9: Use a multi-scale convolutional memory neural network to train fault features and generate a fault database;

[0018] Step S10: input the voiceprint features and frequency domain features into the fault database, and realize intelligent fault diagnosis through fault signal comparison.

[0019] Preferably, in step S1, the microphone array includes a plurality of acoustic sensors for collecting voiceprint signals.

[0020] Preferably, in step S1, the microphone array is arranged according to the structural characteristics of the spacecraft and the test requirements to ensure the effective collection of sound signals in key areas.

[0021] Preferably, in step S2, the voiceprint signal is stored in the device through voiceprint signal processing software.

[0022] Preferably, step S3 includes the following sub-steps:

[0023] Step S3.1, using voiceprint signal processing software to process the collected voiceprint signal and extract voiceprint features;

[0024] In step S3.2, the voiceprint features are analyzed using an acoustic imaging algorithm optimized by eigenvalue decomposition to generate a spectrogram.

[0025] Preferably, step S3.1 includes: performing time-frequency analysis and filtering on the voiceprint signal and performing singular value decomposition on the cross-spectral matrix of the signal measured by the array microphone to extract voiceprint features.

[0026] Preferably, step S3.2 includes: reconstructing the cross-spectral matrix based on the voiceprint features, and generating spectrograms of the main sound source and various levels of secondary sound sources through a cross-spectral delay sum beamforming operation.

[0027] Preferably, in step S5, time-frequency analysis is performed on the time domain data by Laplace transform to obtain frequency domain features.

[0028] Preferably, in step S7, the voiceprint features and frequency domain features in the fault image library are extracted by the convolution layer of the convolutional neural network to abstract high-level features, and then the output feature map is reduced in scale by the downsampling layer according to a certain rule to obtain high-level feature signals.

[0029] The present invention also provides a spacecraft vibration test product fault intelligent diagnosis and analysis system, comprising:

[0030] Module M1: hoist the spacecraft onto the vibration table, deploy the vibration response acquisition device, and arrange the microphone array around the spacecraft;

[0031] Module M2 starts the spacecraft vibration test, uses the microphone array to collect the soundprint signals generated during the spacecraft vibration test, and stores the soundprint signals in the device; at the same time, it uses the vibration response acquisition device to collect time domain data;

[0032] Module M3 processes the voiceprint signal, extracts voiceprint features, and generates a spectrogram;

[0033] Module M4 determines the fault location based on the sound spectrogram;

[0034] Module M5 selects measurement points in the fault area and performs time-frequency analysis on the time domain data of the measurement points to obtain frequency domain features;

[0035] Module M6, builds a fault image library based on the spectrogram and frequency domain features;

[0036] Module M7 inputs the fault image library into the convolution layer to extract high-level feature signals;

[0037] Module M8 processes high-level feature signals through long short-term memory networks to further extract fault features;

[0038] Module M9: Use multi-scale convolutional memory neural network to train fault features and generate a fault database;

[0039] Module M10 inputs the voiceprint features and frequency domain features into the fault database and realizes intelligent fault diagnosis through fault signal comparison.

[0040] Compared with the prior art, the present invention has the following beneficial effects:

[0041] 1. By combining the collection, processing, and analysis of voiceprint signals and vibration signals, the present invention can accurately identify various faults during spacecraft vibration testing, improve the accuracy and efficiency of fault diagnosis, realize intelligent and automated fault diagnosis, quickly locate the fault area, reduce fault investigation and maintenance costs, and enhance the reliability and safety of spacecraft vibration testing.

[0042] 2. The present invention provides a strong technical guarantee for the smooth implementation of spacecraft vibration tests and has important application value. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Other features, objects and advantages of the present invention will become more apparent upon reading the detailed description of non-limiting embodiments with reference to the following drawings:

[0044] Figure 1 This is a flow chart of a method for intelligent fault diagnosis and analysis of a spacecraft vibration test product according to an embodiment of the present invention;

[0045] Figure 2A diagram showing the layout of the microphone array around the spacecraft in an embodiment of the present invention;

[0046] Figure 3 The abnormal noise location area for the sound source in the embodiment of the present invention;

[0047] Figure 4 is a voiceprint feature graph in an embodiment of the present invention;

[0048] Figure 5 is the time domain response of the abnormal response measuring point in the fault area in the embodiment of the present invention;

[0049] Figure 6 Schematic diagram of signal convolution in an embodiment of the present invention (a 6×6 input feature map on the left is convolved by a 3×3 convolution kernel, and the right side shows local features);

[0050] Figure 7 Schematic diagram of feature map sampling extracted by convolution in an embodiment of the present invention. DETAILED DESCRIPTION

[0051] The present invention will be described in detail below with reference to specific embodiments. The following examples will help those skilled in the art to further understand the present invention, but are not intended to limit the present invention in any form. It should be noted that, for those skilled in the art, several changes and improvements can be made without departing from the scope of the present invention. These all fall within the scope of protection of the present invention.

[0052] The present invention discloses a method and system for intelligent fault diagnosis and analysis of spacecraft vibration test products, comprising a microphone array, voiceprint signal processing software, and a vibration response acquisition device. By rationally arranging the microphone array, microphones are used to collect the voiceprint signals generated by the spacecraft during vibration testing on a vibration table. The received signals are then weighted, delayed, and summed to form spatial directivity, thereby rapidly locating the fault area. Time-domain response data collected from measurement points near the fault area is then analyzed, and the fault type is accurately located by comparing the voiceprint time-frequency signal with the acceleration time-frequency signal. Combined with a multi-scale convolutional memory neural network, fault features are extracted and different fault models are trained to achieve intelligent fault diagnosis. The present invention can accurately identify various faults during spacecraft vibration testing, rapidly guide troubleshooting and repair of spacecraft components, and provide technical support for the smooth progress of spacecraft vibration testing. The purpose of the present invention is to provide a method and system for intelligent fault diagnosis based on the combined analysis of voiceprint features and vibration response, which can rapidly locate the location and type of faults during spacecraft vibration testing and improve fault diagnosis capabilities.

[0053] Example 1:

[0054] Figure 1The present invention is a flowchart of a method for intelligent diagnosis and analysis of spacecraft vibration test product faults in an embodiment of the present invention.

[0055] like Figure 1 As shown, this embodiment provides a method for intelligent diagnosis and analysis of spacecraft vibration test product faults, including the following steps:

[0056] Step S1: hoist the spacecraft onto the vibration table, deploy the vibration response acquisition device, and reasonably arrange the microphone array around the spacecraft.

[0057] Specifically, the microphone array includes multiple acoustic sensors for collecting voiceprint signals. The arrangement of the microphone array is based on the structural characteristics of the spacecraft and the test requirements to ensure the effective collection of sound signals in key areas.

[0058] Figure 2 FIG. 4 is a layout diagram of the microphone array around the spacecraft in an embodiment of the present invention.

[0059] like Figure 2 As shown, in this embodiment, multiple acoustic sensors are arranged around the spacecraft, and additional acoustic sensors are arranged at positions that require attention, such as large payloads, antennas, etc.

[0060] Step S2: Start the spacecraft vibration test, use the microphone array to collect the voiceprint signal generated during the spacecraft vibration test, and store the voiceprint signal in the device; at the same time, use the vibration response acquisition device to collect time domain data.

[0061] In this embodiment, the voiceprint signal is stored in the device through voiceprint signal processing software.

[0062] Step S3, processing the voiceprint signal, extracting voiceprint features, and generating a spectrogram;

[0063] Specifically, step S3 includes the following sub-steps:

[0064] Step S3.1: Use voiceprint signal processing software to process the collected voiceprint signal and extract voiceprint features.

[0065] Specifically, the voiceprint signal is subjected to time-frequency analysis and filtering, and the cross-spectral matrix of the signal measured by the array microphone is subjected to singular value decomposition to extract the voiceprint features.

[0066] In step S3.2, the voiceprint features are analyzed using an acoustic imaging algorithm optimized by eigenvalue decomposition to generate a spectrogram.

[0067] Figure 3 This is a voiceprint feature graph in an embodiment of the present invention.

[0068] Specifically, based on the voiceprint features, the cross-spectral matrix is reconstructed, and the spectrograms of the main sound source and the sub-sound sources at all levels are generated through the cross-spectral delay summation beamforming operation (such as Figure 3 shown).

[0069] Step S4, determining the fault occurrence area based on the sound spectrogram;

[0070] Figure 4 This is the abnormal noise location area for sound source positioning in the embodiment of the present invention.

[0071] Specifically, the sound spectrogram is analyzed and the fault location is preliminarily determined based on the voiceprint features. In this embodiment, spatial directivity is generated through weighting, delay, and summation to quickly locate the fault location.

[0072] Step S5: selecting a measurement point in the fault occurrence area, and performing time-frequency analysis on the time domain data of the measurement point to obtain frequency domain features.

[0073] Figure 5 It is the time domain response of the abnormal response measuring point in the fault area in the embodiment of the present invention.

[0074] In this embodiment, time-frequency analysis is performed on time-domain data through Laplace transform to obtain frequency-domain features.

[0075] Step S6, constructing a fault image library based on the spectrogram and frequency domain features;

[0076] Specifically, based on the frequency, sound pressure, attenuation law of the voiceprint signal in the sound spectrogram, and whether the frequency domain characteristics of high-frequency signals and amplitude distortion appear in the vibration frequency domain signal, the type of fault is determined and classified to form a fault library.

[0077] Step S7: input the fault image library into the convolution layer to extract high-level feature signals;

[0078] Figure 6 Schematic diagram of signal convolution in an embodiment of the present invention (a 6×6 input feature map on the left is convolved by a 3×3 convolution kernel, and the right side shows local features); Figure 7 Schematic diagram of feature map sampling extracted by convolution in an embodiment of the present invention.

[0079] Specifically, the voiceprint features and frequency domain features in the fault image library are extracted by the convolutional layer of the convolutional neural network (such as Figure 6 As shown), we abstract the high-level features, and then use the downsampling layer to reduce the scale of the output feature map according to a certain rule to obtain the high-level feature signal (as shown in Figure 7 shown).

[0080] Furthermore, a convolution operation is performed on the voiceprint features of size m×n using a convolution kernel of size k×k. Then, the local output features sensed by each convolution kernel are connected according to a special rule to form the total input. Then, an overall bias is added, and the result is input into a nonlinear activation function to finally form a high-level feature signal.

[0081] Step S8, processing the high-level feature signal through the long short-term memory network to further extract the fault features;

[0082] Step S9: Use a multi-scale convolutional memory neural network to train fault features and generate a fault database;

[0083] Step S10: input the voiceprint features and frequency domain features into the fault database, and realize intelligent fault diagnosis through fault signal comparison.

[0084] Example 2:

[0085] The present invention also provides an intelligent diagnosis and analysis system for spacecraft vibration test product faults. The intelligent diagnosis and analysis system for spacecraft vibration test product faults can be implemented by executing the process steps of the intelligent diagnosis and analysis method for spacecraft vibration test product faults. That is, those skilled in the art can understand the intelligent diagnosis and analysis method for spacecraft vibration test product faults as an optimal implementation of the intelligent diagnosis and analysis system for spacecraft vibration test product faults.

[0086] Specifically, the spacecraft vibration test product fault intelligent diagnosis and analysis system includes:

[0087] Module M1: hoist the spacecraft onto the vibration table, set up the vibration response acquisition device, and arrange the microphone array around the spacecraft;

[0088] Module M2 starts the spacecraft vibration test, uses the microphone array to collect the soundprint signals generated during the spacecraft vibration test, and stores the soundprint signals in the device; at the same time, it uses the vibration response acquisition device to collect time domain data;

[0089] Module M3 processes the voiceprint signal, extracts voiceprint features, and generates a spectrogram;

[0090] Module M4 determines the fault location based on the sound spectrogram;

[0091] Module M5 selects measurement points in the fault area and performs time-frequency analysis on the time domain data of the measurement points to obtain frequency domain features;

[0092] Module M6, builds a fault image library based on the spectrogram and frequency domain features;

[0093] Module M7 inputs the fault image library into the convolution layer to extract high-level feature signals;

[0094] Module M8 processes high-level feature signals through long short-term memory networks to further extract fault features;

[0095] Module M9: Use multi-scale convolutional memory neural network to train fault features and generate a fault database;

[0096] Module M10 inputs the voiceprint features and frequency domain features into the fault database and realizes intelligent fault diagnosis through fault signal comparison.

[0097] Those skilled in the art will appreciate that, in addition to implementing the system and its various devices, modules, and units provided by the present invention in purely computer-readable program code, it is entirely possible to implement the same functions of the system and its various devices, modules, and units provided by the present invention in the form of logic gates, switches, application-specific integrated circuits, programmable logic controllers, and embedded microcontrollers by logically programming the method steps. Therefore, the system and its various devices, modules, and units provided by the present invention can be considered a hardware component, and the devices, modules, and units included therein for implementing various functions can also be considered as structures within the hardware component; the devices, modules, and units for implementing various functions can also be considered as both software modules implementing the method and structures within the hardware component.

[0098] The above describes specific embodiments of the present invention. It should be understood that the present invention is not limited to the specific embodiments described above, and those skilled in the art may make various changes or modifications within the scope of the claims, which do not affect the essence of the present invention. The embodiments of this application and the features in the embodiments may be combined with each other in any manner unless there is a conflict.

Claims

1. A method for intelligent diagnosis and analysis of spacecraft vibration test product faults, characterized in that: The following steps are involved: Step S1: hoisting the spacecraft onto a vibration table, deploying a vibration response acquisition device, and reasonably arranging a microphone array around the spacecraft; Step S2, starting the spacecraft vibration test, using the microphone array to collect the voiceprint signal generated during the spacecraft vibration test, and storing the voiceprint signal in the device; and simultaneously using the vibration response acquisition device to collect time domain data; Step S3, processing the voiceprint signal, extracting voiceprint features, and generating a spectrogram; Step S4, determining the fault occurrence area based on the sound spectrogram; Step S5, selecting a measurement point in the fault occurrence area, and performing time-frequency analysis on the time domain data of the measurement point to obtain frequency domain features; Step S6, constructing a fault map library based on the spectrogram and the frequency domain features; Step S7, inputting the fault image library into the convolution layer to extract high-level feature signals; Step S8, processing the high-level feature signal through a long short-term memory network to further extract fault features; Step S9: using a multi-scale convolutional memory neural network to train the fault features and generate a fault database; Step S10: input the voiceprint features and the frequency domain features into the fault database, and implement intelligent fault diagnosis through fault signal comparison.

2. The method for intelligent diagnosis and analysis of spacecraft vibration test faults according to claim 1, characterized in that: In step S1, the microphone array includes a plurality of acoustic sensors for collecting voiceprint signals.

3. The intelligent diagnosis and analysis method for spacecraft vibration test faults according to claim 2, characterized in that: In step S1, the microphone array is arranged according to the structural characteristics of the spacecraft and the test requirements to ensure the effective collection of sound signals in key areas.

4. The method for intelligent diagnosis and analysis of spacecraft vibration test faults according to claim 1, characterized in that: In step S2, the voiceprint signal is stored in the device through voiceprint signal processing software.

5. The method for intelligent diagnosis and analysis of spacecraft vibration test faults according to claim 1, characterized in that: The step S3 includes the following sub-steps: Step S3.1, using voiceprint signal processing software to process the collected voiceprint signal and extract voiceprint features; Step S3.2: Analyze the voiceprint features using an acoustic imaging algorithm optimized by eigenvalue decomposition to generate a spectrogram.

6. The method for intelligent diagnosis and analysis of spacecraft vibration test faults according to claim 5, characterized in that: The step S3.1 includes: performing time-frequency analysis and filtering on the voiceprint signal and performing singular value decomposition on the cross-spectral matrix of the signal measured by the array microphone to extract voiceprint features.

7. The method for intelligent diagnosis and analysis of spacecraft vibration test faults according to claim 5, characterized in that: The step S3.2 includes: reconstructing the cross-spectral matrix based on the voiceprint features, and generating the spectrograms of the main sound source and the sub-sound sources at various levels through the cross-spectral delay sum beamforming operation.

8. The method for intelligent diagnosis and analysis of spacecraft vibration test faults according to claim 1, characterized in that: In step S5, time-frequency analysis is performed on the time domain data by Laplace transform to obtain frequency domain features.

9. The method for intelligent diagnosis and analysis of spacecraft vibration test faults according to claim 1, characterized in that: In step S7, the voiceprint features and frequency domain features in the fault image library are extracted by the convolution layer of the convolutional neural network to abstract high-level features, and then the output feature map is reduced in scale according to a certain rule through the downsampling layer to obtain high-level feature signals.

10. An intelligent fault diagnosis and analysis system for spacecraft vibration test products, characterized in that: include: Module M1: hoisting the spacecraft onto the vibration table, setting up the vibration response acquisition device, and reasonably arranging the microphone array around the spacecraft; Module M2 starts the spacecraft vibration test, uses the microphone array to collect the voiceprint signal generated during the spacecraft vibration test, and stores the voiceprint signal in the device; and simultaneously uses the vibration response acquisition device to collect time domain data; Module M3 processes the voiceprint signal, extracts voiceprint features, and generates a spectrogram; Module M4, determining a fault occurrence area based on the sound spectrogram; Module M5, selecting a measurement point in the fault occurrence area, and performing time-frequency analysis on the time domain data of the measurement point to obtain frequency domain features; Module M6, constructing a fault image library based on the spectrogram and the frequency domain features; Module M7, inputs the fault image library into the convolution layer to extract high-level feature signals; Module M8 processes the high-level feature signal through a long short-term memory network to further extract fault features; Module M9: using a multi-scale convolutional memory neural network to train the fault features and generate a fault database; Module M10 inputs the voiceprint features and the frequency domain features into the fault database, and implements intelligent fault diagnosis through fault signal comparison.

Images