CN117171547A - Fault diagnosis method, device, equipment and storage medium based on large model - Google Patents
Fault diagnosis method, device, equipment and storage medium based on large model Download PDFInfo
- Publication number
- CN117171547A CN117171547A CN202311445692.6A CN202311445692A CN117171547A CN 117171547 A CN117171547 A CN 117171547A CN 202311445692 A CN202311445692 A CN 202311445692A CN 117171547 A CN117171547 A CN 117171547A
- Authority
- CN
- China
- Prior art keywords
- model
- signal
- vibration signal
- result
- target
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 58
- 238000003745 diagnosis Methods 0.000 title claims abstract description 55
- 238000012545 processing Methods 0.000 claims abstract description 83
- 238000012549 training Methods 0.000 claims description 98
- 238000000605 extraction Methods 0.000 claims description 59
- 238000001228 spectrum Methods 0.000 claims description 11
- 238000012790 confirmation Methods 0.000 claims description 7
- 238000012512 characterization method Methods 0.000 claims description 4
- 238000013507 mapping Methods 0.000 claims description 4
- 238000012795 verification Methods 0.000 claims description 4
- 230000009466 transformation Effects 0.000 claims description 3
- 230000008569 process Effects 0.000 description 11
- 230000002159 abnormal effect Effects 0.000 description 8
- 238000010586 diagram Methods 0.000 description 6
- 238000004364 calculation method Methods 0.000 description 5
- 238000004891 communication Methods 0.000 description 5
- 238000001514 detection method Methods 0.000 description 5
- 230000008901 benefit Effects 0.000 description 3
- 230000008859 change Effects 0.000 description 3
- 125000004122 cyclic group Chemical group 0.000 description 3
- 238000003058 natural language processing Methods 0.000 description 3
- 238000003062 neural network model Methods 0.000 description 3
- 230000000694 effects Effects 0.000 description 2
- 238000013527 convolutional neural network Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 239000000284 extract Substances 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000007781 pre-processing Methods 0.000 description 1
- 238000012163 sequencing technique Methods 0.000 description 1
- 230000003595 spectral effect Effects 0.000 description 1
Classifications
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
- Y02T90/00—Enabling technologies or technologies with a potential or indirect contribution to GHG emissions mitigation
Landscapes
- Measurement Of Mechanical Vibrations Or Ultrasonic Waves (AREA)
Abstract
The application discloses a fault diagnosis method, device and equipment based on a large model and a storage medium, and belongs to the technical field of data processing. According to the application, the vibration signal of the target to be detected is obtained, the signal processing is carried out through the signal processing model for representing the association relation between the vibration signal and the characteristic description result based on the vibration signal, so that the characteristic description result of the vibration signal is obtained, a user can determine the characteristic description result of the target to be detected based on the characteristic description result, the user does not need to analyze the collected vibration signal, even if the user experience is insufficient or the specialty is poor, the user can simply and intuitively judge the characteristic description result of the target to be detected directly based on the characteristic description result of the vibration signal, the occurrence of misjudgment or omission is avoided, and the accuracy of fault diagnosis can be improved.
Description
Technical Field
The present application relates to the field of data processing technologies, and in particular, to a fault diagnosis method, apparatus, device, and storage medium based on a large model.
Background
At present, in the fault diagnosis process, a method for acquiring a vibration signal of an object to be detected and analyzing the vibration signal is generally adopted, so that a characteristic description result is obtained.
However, at present, the waveform diagram of the collected vibration signal needs to be analyzed manually, and if the manual experience is insufficient or the professional is poor, the situation of misjudgment or missed detection is easy to occur, so that the accuracy of fault diagnosis is reduced.
Disclosure of Invention
The application mainly aims to provide a fault diagnosis method, device, equipment and storage medium based on a large model, and aims to solve the technical problems.
In order to achieve the above object, the present application provides a large model-based fault diagnosis method, comprising the steps of:
acquiring a vibration signal of a target to be detected;
and carrying out signal processing through the signal processing model based on the vibration signal to obtain a feature description result of the vibration signal so as to enable a user to determine the feature description result of the target to be detected based on the feature description result, wherein the signal processing model is used for representing the association relationship between the vibration signal and the feature description result.
Optionally, the signal processing model includes a feature extraction model and a feature description model;
the step of performing signal processing through the signal processing model based on the vibration signal to obtain a feature description result of the vibration signal comprises the following steps:
based on the vibration signal, performing feature extraction processing through a feature extraction model to obtain a target signal feature of the vibration signal;
and carrying out feature description on the target signal features through a feature description model based on the target signal features to obtain feature description results of the vibration signals.
Optionally, the step of obtaining the target signal feature of the vibration signal by performing feature extraction processing through a feature extraction model based on the vibration signal includes:
determining a time domain map and a frequency domain map of the vibration signal through Fourier transformation;
and carrying out feature extraction processing through a feature extraction model based on the time domain spectrum and the frequency domain spectrum to obtain target features of the vibration signal.
Optionally, before the step of performing feature extraction processing through a feature extraction model based on the vibration signal to obtain the target signal feature of the vibration signal, the method further includes:
acquiring a first training sample, wherein the first training sample comprises vibration signals obtained by detecting a plurality of historical targets;
determining target signal characteristic labels of vibration signals corresponding to the historical targets based on the mapping relation between the historical targets and the characteristic labels;
and carrying out iterative training on the first model to be trained based on the first training sample and the target signal characteristic label to obtain a characteristic extraction model.
Optionally, before the step of performing feature description on the target signal feature through a feature description model based on the target signal feature to obtain a feature description result of the vibration signal, the method further includes:
acquiring a second training sample and a first result label thereof, wherein the second training sample comprises a plurality of signal features obtained by performing feature extraction processing through the feature extraction model based on a plurality of historical vibration signals;
and carrying out iterative training on the second model to be trained based on the second training sample and the first result label to obtain a feature description model.
Optionally, after the step of obtaining the feature description result of the vibration signal by performing signal processing through the signal processing model based on the vibration signal, the method further includes:
the feature description result and the corresponding vibration signal are sent to a client for a user to check;
receiving a feature description confirmation result fed back by a user through a client;
determining a target vibration signal corresponding to the erroneous signature verification result.
Optionally, after the step of iteratively training the second model to be trained based on the second training sample and the first result label to obtain the feature description model, the method further includes:
based on the target vibration signal, performing feature extraction processing through the feature extraction model to obtain a third training sample;
determining a second result label corresponding to the third training sample;
and optimizing the feature description model based on the third training sample and the second result label to obtain an optimized feature description model.
In addition, in order to achieve the above object, the present application also provides a large model-based fault diagnosis apparatus including:
the signal acquisition module is used for acquiring a vibration signal of a target to be detected;
the description result determining module is used for performing signal processing through the signal processing model based on the vibration signal to obtain a feature description result of the vibration signal so as to enable a user to determine the feature description result of the target to be detected based on the feature description result, wherein the signal processing model is used for representing the association relation between the vibration signal and the feature description result.
In addition, to achieve the above object, the present application also provides an apparatus comprising: a memory, a processor, and a large model-based fault diagnosis program stored on the memory and executable on the processor, the large model-based fault diagnosis program configured to implement the steps of the large model-based fault diagnosis method as described above.
In addition, in order to achieve the above object, the present application also provides a computer-readable storage medium having stored thereon a large model-based fault diagnosis program which, when executed by a processor, implements the steps of the large model-based fault diagnosis method as described above.
According to the application, the vibration signal of the target to be detected is obtained, the signal processing is carried out through the signal processing model for representing the association relation between the vibration signal and the characteristic description result based on the vibration signal, so that the characteristic description result of the vibration signal is obtained, a user can determine the characteristic description result of the target to be detected based on the characteristic description result, the user does not need to analyze the collected vibration signal, even if the user experience is insufficient or the specialty is poor, the user can simply and intuitively judge the characteristic description result of the target to be detected directly based on the characteristic description result of the vibration signal, the occurrence of misjudgment or omission is avoided, and the accuracy of fault diagnosis can be improved.
Drawings
FIG. 1 is a schematic flow chart of a first embodiment of a large model-based fault diagnosis method according to the present application;
FIG. 2 is a schematic diagram of a second flow chart of a second embodiment of the large model-based fault diagnosis method of the present application;
FIG. 3 is a schematic view of a third flow chart of a third embodiment of a large model-based fault diagnosis method according to the present application;
FIG. 4 is a schematic view of a third embodiment of a large model-based fault diagnosis method according to the present application;
FIG. 5 is a block diagram of an embodiment of a large model-based fault diagnosis apparatus of the present application;
FIG. 6 is a schematic diagram of a device architecture of a hardware operating environment according to an embodiment of the present application.
The achievement of the objects, functional features and advantages of the present application will be further described with reference to the accompanying drawings, in conjunction with the embodiments.
Detailed Description
It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the scope of the application.
Referring to fig. 1, fig. 1 is a schematic flow chart of a first embodiment of a fault diagnosis method based on a large model according to the present application.
In a first embodiment, the large model-based fault diagnosis method includes the steps of:
s10: acquiring a vibration signal of a target to be detected;
it should be noted that, the large model-based fault diagnosis method of the present embodiment is applied to a large model-based fault diagnosis device or a cloud server, which is subordinate to a large model-based fault diagnosis apparatus.
Specifically, in order to reduce the calculation pressure of the local area of the device, the fault diagnosis method based on the large model in the embodiment is applied to the cloud server, and the fault diagnosis method based on the large model can be applied to the whole fault detection scene of the device and/or the fault detection scene of the device component, that is, the whole device and/or the device component is the target to be detected.
In general, faults of the whole equipment and/or the parts of the equipment can be diagnosed by detecting and analyzing changes of vibration signals of the equipment, specifically, vibration signals can be acquired through vibration sensors which are arranged on the whole equipment and/or the parts of the equipment in advance, and the vibration signals are returned through a data acquisition instrument.
S20: and carrying out signal processing through the signal processing model based on the vibration signal to obtain a feature description result of the vibration signal so as to enable a user to determine the feature description result of the target to be detected based on the feature description result, wherein the signal processing model is used for representing the association relationship between the vibration signal and the feature description result.
It can be understood that the detected vibration signal is usually presented in the form of a waveform chart, and for a user with insufficient experience or poor professional, the waveform chart is unsmooth and understandable, and if the user directly analyzes the waveform chart of the collected vibration signal, a situation of misjudgment or missed detection may occur, so that in order to reduce the situation of misjudgment or missed detection caused by insufficient experience or poor professional of the user, the embodiment performs signal processing through the signal processing model based on the vibration signal, so as to provide a feature description result of the vibration signal for the user, and enable the user to intuitively determine the meaning represented by the vibration signal from the feature description result.
Specifically, the feature description result includes contents such as the vibration signal is normal and the vibration signal is abnormal.
It should be noted that, the specific implementation manner of the signal processing model for characterizing the association relationship between the vibration signal and the feature description result may be:
obtaining a target training sample and a target result label thereof, wherein the target training sample comprises vibration signals obtained by detecting a plurality of historical targets, the target result label comprises a feature description result corresponding to the vibration signals in the target training sample, and the target model to be trained is subjected to iterative training based on the target training sample and the target result label thereof to obtain a signal processing model.
Based on the target training sample and the target result label thereof, the target model to be trained is iteratively trained, and the specific implementation manner of obtaining the signal processing model can be as follows:
specifically, the target model to be trained is an initial model with the capability of primarily processing the vibration signal, wherein the accuracy of the target model to be trained is low. Inputting a target training sample into the target to-be-trained model, wherein the mode of processing the vibration signal by the target to-be-trained model is to analyze the association relation between the vibration signal in the target training sample and the corresponding characteristic description result, analyze the characteristic description result corresponding to the vibration signal in the target training sample based on factors such as fluctuation amplitude, peak sharpness and the like of the vibration signal in the target training sample, and perform difference calculation on the characteristic description result and the characteristic description result in the target result label to obtain an error result; because the target model to be trained does not have high accuracy before the incomplete iterative training, an error result can be obtained through difference calculation, or an error result can be obtained through loss function convergence, whether the error result meets the error standard indicated by the preset error threshold range is judged, if the error result does not meet the error standard indicated by the preset error threshold range, the step of inputting the target training sample into the target model to be trained is returned until the training error result meets the error standard indicated by the preset error threshold range, and training is stopped, so that a signal processing model is obtained.
Specifically, the target to-be-trained model is a large model, which may be a cyclic neural network model or a convolutional neural network model, and since the cyclic neural network model has a certain advantage when learning the nonlinear characteristics of the sequence data (the vibration signal belongs to the sequence data), the preferred target to-be-trained model is the cyclic neural network model.
It can be understood that the signal processing model obtained through training in the above manner carries out signal processing on the vibration signal, so that the characteristic description result of the vibration signal can be accurately obtained, the user is not required to analyze the collected vibration signal, even if the user experience is insufficient or the specialty is poor, the user can simply and intuitively judge the characteristic description result of the object to be detected directly based on the characteristic description result of the vibration signal, the occurrence of misjudgment or omission is avoided, and the accuracy of fault diagnosis can be improved.
As shown in fig. 2, a second embodiment of fault diagnosis according to the present application is proposed based on the first embodiment, in this embodiment, the step of performing signal processing through the signal processing model based on the vibration signal to obtain a feature description result of the vibration signal includes:
a1: based on the vibration signal, performing feature extraction processing through a feature extraction model to obtain a target signal feature of the vibration signal;
specifically, since the feature description result determined in the above manner includes contents such as the vibration signal is normal and the vibration signal is abnormal, the user can see only one diagnosis result through the above feature description result, and cannot determine whether the diagnosis result is correct.
In order to solve the problem, the signal processing model of the present embodiment may include a feature extraction model for characterizing an association relationship between a vibration signal and a signal feature, and a feature description model for characterizing an association relationship between a signal feature and a feature description result.
It can be understood that the target signal characteristics of the vibration signal can be extracted through the characteristic extraction model, and more detailed characteristic description results can be obtained based on the characteristics of the target signal characteristics through the characteristic description model.
The specific implementation manner of obtaining the feature extraction model may be:
obtaining a first training sample, wherein the first training sample comprises vibration signals obtained by detecting a plurality of historical targets, determining target signal characteristic labels of vibration signals corresponding to the historical targets based on a mapping relation between the historical targets and the characteristic labels, and performing iterative training on a first model to be trained based on the first training sample and the target signal characteristic labels to obtain a characteristic extraction model.
Specifically, the first model to be trained is an initial model with the capability of initially extracting the vibration signal characteristics, wherein the accuracy of the first model to be trained is low. Inputting a first training sample into the first model to be trained, wherein the characteristic extraction mode of the first model to be trained is as follows: analyzing the association relation between the vibration signals in the first training sample and the corresponding signal characteristics, determining the signal characteristics corresponding to the vibration signals in the first training sample based on factors such as the association degree of a plurality of signal characteristics and the vibration signals in the first training sample, and performing difference calculation on the signal characteristics and the signal characteristics in the characteristic labels to obtain an error result; and judging whether the error result meets the error standard indicated by the preset error threshold range, if the error result does not meet the error standard indicated by the preset error threshold range, returning to the step of inputting the first training sample into the first model to be trained until the training error result meets the error standard indicated by the preset error threshold range, and stopping training to obtain the feature extraction model.
Since the vibration signal not only includes the characteristics in the time domain, if the characteristic change of the vibration signal in the time domain is simply analyzed, but also the characteristic change of the vibration signal in the frequency domain is ignored, the accuracy of the characteristic description result is affected.
Based on the above, before analyzing the association relation between the vibration signal and the corresponding signal characteristic in the first training sample, the first model to be trained may determine the time domain spectrum and the frequency domain spectrum of the vibration signal through fast fourier transform (fast Fourier transform, FFT) or wavelet transform (wavelet transform, WT), and then determine the signal characteristic corresponding to the vibration signal in the first training sample based on factors such as the association degree between the signal characteristic corresponding to the time domain spectrum and the frequency domain spectrum and the vibration signal in the first training sample.
Specifically, the signal features include time domain features including mean, variance, peak, etc., spectral features including frequency, amplitude, phase, etc., and other related features including, for example, kurtosis, skewness, etc.
Because the fault phenomena reflected by different signal features are different, when the first model to be trained extracts the features, the signal features corresponding to the vibration signals in the first training sample need to be determined based on factors such as the association degree of a plurality of signal features and the vibration signals in the first training sample. And carrying out feature extraction processing on the vibration signal based on the feature extraction model obtained through training in the mode, so that the target signal feature of the vibration signal can be accurately obtained. For example, for the fault phenomenon of motor misalignment, the fault phenomenon is most relevant to the change rule of vibration frequency and frequency multiplication, but not all signal features can be extracted because of the fact that the signal features which can be extracted are many, but only several signal features with higher association degrees (which can be higher than a preset association degree threshold value or can be a preset number before association degree sequencing) corresponding to different fault phenomena are extracted.
Specifically, in the practical application process, the implementation manner of obtaining the target signal characteristic of the vibration signal by performing the characteristic extraction processing through the characteristic extraction model based on the vibration signal may be:
determining a time domain map and a frequency domain map of the vibration signal through fast Fourier transform (specifically, decomposing the vibration signal into a time domain component and a frequency domain component, obtaining the time domain map based on the time domain component, obtaining the frequency domain map based on the frequency domain component), and performing feature extraction processing through a feature extraction model obtained through the training based on the time domain map and the frequency domain map, so that accurate target signal features of the vibration signal can be obtained.
Specifically, the target signal characteristic comprises at least one of a time domain characteristic, a frequency spectrum characteristic, and other related characteristics.
A2: and carrying out feature description on the target signal features through a feature description model based on the target signal features to obtain feature description results of the vibration signals.
It can be understood that after obtaining the more accurate target signal characteristics, the characteristic description model is used for carrying out characteristic description on the target signal characteristics, so that the characteristic description result of the vibration signal can be obtained, and specifically, the characteristic description result comprises a fault diagnosis process determined based on the vibration signal corresponding to the target signal characteristics, a preliminary diagnosis result and the like. For example, the fault diagnosis process determined based on the target signal features may be that the fluctuation amplitude of the vibration signal at the 200 th data point is abnormal, the preliminary diagnosis result is that the target to be detected is abnormal, and the like.
Through the above feature description result, the user can determine whether the feature description result is correct or not by checking the waveform diagram (or the time domain map or the frequency domain map) of the vibration signal at the 200 th data point, and can provide more information for the user, so that the user can further determine whether the target to be tested is faulty or not based on the above feature description result and experience, the number of times that the user checks faults for many times when the fault condition is light or misjudgment is reduced, and particularly whether the user checks or not can further judge.
In particular, the implementation of obtaining the feature description model may be:
acquiring a second training sample and a first result label thereof, wherein the second training sample comprises a plurality of signal features obtained by performing feature extraction processing through the feature extraction model based on a plurality of historical vibration signals; and carrying out iterative training on the second model to be trained based on the second training sample and the first result label to obtain a feature description model.
Specifically, the second training sample includes a plurality of signal features obtained by performing feature extraction processing through the feature extraction model based on a plurality of historical vibration signals, the expression form of the plurality of signal features may be a digital vector or a structured data table, the first result label and the target result label may be the same, and the second model to be trained is an initial model with the capability of preliminarily converting the signal features into text.
The second model to be trained has low accuracy, and may be a large language model (Large Language Model, LLM) or a natural language processing model (Natural language processing, NLP).
After a second training sample is input into the second model to be trained, the second model to be trained is characterized in the following manner: analyzing the association relation between the signal characteristics in the second training sample and the corresponding first result label, determining a characteristic description result corresponding to the vibration signals in the second training sample based on the association relation, and performing difference calculation on the characteristic description result and the characteristic description result in the first result label to obtain an error result; and judging whether the error result meets the error standard indicated by the preset error threshold range, if the error result does not meet the error standard indicated by the preset error threshold range, returning to the step of inputting a second training sample into the second model to be trained until the training error result meets the error standard indicated by the preset error threshold range, and stopping training to obtain the feature description model.
Specifically, in order to improve accuracy and understandability of the feature description result, it is preferable that the second model to be trained is a large language model, and after the step of determining the feature description result corresponding to the vibration signal in the second training sample based on the association relationship, the feature description result may be checked and edited by a Promopt part in the large language model, so as to improve accuracy and understandability of the feature description result.
Therefore, the feature description model trained based on the above manner can accurately determine the feature description result of the target feature signal, that is, the feature description result of the vibration signal.
It should be noted that, after the training samples (including the target training sample, the first training sample and the second training sample) are obtained in the above scheme, data preprocessing is required to be performed through methods of data standardization, normalization, noise removal and the like, so that accuracy of model training is improved.
In this embodiment, the feature description result can be better determined through the feature extraction model and the feature description model, and the diagnosis process and the diagnosis conclusion can be provided for the user more intuitively and in detail, so that the use experience of the user is improved.
As shown in fig. 3, a third embodiment of the fault diagnosis of the present application is proposed based on the first embodiment and the second embodiment, in this embodiment, after the step of obtaining the feature description result of the vibration signal by performing signal processing through the signal processing model based on the vibration signal, the method further includes:
b1: the feature description result and the corresponding vibration signal are sent to a client for a user to check;
specifically, the feature description result includes a normal feature description result and an abnormal feature description result, as shown in fig. 4, both the normal feature description result and the abnormal feature description result may be sent to the client, so that the user may confirm whether the feature description result is correct through the client, where an abnormal alarm may be sent when both the abnormal feature description result is sent to the client.
For example, the normal characterization result may be: the vibration signal exhibits a peak value with a frequency in the range xHz to yHz, wherein the highest peak value reaches a maximum at time z, identifying that the motor is operating properly.
The feature description result of the anomaly may be: the vibration signal shows a frequency occurrence of a sideband, indicating that the motor is not running smoothly, etc.
B2: receiving a feature description confirmation result fed back by a user through a client;
after the user checks the feature description result and the corresponding vibration signal in the client, the user can click correctly or incorrectly and submit feedback feature description confirmation result.
B3: determining a target vibration signal corresponding to the erroneous signature verification result.
After receiving the feature description confirmation result fed back by the user through the client, the target vibration signal corresponding to the feature description confirmation result considered by the user as an error can be stored in the training database of the cloud server, namely, the target vibration signal is a vibration signal which is easy to misdiagnose or misjudge by the user, and the accuracy of feature description of the feature description model on the signal features of the vibration signal needs to be further enhanced.
Specifically, after the step of performing iterative training on the second model to be trained based on the second training sample and the first result label to obtain a feature description model, feature extraction processing may be performed through the feature extraction model based on the target vibration signal in the training database to obtain a third training sample (including extracted signal features); determining a second result label corresponding to the third training sample (can be to obtain a correct feature description result redetermined based on expert); and optimizing the feature description model based on the third training sample and the second result label to obtain an optimized feature description model.
It can be understood that, based on the third training sample and the second result label, the iterative training is continuously performed on the feature description model, so as to obtain an optimized feature description model meeting a preset precision condition.
In the subsequent application process, the target to be detected can be processed based on the optimized feature description model, so that a more accurate feature description result is obtained, the fault diagnosis efficiency is improved, and because the training sample is larger, more resources are occupied in the model training process, but because the fault diagnosis method based on the large model is applied to the cloud server, the influence of resource pressure on the fault diagnosis equipment based on the large model cannot be caused locally.
In addition, an embodiment of the present application further provides a large model-based fault diagnosis apparatus, referring to fig. 5, including:
a signal acquisition module 10 for acquiring a vibration signal of a target to be measured;
the description result determining module 20 is configured to perform signal processing through the signal processing model based on the vibration signal, so as to obtain a feature description result of the vibration signal, so that a user can determine a feature description result of the object to be measured based on the feature description result, where the signal processing model is used to characterize an association relationship between the vibration signal and the feature description result.
According to the method, the vibration signal of the target to be detected is obtained, the signal processing is carried out through the signal processing model for representing the association relation between the vibration signal and the characteristic description result based on the vibration signal, the characteristic description result of the vibration signal is obtained, a user can determine the characteristic description result of the target to be detected based on the characteristic description result, the user does not need to analyze the collected vibration signal, even if the user experience is insufficient or the specialty is poor, the user can simply and intuitively judge the characteristic description result of the target to be detected directly based on the characteristic description result of the vibration signal, misjudgment or omission is avoided, and the accuracy of fault diagnosis can be improved.
It should be noted that each module in the above apparatus may be used to implement each step in the above method, and achieve a corresponding technical effect, which is not described herein again.
Referring to fig. 6, fig. 6 is a schematic structural diagram of a device of a hardware running environment according to an embodiment of the present application.
As shown in fig. 6, the apparatus may include: a processor 1001, such as a CPU, a communication bus 1002, a user interface 1003, a network interface 1004, and a memory 1005. Wherein the communication bus 1002 is used to enable connected communication between these components. The user interface 1003 may include a Display, an input unit such as a Keyboard (Keyboard), and the optional user interface 1003 may further include a standard wired interface, a wireless interface. The network interface 1004 may optionally include a standard wired interface, a wireless interface (e.g., WI-FI interface). The memory 1005 may be a high-speed RAM memory or a stable memory (non-volatile memory), such as a disk memory. The memory 1005 may also optionally be a storage device separate from the processor 1001 described above.
It will be appreciated by those skilled in the art that the structure shown in fig. 6 is not limiting of the apparatus and may include more or fewer components than shown, or certain components may be combined, or a different arrangement of components.
As shown in fig. 6, an operating system, a network communication module, a user interface module, and a large model-based fault diagnosis program may be included in a memory 1005 as one type of computer storage medium.
In the device shown in fig. 6, the network interface 1004 is mainly used for data communication with an external network; the user interface 1003 is mainly used for receiving an input instruction of a user; the apparatus calls a large model-based fault diagnosis program stored in the memory 1005 by the processor 1001, and performs the following operations:
acquiring a vibration signal of a target to be detected;
and carrying out signal processing through the signal processing model based on the vibration signal to obtain a feature description result of the vibration signal so as to enable a user to determine the feature description result of the target to be detected based on the feature description result, wherein the signal processing model is used for representing the association relationship between the vibration signal and the feature description result.
Further, the signal processing model includes a feature extraction model and a feature description model, and the processor 1001 may call the large model-based fault diagnosis program stored in the memory 1005, and further perform the following operations:
based on the vibration signal, performing feature extraction processing through a feature extraction model to obtain a target signal feature of the vibration signal;
and carrying out feature description on the target signal features through a feature description model based on the target signal features to obtain feature description results of the vibration signals.
Further, the processor 1001 may call the large model-based fault diagnosis program stored in the memory 1005, and further perform the following operations:
determining a time domain map and a frequency domain map of the vibration signal through Fourier transformation;
and carrying out feature extraction processing through a feature extraction model based on the time domain spectrum and the frequency domain spectrum to obtain target features of the vibration signal.
Further, the processor 1001 may call the large model-based fault diagnosis program stored in the memory 1005, and further perform the following operations:
acquiring a first training sample, wherein the first training sample comprises vibration signals obtained by detecting a plurality of historical targets;
determining target signal characteristic labels of vibration signals corresponding to the historical targets based on the mapping relation between the historical targets and the characteristic labels;
and carrying out iterative training on the first model to be trained based on the first training sample and the target signal characteristic label to obtain a characteristic extraction model.
Further, the processor 1001 may call the large model-based fault diagnosis program stored in the memory 1005, and further perform the following operations:
acquiring a second training sample and a first result label thereof, wherein the second training sample comprises a plurality of signal features obtained by performing feature extraction processing through the feature extraction model based on a plurality of historical vibration signals;
and carrying out iterative training on the second model to be trained based on the second training sample and the first result label to obtain a feature description model.
Further, the processor 1001 may call the large model-based fault diagnosis program stored in the memory 1005, and further perform the following operations:
the feature description result and the corresponding vibration signal are sent to a client for a user to check;
receiving a feature description confirmation result fed back by a user through a client;
determining a target vibration signal corresponding to the erroneous signature verification result.
Further, the processor 1001 may call the large model-based fault diagnosis program stored in the memory 1005, and further perform the following operations:
based on the target vibration signal, performing feature extraction processing through the feature extraction model to obtain a third training sample;
determining a second result label corresponding to the third training sample;
and optimizing the feature description model based on the third training sample and the second result label to obtain an optimized feature description model.
According to the method, the vibration signal of the target to be detected is obtained, the signal processing is carried out through the signal processing model for representing the association relation between the vibration signal and the characteristic description result based on the vibration signal, the characteristic description result of the vibration signal is obtained, a user can determine the characteristic description result of the target to be detected based on the characteristic description result, the user does not need to analyze the collected vibration signal, even if the user experience is insufficient or the specialty is poor, the user can simply and intuitively judge the characteristic description result of the target to be detected directly based on the characteristic description result of the vibration signal, misjudgment or omission is avoided, and the accuracy of fault diagnosis can be improved.
In addition, an embodiment of the present application also proposes a computer-readable storage medium having stored thereon a large model-based fault diagnosis program that, when executed by a processor, realizes the following operations:
acquiring a vibration signal of a target to be detected;
and carrying out signal processing through the signal processing model based on the vibration signal to obtain a feature description result of the vibration signal so as to enable a user to determine the feature description result of the target to be detected based on the feature description result, wherein the signal processing model is used for representing the association relationship between the vibration signal and the feature description result.
According to the method, the vibration signal of the target to be detected is obtained, the signal processing is carried out through the signal processing model for representing the association relation between the vibration signal and the characteristic description result based on the vibration signal, the characteristic description result of the vibration signal is obtained, a user can determine the characteristic description result of the target to be detected based on the characteristic description result, the user does not need to analyze the collected vibration signal, even if the user experience is insufficient or the specialty is poor, the user can simply and intuitively judge the characteristic description result of the target to be detected directly based on the characteristic description result of the vibration signal, misjudgment or omission is avoided, and the accuracy of fault diagnosis can be improved.
It should be noted that, when the computer readable storage medium is executed by the processor, each step in the method may be further implemented, and meanwhile, the corresponding technical effects are achieved, which is not described herein.
It should be noted that, in this document, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or system that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or system. Without further limitation, an element defined by the phrase "comprising one … …" does not exclude the presence of other like elements in a process, method, article, or system that comprises the element.
The foregoing embodiment numbers of the present application are merely for the purpose of description, and do not represent the advantages or disadvantages of the embodiments.
From the above description of embodiments, it will be clear to a person skilled in the art that the above embodiment method may be implemented by means of software plus a necessary general hardware platform, but may of course also be implemented by means of hardware, but in many cases the former is a preferred embodiment. Based on this understanding, the technical solution of the present application may be embodied essentially or in a part contributing to the prior art in the form of a software product stored in a storage medium (e.g. ROM/RAM, magnetic disk, optical disk) as described above, comprising instructions for causing a terminal device (which may be a mobile phone, a computer, a server, an air conditioner, or a network device, etc.) to perform the method according to the embodiments of the present application.
The foregoing description is only of the preferred embodiments of the present application, and is not intended to limit the scope of the application, but rather is intended to cover any equivalents of the structures or equivalent processes disclosed herein or in the alternative, which may be employed directly or indirectly in other related arts.
Claims (10)
1. A large model-based fault diagnosis method, characterized in that the large model-based fault diagnosis method comprises the steps of:
acquiring a vibration signal of a target to be detected;
and carrying out signal processing through the signal processing model based on the vibration signal to obtain a feature description result of the vibration signal so as to enable a user to determine the feature description result of the target to be detected based on the feature description result, wherein the signal processing model is used for representing the association relationship between the vibration signal and the feature description result.
2. The large model-based fault diagnosis method according to claim 1, wherein said signal processing model includes a feature extraction model and a feature description model;
the step of performing signal processing through the signal processing model based on the vibration signal to obtain a feature description result of the vibration signal comprises the following steps:
based on the vibration signal, performing feature extraction processing through a feature extraction model to obtain a target signal feature of the vibration signal;
and carrying out feature description on the target signal features through a feature description model based on the target signal features to obtain feature description results of the vibration signals.
3. The large model-based fault diagnosis method according to claim 2, wherein the step of obtaining the target signal characteristics of the vibration signal by performing a characteristic extraction process through a characteristic extraction model based on the vibration signal comprises:
determining a time domain map and a frequency domain map of the vibration signal through Fourier transformation;
and carrying out feature extraction processing through a feature extraction model based on the time domain spectrum and the frequency domain spectrum to obtain target features of the vibration signal.
4. The large model-based fault diagnosis method according to claim 2, wherein, before the step of obtaining the target signal characteristics of the vibration signal by performing a characteristic extraction process by a characteristic extraction model based on the vibration signal, the method further comprises:
acquiring a first training sample, wherein the first training sample comprises vibration signals obtained by detecting a plurality of historical targets;
determining target signal characteristic labels of vibration signals corresponding to the historical targets based on the mapping relation between the historical targets and the characteristic labels;
and carrying out iterative training on the first model to be trained based on the first training sample and the target signal characteristic label to obtain a characteristic extraction model.
5. The large model-based fault diagnosis method according to claim 4, wherein said step of characterizing said target signal characteristics by a characterization model based on said target signal characteristics, further comprises, before the step of obtaining a characterization result of said vibration signal:
acquiring a second training sample and a first result label thereof, wherein the second training sample comprises a plurality of signal features obtained by performing feature extraction processing through the feature extraction model based on a plurality of historical vibration signals;
and carrying out iterative training on the second model to be trained based on the second training sample and the first result label to obtain a feature description model.
6. The large model-based fault diagnosis method according to claim 5, wherein said step of obtaining a characterization result of said vibration signal by signal processing through said signal processing model based on said vibration signal further comprises:
the feature description result and the corresponding vibration signal are sent to a client for a user to check;
receiving a feature description confirmation result fed back by a user through a client;
determining a target vibration signal corresponding to the erroneous signature verification result.
7. The large model-based fault diagnosis method according to claim 6, wherein, after said step of iteratively training a second model to be trained based on said second training sample and said first result label to obtain a feature description model, said method further comprises:
based on the target vibration signal, performing feature extraction processing through the feature extraction model to obtain a third training sample;
determining a second result label corresponding to the third training sample;
and optimizing the feature description model based on the third training sample and the second result label to obtain an optimized feature description model.
8. A large model-based fault diagnosis apparatus, characterized in that the large model-based fault diagnosis apparatus includes:
the signal acquisition module is used for acquiring a vibration signal of a target to be detected;
the description result determining module is used for performing signal processing through the signal processing model based on the vibration signal to obtain a feature description result of the vibration signal so as to enable a user to determine the feature description result of the target to be detected based on the feature description result, wherein the signal processing model is used for representing the association relation between the vibration signal and the feature description result.
9. An apparatus, the apparatus comprising: a memory, a processor, and a large model-based fault diagnosis program stored on the memory and executable on the processor, the large model-based fault diagnosis program configured to implement the steps of the large model-based fault diagnosis method of any one of claims 1 to 7.
10. A computer-readable storage medium, characterized in that the computer-readable storage medium has stored thereon a large model-based fault diagnosis program, which when executed by a processor, implements the steps of the large model-based fault diagnosis method according to any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311445692.6A CN117171547B (en) | 2023-11-02 | Fault diagnosis method, device, equipment and storage medium based on large model |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311445692.6A CN117171547B (en) | 2023-11-02 | Fault diagnosis method, device, equipment and storage medium based on large model |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN117171547A true CN117171547A (en) | 2023-12-05 |
| CN117171547B CN117171547B (en) | 2024-05-24 |
Family
ID=
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114548254A (en) * | 2022-02-17 | 2022-05-27 | 中国科学技术大学先进技术研究院 | Equipment fault classification method, device, equipment and medium |
| CN115238740A (en) * | 2022-07-22 | 2022-10-25 | 中煤科工集团信息技术有限公司 | Fault diagnosis method and device based on time-frequency distribution and deep learning |
| CN116186520A (en) * | 2023-03-10 | 2023-05-30 | 沈阳建筑大学 | Multi-task combined high-voltage isolating switch fault diagnosis method |
| CN116223009A (en) * | 2023-03-07 | 2023-06-06 | 国网江苏省电力有限公司超高压分公司 | Fault diagnosis method, system, device and medium for main circulating pump in converter valve |
| CN116304784A (en) * | 2022-11-29 | 2023-06-23 | 中联重科股份有限公司 | Hydraulic cylinder fault diagnosis model training method and hydraulic cylinder fault diagnosis method |
| CN116659856A (en) * | 2023-04-28 | 2023-08-29 | 北京科锐特科技有限公司 | Fault diagnosis method of motor bearing based on WPT-1DCNN |
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114548254A (en) * | 2022-02-17 | 2022-05-27 | 中国科学技术大学先进技术研究院 | Equipment fault classification method, device, equipment and medium |
| CN115238740A (en) * | 2022-07-22 | 2022-10-25 | 中煤科工集团信息技术有限公司 | Fault diagnosis method and device based on time-frequency distribution and deep learning |
| CN116304784A (en) * | 2022-11-29 | 2023-06-23 | 中联重科股份有限公司 | Hydraulic cylinder fault diagnosis model training method and hydraulic cylinder fault diagnosis method |
| CN116223009A (en) * | 2023-03-07 | 2023-06-06 | 国网江苏省电力有限公司超高压分公司 | Fault diagnosis method, system, device and medium for main circulating pump in converter valve |
| CN116186520A (en) * | 2023-03-10 | 2023-05-30 | 沈阳建筑大学 | Multi-task combined high-voltage isolating switch fault diagnosis method |
| CN116659856A (en) * | 2023-04-28 | 2023-08-29 | 北京科锐特科技有限公司 | Fault diagnosis method of motor bearing based on WPT-1DCNN |
Non-Patent Citations (1)
| Title |
|---|
| 张庆 等: "基于支持向量域描述的多故障诊断动态模型", 西安交通大学学报, no. 05, 10 May 2007 (2007-05-10) * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US9489292B2 (en) | Devices and methods for acquiring abnormal information | |
| CN111382740B (en) | Text picture analysis method, text picture analysis device, computer equipment and storage medium | |
| CN110716843B (en) | System fault analysis processing method and device, storage medium and electronic equipment | |
| CN110781612A (en) | Fault diagnosis method and device for ball screw, computer device and storage medium | |
| CN112268696B (en) | Clutch detection method, device, storage medium and device | |
| CN117171547B (en) | Fault diagnosis method, device, equipment and storage medium based on large model | |
| CN113468035A (en) | Log anomaly detection method and device, training method and device and electronic equipment | |
| CN112508388A (en) | Method and system for inputting product quality detection data, server side and storage medium | |
| CN117171547A (en) | Fault diagnosis method, device, equipment and storage medium based on large model | |
| CN113447572B (en) | Steel rail flaw detection method, electronic device, steel rail flaw detection vehicle and readable storage medium | |
| CN115629930A (en) | Fault detection method, device and equipment based on DSP system and storage medium | |
| CN115859065A (en) | Model evaluation method, device, equipment and storage medium | |
| CN112989256B (en) | Method and device for identifying web fingerprint in response information | |
| CN110083807B (en) | Contract modification influence automatic prediction method, device, medium and electronic equipment | |
| CN110046067B (en) | Interface testing method and device | |
| CN113239075A (en) | Construction data self-checking method and system | |
| CN112242929A (en) | Log detection method and device | |
| CN115398442A (en) | Device and automation method for evaluating sensor measurement values and use of the device | |
| CN115588439B (en) | Fault detection method and device of voiceprint acquisition device based on deep learning | |
| CN114580982B (en) | Method, device and equipment for evaluating data quality of industrial equipment | |
| CN111953544B (en) | Fault detection method, device, equipment and storage medium of server | |
| US20240004747A1 (en) | Processor System and Failure Diagnosis Method | |
| KR102401296B1 (en) | Estimation calculation system using artificial intelligence and its method and display module for monitoring it | |
| CN114723666A (en) | Vehicle part state detection method, device, equipment and storage medium | |
| CN117972508A (en) | Fault diagnosis method, device, terminal equipment and storage medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant |