Rotating equipment state diagnosis method and device, storage medium and computer device
By collecting sound and vibration data from rotating equipment, calculating cosine similarity and feature fusion vectors, and using a fault diagnosis model to determine the equipment status, the problem of large diagnostic errors in existing technologies is solved, and more accurate status diagnosis is achieved.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- CHINA NUCLEAR IND MAINTENANCE
- Filing Date
- 2023-02-23
- Publication Date
- 2026-06-26
AI Technical Summary
Existing diagnostic technologies for rotating equipment cannot fully reflect the true operating status of the equipment, resulting in significant errors in the diagnostic results.
The system collects sound and vibration data from rotating equipment, obtains sound and vibration feature vectors, calculates cosine similarity and feature fusion vectors, and uses a fault diagnosis model to determine the operating condition of the equipment.
It improves the accuracy of rotating equipment status diagnosis, comprehensively reflects the actual operating status of the equipment through multimodal information fusion, and reduces diagnostic errors.
Smart Images
Figure CN116164972B_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of fault diagnosis technology, and in particular to a method and apparatus for diagnosing the condition of rotating equipment, a storage medium, and a computer device. Background Technology
[0002] The continuous development of modern industrial technology has led to the widespread application of rotating machinery in many important industrial sectors, such as power generation, aerospace, and rail transportation. Rotating equipment is a key component of the shaft system support in mechanical equipment. In actual working conditions, since most mechanical equipment rotates periodically under varying operating conditions, even small wear or defects can deteriorate into malfunctions over long-term operation, causing significant property damage. Its normal operating condition directly affects the operation of the mechanical equipment. Existing related technologies for rotating equipment condition diagnosis are relatively limited and cannot comprehensively reflect the true operating state of rotating equipment, resulting in significant errors in the diagnostic results. Summary of the Invention
[0003] In view of this, this application provides a method and apparatus for diagnosing the condition of rotating equipment, a storage medium, and a computer device, which helps to improve the accuracy of diagnosing the condition of rotating equipment.
[0004] According to one aspect of this application, a method for diagnosing the condition of a rotating device is provided, the method comprising:
[0005] Collect sound and vibration data of the rotating device in its working state, and obtain the sound feature vector corresponding to the sound data and the vibration feature vector corresponding to the vibration data;
[0006] Based on the image space of the sound feature vector and the vibration feature vector, the sound cosine similarity is obtained, and based on the image space of the vibration feature vector and the sound feature vector, the vibration cosine similarity is obtained.
[0007] The feature fusion vector is obtained based on the sound cosine similarity and the vibration cosine similarity;
[0008] The operating condition of the rotating equipment is determined based on the feature fusion vector.
[0009] Optionally, obtaining the feature fusion vector based on the sound cosine similarity and the vibration cosine similarity includes:
[0010] Determine whether the sound cosine similarity and the vibration cosine similarity are equal;
[0011] When the sound cosine similarity and the vibration cosine similarity are equal, a feature fusion vector is calculated based on the sound feature vector, the vibration feature vector, and preset parameters.
[0012] When the sound cosine similarity and the vibration cosine similarity are not equal, a feature fusion vector is calculated based on the sound feature vector, the vibration feature vector, the image space vector of the sound feature vector, and the image space vector of the vibration feature vector.
[0013] Optionally, obtaining the sound cosine similarity based on the image space of the sound feature vector and the vibration feature vector, and obtaining the vibration cosine similarity based on the image space of the vibration feature vector and the sound feature vector, includes:
[0014] The sound feature vector is mapped to the image space of the vibration feature vector to obtain a first spatial mapping vector. Based on the first spatial mapping vector and the sound feature vector, the sound cosine similarity is obtained by multiplication.
[0015] The vibration feature vector is mapped to the image space of the sound feature vector to obtain a second spatial mapping vector. Based on the second spatial mapping vector and the vibration feature vector, the vibration cosine similarity is obtained by multiplication.
[0016] Optionally, determining the operating state of the rotating equipment based on the feature fusion vector includes:
[0017] Based on the feature fusion vector, the operating condition of the rotating equipment is obtained using a fault diagnosis model.
[0018] The sound data, vibration data, and operating condition status are saved to the database.
[0019] Optionally, obtaining the operating condition of the rotating equipment based on the feature fusion vector using a fault diagnosis model includes:
[0020] Based on the feature fusion vector, the preliminary diagnostic status of the rotating equipment and the predicted probability value of the preliminary diagnostic status are obtained using the fault diagnosis model.
[0021] When the predicted probability value is greater than or equal to the preset probability value, the operating condition of the rotating equipment is determined to be the preliminary diagnostic state.
[0022] When the predicted probability value is less than the preset probability value, the operating condition of the rotating equipment is determined to be normal.
[0023] Optionally, obtaining the sound feature vector corresponding to the sound data and the vibration feature vector corresponding to the vibration data includes:
[0024] The sound data is preprocessed, and a fast Fourier transform is performed on the preprocessed sound data to generate a speech spectrogram. Sound pixels are obtained based on the speech spectrogram, and sound feature vectors of the sound data are obtained based on the sound pixels.
[0025] The vibration data is preprocessed, and a fast Fourier transform is performed on the preprocessed vibration data to generate a vibration spectrum. Vibration pixels are obtained based on the vibration spectrum, and the vibration feature vector of the vibration data is obtained based on the vibration pixels.
[0026] Optionally, before acquiring the sound and vibration data of the rotating device in its working state, the method further includes:
[0027] Each working condition is used as a preliminary diagnostic state label, and a feature fusion vector sample of the working condition is obtained based on the sound data sample and vibration data sample corresponding to the working condition.
[0028] The model is trained based on the preliminary diagnostic status labels and the feature fusion vector samples to obtain a fault diagnosis model.
[0029] According to another aspect of this application, a rotating equipment condition diagnostic device is provided, the device comprising:
[0030] The acquisition module is used to acquire sound data and vibration data of the rotating equipment in the working state, and to obtain the sound feature vector corresponding to the sound data and the vibration feature vector corresponding to the vibration data.
[0031] The acquisition module is used to obtain sound cosine similarity based on the image space of the sound feature vector and the vibration feature vector, and to obtain vibration cosine similarity based on the image space of the vibration feature vector and the sound feature vector.
[0032] The fusion module is used to obtain a feature fusion vector based on the sound cosine similarity and the vibration cosine similarity;
[0033] The determination module is used to determine the operating condition of the rotating equipment based on the feature fusion vector.
[0034] Optionally, the fusion module is further configured to:
[0035] Determine whether the sound cosine similarity and the vibration cosine similarity are equal;
[0036] When the sound cosine similarity and the vibration cosine similarity are equal, a feature fusion vector is calculated based on the sound feature vector, the vibration feature vector, and preset parameters.
[0037] When the sound cosine similarity and the vibration cosine similarity are not equal, a feature fusion vector is calculated based on the sound feature vector, the vibration feature vector, the image space vector of the sound feature vector, and the image space vector of the vibration feature vector.
[0038] Optionally, the acquisition module is further configured to:
[0039] The sound feature vector is mapped to the image space of the vibration feature vector to obtain a first spatial mapping vector. Based on the first spatial mapping vector and the sound feature vector, the sound cosine similarity is obtained by multiplication.
[0040] The vibration feature vector is mapped to the image space of the sound feature vector to obtain a second spatial mapping vector. Based on the second spatial mapping vector and the vibration feature vector, the vibration cosine similarity is obtained by multiplication.
[0041] Optionally, the determining module is further configured to:
[0042] Based on the feature fusion vector, the operating condition of the rotating equipment is obtained using a fault diagnosis model.
[0043] The sound data, vibration data, and operating condition status are saved to the database.
[0044] Optionally, the determining module is further configured to:
[0045] Based on the feature fusion vector, the preliminary diagnostic status of the rotating equipment and the predicted probability value of the preliminary diagnostic status are obtained using the fault diagnosis model.
[0046] When the predicted probability value is greater than or equal to the preset probability value, the operating condition of the rotating equipment is determined to be the preliminary diagnostic state.
[0047] When the predicted probability value is less than the preset probability value, the operating condition of the rotating equipment is determined to be normal.
[0048] Optionally, the acquisition module is further configured to:
[0049] The sound data is preprocessed, and a fast Fourier transform is performed on the preprocessed sound data to generate a speech spectrogram. Sound pixels are obtained based on the speech spectrogram, and sound feature vectors of the sound data are obtained based on the sound pixels.
[0050] The vibration data is preprocessed, and a fast Fourier transform is performed on the preprocessed vibration data to generate a vibration spectrum. Vibration pixels are obtained based on the vibration spectrum, and the vibration feature vector of the vibration data is obtained based on the vibration pixels.
[0051] Optionally, the device further includes: a training module, used for:
[0052] Each working condition is used as a preliminary diagnostic state label, and a feature fusion vector sample of the working condition is obtained based on the sound data sample and vibration data sample corresponding to the working condition.
[0053] The model is trained based on the preliminary diagnostic status labels and the feature fusion vector samples to obtain a fault diagnosis model.
[0054] According to another aspect of this application, a storage medium is provided that stores a computer program thereon, which, when executed by a processor, implements the above-described method for diagnosing the state of rotating equipment.
[0055] According to another aspect of this application, a computer device is provided, including a storage medium, a processor, and a computer program stored on the storage medium and executable on the processor, wherein the processor executes the program to implement the above-described rotating equipment state diagnosis method.
[0056] By employing the above technical solution, this application provides a method and apparatus for diagnosing the condition of rotating equipment, a storage medium, and a computer device. This method collects sound and vibration data of the rotating equipment during its operating state, and obtains the sound feature vector corresponding to the sound data and the vibration feature vector corresponding to the vibration data. Based on the image space of the sound and vibration feature vectors, it obtains the sound cosine similarity and the vibration cosine similarity. Based on the sound and vibration cosine similarity, it obtains a feature fusion vector. Based on the feature fusion vector, it determines the operating state of the rotating equipment. This achieves the effect of diagnosing the condition of rotating equipment by fusing the sound feature vector corresponding to the sound data and the vibration feature vector corresponding to the vibration data to generate a feature fusion vector, thereby improving the accuracy of rotating equipment condition diagnosis.
[0057] The above description is only an overview of the technical solution of this application. In order to better understand the technical means of this application and to implement it in accordance with the contents of the specification, and to make the above and other objects, features and advantages of this application more obvious and understandable, the following are specific embodiments of this application. Attached Figure Description
[0058] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings:
[0059] Figure 1A flowchart illustrating a method for diagnosing the condition of a rotating device according to an embodiment of this application is shown.
[0060] Figure 2 A flowchart illustrating another method for diagnosing the condition of rotating equipment provided in an embodiment of this application is shown;
[0061] Figure 3 A schematic diagram of the structure of a rotating equipment condition diagnosis device provided in an embodiment of this application is shown. Detailed Implementation
[0062] The present application will be described in detail below with reference to the accompanying drawings and embodiments. It should be noted that, unless otherwise specified, the embodiments and features described in the embodiments of the present application can be combined with each other.
[0063] This embodiment provides a method for diagnosing the condition of rotating equipment, such as... Figure 1 As shown, the method includes:
[0064] Step 101: Collect sound data and vibration data of the rotating device in working condition, and obtain the sound feature vector corresponding to the sound data and the vibration feature vector corresponding to the vibration data.
[0065] The above embodiments of this application can be applied to the condition diagnosis of rotating equipment, and can be used to monitor the condition of rotating equipment in real time. First, a professional sound acquisition device or acquisition card is used to collect sound data and vibration data of the rotating equipment when it is in operation. Then, the sound feature vectors corresponding to the sound data and the vibration feature vectors corresponding to the vibration data are obtained. The rotating equipment includes pumps, compressors, and other rotating equipment, in preparation for subsequent condition diagnosis.
[0066] Step 102: Obtain sound cosine similarity based on the image space of the sound feature vector and the vibration feature vector, and obtain vibration cosine similarity based on the image space of the vibration feature vector and the sound feature vector.
[0067] Next, based on the aforementioned sound feature vectors and vibration feature vectors, image spaces for sound feature vectors and vibration feature vectors are constructed respectively. Then, based on the image spaces of the aforementioned sound feature vectors and vibration feature vectors, sound cosine similarity is calculated, and vibration cosine similarity is calculated based on the image spaces of the aforementioned vibration feature vectors and sound feature vectors, in order to facilitate subsequent state diagnosis.
[0068] Step 103: Obtain the feature fusion vector based on the sound cosine similarity and the vibration cosine similarity;
[0069] Step 104: Determine the operating condition of the rotating equipment based on the feature fusion vector.
[0070] Next, based on the aforementioned sound cosine similarity and vibration cosine similarity, a feature fusion vector is calculated. The two feature vectors are then fused using multimodal information fusion to obtain a feature fusion vector. Finally, the operating condition of the rotating equipment is determined based on the aforementioned feature fusion vector. The entire process is simple and novel, improving upon traditional diagnostic methods for rotating equipment condition diagnosis and achieving simplified condition diagnosis of rotating equipment.
[0071] By applying the technical solution of this embodiment, sound data and vibration data of a rotating device in its working state are collected, and sound feature vectors and vibration feature vectors corresponding to the sound data and vibration data are obtained. Based on the image space of the sound and vibration feature vectors, sound cosine similarity is obtained, and based on the image space of the vibration and sound feature vectors, vibration cosine similarity is obtained. A feature fusion vector is obtained based on the sound and vibration cosine similarity. The operating state of the rotating device is determined based on the feature fusion vector. This achieves the effect of diagnosing the state of the rotating device by fusing the sound feature vectors corresponding to the sound data and the vibration feature vectors corresponding to the vibration data, thereby improving the accuracy of rotating device state diagnosis.
[0072] Furthermore, as a refinement and extension of the specific implementation of the above embodiments, and to fully illustrate the specific implementation process of this embodiment, another method for diagnosing the condition of rotating equipment is provided, such as... Figure 2 As shown, the method includes:
[0073] Step 201: Collect sound data and vibration data of the rotating equipment in working condition; preprocess the sound data; perform a Fast Fourier Transform on the preprocessed sound data to generate a sound spectrum; obtain sound pixels based on the sound spectrum; and obtain the sound feature vector of the sound data based on the sound pixels. Similarly, preprocess the vibration data; perform a Fast Fourier Transform on the preprocessed vibration data to generate a vibration spectrum; obtain vibration pixels based on the vibration spectrum; and obtain the vibration feature vector of the vibration data based on the vibration pixels.
[0074] In the above embodiments of this application, firstly, sound data and vibration data of the rotating equipment in its working state are collected. The rotating equipment includes sliding bearings, joint bearings, and rolling bearings, etc. Then, the sound data is preprocessed, including pre-emphasis, framing, and windowing operations to remove invalid data and reduce its impact on the diagnostic status. A fast Fourier transform is performed on the preprocessed sound data to generate a sound spectrogram, and sound pixels are obtained from the sound spectrogram according to preset rules. Based on the sound pixels, a sound feature vector is obtained from the sound data. Next, the vibration data is preprocessed, including trend term removal and windowing operations. A fast Fourier transform is performed on the preprocessed vibration data to generate a vibration spectrogram to unify it with the sound data. Vibration pixels are obtained from the vibration spectrogram according to preset rules, and a vibration feature vector is obtained from the vibration pixels. Specifically, the sound feature vector F is obtained using the following formula. x and vibration eigenvector F y :
[0075] F x =Encode r x
[0076] F y =Encode r y
[0077] Among them, F x F represents the sound feature vector. y Represents the vibration eigenvector, r x Represents a sound pixel, r y This represents a vibration pixel. Encode is a function in the Python programming language, which here is used to convert acoustic pixels or vibration pixels into binary data. By preprocessing the sound data and vibration data and using Fast Fourier Transform to align the features of the two types of data, the accuracy of the state diagnosis data is improved, preparing for the subsequent state diagnosis.
[0078] Step 202: Map the sound feature vector to the image space of the vibration feature vector to obtain a first spatial mapping vector; based on the first spatial mapping vector and the sound feature vector, use multiplication to obtain the sound cosine similarity; map the vibration feature vector to the image space of the sound feature vector to obtain a second spatial mapping vector; based on the second spatial mapping vector and the vibration feature vector, use multiplication to obtain the vibration cosine similarity.
[0079] Next, an image space is constructed based on the aforementioned sound feature vectors and an image space is constructed based on the aforementioned vibration feature vectors. The sound feature vectors are mapped to the image space of the aforementioned vibration feature vectors to obtain a first spatial mapping vector. Based on the first spatial mapping vector and the aforementioned sound feature vectors, a multiplication operation is used to obtain the sound cosine similarity. The vibration feature vectors are then mapped to the image space of the aforementioned sound feature vectors to obtain a second spatial mapping vector. Based on the second spatial mapping vector and the aforementioned vibration feature vectors, a multiplication operation is used to obtain the vibration cosine similarity. Specifically, the sound cosine similarity E is obtained using the following formula. x Similarity E to the above-mentioned vibration cosine y :
[0080] E x =Normalize(dot(F x W y ))
[0081] E y =Normalize(dot(F y W x ))
[0082] Among them, E x E represents the cosine similarity of the sounds, which is the matrix dot product of the first spatial mapping vector and the sound feature vector. y Representing the vibration cosine similarity, the matrix dot product of the aforementioned second spatial mapping vector and the aforementioned vibration feature vector; W x The image space representing the sound feature vectors, W y The image space represents the vibration feature vector. `dot` is a function in Python representing matrix multiplication, and `Normalize` is a normalization function in Python that normalizes the values to the [0, 1] interval based on different calculations. By constructing the corresponding image space and using a mapping method, the cosine similarity of sound and vibration are calculated respectively. This method is simple and prepares for subsequent state diagnosis.
[0083] Step 203: Determine whether the sound cosine similarity and the vibration cosine similarity are equal; when the sound cosine similarity and the vibration cosine similarity are equal, calculate the feature fusion vector based on the sound feature vector, the vibration feature vector, and preset parameters; when the sound cosine similarity and the vibration cosine similarity are not equal, calculate the feature fusion vector based on the sound feature vector, the vibration feature vector, the image space vector of the sound feature vector, and the image space vector of the vibration feature vector.
[0084] Next, it is determined whether the aforementioned sound cosine similarity and the aforementioned vibration cosine similarity are equal; when the aforementioned sound cosine similarity and the aforementioned vibration cosine similarity are equal, a feature fusion vector is calculated based on the aforementioned sound feature vector, the aforementioned vibration feature vector, and preset parameters; specifically, E x E represents the cosine similarity of sounds. y Represents the cosine similarity of vibrations, when E x =E y The feature fusion vector can be calculated using the following formula:
[0085] C = 0.5 * F x +0.5*F y
[0086] Where C represents the feature fusion vector, F x F represents the sound feature vector. y 0.5 represents the vibration feature vector, and 0.5 represents the preset parameters for the sound feature vector and the vibration feature vector;
[0087] Next, when the aforementioned sound cosine similarity and the aforementioned vibration cosine similarity are not equal, a feature fusion vector is calculated based on the aforementioned sound feature vector, the aforementioned vibration feature vector, the aforementioned sound feature vector's image space vector, and the aforementioned vibration feature vector's image space vector. Specifically, E x E represents the cosine similarity of sounds. y Represents the cosine similarity of vibrations, when E x ≠E y The feature fusion vector can be calculated using the following formula:
[0088] e x =ω t F x
[0089] e y =ω t′ F y
[0090]
[0091]
[0092] C = a * F x +b*F y
[0093] Where C represents the feature fusion vector, and a and b are parameters of C; ω t and ω t′ They are respectively with F x and F y Image space vectors of the same dimension; e xE represents the cosine similarity of sounds. y Represents the cosine similarity of vibrations; exp is the exponential function operation in the Python programming language, expe x e is represented as e x exponentiation, expe y e is represented as e y By judging the cosine similarity of sound and vibration, feature fusion vectors are obtained under different conditions. Multimodal information fusion is then performed to comprehensively integrate the real operating state characteristics of rotating equipment, reducing the error of state diagnosis and improving the accuracy of state diagnosis.
[0094] Step 204: Based on the feature fusion vector, obtain the preliminary diagnostic state of the rotating equipment and the predicted probability value of the preliminary diagnostic state using the fault diagnosis model; when the predicted probability value is greater than or equal to a preset probability value, determine the operating condition of the rotating equipment as the preliminary diagnostic state; when the predicted probability value is less than the preset probability value, determine the operating condition of the rotating equipment as normal operating condition; save the sound data, the vibration data, and the operating condition to the database.
[0095] Next, based on the aforementioned feature fusion vector, a fault diagnosis model is used to obtain the preliminary diagnostic state of the rotating equipment and the predicted probability value of the preliminary diagnostic state. When the predicted probability value is greater than or equal to a preset probability value, the operating condition of the rotating equipment is determined to be the preliminary diagnostic state. When the predicted probability value is less than the preset probability value, the operating condition of the rotating equipment is determined to be normal operating condition. The aforementioned sound data, vibration data, and operating condition states are saved to the database. Specifically, the operating conditions include: normal operating condition, inner raceway fault, outer raceway fault, rolling element fault, and inner raceway fault superimposed on outer raceway fault. Fault conditions such as inner raceway fault, inner raceway fault combined with rolling element fault, and outer raceway fault combined with rolling element fault are identified. The aforementioned feature fusion vectors are input into the fault diagnosis model. The model outputs a preliminary diagnostic state of the rotating equipment and a predicted probability value for that state. For example, if the preliminary diagnostic state is an outer raceway fault with a preset probability value of 0.8, and the predicted probability value is 0.9 (0.9 is greater than 0.8), then the operating condition of the rotating equipment is an outer raceway fault. If the predicted probability value is 0.7 (0.7 is less than 0.8), then the operating condition of the rotating equipment is normal. After diagnosing the operating condition of the rotating equipment, the corresponding sound data, vibration data, and operating condition status are saved to a database for training, validation, and testing of the fault diagnosis model. By further judging the predicted probability values output by the fault diagnosis model, the operating condition of the rotating equipment is determined, improving the accuracy of the condition diagnosis.
[0096] Optionally, before step 201, the method further includes: using each working condition as a preliminary diagnostic state label, and obtaining a feature fusion vector sample of the working condition based on the sound data sample and vibration data sample corresponding to the working condition; training the model based on the preliminary diagnostic state label and the feature fusion vector sample to obtain a fault diagnosis model.
[0097] In the above embodiments of this application, the database contains sound feature vectors and vibration feature vectors, as well as the operating conditions that occur when the sound feature vectors and vibration feature vectors appear. During training, the sound feature vectors and vibration feature vectors are used as samples, and the operating conditions are used as preliminary diagnostic state labels to train the model so that the model can receive feature fusion vectors, output preliminary diagnostic states and predicted probability values, and achieve the effect of fault diagnosis.
[0098] By applying the technical solution of this embodiment, sound and vibration data of rotating equipment in its working state are collected. The sound and vibration data are preprocessed, and then subjected to Fast Fourier Transform (FFT) to generate uniform sound and vibration spectrograms. Sound pixels are obtained from the sound spectrograms, and vibration pixels are obtained from the vibration spectrograms. Sound feature vectors are obtained from the sound pixels, and vibration feature vectors are obtained from the vibration pixels. Graphical spaces for sound and vibration feature vectors are constructed, and the cosine similarity of the mappings of the two feature vectors in each other's image spaces is calculated. Based on the judgment of the sound and vibration cosine similarity, multimodal information fusion is used to fuse the sound and vibration feature vectors to obtain a feature fusion vector. This feature fusion vector is then input into a fault diagnosis model, which outputs an initial diagnostic state and a predicted probability value. The state of the rotating equipment is determined based on the predicted probability value. This achieves the fusion of multimodal information, more comprehensively integrating the actual operating state characteristics of the rotating equipment, and improving the accuracy of rotating equipment state diagnosis.
[0099] Furthermore, as Figure 1 To specifically implement the method, this application provides a rotating equipment condition diagnosis device, such as... Figure 3 As shown, the device includes:
[0100] The acquisition module is used to acquire sound data and vibration data of the rotating equipment in the working state, and to obtain the sound feature vector corresponding to the sound data and the vibration feature vector corresponding to the vibration data.
[0101] The acquisition module is used to obtain sound cosine similarity based on the image space of the sound feature vector and the vibration feature vector, and to obtain vibration cosine similarity based on the image space of the vibration feature vector and the sound feature vector.
[0102] The fusion module is used to obtain a feature fusion vector based on the sound cosine similarity and the vibration cosine similarity;
[0103] The determination module is used to determine the operating condition of the rotating equipment based on the feature fusion vector.
[0104] Optionally, the fusion module is further configured to:
[0105] Determine whether the sound cosine similarity and the vibration cosine similarity are equal;
[0106] When the sound cosine similarity and the vibration cosine similarity are equal, a feature fusion vector is calculated based on the sound feature vector, the vibration feature vector, and preset parameters.
[0107] When the sound cosine similarity and the vibration cosine similarity are not equal, a feature fusion vector is calculated based on the sound feature vector, the vibration feature vector, the image space vector of the sound feature vector, and the image space vector of the vibration feature vector.
[0108] Optionally, the acquisition module is further configured to:
[0109] The sound feature vector is mapped to the image space of the vibration feature vector to obtain a first spatial mapping vector. Based on the first spatial mapping vector and the sound feature vector, the sound cosine similarity is obtained by multiplication.
[0110] The vibration feature vector is mapped to the image space of the sound feature vector to obtain a second spatial mapping vector. Based on the second spatial mapping vector and the vibration feature vector, the vibration cosine similarity is obtained by multiplication.
[0111] Optionally, the determining module is further configured to:
[0112] Based on the feature fusion vector, the operating condition of the rotating equipment is obtained using a fault diagnosis model.
[0113] The sound data, vibration data, and operating condition status are saved to the database.
[0114] Optionally, the determining module is further configured to:
[0115] Based on the feature fusion vector, the preliminary diagnostic status of the rotating equipment and the predicted probability value of the preliminary diagnostic status are obtained using the fault diagnosis model.
[0116] When the predicted probability value is greater than or equal to the preset probability value, the operating condition of the rotating equipment is determined to be the preliminary diagnostic state.
[0117] When the predicted value is less than the preset probability value, the operating condition of the rotating equipment is determined to be normal.
[0118] Optionally, the acquisition module is further configured to:
[0119] The sound data is preprocessed, and a fast Fourier transform is performed on the preprocessed sound data to generate a speech spectrogram. Sound pixels are obtained based on the speech spectrogram, and sound feature vectors of the sound data are obtained based on the sound pixels.
[0120] The vibration data is preprocessed, and a fast Fourier transform is performed on the preprocessed vibration data to generate a vibration spectrum. Vibration pixels are obtained based on the vibration spectrum, and the vibration feature vector of the vibration data is obtained based on the vibration pixels.
[0121] Optionally, the device further includes: a training module, used for:
[0122] Each working condition is used as a preliminary diagnostic state label, and a feature fusion vector sample of the working condition is obtained based on the sound data sample and vibration data sample corresponding to the working condition.
[0123] The model is trained based on the preliminary diagnostic status labels and the feature fusion vector samples to obtain a fault diagnosis model.
[0124] It should be noted that other corresponding descriptions of the functional units involved in the rotating equipment condition diagnosis device provided in this application embodiment can be found by referring to... Figures 1 to 2 The corresponding descriptions in the method will not be repeated here.
[0125] Based on the above, Figures 1 to 2 Accordingly, this application also provides a storage medium storing a computer program, which, when executed by a processor, implements the above-described method. Figures 1 to 2 The method for diagnosing the condition of rotating equipment is shown.
[0126] Based on this understanding, the technical solution of this application can be embodied in the form of a software product. This software product can be stored in a non-volatile storage medium (such as a CD-ROM, USB flash drive, or portable hard drive), and includes several instructions to cause a computer device (such as a personal computer, server, or network device) to execute the methods described in the various implementation scenarios of this application.
[0127] Based on the above, Figures 1 to 2 The method shown, and Figure 3To achieve the above objectives, the present application also provides a computer device, specifically a personal computer, server, network device, etc., as shown in the virtual device embodiment. This computer device includes a storage medium and a processor; the storage medium stores a computer program; the processor executes the computer program to achieve the above-described objectives. Figures 1 to 2 The method for diagnosing the condition of rotating equipment is shown.
[0128] Optionally, the computer device may also include a user interface, a network interface, a camera, radio frequency (RF) circuitry, sensors, audio circuitry, a Wi-Fi module, etc. The user interface may include a display screen, input units such as a keyboard, etc., and optional user interfaces may also include USB interfaces, card reader interfaces, etc. The network interface may optionally include standard wired interfaces, wireless interfaces (such as Bluetooth interfaces, Wi-Fi interfaces), etc.
[0129] Those skilled in the art will understand that the computer device structure provided in this embodiment does not constitute a limitation on the computer device, and may include more or fewer components, or combine certain components, or have different component arrangements.
[0130] The storage medium may also include an operating system and a network communication module. The operating system is a program that manages and stores the hardware and software resources of a computer device, supporting the operation of information processing programs and other software and / or programs. The network communication module is used to enable communication between the various components within the storage medium, as well as communication with other hardware and software within the physical device.
[0131] Through the above description of the embodiments, those skilled in the art can clearly understand that this application can be implemented by means of software plus necessary general-purpose hardware platforms, or by hardware. It collects sound data and vibration data of the rotating equipment in its working state, and obtains the sound feature vector corresponding to the sound data and the vibration feature vector corresponding to the vibration data; obtains the sound cosine similarity based on the image space of the sound feature vector and the vibration feature vector, and obtains the vibration cosine similarity based on the image space of the vibration feature vector and the sound feature vector; obtains a feature fusion vector based on the sound cosine similarity and the vibration cosine similarity; and determines the working state of the rotating equipment based on the feature fusion vector. This achieves the effect of diagnosing the state of the rotating equipment by fusing the sound feature vector corresponding to the sound data and the vibration feature vector corresponding to the vibration data, thereby improving the accuracy of rotating equipment state diagnosis.
[0132] Those skilled in the art will understand that the accompanying drawings are merely schematic diagrams of a preferred embodiment, and the modules or processes shown in the drawings are not necessarily essential for implementing this application. Those skilled in the art will understand that the modules in the apparatus of the embodiment can be distributed within the apparatus of the embodiment as described, or can be modified to be located in one or more apparatuses different from this embodiment. The modules of the above-described embodiment can be combined into one module, or further divided into multiple sub-modules.
[0133] The serial numbers in this application are for descriptive purposes only and do not represent the superiority or inferiority of any particular implementation scenario. The above disclosures are merely a few specific implementation scenarios of this application; however, this application is not limited thereto, and any variations conceived by those skilled in the art should fall within the protection scope of this application.
Claims
1. A method for diagnosing the condition of rotating equipment, characterized in that, The method includes: Collect sound and vibration data of the rotating device in operation, and obtain the sound feature vector corresponding to the sound data and the vibration feature vector corresponding to the vibration data; Based on the image space of the sound feature vector and the vibration feature vector, the sound cosine similarity is obtained, and based on the image space of the vibration feature vector and the sound feature vector, the vibration cosine similarity is obtained. The feature fusion vector is obtained based on the sound cosine similarity and the vibration cosine similarity; The operating status of the rotating equipment is determined based on the feature fusion vector. The step of obtaining sound cosine similarity based on the image space of the sound feature vector and the vibration feature vector, and obtaining vibration cosine similarity based on the image space of the vibration feature vector and the sound feature vector, includes: The sound feature vector is mapped to the image space of the vibration feature vector to obtain a first spatial mapping vector. Based on the first spatial mapping vector and the sound feature vector, the sound cosine similarity is obtained by multiplication. The vibration feature vector is mapped to the image space of the sound feature vector to obtain a second spatial mapping vector. Based on the second spatial mapping vector and the vibration feature vector, the vibration cosine similarity is obtained by multiplication.
2. The method according to claim 1, characterized in that, The step of obtaining the feature fusion vector based on the sound cosine similarity and the vibration cosine similarity includes: Determine whether the sound cosine similarity and the vibration cosine similarity are equal; When the sound cosine similarity and the vibration cosine similarity are equal, a feature fusion vector is calculated based on the sound feature vector, the vibration feature vector, and preset parameters. When the sound cosine similarity and the vibration cosine similarity are not equal, a feature fusion vector is calculated based on the sound feature vector, the vibration feature vector, the image space vector of the sound feature vector, and the image space vector of the vibration feature vector.
3. The method according to claim 1, characterized in that, Determining the operating state of the rotating equipment based on the feature fusion vector includes: Based on the feature fusion vector, the operating condition of the rotating equipment is obtained using a fault diagnosis model. The sound data, vibration data, and operating condition status are saved to the database.
4. The method according to claim 3, characterized in that, The step of obtaining the operating condition of the rotating equipment based on the feature fusion vector and using a fault diagnosis model includes: Based on the feature fusion vector, the preliminary diagnostic status of the rotating equipment and the predicted probability value of the preliminary diagnostic status are obtained using the fault diagnosis model. When the predicted probability value is greater than or equal to the preset probability value, the operating condition of the rotating equipment is determined to be the preliminary diagnostic state. When the predicted probability value is less than the preset probability value, the operating condition of the rotating equipment is determined to be normal.
5. The method according to claim 1, characterized in that, The step of obtaining the sound feature vector corresponding to the sound data and the vibration feature vector corresponding to the vibration data includes: The sound data is preprocessed, and a fast Fourier transform is performed on the preprocessed sound data to generate a speech spectrogram. Sound pixels are obtained based on the speech spectrogram, and sound feature vectors of the sound data are obtained based on the sound pixels. The vibration data is preprocessed, and a fast Fourier transform is performed on the preprocessed vibration data to generate a vibration spectrum. Vibration pixels are obtained based on the vibration spectrum, and the vibration feature vector of the vibration data is obtained based on the vibration pixels.
6. The method according to claim 1, characterized in that, Before acquiring the sound and vibration data of the rotating device in its working state, the method further includes: Each working condition is used as a preliminary diagnostic state label, and a feature fusion vector sample of the working condition is obtained based on the sound data sample and vibration data sample corresponding to the working condition. The model is trained based on the preliminary diagnostic status labels and the feature fusion vector samples to obtain a fault diagnosis model.
7. A rotating equipment condition diagnosis device, characterized in that, The device includes: The acquisition module is used to acquire sound data and vibration data of the rotating equipment in the working state, and to obtain the sound feature vector corresponding to the sound data and the vibration feature vector corresponding to the vibration data. The acquisition module is used to obtain sound cosine similarity based on the image space of the sound feature vector and the vibration feature vector, and to obtain vibration cosine similarity based on the image space of the vibration feature vector and the sound feature vector. The fusion module is used to obtain a feature fusion vector based on the sound cosine similarity and the vibration cosine similarity; The determination module is used to determine the operating condition of the rotating equipment based on the feature fusion vector; The acquisition module is used for: The sound feature vector is mapped to the image space of the vibration feature vector to obtain a first spatial mapping vector. Based on the first spatial mapping vector and the sound feature vector, the sound cosine similarity is obtained by multiplication. The vibration feature vector is mapped to the image space of the sound feature vector to obtain a second spatial mapping vector. Based on the second spatial mapping vector and the vibration feature vector, the vibration cosine similarity is obtained by multiplication.
8. A storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the method for diagnosing the state of the rotating equipment as described in any one of claims 1 to 6.
9. A computer device, comprising a storage medium, a processor, and a computer program stored on the storage medium and executable on the processor, characterized in that, When the processor executes the computer program, it implements the method for diagnosing the state of rotating equipment as described in any one of claims 1 to 6.