Bearing state evaluation method and system based on vibration signals and medium
By simultaneously extracting multi-dimensional features and constructing a deep learning model, the problems of poor signal preprocessing and incomplete feature extraction in existing bearing condition assessments are solved, achieving efficient and accurate bearing condition assessment.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- RUIHU ZHIKE DATA (SUZHOU) CO LTD
- Filing Date
- 2026-01-27
- Publication Date
- 2026-05-08
AI Technical Summary
Existing bearing condition assessment methods suffer from poor vibration signal preprocessing, low signal-to-noise ratio, incomplete feature extraction, and poor adaptability and generalization ability of the assessment model, making it difficult to comprehensively characterize the complex operating conditions of bearings.
By simultaneously extracting multi-dimensional features in the time domain, frequency domain, and wavelet domain, a combined feature filtering algorithm is used to remove redundant and invalid features. Combined with feature fusion technology, a low-dimensional and efficient fused feature vector is generated, and a bearing condition assessment model based on deep learning is constructed.
It improves the distinguishability of bearing fault characteristics, reduces the computational load of the model, enhances the evaluation efficiency and accuracy, and enables accurate and real-time evaluation of different types of bearing faults.
Smart Images
Figure CN121994487A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of bearing condition assessment, and more specifically, to a bearing condition assessment method, system, and medium based on vibration signals. Background Technology
[0002] Bearings, as core transmission components of rotating machinery, are widely used in critical fields such as industrial manufacturing, rail transportation, aerospace, and wind power generation. Their operating status directly determines the reliability, safety, and service life of equipment. According to industrial equipment failure statistics, over 40% of rotating machinery failures originate from bearing failure. Bearing failures not only lead to equipment downtime and production stoppages, causing huge economic losses, but in extreme cases, they can also trigger safety accidents, threatening personnel lives. Therefore, achieving accurate and real-time assessment of bearing operating status, early warning of potential failures, and determination of failure severity are of significant practical importance for ensuring stable equipment operation, reducing maintenance costs, and improving industrial production safety.
[0003] Existing bearing condition assessment methods have the following drawbacks:
[0004] First, the preprocessing of vibration signals is inadequate. The complex industrial environment makes the raw vibration signals easily affected by equipment operating noise, electromagnetic interference, and environmental vibration, resulting in a low signal-to-noise ratio.
[0005] Secondly, the comprehensiveness and effectiveness of feature extraction are insufficient. Bearing faults are diverse, and different faults exhibit different characteristics in vibration signals. Moreover, fault features are often masked by noise. Existing methods mostly extract features only in a single dimension, such as the time domain or frequency domain, which makes it difficult to comprehensively characterize the complex operating state of bearings. At the same time, the extracted features are not effectively screened, and redundant and invalid features increase the computational load of subsequent models and reduce evaluation efficiency.
[0006] Third, state evaluation models have poor adaptability and generalization ability. Existing evaluation models are mostly built based on a single algorithm, making it difficult to take into account both the local representation ability of features and the ability to capture temporal dependencies. Summary of the Invention
[0007] The purpose of this application is to provide a bearing condition assessment method, system, and medium based on vibration signals. By simultaneously extracting multi-dimensional features in the time domain, frequency domain, and wavelet domain, it can comprehensively capture the feature information of different types of bearing faults. It adopts a combined feature screening algorithm to eliminate redundant and invalid features, and combines feature fusion technology to generate low-dimensional and efficient fused feature vectors. While improving feature discrimination, it reduces the computational load of subsequent models and improves assessment efficiency.
[0008] This application also provides a bearing condition assessment method based on vibration signals, including:
[0009] The vibration sensor collects the original vibration signals of the bearing under different operating conditions, and preprocesses the original vibration signals.
[0010] Based on the preprocessed original vibration information, time-domain features, frequency-domain features, and wavelet-domain features are extracted respectively.
[0011] A combined feature filtering algorithm is used to filter time-domain features, frequency-domain features, and wavelet-domain features, eliminating redundant and invalid features to obtain the optimal feature set;
[0012] The optimal feature set is subjected to feature fusion processing to generate a fused feature vector;
[0013] A bearing condition assessment model based on deep learning is constructed. The fused feature vector is input into the bearing condition assessment model, and the bearing condition assessment result is output.
[0014] The bearing condition assessment results are transmitted to the terminal in real time.
[0015] Optionally, in the bearing condition assessment method based on vibration signals described in the embodiments of this application, the original vibration signal is preprocessed, specifically including:
[0016] The original vibration signal is acquired, and the original vibration signal is analyzed based on the information of abnormal change conditions. Abnormal signal segments caused by sensor malfunction or electromagnetic interference are removed from the original vibration signal, and valid signal segments are retained.
[0017] An adaptive filtering algorithm is used to denoise the effective signal segment, filtering out environmental noise and equipment operation interference noise to obtain the denoised signal.
[0018] The denoised signal is subjected to amplitude normalization processing to map the signal amplitude to a preset range, thereby eliminating the influence of signal amplitude differences under different operating conditions and obtaining a standardized signal.
[0019] Calculate the signal-to-noise ratio (SNR) of the standardized signal and compare it with a preset SNR threshold.
[0020] If the signal-to-noise ratio is lower than the preset threshold, the original vibration signal under the corresponding working condition will be re-acquired until a valid signal is obtained.
[0021] Optionally, in the bearing condition assessment method based on vibration signals described in the embodiments of this application, the time-domain features include peak value, peak factor, kurtosis, skewness, variance, standard deviation, mean, impulse factor, margin factor, and waveform factor.
[0022] The frequency domain features include centroid frequency, mean square frequency, frequency variance, frequency standard deviation, peak energy, and peak frequency.
[0023] Optionally, in the bearing condition assessment method based on vibration signals described in the embodiments of this application, a combined feature screening algorithm is used to screen time-domain features, frequency-domain features, and wavelet-domain features, eliminating redundant and invalid features to obtain the optimal feature set, specifically including:
[0024] The combined feature filtering algorithm is a combination of the ReliefF algorithm and the L1 regularization algorithm;
[0025] The ReliefF algorithm is used to calculate the corresponding weight values of time-domain features, frequency-domain features and wavelet-domain features, and features with weight values less than a preset weight threshold are removed.
[0026] Features with weight values greater than or equal to a preset weight threshold are subjected to a second screening using the L1 regularization algorithm to remove features with a coefficient of 0, thereby obtaining the optimal feature set.
[0027] Optionally, in the bearing condition assessment method based on vibration signals described in the embodiments of this application, the feature fusion processing adopts the principal component analysis (PCA) algorithm to map the optimal feature set to a low-dimensional space and generate a fused feature vector with a dimension of 10-20.
[0028] Optionally, in the bearing condition assessment method based on vibration signals described in the embodiments of this application, the bearing condition assessment model based on deep learning is a CNN-LSTM hybrid model, which includes a convolutional layer, a pooling layer, an LSTM layer and a fully connected layer connected in sequence.
[0029] Among them, the convolutional layer is used to extract local features from the fused feature vector, the pooling layer is used to reduce the dimensionality of the features output by the convolutional layer, the LSTM layer is used to capture the temporal dependencies in the feature sequence, and the fully connected layer is used to output the classification results of the bearing state.
[0030] The loss function of the bearing condition assessment model is the cross-entropy loss function, and the optimizer is the Adam optimizer.
[0031] Secondly, embodiments of this application provide a bearing condition assessment system based on vibration signals. The system includes a memory and a processor. The memory includes a program for a bearing condition assessment method based on vibration signals. When the processor executes the program for the bearing condition assessment method based on vibration signals, it performs the following steps:
[0032] The vibration sensor collects the original vibration signals of the bearing under different operating conditions, and preprocesses the original vibration signals.
[0033] Based on the preprocessed original vibration information, time-domain features, frequency-domain features, and wavelet-domain features are extracted respectively.
[0034] A combined feature filtering algorithm is used to filter time-domain features, frequency-domain features, and wavelet-domain features, eliminating redundant and invalid features to obtain the optimal feature set;
[0035] The optimal feature set is subjected to feature fusion processing to generate a fused feature vector;
[0036] A bearing condition assessment model based on deep learning is constructed. The fused feature vector is input into the bearing condition assessment model, and the bearing condition assessment result is output.
[0037] The bearing condition assessment results are transmitted to the terminal in real time.
[0038] Optionally, in the bearing condition assessment system based on vibration signals described in this application embodiment, the original vibration signal is preprocessed, specifically including:
[0039] The original vibration signal is acquired, and the original vibration signal is analyzed based on the information of abnormal change conditions. Abnormal signal segments caused by sensor malfunction or electromagnetic interference are removed from the original vibration signal, and valid signal segments are retained.
[0040] An adaptive filtering algorithm is used to denoise the effective signal segment, filtering out environmental noise and equipment operation interference noise to obtain the denoised signal.
[0041] The denoised signal is subjected to amplitude normalization processing to map the signal amplitude to a preset range, thereby eliminating the influence of signal amplitude differences under different operating conditions and obtaining a standardized signal.
[0042] Calculate the signal-to-noise ratio (SNR) of the standardized signal and compare it with a preset SNR threshold.
[0043] If the signal-to-noise ratio is lower than the preset threshold, the original vibration signal under the corresponding working condition will be re-acquired until a valid signal is obtained.
[0044] Optionally, in the bearing condition assessment system based on vibration signals described in the embodiments of this application, the time-domain features include peak value, peak factor, kurtosis, skewness, variance, standard deviation, mean, impulse factor, margin factor, and waveform factor.
[0045] The frequency domain features include centroid frequency, mean square frequency, frequency variance, frequency standard deviation, peak energy, and peak frequency.
[0046] Thirdly, embodiments of this application also provide a computer-readable storage medium, which includes a bearing condition assessment method program based on vibration signals. When the bearing condition assessment method program based on vibration signals is executed by a processor, it implements the steps of the bearing condition assessment method based on vibration signals as described in any of the above claims.
[0047] As can be seen from the above, the bearing condition assessment method, system, and medium based on vibration signals provided in this application collect raw vibration signals of the bearing under different operating conditions using vibration sensors, and preprocess the raw vibration signals; extract time-domain features, frequency-domain features, and wavelet-domain features based on the preprocessed raw vibration information; use a combined feature filtering algorithm to filter the time-domain features, frequency-domain features, and wavelet-domain features, eliminating redundant and invalid features to obtain an optimal feature set; perform feature fusion processing on the optimal feature set to generate a fused feature vector; construct a bearing condition assessment model based on deep learning, input the fused feature vector into the bearing condition assessment model, and output the bearing condition assessment result; transmit the bearing condition assessment result to the terminal in real time; by simultaneously extracting multi-dimensional features in the time domain, frequency domain, and wavelet domain, the feature information of different types of bearing faults can be comprehensively captured; the combined feature filtering algorithm eliminates redundant and invalid features; and the feature fusion technology generates a low-dimensional and efficient fused feature vector, which improves feature discrimination while reducing the computational load of subsequent models and improving assessment efficiency.
[0048] Other features and advantages of this application will be set forth in the following description, and the advantages of this application will be apparent in part from the description, or may be learned by practicing the embodiments of this application. The objectives and other advantages of this application may be realized and obtained by means of the structures particularly pointed out in the written description, claims, and drawings. Attached Figure Description
[0049] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments of this application will be briefly introduced below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.
[0050] Figure 1 A flowchart illustrating a bearing condition assessment method based on vibration signals, provided for an embodiment of this application;
[0051] Figure 2 The flowchart shows the preprocessing of the original vibration signal for a bearing condition assessment method based on vibration signal provided in this application embodiment. Detailed Implementation
[0052] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of the embodiments. The components of the embodiments of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely represents selected embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.
[0053] It should be noted that similar reference numerals and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. Furthermore, in the description of this application, the terms "first," "second," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.
[0054] Please refer to Figure 1 , Figure 1 This is a flowchart of a bearing condition assessment method based on vibration signals, as described in some embodiments of this application. This bearing condition assessment method based on vibration signals is used in a terminal device and includes the following steps:
[0055] S101 collects the original vibration signals of the bearing under different operating conditions through a vibration sensor and preprocesses the original vibration signals.
[0056] Specifically, the operating conditions include no-load condition, rated load condition, and variable load condition;
[0057] S102, extract time-domain features, frequency-domain features and wavelet-domain features based on the preprocessed original vibration information;
[0058] Specifically, when extracting time-domain features, the preprocessed vibration signal is first segmented, with each segment consisting of 2048-4096 sampling points. After extracting the time-domain features of each segment, the mean is calculated and used as the final time-domain feature value.
[0059] S103, a combined feature filtering algorithm is used to filter time-domain features, frequency-domain features and wavelet-domain features, remove redundant and invalid features, and obtain the optimal feature set;
[0060] S104, Perform feature fusion processing on the optimal feature set to generate a fused feature vector;
[0061] S105: Construct a bearing condition assessment model based on deep learning, input the fused feature vector into the bearing condition assessment model, output the bearing condition assessment result, and transmit the bearing condition assessment result to the terminal in real time.
[0062] Please refer to Figure 2 , Figure 2 This is a flowchart illustrating the preprocessing of the original vibration signal in a bearing condition assessment method based on vibration signals, as described in some embodiments of this application. According to embodiments of the present invention, the preprocessing of the original vibration signal specifically includes:
[0063] S201, acquire the original vibration signal, analyze the original vibration signal based on the abnormal change condition information, remove abnormal signal segments caused by sensor abnormality or electromagnetic interference in the original vibration signal, and retain the valid signal segments;
[0064] S202 uses an adaptive filtering algorithm to denoise the effective signal segment, filtering out environmental noise and equipment operation interference noise to obtain the denoised signal.
[0065] S203 performs amplitude standardization processing on the denoised signal, mapping the signal amplitude to a preset range to eliminate the influence of signal amplitude differences under different operating conditions and obtain a standardized signal.
[0066] S204, Calculate the signal-to-noise ratio (SNR) of the standardized signal and compare it with a preset SNR threshold;
[0067] S205 If the signal-to-noise ratio is lower than the preset threshold, the original vibration signal under the corresponding working condition will be re-acquired until a valid signal is obtained.
[0068] According to embodiments of the present invention, the time-domain features include peak value, peak factor, kurtosis, skewness, variance, standard deviation, mean, impulse factor, margin factor, and waveform factor;
[0069] Frequency domain characteristics include centroid frequency, mean square frequency, frequency variance, frequency standard deviation, peak energy, and peak frequency.
[0070] Specifically, the time-domain features include at least six of the following: peak value, peak factor, kurtosis, skewness, variance, standard deviation, mean, impulse factor, margin factor, and waveform factor; the frequency-domain features include at least four of the following: centroid frequency, mean square frequency, frequency variance, frequency standard deviation, spectral peak energy, and spectral peak frequency; the wavelet domain features are obtained by performing db4 wavelet decomposition on the preprocessed vibration signal, specifically by performing 3-5 levels of wavelet decomposition on the preprocessed vibration signal and extracting the energy value and entropy value of the decomposition coefficients at each level as wavelet domain features.
[0071] According to an embodiment of the present invention, a combined feature filtering algorithm is used to filter time-domain features, frequency-domain features, and wavelet-domain features, eliminating redundant and invalid features to obtain an optimal feature set, specifically including:
[0072] The combined feature selection algorithm is a combination of the ReliefF algorithm and the L1 regularization algorithm;
[0073] The ReliefF algorithm is used to calculate the corresponding weight values of time-domain features, frequency-domain features and wavelet-domain features, and features with weight values less than a preset weight threshold are removed.
[0074] Features with weight values greater than or equal to a preset weight threshold are subjected to a second screening using the L1 regularization algorithm to remove features with a coefficient of 0, thereby obtaining the optimal feature set.
[0075] According to an embodiment of the present invention, the feature fusion processing employs the principal component analysis (PCA) algorithm to map the optimal feature set to a low-dimensional space, generating a fusion feature vector with a dimension of 10-20.
[0076] According to an embodiment of the present invention, the bearing state evaluation model based on deep learning is a CNN-LSTM hybrid model, which includes a convolutional layer, a pooling layer, an LSTM layer and a fully connected layer connected in sequence.
[0077] Among them, the convolutional layer is used to extract local features from the fused feature vector, the pooling layer is used to reduce the dimensionality of the features output by the convolutional layer, the LSTM layer is used to capture the temporal dependencies in the feature sequence, and the fully connected layer is used to output the classification results of the bearing state.
[0078] The loss function of the bearing condition assessment model is the cross-entropy loss function, and the optimizer is the Adam optimizer.
[0079] It should be noted that the obtained fused feature vectors are divided into training set, validation set and test set. The bearing condition assessment model is trained using the training set and the hyperparameters of the model are adjusted using the validation set until the model converges and meets the preset accuracy requirements. The bearing condition includes normal condition, inner ring fault condition, outer ring fault condition and rolling element fault condition, and different fault conditions correspond to different fault severity levels.
[0080] Secondly, embodiments of this application provide a bearing condition assessment system based on vibration signals. The system includes a memory and a processor. The memory includes a program for a bearing condition assessment method based on vibration signals. When the processor executes the program for the bearing condition assessment method based on vibration signals, it implements the following steps:
[0081] The original vibration signals of the bearing under different operating conditions are collected by vibration sensors, and the original vibration signals are preprocessed.
[0082] Based on the preprocessed original vibration information, time-domain features, frequency-domain features, and wavelet-domain features are extracted respectively.
[0083] A combined feature filtering algorithm is used to filter time-domain features, frequency-domain features, and wavelet-domain features, eliminating redundant and invalid features to obtain the optimal feature set;
[0084] The optimal feature set is subjected to feature fusion processing to generate a fused feature vector;
[0085] A bearing condition assessment model based on deep learning is constructed. The fused feature vector is input into the bearing condition assessment model, and the bearing condition assessment result is output.
[0086] The bearing condition assessment results are transmitted to the terminal in real time.
[0087] According to an embodiment of the present invention, the preprocessing of the original vibration signal specifically includes:
[0088] The original vibration signal is acquired, and the original vibration signal is analyzed based on the information of abnormal change conditions. Abnormal signal segments caused by sensor malfunction or electromagnetic interference are removed from the original vibration signal, and valid signal segments are retained.
[0089] An adaptive filtering algorithm is used to denoise the effective signal segment, filtering out environmental noise and equipment operation interference noise to obtain the denoised signal.
[0090] The denoised signal is subjected to amplitude normalization processing to map the signal amplitude to a preset range, thereby eliminating the influence of signal amplitude differences under different operating conditions and obtaining a standardized signal.
[0091] Calculate the signal-to-noise ratio (SNR) of the standardized signal and compare it with a preset SNR threshold.
[0092] If the signal-to-noise ratio is lower than the preset threshold, the original vibration signal under the corresponding working condition will be re-acquired until a valid signal is obtained.
[0093] According to embodiments of the present invention, the time-domain features include peak value, peak factor, kurtosis, skewness, variance, standard deviation, mean, impulse factor, margin factor, and waveform factor;
[0094] Frequency domain characteristics include centroid frequency, mean square frequency, frequency variance, frequency standard deviation, peak energy, and peak frequency.
[0095] A third aspect of the present invention provides a computer-readable storage medium including a bearing condition assessment method program based on vibration signals. When the bearing condition assessment method program based on vibration signals is executed by a processor, it implements the steps of the bearing condition assessment method based on vibration signals as described in any of the above claims.
[0096] This invention discloses a bearing condition assessment method, system, and medium based on vibration signals. The method involves acquiring raw vibration signals of the bearing under different operating conditions using vibration sensors, preprocessing the raw vibration signals, extracting time-domain, frequency-domain, and wavelet-domain features based on the preprocessed raw vibration information, and using a combined feature filtering algorithm to filter these features, eliminating redundant and invalid features to obtain an optimal feature set. The optimal feature set is then subjected to feature fusion processing to generate a fused feature vector. A deep learning-based bearing condition assessment model is constructed, and the fused feature vector is input into the model to output the bearing condition assessment result. The bearing condition assessment result is transmitted to a terminal in real time. By simultaneously extracting multi-dimensional features in the time, frequency, and wavelet domains, the method can comprehensively capture the feature information of different types of bearing faults. The combined feature filtering algorithm eliminates redundant and invalid features, and the feature fusion technology generates a low-dimensional and efficient fused feature vector, improving feature discrimination while reducing the computational load of subsequent models and increasing assessment efficiency.
[0097] In the several embodiments provided in this application, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are merely illustrative. For example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods, such as: multiple units or components can be combined, or integrated into another system, or some features can be ignored or not executed. In addition, the coupling, direct coupling, or communication connection between the various components shown or discussed can be through some interfaces, and the indirect coupling or communication connection between devices or units can be electrical, mechanical, or other forms.
[0098] The units described above as separate components may or may not be physically separate. The components shown as units may or may not be physical units. They may be located in one place or distributed across multiple network units. Some or all of the units may be selected to achieve the purpose of this embodiment according to actual needs.
[0099] In addition, in the various embodiments of the present invention, each functional unit can be integrated into one processing unit, or each unit can be a separate unit, or two or more units can be integrated into one unit; the integrated unit can be implemented in hardware or in the form of hardware plus software functional units.
[0100] Those skilled in the art will understand that all or part of the steps of the above method embodiments can be implemented by hardware related to program instructions. The aforementioned program can be stored in a readable storage medium. When the program is executed, it performs the steps of the above method embodiments. The aforementioned storage medium includes various media capable of storing program code, such as mobile storage devices, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.
[0101] Alternatively, if the integrated units of this invention are implemented as software functional modules and sold or used as independent products, they can also be stored in a readable storage medium. Based on this understanding, the technical solutions of the embodiments of this invention, or the parts that contribute to the prior art, can be embodied in the form of a software product. This software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as mobile storage devices, ROM, RAM, magnetic disks, or optical disks.
Claims
1. A bearing condition assessment method based on vibration signals, characterized in that, include: The vibration sensor collects the original vibration signals of the bearing under different operating conditions, and preprocesses the original vibration signals. Based on the preprocessed original vibration information, time-domain features, frequency-domain features, and wavelet-domain features are extracted respectively. A combined feature filtering algorithm is used to filter time-domain features, frequency-domain features, and wavelet-domain features, eliminating redundant and invalid features to obtain the optimal feature set; The optimal feature set is subjected to feature fusion processing to generate a fused feature vector; A bearing condition assessment model based on deep learning is constructed. The fused feature vector is input into the bearing condition assessment model, and the bearing condition assessment result is output. The bearing condition assessment results are transmitted to the terminal in real time.
2. The bearing condition assessment method based on vibration signals according to claim 1, characterized in that, The original vibration signal is preprocessed, specifically including: The original vibration signal is acquired, and the original vibration signal is analyzed based on the information of abnormal change conditions. Abnormal signal segments caused by sensor malfunction or electromagnetic interference are removed from the original vibration signal, and valid signal segments are retained. An adaptive filtering algorithm is used to denoise the effective signal segment, filtering out environmental noise and equipment operation interference noise to obtain the denoised signal. The denoised signal is subjected to amplitude normalization processing to map the signal amplitude to a preset range, thereby eliminating the influence of signal amplitude differences under different operating conditions and obtaining a standardized signal. Calculate the signal-to-noise ratio (SNR) of the standardized signal and compare it with a preset SNR threshold. If the signal-to-noise ratio is lower than the preset threshold, the original vibration signal under the corresponding working condition will be re-acquired until a valid signal is obtained.
3. The bearing condition assessment method based on vibration signals according to claim 2, characterized in that, The time-domain features include peak value, peak factor, kurtosis, skewness, variance, standard deviation, mean, impulse factor, margin factor, and waveform factor. The frequency domain features include centroid frequency, mean square frequency, frequency variance, frequency standard deviation, peak energy, and peak frequency.
4. The bearing condition assessment method based on vibration signals according to claim 3, characterized in that, A combined feature filtering algorithm is used to filter time-domain features, frequency-domain features, and wavelet-domain features, eliminating redundant and invalid features to obtain the optimal feature set, which specifically includes: The combined feature filtering algorithm is a combination of the ReliefF algorithm and the L1 regularization algorithm; The ReliefF algorithm is used to calculate the corresponding weight values of time-domain features, frequency-domain features and wavelet-domain features, and features with weight values less than a preset weight threshold are removed. Features with weight values greater than or equal to a preset weight threshold are subjected to a second screening using the L1 regularization algorithm to remove features with a coefficient of 0, thereby obtaining the optimal feature set.
5. The bearing condition assessment method based on vibration signals according to claim 4, characterized in that, The feature fusion process employs Principal Component Analysis (PCA) to map the optimal feature set to a low-dimensional space, generating a fusion feature vector with a dimension of 10-20.
6. The bearing condition assessment method based on vibration signals according to claim 5, characterized in that, The deep learning-based bearing state evaluation model is a CNN-LSTM hybrid model, which includes a convolutional layer, a pooling layer, an LSTM layer, and a fully connected layer connected in sequence. Among them, the convolutional layer is used to extract local features from the fused feature vector, the pooling layer is used to reduce the dimensionality of the features output by the convolutional layer, the LSTM layer is used to capture the temporal dependencies in the feature sequence, and the fully connected layer is used to output the classification results of the bearing state. The loss function of the bearing condition assessment model is the cross-entropy loss function, and the optimizer is the Adam optimizer.
7. A bearing condition assessment system based on vibration signals, characterized in that, The system includes a memory and a processor. The memory includes a program for a bearing condition assessment method based on vibration signals. When the processor executes the program for the bearing condition assessment method based on vibration signals, it performs the following steps: The vibration sensor collects the original vibration signals of the bearing under different operating conditions, and preprocesses the original vibration signals. Based on the preprocessed original vibration information, time-domain features, frequency-domain features, and wavelet-domain features are extracted respectively. A combined feature filtering algorithm is used to filter time-domain features, frequency-domain features, and wavelet-domain features, eliminating redundant and invalid features to obtain the optimal feature set; The optimal feature set is subjected to feature fusion processing to generate a fused feature vector; A bearing condition assessment model based on deep learning is constructed. The fused feature vector is input into the bearing condition assessment model, and the bearing condition assessment result is output. The bearing condition assessment results are transmitted to the terminal in real time.
8. The bearing condition assessment system based on vibration signals according to claim 7, characterized in that, The original vibration signal is preprocessed, specifically including: The original vibration signal is acquired, and the original vibration signal is analyzed based on the information of abnormal change conditions. Abnormal signal segments caused by sensor malfunction or electromagnetic interference are removed from the original vibration signal, and valid signal segments are retained. An adaptive filtering algorithm is used to denoise the effective signal segment, filtering out environmental noise and equipment operation interference noise to obtain the denoised signal. The denoised signal is subjected to amplitude normalization processing to map the signal amplitude to a preset range, thereby eliminating the influence of signal amplitude differences under different operating conditions and obtaining a standardized signal. Calculate the signal-to-noise ratio (SNR) of the standardized signal and compare it with a preset SNR threshold. If the signal-to-noise ratio is lower than the preset threshold, the original vibration signal under the corresponding working condition will be re-acquired until a valid signal is obtained.
9. A bearing condition assessment system based on vibration signals according to claim 8, characterized in that, The time-domain features include peak value, peak factor, kurtosis, skewness, variance, standard deviation, mean, impulse factor, margin factor, and waveform factor. The frequency domain features include centroid frequency, mean square frequency, frequency variance, frequency standard deviation, peak energy, and peak frequency.
10. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes a bearing condition assessment method program based on vibration signals. When the bearing condition assessment method program based on vibration signals is executed by a processor, it implements the steps of the bearing condition assessment method based on vibration signals as described in any one of claims 1 to 6.