Transmission system bearing health state prediction method and system based on digital twinning

Abstract

The invention discloses a transmission system bearing health state prediction method and system based on digital twinning, and relates to the technical field of fault prediction.The method comprises the steps that firstly, geometric and finite element models and behavior rules of a bearing are fused to construct a digital twinning body of the bearing, and a real-time data interface is established to achieve dynamic mapping of physical parameters; secondly, vibration, temperature and rotating speed signals collected by a sensor are subjected to preprocessing and multi-dimensional feature extraction, and feature vectors representing the bearing state are formed; and then, inputting the vector into a prediction model composed of a depth time sequence convolutional network-attention mechanism and a bidirectional long and short time memory network, so as to output a health index of the bearing and calculate the remaining service life. And finally, the prediction result is visualized in a digital twinborn interface, and graded early warning information is automatically generated according to a preset health threshold. According to the invention, real-time and accurate prediction and visual monitoring of the health state of the bearing are realized, and decision support is provided for predictive maintenance of a transmission system.

Description

Technical Field

[0001] This application relates to the field of fault prediction technology, and in particular to a method and system for predicting the health status of bearings in transmission systems based on digital twins. Background Technology

[0002] Traditional methods for predicting bearing health status often rely on a single data-driven model. These models typically operate as "black boxes," lacking an understanding of the internal physical failure mechanisms of bearings. This results in poor generalization ability and inaccurate predictions when faced with non-stationary operating conditions such as load and speed changes.

[0003] The above content is only used to help understand the technical solution of the present invention and does not represent an admission that the above content is prior art. Summary of the Invention

[0004] The main objective of this application is to provide a method and system for predicting the health status of bearings in a transmission system based on digital twins, aiming to solve the technical problem of insufficient accuracy in predicting the health status of bearings in the prior art.

[0005] To achieve the above objectives, this application provides a method for predicting the health status of bearings in a transmission system based on digital twins. The method includes: A digital twin of the bearing is constructed based on its geometric model, finite element physical model, and behavioral rule library. Feature extraction is performed on the original vibration signal, original temperature signal and rotation speed signal of the bearing acquired by the sensor network to obtain a multi-dimensional feature vector; The feature vector is input into a pre-trained prediction model, which outputs the health index and remaining service life of the bearing. The health index and remaining lifespan are visualized in the digital twin, and early warning information is generated based on preset thresholds.

[0006] In one embodiment, the step of constructing a digital twin of the bearing based on its geometric model, finite element physical model, and behavioral rule library includes: The bearing's geometric model, finite element model, and behavioral rule library are aligned and coupled to obtain a static basic model; A data interface is established for the static basic model based on a standard communication protocol, wherein the data interface is used to receive real-time sensing data streams from the bearing; Based on the real-time sensing data stream received by the data interface, a dynamic mapping relationship is established between the real-time sensing data stream and the physical field parameters and state variables in the static basic model. A digital twin of the bearing is constructed based on the dynamic mapping relationship.

[0007] In one embodiment, the step of establishing a dynamic mapping relationship between the real-time sensing data stream received based on the data interface and the physical field parameters and state variables in the static basic model includes: Extract the vibration signal amplitude, temperature gradient, and rotational speed signal from the real-time sensing data stream received based on the data interface; The amplitude of the vibration signal is determined as the dynamic load boundary condition; The temperature gradient is mapped to the equivalent thermal stress influence factor in the material property parameters; The rotational speed signal is mapped to a time-varying distribution of contact stress between the roller and the raceway; Based on the dynamic load boundary conditions, the equivalent thermal stress influence factor, and the time-varying distribution of contact stress between the rollers and raceways, a dynamic mapping relationship is established between the real-time sensing data stream and the physical field parameters and state variables in the static basic model.

[0008] In one embodiment, the step of extracting features from the original vibration signal, original temperature signal, and rotational speed signal obtained by the sensor network from the bearing to obtain a multi-dimensional feature vector includes: The original vibration signal of the bearing acquired by the sensor network is obtained by wavelet threshold denoising to obtain the vibration signal. The original temperature signal obtained from the bearing by the sensor network is obtained by performing a sliding smoothing filter to obtain the temperature signal; The time-domain statistical characteristics, frequency-domain spectral characteristics, and envelope spectrum characteristics of the vibration signal are calculated in parallel. The time-domain statistical features, frequency-domain spectral features, envelope spectral features, temperature signal, and rotational speed signal obtained from the bearing by the sensor network are time-aligned and spliced ​​to obtain a multi-dimensional feature vector.

[0009] In one embodiment, the step of performing wavelet threshold denoising on the original vibration signal obtained from the bearing by the sensor network to obtain the vibration signal includes: Determine the current operating condition based on the speed signal; When the current working condition is a non-stationary working condition, the original vibration signal of the bearing collected by the sensor network is decomposed into multiple sub-bands based on adaptive wavelet packet decomposition, and noise is removed in each sub-band according to the wavelet threshold to obtain the vibration signal. When the current operating condition is a stable operating condition, the original vibration signal of the bearing acquired by the sensor network is subjected to global threshold denoising based on a fixed wavelet basis function to obtain the vibration signal.

[0010] In one embodiment, the step of inputting the feature vector into a pre-trained prediction model and outputting the health index and remaining service life of the bearing includes: The feature vector is input into the deep temporal convolutional network-attention mechanism layer of the pre-trained prediction model to extract deep temporal features. The prediction model includes a deep temporal convolutional network-attention mechanism layer, a bidirectional long short-term memory network layer, and a fully connected regression layer. The deep temporal convolutional network-attention mechanism layer is obtained by fusing and training a deep temporal convolutional network and an attention mechanism. The deep temporal features are input into a bidirectional long short-term memory network layer to capture the temporal dependency between the deep temporal features and the health status; The temporal dependency relationship between the deep temporal features and the health status is input into a fully connected regression layer, and the health index of the bearing is output. A degradation trajectory is generated based on the historical health index and the health index, and the remaining service life of the bearing is determined based on the degradation trajectory.

[0011] In one embodiment, the step of training a deep temporal convolutional network-attention mechanism layer includes: A training sample set was constructed based on historical full-life-cycle bearing monitoring data; Multi-scale features of the training sample set are extracted based on a temporal convolutional network. Key features of the multi-scale features are determined through an attention mechanism; The key features are adaptively weighted to obtain weighted key features; The weighted key features are trained based on the mean squared error loss function and the Adam optimizer, and a deep temporal convolutional network-attention mechanism layer is obtained when the convergence condition is met.

[0012] In one embodiment, the step of visualizing the health index and remaining lifespan in the digital twin and generating early warning information based on a preset threshold includes: In the three-dimensional visualization interface of the digital twin, the health index change trend and remaining service life prediction results of the bearing are displayed in real time in the form of dynamic curves. When the health index falls below a preset first threshold, a primary warning indicator is triggered in the digital twin interface; When the health index is lower than the preset second threshold or the remaining service life is lower than the set safety margin, an advanced warning sign is triggered, and a warning message containing bearing positioning information and recommended maintenance measures is automatically generated, wherein the first threshold is greater than the second threshold.

[0013] In one embodiment, the step of triggering an advanced warning indicator and automatically generating warning information including bearing positioning information and recommended maintenance measures when the health index is lower than a preset second threshold or the remaining service life is lower than a set safety margin includes: When the health index is lower than the preset second threshold or the remaining service life is lower than the set safety margin, an early warning event is generated, triggering an advanced early warning flag. The early warning event information includes the current time information, trigger condition type, health index, and remaining service life. When an advanced warning sign is detected, the positioning information of the bearing is determined, including the equipment code, installation location, production line number, and three-dimensional spatial coordinates. Determine the degradation information of the health index or the remaining service life, and determine recommended maintenance measures based on the degradation information; The warning information is generated based on the warning event information, the bearing positioning information, and the recommended maintenance measures.

[0014] Furthermore, to achieve the above objectives, this application also proposes a digital twin-based transmission system bearing health condition prediction device, which includes: The bearing modeling module is used to construct a digital twin of the bearing based on its geometric model, finite element physical model, and behavioral rule library. The feature extraction module is used to extract features from the original vibration signal, original temperature signal and rotation speed signal of the bearing acquired by the sensor network, and obtain a multi-dimensional feature vector. The status prediction module is used to input the feature vector into a pre-trained prediction model and output the health index and remaining service life of the bearing. The early warning display module is used to visualize the health index and remaining lifespan in the digital twin and generate early warning information based on a preset threshold.

[0015] Furthermore, to achieve the above objectives, this application also proposes a digital twin-based transmission system bearing health status prediction device, which includes: a memory, a processor, and a computer program stored in the memory and executable on the processor. The computer program is configured to implement the steps of the digital twin-based transmission system bearing health status prediction method described above.

[0016] In addition, to achieve the above objectives, the present invention also proposes a storage medium, which is a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, it implements the steps of the above-described method for predicting the health status of transmission system bearings based on digital twins.

[0017] In addition, to achieve the above objectives, this application also provides a computer program product, which includes a computer program that, when executed by a processor, implements the steps of the digital twin-based transmission system bearing health status prediction method described above.

[0018] This application provides a digital twin-based method for predicting the health status of bearings in transmission systems. First, it constructs a digital twin of the bearing by integrating its geometry, finite element model, and behavioral rules, and establishes a real-time data interface to achieve dynamic mapping of physical parameters. Second, it preprocesses and extracts multi-dimensional features from vibration, temperature, and speed signals collected by sensors to form a feature vector characterizing the bearing's state. Then, this vector is input into a prediction model composed of a deep temporal convolutional network with an attention mechanism and a bidirectional long short-term memory network to output a bearing health index and calculate its remaining service life. Finally, the prediction results are visualized in the digital twin interface, and graded early warning information is automatically generated based on preset health thresholds. This application achieves real-time, accurate prediction and visualized monitoring of bearing health status, providing decision support for predictive maintenance of transmission systems. Attached Figure Description

[0019] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0020] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0021] Figure 1 This is a flowchart illustrating Embodiment 1 of the transmission system bearing health status prediction method based on digital twins of this application; Figure 2 This is a schematic diagram of the trend curve of an embodiment of the transmission system bearing health status prediction method based on digital twins in this application; Figure 3 This is a schematic diagram of the module structure of the transmission system bearing health status prediction device based on digital twin according to an embodiment of this application; Figure 4 This is a schematic diagram of the equipment structure of the hardware operating environment involved in the digital twin-based transmission system bearing health status prediction method in this application embodiment.

[0022] The realization of the purpose, functional features and advantages of this application will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0023] It should be understood that the specific embodiments described herein are merely illustrative of the technical solutions of this application and are not intended to limit this application.

[0024] To better understand the technical solution of this application, a detailed description will be provided below in conjunction with the accompanying drawings and specific implementation methods.

[0025] The main solution of this application embodiment is: to construct a digital twin of the bearing based on the bearing's geometric model, finite element physical model, and behavioral rule library; Feature extraction is performed on the original vibration signal, original temperature signal and rotation speed signal of the bearing acquired by the sensor network to obtain a multi-dimensional feature vector; The feature vector is input into a pre-trained prediction model, which outputs the health index and remaining service life of the bearing. The health index and remaining lifespan are visualized in the digital twin, and early warning information is generated based on preset thresholds.

[0026] Currently, traditional methods for predicting bearing health status often rely on a single data-driven model. These models typically operate as "black boxes," lacking integration of internal physical failure mechanisms of bearings, resulting in poor generalization ability and inaccurate predictions when faced with non-stationary operating conditions such as load and speed changes.

[0027] This application provides a solution that firstly, integrates the bearing's geometry, finite element model, and behavioral rules to construct its digital twin, and establishes a real-time data interface to achieve dynamic mapping of physical parameters. Secondly, it preprocesses and extracts multi-dimensional features from vibration, temperature, and speed signals collected by sensors to form a feature vector characterizing the bearing's state. Then, this vector is input into a prediction model composed of a deep temporal convolutional network with an attention mechanism and a bidirectional long short-term memory network to output the bearing's health index and calculate its remaining service life. Finally, the prediction results are visualized in the digital twin interface, and graded early warning information is automatically generated based on preset health thresholds. This application achieves real-time, accurate prediction and visualized monitoring of bearing health status, providing decision support for predictive maintenance of transmission systems.

[0028] It should be noted that the executing entity in this embodiment can be a computing service device with data processing, network communication, and program execution functions, such as a tablet computer, personal computer, or mobile phone, or an electronic device capable of performing the above functions, such as a digital twin-based transmission system bearing health status prediction device. This embodiment does not specifically limit it in this regard. The following uses a digital twin-based transmission system bearing health status prediction device as an example to describe this embodiment and the following embodiments.

[0029] All actions involving the acquisition of signals, information, or data in this application are carried out in accordance with the relevant data protection laws and policies of the country where the application is located, and with the authorization of the owner of the relevant device.

[0030] This application provides a method for predicting the health status of bearings in a transmission system based on digital twins, referring to... Figure 1 , Figure 1 This is a flowchart illustrating the first embodiment of the transmission system bearing health status prediction method based on digital twins of this application.

[0031] In this embodiment, the transmission system bearing health status prediction method based on digital twins includes steps S10~S40: Step S10: Construct a digital twin of the bearing based on its geometric model, finite element physical model, and behavioral rule library.

[0032] It should be noted that a digital twin refers to a dynamic mapping model of the entire lifecycle of a physical bearing in virtual space.

[0033] Understandably, when creating a digital twin of a bearing, the first step is to use CAD software to generate a precise bearing geometric model, which is then imported into finite element analysis software to establish a parametric mechanical and thermal simulation model, forming the static foundation. Subsequently, a data interface is established using standard industrial communication protocols such as OPC UA or MQTT to connect real-time sensor data from the physical bearing. A dynamic mapping relationship is then established between the real-time data stream and the physical parameters in the simulation model; for example, the amplitude of vibration signals is mapped to dynamic loads in the finite element model, enabling the digital twin to simulate real-world operating conditions in real time.

[0034] In one feasible implementation, the step of constructing a digital twin of the bearing based on its geometric model, finite element physical model, and behavioral rule library includes: The bearing's geometric model, finite element model, and behavioral rule library are aligned and coupled to obtain a static basic model; A data interface is established for the static basic model based on a standard communication protocol, wherein the data interface is used to receive real-time sensing data streams from the bearing; Based on the real-time sensing data stream received by the data interface, a dynamic mapping relationship is established between the real-time sensing data stream and the physical field parameters and state variables in the static basic model. A digital twin of the bearing is constructed based on the dynamic mapping relationship.

[0035] In practical implementation, the first step is to establish a data interface for the static foundation model based on a standard communication protocol. This interface receives real-time sensor data streams from the bearing and ensures, through data alignment, that the coordinate system, mesh generation, and boundary conditions of the geometric model are completely consistent with those of the finite element analysis. Furthermore, the logic in the behavioral rule base is encoded into functions that can be called by the model, forming the static foundation model. Subsequently, the real-time sensor data streams are used to solve for or correct key physical field parameters in the static foundation model through physical equations or data-driven algorithms. For example, the relationship between force and acceleration can be used... Direct load mapping. During data mapping, when the mapped state variable triggers its preset rules, the digital twin can automatically perform fault diagnosis or lifespan prediction through a behavioral rule library. A digital twin of the bearing is constructed based on this dynamic mapping relationship.

[0036] In one feasible implementation, the step of establishing a dynamic mapping relationship between the real-time sensing data stream received based on the data interface and the physical field parameters and state variables in the static basic model includes: Extract the vibration signal amplitude, temperature gradient, and rotational speed signal from the real-time sensing data stream received based on the data interface; The amplitude of the vibration signal is determined as the dynamic load boundary condition; The temperature gradient is mapped to the equivalent thermal stress influence factor in the material property parameters; The rotational speed signal is mapped to a time-varying distribution of contact stress between the roller and the raceway; Based on the dynamic load boundary conditions, the equivalent thermal stress influence factor, and the time-varying distribution of contact stress between the rollers and raceways, a dynamic mapping relationship is established between the real-time sensing data stream and the physical field parameters and state variables in the static basic model.

[0037] In practical implementation, the raw sensor signal first needs to be preprocessed and its features extracted. For example, the amplitude of the vibration signal. This can be directly mapped to the dynamic load boundary conditions acting on the bearing using Newton's second law. The mapping relationship can be simplified as follows:

[0038] in, For equivalent quality, This represents the root mean square value of the real-time vibration acceleration. Temperature gradient. It affects properties such as the elastic modulus of a material, which can be determined by the formula:

[0039] Map it to the equivalent thermal stress influence factor This corrects the material constitutive relations in the finite element model.

[0040] For the more complex time-varying distribution of contact stress in rollers and raceways, the mapping relationship needs to be combined with kinematics and contact mechanics. The rotational speed signal n(t) determines the passing frequency of the rolling element, and the number of contact cycles is calculated based on this. Combining the dynamic load derived from vibration, the maximum contact stress can be calculated using Hertzian contact theory. And its distribution. The formula is:

[0041] in, For contact length, The equivalent radius of curvature is used. In this way, a simple rotational speed signal can be transformed into a detailed physical field distribution that varies over time and reflects the internal microscopic mechanical state. Finally, these real-time mapped parameters are used as updated boundary conditions and material properties and substituted into the static basic model for solution, establishing a dynamic mapping relationship between the real-time sensing data stream and the physical field parameters and state variables in the static basic model.

[0042] Step S20: Extract features from the original vibration signal, original temperature signal and rotational speed signal of the bearing acquired by the sensor network to obtain a multidimensional feature vector.

[0043] Understandably, the original signal is first preprocessed, such as using a bandpass filter to eliminate low-frequency trend terms and high-frequency noise. Then, for vibration signals, statistical quantities such as mean, root mean square (RMS), peak value, and kurtosis are typically calculated in the time domain to characterize the overall energy and impact characteristics; in the frequency domain, the characteristic frequencies of each bearing component and the amplitude of its harmonics are extracted using a Fast Fourier Transform (FFT); for temperature signals, their instantaneous values, rate of change, and temperature difference with the environment are calculated; and for speed signals, the speed itself and its fluctuation rate are calculated. Finally, these feature values ​​extracted from different signal sources and dimensions are arranged in order and combined into a fixed-length multidimensional feature vector.

[0044] In one feasible implementation, the step of extracting features from the original vibration signal, original temperature signal, and rotational speed signal of the bearing acquired by the sensor network to obtain a multidimensional feature vector includes: The original vibration signal of the bearing acquired by the sensor network is obtained by wavelet threshold denoising to obtain the vibration signal. The original temperature signal obtained from the bearing by the sensor network is obtained by performing a sliding smoothing filter to obtain the temperature signal; The time-domain statistical characteristics, frequency-domain spectral characteristics, and envelope spectrum characteristics of the vibration signal are calculated in parallel. The time-domain statistical features, frequency-domain spectral features, envelope spectral features, temperature signal, and rotational speed signal obtained from the bearing by the sensor network are time-aligned and spliced ​​to obtain a multi-dimensional feature vector.

[0045] In practical implementation, targeted signal preprocessing is first performed. For non-stationary, noisy original vibration signals, wavelet thresholding denoising is used. The core of this method is to select a suitable wavelet basis (such as Db4) to decompose the signal, and then apply a threshold function to the obtained wavelet coefficients. A smooth vibration signal is obtained by reconstructing the noise figure. The formula is:

[0046] Where T is the threshold.

[0047] For the raw temperature signal that changes relatively slowly, a simple sliding smoothing filter is used, with the following formula:

[0048] After preprocessing, the clean vibration signal is subjected to parallel computation to efficiently extract multi-dimensional features. Time-domain statistical features include root mean square value and kurtosis. Frequency-domain spectral features are calculated using Fast Fourier Transform (FFT) to extract the amplitude of bearing fault characteristic frequencies; envelope spectrum analysis first performs a Hilbert transform on the vibration signal to obtain the envelope. Then, performing an FFT on the envelope effectively demodulates the fault characteristic frequencies modulated by high-frequency resonance, which is particularly sensitive to early damage. Finally, all these features are strictly time-stamped with the speed and temperature signals and concatenated into a long multidimensional feature vector.

[0049] In one feasible implementation, the step of performing wavelet threshold denoising on the original vibration signal of the bearing acquired by the sensor network to obtain the vibration signal includes: Determine the current operating condition based on the speed signal; When the current working condition is a non-stationary working condition, the original vibration signal of the bearing collected by the sensor network is decomposed into multiple sub-bands based on adaptive wavelet packet decomposition, and noise is removed in each sub-band according to the wavelet threshold to obtain the vibration signal. When the current operating condition is a stable operating condition, the original vibration signal of the bearing acquired by the sensor network is subjected to global threshold denoising based on a fixed wavelet basis function to obtain the vibration signal.

[0050] In practical implementation, the current operating condition needs to be determined based on the speed signal. This is typically achieved by calculating the standard deviation or rate of change of the speed signal within a time window. If this value is below a preset threshold (i.e., the speed is stable), it is determined to be a stationary operating condition; conversely, if the speed fluctuates significantly, it is determined to be a non-stationary operating condition. For stationary operating conditions, the signal frequency components are relatively stable. A computationally efficient fixed wavelet basis function global threshold denoising method is used. The signal is decomposed using pre-selected basis functions, and a uniform threshold is applied to the detail coefficients at all levels. The signal is processed and finally reconstructed to obtain the denoised signal, where For noise standard deviation estimation, This is the signal length.

[0051] For non-stationary operating conditions, the frequency components of a signal change rapidly over time, and standard wavelet decomposition may obscure fault characteristics due to fixed frequency band division. Therefore, adaptive wavelet packet decomposition is used, which can perform a more refined secondary decomposition of the high-frequency components, forming a complete wavelet packet decomposition tree, thereby providing higher time-frequency resolution. During denoising, each sub-band obtained from the decomposition... (j is the scale, k is the frequency index) Calculate and apply an adaptive threshold related to the energy of the subband separately, and then perform wavelet packet reconstruction.

[0052] Step S30: Input the feature vector into the pre-trained prediction model and output the health index and remaining service life of the bearing.

[0053] It should be noted that the health index refers to the degree of degradation of a bearing from healthy to complete failure. Remaining service life is a specific value, expressed in time (e.g., hours) or operating cycles, based on the current condition and future degradation trends. It represents the predicted remaining safe operating time of the bearing before reaching the failure threshold.

[0054] In its implementation, the pre-trained prediction model constructs a regression predictor of the bearing's health status by learning the complex nonlinear mapping relationship between feature vectors from a large amount of historical operating data and the actual degradation trajectory of the bearing. The model takes the feature vector sequence from the previous time window as input, and its output layer is typically designed with two branches: one branch outputs an instantaneous health index between 0 and 1 through a sigmoid activation function, reflecting the current state assessment; the other branch combines the degradation trend of the health index and directly regresses the specific remaining service life value through a fully connected layer or sequence prediction. This prediction comprehensively considers factors such as operating load and historical degradation rate, achieving end-to-end computation from state assessment to lifespan prediction.

[0055] In one feasible implementation, the step of inputting the feature vector into a pre-trained prediction model and outputting the health index and remaining service life of the bearing includes: The feature vector is input into the deep temporal convolutional network-attention mechanism layer of the pre-trained prediction model to extract deep temporal features. The prediction model includes a deep temporal convolutional network-attention mechanism layer, a bidirectional long short-term memory network layer, and a fully connected regression layer. The deep temporal convolutional network-attention mechanism layer is obtained by fusing and training a deep temporal convolutional network and an attention mechanism. The deep temporal features are input into a bidirectional long short-term memory network layer to capture the temporal dependency between the deep temporal features and the health status; The temporal dependency relationship between the deep temporal features and the health status is input into a fully connected regression layer, and the health index of the bearing is output. A degradation trajectory is generated based on the historical health index and the health index, and the remaining service life of the bearing is determined based on the degradation trajectory.

[0056] In its implementation, the feature vector sequence is first input into a deep temporal convolutional network (TCN) with an attention mechanism layer. The TCN expands the receptive field through dilated causal convolution, and its output... It can efficiently capture long-term local patterns in sequences. Subsequently, an attention mechanism was introduced, which computes a weight for the features of each time step in the TCN output. And perform a weighted summation of the features at all time steps:

[0057] This allows the model to dynamically focus on historical moments most relevant to the current bearing degradation state, highlighting key information and suppressing noise interference.

[0058] Next, the features are fed into a Bidirectional Long Short-Term Memory (Bi-LSTM) layer. Bi-LSTM, through its unique gating mechanism (input gate, forget gate, output gate) and cell state, scans the sequence from both forward and reverse directions, thereby capturing complex temporal dependencies between historical and future contextual information, such as the inertia, acceleration points, or plateaus of the degradation process. The final hidden state of the Bi-LSTM... It incorporates all the contextual information of the sequence.

[0059] at last, The data is fed into a fully connected regression layer, which is typically one or more fully connected neural networks. Through linear transformations and non-linear activation functions (such as the sigmoid function), the high-dimensional features are mapped to a scalar value, directly outputting a health index (HI) ranging from [0,1]. Determining the remaining useful life (RUL) is a post-processing step: the system records and fits a series of continuous health index points, forming a degradation trajectory. When this trajectory crosses a preset failure threshold (e.g., HI=0.2), the time from the current moment to that crossing point is predicted as the RUL, usually using simple models such as linear or exponential regression for extrapolation.

[0060] In one feasible implementation, the steps for training a deep temporal convolutional network-attention mechanism layer include: A training sample set was constructed based on historical full-life-cycle bearing monitoring data; Multi-scale features of the training sample set are extracted based on a temporal convolutional network. Key features of the multi-scale features are determined through an attention mechanism; The key features are adaptively weighted to obtain weighted key features; The weighted key features are trained based on the mean squared error loss function and the Adam optimizer, and a deep temporal convolutional network-attention mechanism layer is obtained when the convergence condition is met.

[0061] In the specific implementation, a training sample set is first constructed based on historical full-lifecycle bearing monitoring data. Each sample is a time-series feature vector segment and its corresponding health index label. The samples are input into the TCN network, which achieves multi-scale feature extraction by stacking multiple dilated convolutional layers. Let the... The expansion factor of the layer is The kernel size is Then the layer at time step The output is:

[0062] in The activation function is used. By stacking multiple layers with different inflation factors, TCN can obtain multi-scale feature representations from short-term fluctuations to long-term trends while maintaining temporal causality. .

[0063] Subsequently, an attention mechanism is introduced to adaptively weight the multi-scale features output by the TCN. First, the attention score at each time step t is calculated. ,in and These are learnable parameters. The attention weights are then obtained by normalization using the Softmax function.

[0064] Attention weights reflect the importance of features at different time points for predicting health status. Finally, a weighted sum of all features is performed to obtain the focused key feature representation:

[0065] The entire network uses mean squared error (MSE) as the loss function:

[0066] The Adam optimizer is used for end-to-end training. The network parameters of the TCN convolution weights and attention mechanism are optimized simultaneously through backpropagation until the loss function converges, and finally the trained deep temporal convolutional network-attention mechanism layer is obtained.

[0067] Step S40: Visualize the health index and remaining lifespan in the digital twin, and generate early warning information based on a preset threshold.

[0068] It should be noted that the early warning information is a status notification or alarm that is automatically triggered by the system based on a preset health index or remaining lifespan threshold, and is used to indicate potential risks.

[0069] Understandably, the health index and remaining lifespan output by the predictive model are synchronized in real time to the digital twin via a data interface, and then rendered and displayed intuitively in its 3D visualization interface using dashboards, progress bars, color changes (such as from green to red), or trend curves. (Refer to...) Figure 2 , Figure 2 This is a schematic diagram of the trend curve. Simultaneously, the system backend continuously compares the real-time health index with preset attention thresholds and alarm thresholds, or compares the remaining service life with preset safety margin thresholds. Once the monitored data exceeds the corresponding threshold, an early warning process is automatically triggered, highlighting the warning on the visualization interface and generating structured early warning information containing the device number, warning level, and suggested measures, which is then pushed to the operation and maintenance management system via API.

[0070] In one feasible implementation, the step of visualizing the health index and remaining lifespan in the digital twin and generating early warning information based on a preset threshold includes: In the three-dimensional visualization interface of the digital twin, the health index change trend and remaining service life prediction results of the bearing are displayed in real time in the form of dynamic curves. When the health index falls below a preset first threshold, a primary warning indicator is triggered in the digital twin interface; When the health index is lower than the preset second threshold or the remaining service life is lower than the set safety margin, an advanced warning sign is triggered, and a warning message containing bearing positioning information and recommended maintenance measures is automatically generated, wherein the first threshold is greater than the second threshold.

[0071] In its implementation, the system uses a constructed digital twin 3D scene and integrates a chart library to plot the dynamic curve HI(t) of the health index, prominently displaying the current predicted RUL value. The horizontal axis of the curve represents time, and the vertical axis represents the health index (usually normalized to [0,1]), visually representing the bearing's performance degradation trajectory. The early warning logic is triggered by real-time monitoring of the health index sequence and RUL value. The system presets two decreasing thresholds: a first threshold T1 (e.g., HI=0.7) and a second threshold T2 (e.g., HI=0.3), and sets a RUL safety margin threshold Trul (e.g., 1.5 years).

[0072] The specific warning trigger conditions can be formally represented as: Primary warning: When HI(t) ≤ T1 and HI(t) > T2, the system activates a yellow flashing mark or icon on the components associated with the 3D model.

[0073] Advanced Warning: When HI(t) ≤ T2 or RUL(t) ≤ Trul, the system triggers a red alarm indicator and immediately generates structured warning information. This information not only includes the warning level and trigger time, but also automatically associates the bearing's precise location in three-dimensional space (such as equipment number and installation location coordinates), and invokes preset maintenance suggestions based on historical fault data or expert knowledge base.

[0074] In one feasible implementation, the step of triggering an advanced warning indicator and automatically generating warning information containing bearing positioning information and recommended maintenance measures when the health index is lower than a preset second threshold or the remaining service life is lower than a set safety margin includes: When the health index is lower than the preset second threshold or the remaining service life is lower than the set safety margin, an early warning event is generated, triggering an advanced early warning flag. The early warning event information includes the current time information, trigger condition type, health index, and remaining service life. When an advanced warning sign is detected, the positioning information of the bearing is determined, including the equipment code, installation location, production line number, and three-dimensional spatial coordinates. Determine the degradation information of the health index or the remaining service life, and determine recommended maintenance measures based on the degradation information; The warning information is generated based on the warning event information, the bearing positioning information, and the recommended maintenance measures.

[0075] In terms of technical implementation, triggering an advanced early warning is a multi-step automated process. The system continuously monitors the health index HI(t) and the predicted remaining useful life RUL(t). The warning is triggered when the logical OR condition HI(t) ≤ T2 or RUL(t) ≤ T2 is met. Upon receiving the warning, the system immediately instantiates a warning event object. The object's core attributes (warning event information) are populated, including: timestamp ( Triggering condition type (Value is "HI" or "RUL") Health Index at the trigger time HI ( ) and remaining useful life RUL ( Meanwhile, in the 3D scene of the digital twin, the system triggers advanced warning indicators by changing the appearance and material of the corresponding bearing model (e.g., turning it red and flashing at high frequency) or attaching warning icons. Subsequently, based on the bearing's unique equipment code, the system retrieves its location information from the equipment asset database, including precise 3D spatial coordinates (x, y, z), production line number, and installation location description. Next, the system analyzes recent window data... Historical data within t is used to calculate deterioration information, such as the average rate of decline in health indicators. Or, the rate of consumption of the remaining useful life. According to and degradation rate (e.g.) The system performs a matching query from a predefined maintenance strategy knowledge base, considering the size of the data. Finally, the system encapsulates the three types of data—early warning event information, location information, and suggested maintenance measures—into a complete and readable early warning message (usually in JSON or XML format), and pushes it to the operations and maintenance management system and relevant personnel via message queue or API interface, thus forming a closed loop from automatic detection to the generation of executable guidance.

[0076] This embodiment provides a method for predicting the health status of bearings in a transmission system based on digital twins. First, a digital twin of the bearing is constructed by integrating its geometry, finite element model, and behavioral rules, and a real-time data interface is established to achieve dynamic mapping of physical parameters. Second, vibration, temperature, and speed signals collected by sensors are preprocessed and multi-dimensional features are extracted to form a feature vector characterizing the bearing's state. Then, this vector is input into a prediction model composed of a deep temporal convolutional network with an attention mechanism and a bidirectional long short-term memory network to output a bearing health index and calculate the remaining service life. Finally, the prediction results are visualized in the digital twin interface, and graded early warning information is automatically generated based on preset health thresholds. This application achieves real-time, accurate prediction and visualized monitoring of bearing health status, providing decision support for predictive maintenance of transmission systems.

[0077] It should be noted that the above examples are only for understanding this application and do not constitute a limitation on the digital twin-based transmission system bearing health status prediction method of this application. Any simple modifications based on this technical concept are within the protection scope of this application.

[0078] This application also provides a transmission system bearing health status prediction device based on digital twins, please refer to... Figure 3 The digital twin-based transmission system bearing health status prediction device includes: The bearing modeling module 10 is used to construct a digital twin of the bearing based on the bearing's geometric model, finite element physical model, and behavioral rule library. Feature extraction module 20 is used to extract features from the original vibration signal, original temperature signal and rotation speed signal of the bearing acquired by the sensor network to obtain a multi-dimensional feature vector; The state prediction module 30 is used to input the feature vector into a pre-trained prediction model and output the health index and remaining service life of the bearing. The early warning display module 40 is used to visualize the health index and remaining lifespan in the digital twin and generate early warning information based on a preset threshold.

[0079] In one feasible implementation, the bearing modeling module 10 is further used to align and couple the bearing's geometric model, finite element model, and behavioral rule library to obtain a static basic model. A data interface is established for the static basic model based on a standard communication protocol, wherein the data interface is used to receive real-time sensing data streams from the bearing; Based on the real-time sensing data stream received by the data interface, a dynamic mapping relationship is established between the real-time sensing data stream and the physical field parameters and state variables in the static basic model. A digital twin of the bearing is constructed based on the dynamic mapping relationship.

[0080] In one feasible implementation, the bearing modeling module 10 is further used to extract the vibration signal amplitude, temperature gradient, and rotational speed signal from the real-time sensing data stream received based on the data interface. The amplitude of the vibration signal is determined as the dynamic load boundary condition; The temperature gradient is mapped to the equivalent thermal stress influence factor in the material property parameters; The rotational speed signal is mapped to a time-varying distribution of contact stress between the roller and the raceway; Based on the dynamic load boundary conditions, the equivalent thermal stress influence factor, and the time-varying distribution of contact stress between the rollers and raceways, a dynamic mapping relationship is established between the real-time sensing data stream and the physical field parameters and state variables in the static basic model.

[0081] In one feasible implementation, the feature extraction module 20 is further used to perform wavelet threshold denoising on the original vibration signal obtained by the sensor network from the bearing to obtain a vibration signal; The original temperature signal obtained from the bearing by the sensor network is obtained by performing a sliding smoothing filter to obtain the temperature signal; The time-domain statistical characteristics, frequency-domain spectral characteristics, and envelope spectrum characteristics of the vibration signal are calculated in parallel. The time-domain statistical features, frequency-domain spectral features, envelope spectral features, temperature signal, and rotational speed signal obtained from the bearing by the sensor network are time-aligned and spliced ​​to obtain a multi-dimensional feature vector.

[0082] In one feasible implementation, the feature extraction module 20 is further used to determine the current operating condition based on the rotation speed signal; When the current working condition is a non-stationary working condition, the original vibration signal of the bearing collected by the sensor network is decomposed into multiple sub-bands based on adaptive wavelet packet decomposition, and noise is removed in each sub-band according to the wavelet threshold to obtain the vibration signal. When the current operating condition is a stable operating condition, the original vibration signal of the bearing acquired by the sensor network is subjected to global threshold denoising based on a fixed wavelet basis function to obtain the vibration signal.

[0083] In one feasible implementation, the state prediction module 30 is further configured to input the feature vector into the deep temporal convolutional network-attention mechanism layer of the pre-trained prediction model to extract deep temporal features. The prediction model includes a deep temporal convolutional network-attention mechanism layer, a bidirectional long short-term memory network layer, and a fully connected regression layer. The deep temporal convolutional network-attention mechanism layer is obtained by fusing and training a deep temporal convolutional network and an attention mechanism. The deep temporal features are input into a bidirectional long short-term memory network layer to capture the temporal dependency between the deep temporal features and the health status; The temporal dependency relationship between the deep temporal features and the health status is input into a fully connected regression layer, and the health index of the bearing is output. A degradation trajectory is generated based on the historical health index and the health index, and the remaining service life of the bearing is determined based on the degradation trajectory.

[0084] In one feasible implementation, the state prediction module 30 is further configured to construct a training sample set based on historical full life cycle bearing monitoring data; Multi-scale features of the training sample set are extracted based on a temporal convolutional network. Key features of the multi-scale features are determined through an attention mechanism; The key features are adaptively weighted to obtain weighted key features; The weighted key features are trained based on the mean squared error loss function and the Adam optimizer, and a deep temporal convolutional network-attention mechanism layer is obtained when the convergence condition is met.

[0085] In one feasible implementation, the early warning display module 40 is also used to display the health index change trend and remaining service life prediction results of the bearing in real time in the form of dynamic curves in the three-dimensional visualization interface of the digital twin. When the health index falls below a preset first threshold, a primary warning indicator is triggered in the digital twin interface; When the health index is lower than the preset second threshold or the remaining service life is lower than the set safety margin, an advanced warning sign is triggered, and a warning message containing bearing positioning information and recommended maintenance measures is automatically generated, wherein the first threshold is greater than the second threshold.

[0086] In one feasible implementation, the early warning display module 40 is further configured to generate early warning event information and trigger an advanced early warning identifier when the health index is lower than a preset second threshold or the remaining service life is lower than a set safety margin. The early warning event information includes the current time information, trigger condition type, health index, and remaining service life. When an advanced warning sign is detected, the positioning information of the bearing is determined, including the equipment code, installation location, production line number, and three-dimensional spatial coordinates. Determine the degradation information of the health index or the remaining service life, and determine recommended maintenance measures based on the degradation information; The warning information is generated based on the warning event information, the bearing positioning information, and the recommended maintenance measures.

[0087] The digital twin-based transmission system bearing health status prediction device provided in this application employs the digital twin-based transmission system bearing health status prediction method described in the above embodiments, and can solve the technical problem of insufficient bearing health status prediction accuracy. Compared with the prior art, the beneficial effects of the digital twin-based transmission system bearing health status prediction device provided in this application are the same as those of the digital twin-based transmission system bearing health status prediction method provided in the above embodiments, and other technical features in the digital twin-based transmission system bearing health status prediction device are the same as those disclosed in the methods of the above embodiments, and will not be repeated here.

[0088] This application provides a digital twin-based transmission system bearing health status prediction device. The digital twin-based transmission system bearing health status prediction device includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to execute the digital twin-based transmission system bearing health status prediction method in Embodiment 1 above.

[0089] The following is for reference. Figure 4 The diagram illustrates a structural schematic of a digital twin-based transmission system bearing health status prediction device suitable for implementing embodiments of this application. The digital twin-based transmission system bearing health status prediction device in the embodiments of this application may include, but is not limited to, mobile terminals such as mobile phones, laptops, digital radio receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Descriptions), PMPs (Portable Media Players), and in-vehicle terminals (e.g., in-vehicle navigation terminals), as well as fixed terminals such as digital TVs and desktop computers. Figure 4 The digital twin-based transmission system bearing health status prediction device shown is merely an example and should not impose any limitations on the functionality and scope of use of the embodiments of this application.

[0090] like Figure 4As shown, the digital twin-based transmission system bearing health condition prediction device may include a processing unit 1001 (e.g., a central processing unit, a graphics processing unit, etc.), which can perform various appropriate actions and processes according to a program stored in ROM (Read Only Memory) 1002 or a program loaded from storage device 1003 into RAM (Random Access Memory) 1004. RAM 1004 also stores various programs and data required for the operation of the digital twin-based transmission system bearing health condition prediction device. The processing unit 1001, ROM 1002, and RAM 1004 are interconnected via bus 1005. Input / output (I / O) interface 1006 is also connected to the bus. Typically, the following systems can be connected to I / O interface 1006: input devices 1007 including, for example, touchscreens, touchpads, keyboards, mice, image sensors, microphones, accelerometers, gyroscopes, etc.; output devices 1008 including, for example, LCDs (Liquid Crystal Displays), speakers, vibrators, etc.; storage devices 1003 including, for example, magnetic tapes, hard disks, etc.; and communication devices 1009. Communication device 1009 allows the digital twin-based transmission system bearing health condition prediction device to communicate wirelessly or wiredly with other devices to exchange data. Although the figure shows a digital twin-based transmission system bearing health condition prediction device with various systems, it should be understood that it is not required to implement or possess all the systems shown. More or fewer systems can be implemented alternatively.

[0091] Specifically, according to the embodiments disclosed in this application, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments disclosed in this application include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication device, or installed from storage device 1003, or installed from ROM 1002. When the computer program is executed by processing device 1001, it performs the functions defined in the methods of the embodiments disclosed in this application.

[0092] The digital twin-based transmission system bearing health status prediction device provided in this application employs the digital twin-based transmission system bearing health status prediction method described in the above embodiments, and can solve the technical problem of predicting the health status of transmission system bearings based on digital twins. Compared with the prior art, the beneficial effects of the digital twin-based transmission system bearing health status prediction device provided in this application are the same as those of the digital twin-based transmission system bearing health status prediction method provided in the above embodiments, and other technical features in this digital twin-based transmission system bearing health status prediction device are the same as those disclosed in the previous embodiment method, and will not be repeated here.

[0093] It should be understood that the various parts disclosed in this application can be implemented using hardware, software, firmware, or a combination thereof. In the description of the above embodiments, specific features, structures, materials, or characteristics can be combined in any suitable manner in one or more embodiments or examples.

[0094] The above are merely specific embodiments of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0095] This application provides a computer-readable storage medium having computer-readable program instructions (i.e., a computer program) stored thereon, which are used to execute the transmission system bearing health status prediction method based on digital twins in the above embodiments.

[0096] The computer-readable storage medium provided in this application may be, for example, a USB flash drive, but is not limited to, electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices, or any combination thereof. More specific examples of computer-readable storage media may include, but are not limited to: electrical connections having one or more wires, portable computer disks, hard disks, RAM (Random Access Memory), ROM (Read Only Memory), EPROM (Erasable Programmable Read Only Memory or Flash Memory), optical fibers, CD-ROM (CD-Read Only Memory), optical storage devices, magnetic storage devices, or any suitable combination thereof. In this embodiment, the computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, system, or device. The program code contained on the computer-readable storage medium may be transmitted using any suitable medium, including but not limited to: wires, optical cables, RF (Radio Frequency), etc., or any suitable combination thereof.

[0097] The aforementioned computer-readable storage medium may be included in a digital twin-based transmission system bearing health condition prediction device; or it may exist independently and not be assembled into a digital twin-based transmission system bearing health condition prediction device.

[0098] The aforementioned computer-readable storage medium carries one or more programs, which, when executed by the digital twin-based transmission system bearing health status prediction device, cause the digital twin-based transmission system bearing health status prediction device to: construct a digital twin of the bearing based on the bearing's geometric model, finite element physical model, and behavioral rule library; Feature extraction is performed on the original vibration signal, original temperature signal and rotation speed signal of the bearing acquired by the sensor network to obtain a multi-dimensional feature vector; The feature vector is input into a pre-trained prediction model, which outputs the health index and remaining service life of the bearing. The health index and remaining lifespan are visualized in the digital twin, and early warning information is generated based on preset thresholds.

[0099] Computer program code for performing the operations of this application can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, and C++, as well as conventional procedural programming languages ​​such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including LAN (Local Area Network) or WAN (Wide Area Network)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0100] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0101] The modules described in the embodiments of this application can be implemented in software or hardware. The names of the modules do not necessarily limit the functionality of the unit itself.

[0102] The readable storage medium provided in this application is a computer-readable storage medium that stores computer-readable program instructions (i.e., a computer program) for executing the above-described digital twin-based transmission system bearing health status prediction method, thereby solving the technical problem of digital twin-based transmission system bearing health status prediction. Compared with the prior art, the beneficial effects of the computer-readable storage medium provided in this application are the same as those of the digital twin-based transmission system bearing health status prediction method provided in the above embodiments, and will not be repeated here.

[0103] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the steps of the above-described method for predicting the health status of transmission system bearings based on digital twins.

[0104] The computer program product provided in this application can solve the technical problem of predicting the health status of transmission system bearings based on digital twins. Compared with the prior art, the beneficial effects of the computer program product provided in this application are the same as those of the transmission system bearing health status prediction method based on digital twins provided in the above embodiments, and will not be repeated here.

[0105] The above are only some embodiments of this application and do not limit the patent scope of this application. All equivalent structural transformations made under the technical concept of this application and using the contents of the specification and drawings of this application, or direct / indirect applications in other related technical fields, are included in the patent protection scope of this application.

Claims

1. A method for predicting the health status of bearings in a transmission system based on digital twins, characterized in that, The method for predicting the health status of bearings in a transmission system based on digital twins includes: A digital twin of the bearing is constructed based on its geometric model, finite element physical model, and behavioral rule library. Feature extraction is performed on the original vibration signal, original temperature signal and rotation speed signal of the bearing acquired by the sensor network to obtain a multi-dimensional feature vector; The feature vector is input into a pre-trained prediction model, which outputs the health index and remaining service life of the bearing. The health index and remaining lifespan are visualized in the digital twin, and early warning information is generated based on preset thresholds.

2. The method as described in claim 1, characterized in that, The step of constructing a digital twin of the bearing based on its geometric model, finite element physical model, and behavioral rule library includes: The bearing's geometric model, finite element model, and behavioral rule library are aligned and coupled to obtain a static basic model; A data interface is established for the static basic model based on a standard communication protocol, wherein the data interface is used to receive real-time sensing data streams from the bearing; Based on the real-time sensing data stream received by the data interface, a dynamic mapping relationship is established between the real-time sensing data stream and the physical field parameters and state variables in the static basic model. A digital twin of the bearing is constructed based on the dynamic mapping relationship.

3. The method as described in claim 2, characterized in that, The step of establishing a dynamic mapping relationship between the real-time sensing data stream received based on the data interface and the physical field parameters and state variables in the static basic model includes: Extract the vibration signal amplitude, temperature gradient, and rotational speed signal from the real-time sensing data stream received based on the data interface; The amplitude of the vibration signal is determined as the dynamic load boundary condition; The temperature gradient is mapped to the equivalent thermal stress influence factor in the material property parameters; The rotational speed signal is mapped to a time-varying distribution of contact stress between the roller and the raceway; Based on the dynamic load boundary conditions, the equivalent thermal stress influence factor, and the time-varying distribution of contact stress between the rollers and raceways, a dynamic mapping relationship is established between the real-time sensing data stream and the physical field parameters and state variables in the static basic model.

4. The method as described in claim 1, characterized in that, The step of extracting features from the original vibration signal, original temperature signal, and rotational speed signal of the bearing acquired by the sensor network to obtain a multidimensional feature vector includes: The original vibration signal of the bearing acquired by the sensor network is obtained by wavelet threshold denoising to obtain the vibration signal. The original temperature signal obtained from the bearing by the sensor network is obtained by performing a sliding smoothing filter to obtain the temperature signal; The time-domain statistical characteristics, frequency-domain spectral characteristics, and envelope spectrum characteristics of the vibration signal are calculated in parallel. The time-domain statistical features, frequency-domain spectral features, envelope spectral features, temperature signal, and rotational speed signal obtained from the bearing by the sensor network are time-aligned and spliced ​​to obtain a multi-dimensional feature vector.

5. The method as described in claim 4, characterized in that, The steps for obtaining the vibration signal by performing wavelet threshold denoising on the original vibration signal of the bearing acquired by the sensor network include: Determine the current operating condition based on the speed signal; When the current working condition is a non-stationary working condition, the original vibration signal of the bearing collected by the sensor network is decomposed into multiple sub-bands based on adaptive wavelet packet decomposition, and noise is removed in each sub-band according to the wavelet threshold to obtain the vibration signal. When the current operating condition is a stable operating condition, the original vibration signal of the bearing acquired by the sensor network is subjected to global threshold denoising based on a fixed wavelet basis function to obtain the vibration signal.

6. The method as described in claim 1, characterized in that, The step of inputting the feature vector into a pre-trained prediction model and outputting the health index and remaining service life of the bearing includes: The feature vector is input into the deep temporal convolutional network-attention mechanism layer of the pre-trained prediction model to extract deep temporal features. The prediction model includes a deep temporal convolutional network-attention mechanism layer, a bidirectional long short-term memory network layer, and a fully connected regression layer. The deep temporal convolutional network-attention mechanism layer is obtained by fusing and training a deep temporal convolutional network and an attention mechanism. The deep temporal features are input into a bidirectional long short-term memory network layer to capture the temporal dependency between the deep temporal features and the health status; The temporal dependency relationship between the deep temporal features and the health status is input into a fully connected regression layer, and the health index of the bearing is output. A degradation trajectory is generated based on the historical health index and the health index, and the remaining service life of the bearing is determined based on the degradation trajectory.

7. The method as described in claim 6, characterized in that, The steps for training a deep temporal convolutional network with attention mechanism layers include: A training sample set was constructed based on historical full-life-cycle bearing monitoring data; Multi-scale features of the training sample set are extracted based on a temporal convolutional network. Key features of the multi-scale features are determined through an attention mechanism; The key features are adaptively weighted to obtain weighted key features; The weighted key features are trained based on the mean squared error loss function and the Adam optimizer, and a deep temporal convolutional network-attention mechanism layer is obtained when the convergence condition is met.

8. The method as described in claim 1, characterized in that, The step of visualizing the health index and remaining lifespan in the digital twin and generating early warning information based on a preset threshold includes: In the three-dimensional visualization interface of the digital twin, the health index change trend and remaining service life prediction results of the bearing are displayed in real time in the form of dynamic curves. When the health index falls below a preset first threshold, a primary warning indicator is triggered in the digital twin interface; When the health index is lower than the preset second threshold or the remaining service life is lower than the set safety margin, an advanced warning sign is triggered, and a warning message containing bearing positioning information and recommended maintenance measures is automatically generated, wherein the first threshold is greater than the second threshold.

9. The method as described in claim 8, characterized in that, The steps for triggering an advanced warning indicator and automatically generating warning information containing bearing positioning information and recommended maintenance measures when the health index falls below a preset second threshold or the remaining service life falls below a set safety margin include: When the health index is lower than the preset second threshold or the remaining service life is lower than the set safety margin, an early warning event is generated, triggering an advanced early warning flag. The early warning event information includes the current time information, trigger condition type, health index, and remaining service life. When an advanced warning sign is detected, the positioning information of the bearing is determined, including the equipment code, installation location, production line number, and three-dimensional spatial coordinates. Determine the degradation information of the health index or the remaining service life, and determine recommended maintenance measures based on the degradation information; The warning information is generated based on the warning event information, the bearing positioning information, and the recommended maintenance measures.

10. A transmission system bearing health status prediction system based on digital twin, characterized in that, The digital twin-based transmission system bearing health status prediction system includes: The bearing modeling module is used to construct a digital twin of the bearing based on its geometric model, finite element physical model, and behavioral rule library. The feature extraction module is used to extract features from the original vibration signal, original temperature signal and rotation speed signal of the bearing acquired by the sensor network, and obtain a multi-dimensional feature vector. The status prediction module is used to input the feature vector into a pre-trained prediction model and output the health index and remaining service life of the bearing. The early warning display module is used to visualize the health index and remaining lifespan in the digital twin and generate early warning information based on a preset threshold.

Images