A real-time monitoring method and device for the working state of a gearbox

Abstract

The present invention provides a method and device for real-time monitoring of the working state of a gearbox, relating to the technical field of gearbox testing. In the present invention, all characteristic information in time domain and frequency domain and other scale spaces is integrated through wavelet transform. Then, during the training process of the artificial intelligence model, methods such as attention mechanism and soft thresholding are adopted to screen the characteristics of the time domain information and frequency domain information, realize noise reduction, obtain key characteristics related to the fault type, and train an artificial intelligence model based on the key characteristics and fault type. By using the thus obtained artificial intelligence model to conduct real-time monitoring on the gearbox, the noise in the vibration signal can be effectively reduced, the influence of the noise on the gearbox fault prediction can be weakened, and the accuracy and reliability of the shape-integrated gearbox fault diagnosis can be improved.

Description

Technical Field

[0001] The present invention relates to the technical field of gearbox testing, and particularly to a method and device for real-time monitoring of the working state of a gearbox. Background Art

[0002] In the field of intelligent coal mine tunneling equipment, the shape-integrated gearbox, as a key transmission component, its operating state directly affects the overall performance and operation efficiency of the tunneling equipment. In order to ensure the safe and stable operation of the tunneling equipment, it is particularly important to perform real-time and accurate fault detection on the shape-integrated gearbox. However, in the actual application process, the fault detection process faces many technical challenges, and the problem of noise interference is particularly prominent.

[0003] The shape-integrated gearbox generates complex vibration and noise signals during operation. These signals not only contain the operating state information of the gearbox itself, such as the interaction and friction between components, generating a large amount of mechanical noise, but also are mixed with various interference factors from the external environment, such as the vibration during the operation of the roadheader, the noise in the underground environment, electromagnetic noise, etc. In addition, changes in environmental factors such as air humidity and temperature in the mine may also cause instability in the performance of sensors, thereby introducing additional noise. These factors result in the collected fault signals being often severely contaminated by noise, making it extremely difficult to extract and identify fault features.

[0004] Traditional fault detection methods mainly rely on technical means such as spectrum analysis and wavelet analysis, attempting to separate fault features from the mixed signals. However, in actual applications, these methods are often limited by noise interference, resulting in a significant reduction in the accuracy and reliability of the detection results. Especially in complex working conditions such as roadheaders, the diversity and uncertainty of noise interference greatly reduce the application effect of traditional methods.

[0005] Currently, the problem of noise interference in the fault detection process of the shape-integrated gearbox leads to low accuracy and reliability of the fault detection of the shape-integrated gearbox. Summary of the Invention

[0006] The present invention provides a method and device for real-time monitoring of the working state of a gearbox, which can improve the accuracy and reliability of the fault detection results of the shape-integrated gearbox.

[0007] In a first aspect, the present invention provides a method for real-time monitoring of the working state of a gearbox. The method includes: acquiring vibration signals of a shape-performance integrated gearbox under various fault states and recording the fault types of the vibration signals; performing wavelet transform processing on the vibration signals to obtain a two-dimensional time-frequency image; the two-dimensional time-frequency image is used to reflect the time-domain information and frequency-domain information of the vibration signals; based on the two-dimensional time-frequency image and the fault types of the vibration signals, neural network training is carried out to obtain an artificial intelligence model. During the training process, an attention mechanism and a soft thresholding method are used to denoise and screen the two-dimensional time-frequency image, and key features are extracted for fault type recognition; based on the artificial intelligence model, the working state of the shape-performance integrated gearbox is monitored in real time to obtain a fault detection result.

[0008] In a possible implementation, performing wavelet transform processing on the vibration signals to obtain a two-dimensional time-frequency image includes: analyzing the frequency range, time-domain characteristics, and noise composition of the vibration signals; determining a wavelet basis function based on the frequency range, time-domain characteristics, and noise composition; decomposing the vibration signals based on the wavelet basis function to obtain wavelet coefficients at multiple scales; and reconstructing the wavelet coefficients at multiple scales to obtain a two-dimensional time-frequency image.

[0009] In a possible implementation, based on the two-dimensional time-frequency image and the fault types of the vibration signals, neural network training is carried out to obtain an artificial intelligence model, including: generating a plurality of training samples based on the two-dimensional time-frequency image and the fault types of the vibration signals; and performing neural network training based on the plurality of training samples to obtain an artificial intelligence model.

[0010] In a possible implementation, based on multiple training samples, neural network training is performed to obtain an artificial intelligence model, including: Step 1, build an initial model and set the model parameters of the initial model as the model parameters of the current iteration process; the initial model includes a deep residual shrinkage module and a convolutional neural network module, and the model parameters of the deep residual shrinkage module include attention parameters and soft threshold parameters; the model parameters of the convolutional neural network model include the number of layers, quantity, and weights of neurons; Step 2, based on the two-dimensional time-frequency image of the current training sample, as well as the deep residual shrinkage module and model parameters in the current iteration process, obtain key features; Step 3, based on the convolutional neural network model and model parameters in the current iteration process, as well as the key features, output the predicted fault type; Step 4, compare the deviation between the fault type of the current training sample and the predicted fault type; Step 5, if the deviation meets the preset condition, exit the iteration and execute Step 6; if the deviation does not meet the preset condition, change the model parameters of the deep residual shrinkage module and the convolutional neural network module, and repeat Steps 2 to 5 until the deviation meets the preset condition; Step 6, determine whether the training set in the training samples is trained. If it is trained, build an artificial intelligence model based on the model parameters of the current deep residual shrinkage module and convolutional neural network module; if the training is not completed, change the training samples and repeat Steps 2 to 6 until the training set is trained.

[0011] In a possible implementation, based on the two-dimensional time-frequency image of the current training sample, as well as the deep residual shrinkage module and model parameters in the current iteration process, obtaining key features includes: based on the two-dimensional time-frequency image of the current training sample and the residual block of the deep residual shrinkage module, perform preliminary shrinkage on the two-dimensional time-frequency image to obtain the high-order features of the current training sample; based on the attention parameters of the current iteration process, determine the weight coefficients of the channel features in the high-order features of the current training sample; based on the high-order features of the current training sample and the weight coefficients of the channel features, generate attention features; based on the soft threshold parameters in the current iteration process, perform deep shrinkage on the attention features to obtain key features.

[0012] In a possible implementation, based on the artificial intelligence model, perform real-time monitoring on the working state of the shape-integrated gearbox to obtain a fault detection result, including: obtain the real-time vibration signal of the shape-integrated gearbox in the current time period; based on the real-time vibration signal, perform wavelet transform processing to obtain the two-dimensional time-frequency image in the current time period; input the two-dimensional time-frequency image in the current time period into the deep residual shrinkage module of the artificial intelligence model to obtain the key features in the current time period; based on the key features in the current time period and the convolutional neural network module of the artificial intelligence model, output the fault detection result in the current time period.

[0013] In a possible implementation, based on an artificial intelligence model, the working state of the form-fit integrated gearbox is monitored in real time. After obtaining the fault detection result, it further includes: if the fault detection result is a minor fault, a fault monitoring instruction is generated, and the fault monitoring instruction is used to indicate monitoring the gear system running with a fault; minor faults include wear, pitting, and scratches; if the fault detection result is a severe fault, a shutdown instruction is generated, and the shutdown instruction is used to indicate the gear system to shut down, and severe faults include broken teeth.

[0014] In a possible implementation, if the fault detection result is a minor fault, a fault monitoring instruction is generated. After that, it further includes: recording the infrared data of the gear system after the fault; dividing the infrared data of the gear system after the fault by a sliding time window to obtain the infrared data of multiple sliding time windows; based on the infrared data of multiple sliding time windows and a stress analysis model, determining the analysis results of multiple sliding time windows, and the analysis results include fault types, the fault probabilities of each fault type, and the fault positions; based on the analysis results of multiple sliding time windows, analyzing the fault at the fault position in the fault detection result to determine the fault trend; based on the fault detection result, the analysis results of multiple sliding time windows, and the fault trend, determining whether the gear system shuts down.

[0015] In a possible implementation, based on an artificial intelligence model, the working state of the form-fit integrated gearbox is monitored in real time. After obtaining the fault detection result, it further includes: if the fault detection result is a gearbox fault, obtaining the real-time working condition parameters of the form-fit integrated gearbox at the current time period; the real-time working condition parameters include rotational speed, temperature, pressure, and load; based on the real-time working condition parameters at the current time period and the two-dimensional time-frequency image at the current time period, generating the working condition characteristics at the current time period; based on the working condition characteristics at the current time period and a preset Gaussian process regression model, performing stress prediction to obtain the stress characteristics of the form-fit integrated gearbox; based on the stress characteristics, checking the fault detection result to determine the fault type of the form-fit integrated gearbox.

[0016] In a possible implementation, before performing stress prediction based on the working condition characteristics at the current time period and a preset Gaussian process regression model to obtain the stress characteristics of the form-fit integrated gearbox, it further includes: obtaining the vibration signals, working condition parameters, and stress characteristics of the form-fit integrated gearbox in the normal state and various fault states; generating multiple working condition characteristics based on the vibration signals and working condition parameters; performing Gaussian regression fitting based on the multiple working condition characteristics and the stress characteristics corresponding to each working condition characteristic to obtain the Gaussian process regression model.

[0017] In a second aspect, an embodiment of the present invention provides a real-time monitoring device for the working state of a gearbox. The monitoring device includes a communication module and a processing module. The communication module is configured to acquire vibration signals of the shape-integrated gearbox under various fault states and record the fault types of the vibration signals. The processing module is configured to perform wavelet transform processing on the vibration signals to obtain a two-dimensional time-frequency image. The two-dimensional time-frequency image is used to reflect the time-domain information and frequency-domain information of the vibration signals. Based on the two-dimensional time-frequency image and the fault types of the vibration signals, neural network training is performed to obtain an artificial intelligence model. During the training process, an attention mechanism and a soft thresholding method are used to denoise and screen the two-dimensional time-frequency image, and key features are extracted for fault type recognition. Based on the artificial intelligence model, the working state of the shape-integrated gearbox is monitored in real time to obtain a fault detection result.

[0018] In a third aspect, an embodiment of the present invention provides an electronic device. The electronic device includes a memory and a processor. The memory stores a computer program, and the processor is configured to call and run the computer program stored in the memory to execute the steps of the method described in the first aspect and any possible implementation manner in the first aspect as above.

[0019] In a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium. The computer-readable storage medium stores a computer program, and is characterized in that when the computer program is executed by a processor, it implements the steps of the method described in the first aspect and any possible implementation manner in the first aspect as above.

[0020] The present invention provides a real-time monitoring method and device for the working state of a gearbox. The present invention integrates all feature information in scale spaces such as time domain and frequency domain through wavelet transform. Then, during the training process of the artificial intelligence model, an attention mechanism and a soft thresholding method are used to perform feature screening on the time-domain information and frequency-domain information to achieve denoising, and key features related to the fault type are obtained. Based on the key features and the fault type, an artificial intelligence model is trained. Based on the obtained artificial intelligence model, the gearbox is monitored in real time, which can effectively reduce the noise in the vibration signal, weaken the influence of the noise on the gearbox fault prediction, and improve the accuracy and reliability of the fault diagnosis of the shape-integrated gearbox. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following will briefly introduce the drawings required for use in the embodiments or the description of the prior art. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0022] Figure 1It is a schematic flowchart of a method for real-time monitoring of the working state of a gearbox provided by an embodiment of the present invention;

[0023] Figure 2 It is a schematic structural diagram of a device for real-time monitoring of the working state of a gearbox provided by an embodiment of the present invention;

[0024] Figure 3 It is a schematic structural diagram of an electronic device provided by an embodiment of the present invention. Detailed implementation manners

[0025] In the following description, specific details such as specific system structures and technologies are presented for the purpose of illustration rather than limitation, so as to thoroughly understand the embodiments of the present invention. However, those skilled in the art should clearly understand that the present invention can also be implemented in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits, and methods are omitted to avoid unnecessary details from interfering with the description of the present invention.

[0026] In the description of the present invention, unless otherwise specified, " / " means "or". For example, A / B can represent A or B. The "and / or" in this article is only a description of the association relationship of associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist simultaneously, and B exists alone. In addition, "at least one" and "multiple" mean two or more. The words such as "first" and "second" do not limit the quantity and execution order, and the words such as "first" and "second" do not necessarily limit to be different.

[0027] In the embodiments of the present application, words such as "exemplary" or "for example" are used to represent examples, illustrations, or explanations. Any embodiment or design solution described as "exemplary" or "for example" in the embodiments of the present application should not be construed as being more preferred or more advantageous than other embodiments or design solutions. Exactly speaking, using words such as "exemplary" or "for example" aims to present relevant concepts in a specific way for easy understanding.

[0028] In addition, the terms "including" and "having" mentioned in the description of the present application and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or device that includes a series of steps or modules is not limited to the listed steps or modules, but optionally further includes other unlisted steps or modules, or optionally further includes other steps or modules inherent to these processes, methods, products, or devices.

[0029] To make the purpose, technical solutions, and advantages of the present invention clearer, the following will be described through specific embodiments in conjunction with the drawings of the present invention.

[0030] As shown Figure 1 in the figure, an embodiment of the present invention provides a method for real-time monitoring of the working state of a gearbox. The method includes steps S101 - S104.

[0031] S101. Obtain the vibration signals of the shape-property integrated gearbox under various fault states, and record the fault types of each vibration signal.

[0032] Exemplarily, in an embodiment of the present invention, high-precision vibration sensors can be installed at key parts of the gearbox, such as bearings, gears, etc., to capture the vibration signals under different operating states. In order to obtain comprehensive fault data, different types of faults can be simulated in a laboratory environment, such as gear wear, bearing damage, shaft misalignment, etc. For each fault state, it is necessary to record in detail the labels of the vibration signals, including information such as fault type, fault degree, occurrence time, etc., for subsequent data processing and model training.

[0033] S102. Perform wavelet transform processing on the vibration signals to obtain a two-dimensional time-frequency image.

[0034] In an embodiment of the present application, the two-dimensional time-frequency image is used to reflect the time-domain information and frequency-domain information of the vibration signals.

[0035] Exemplarily, in an embodiment of the present invention, appropriate wavelet basis functions can be selected for transformation according to the characteristics of the vibration signals, such as Morlet wavelet, Db wavelet, etc. Perform multi-scale decomposition on the wavelet transform to obtain the time-frequency information of different frequency components. Represent the decomposed time-frequency information in the form of a two-dimensional image, where the horizontal axis represents time, the vertical axis represents frequency, and the color or brightness of the image represents the intensity or energy of the signal.

[0036] As a possible implementation manner, step S102 can be specifically implemented as steps S1021 - S1024.

[0037] S1021. Analyze the frequency range, time-domain characteristics, and noise composition of the vibration signals.

[0038] Exemplarily, in an embodiment of the present invention, the fast Fourier transform (FFT) or other spectrum analysis methods can be used to determine the main frequency components and their distribution ranges of the vibration signals. This helps to select appropriate wavelet basis functions and decomposition scales. Observe the time-domain waveform of the vibration signals and analyze the characteristics such as periodicity and non-stationarity of the signals. These characteristics are crucial for understanding the time-varying behavior of the signals and subsequent wavelet transform processing. Identify the noise components in the vibration signals, including white noise, colored noise, etc. The presence of noise may affect the effect of the wavelet transform, so appropriate processing needs to be performed in subsequent steps.

[0039] S1022. Determine the wavelet basis function based on the frequency range, time-domain characteristics, and noise composition.

[0040] Exemplarily, embodiments of the present invention can select the most suitable wavelet basis function according to the frequency range, time-domain characteristics, and noise composition of the vibration signal. Common wavelet basis functions include Haar wavelet, Daubechies wavelet, Morlet wavelet, etc. When selecting, characteristics such as orthogonality, compact support, and symmetry of the wavelet basis function need to be considered. According to the frequency resolution requirement of the vibration signal, determine the decomposition scale of the wavelet transform. The more decomposition scales, the higher the frequency resolution, but the computational complexity will also increase. Therefore, a trade-off needs to be made between resolution and computational efficiency.

[0041] S1023. Decompose the vibration signal based on the wavelet basis function to obtain wavelet coefficients at multiple scales.

[0042] Exemplarily, embodiments of the present invention can perform discrete wavelet transform on the vibration signal using the selected wavelet basis function and decomposition scale. This step decomposes the signal into wavelet coefficients at multiple scales, and the coefficients at each scale represent the frequency components of the signal at that scale. Store the decomposed wavelet coefficients in an appropriate data structure for subsequent reconstruction processing.

[0043] S1024. Reconstruct the wavelet coefficients at multiple scales to obtain a two-dimensional time-frequency image.

[0044] Exemplarily, embodiments of the present invention can use the inverse discrete wavelet transform (IDWT) or other reconstruction algorithms to reconstruct the stored wavelet coefficients back into the time-domain signal. This step usually involves inverse transformation and superposition operations on the wavelet coefficients at each scale. Represent the reconstructed signal in the form of a two-dimensional image, where the horizontal axis represents time, the vertical axis represents frequency (or scale), and the color or brightness of the image represents the intensity or energy of the signal. This step can be implemented by programming, using an image processing library (such as OpenCV) to generate and display the two-dimensional time-frequency image.

[0045] S103. Based on the two-dimensional time-frequency image and the fault types of each vibration signal, perform neural network training to obtain an artificial intelligence model.

[0046] In the embodiments of the present application, during the training process, an attention mechanism and a soft thresholding method are used to denoise and screen the two-dimensional time-frequency image, and key features are extracted for fault type recognition.

[0047] Exemplarily, embodiments of the present invention can perform preprocessing operations such as normalization and cropping on two-dimensional time-frequency images to improve the training efficiency and accuracy of the model. An attention mechanism is introduced into the neural network to enable the model to automatically focus on key regions in the image, thereby improving the accuracy of fault recognition. The soft thresholding method is used to denoise the image and reduce the impact of noise on model training. At the same time, by screening key features, the model input is further simplified to improve the generalization ability of the model. The preprocessed two-dimensional time-frequency images and corresponding fault type labels are used to train the neural network. During the training process, the network parameters are continuously adjusted to minimize the loss function and improve the recognition accuracy of the model.

[0048] As a possible implementation manner, step S103 can be specifically implemented as steps S1031 - S1032.

[0049] S1031. Generate multiple training samples based on the two-dimensional time-frequency images and the fault types of each vibration signal.

[0050] Exemplarily, embodiments of the present invention can pair each two-dimensional time-frequency image with its corresponding vibration signal fault type to form a training sample. Ensure that each sample contains sufficient fault information for the neural network to learn fault features from it. To improve the generalization ability of the model, data augmentation processing can be performed on the original training samples. This includes operations such as rotation, scaling, translation, and adding noise to generate more variant samples. Data augmentation helps the model to be more robust when processing vibration signals under different conditions. The generated training samples are divided into a training set, a validation set, and a test set. The training set is used to train the neural network; the validation set is used to adjust the model parameters during the training process to avoid overfitting; the test set is used to evaluate the performance of the final model.

[0051] S1032. Based on the multiple training samples, perform neural network training to obtain an artificial intelligence model.

[0052] Exemplarily, for the processing of two-dimensional time-frequency images, common network architectures include convolutional neural networks (CNNs) and their variants, such as Deep CNN, ResNet, etc. These networks can effectively extract features from images. Select an appropriate loss function to measure the difference between the model prediction results and the true fault types. For classification problems, common loss functions include Cross-Entropy Loss, Focal Loss, etc. The choice of loss function will directly affect the training effect and performance of the model. Adopt an appropriate optimization algorithm to update the weights and biases of the neural network to minimize the loss function. Common optimization algorithms include Stochastic Gradient Descent (SGD), Adam, RMSprop, etc. During the training process, hyperparameters such as the learning rate and momentum can be adjusted according to the performance of the validation set. During the training process, regularly monitor metrics such as the loss function values and accuracy of the training set and validation set to evaluate the training progress and performance of the model. If it is found that the performance of the model on the validation set begins to decline (i.e., overfitting occurs), methods such as Early Stopping and Regularization can be adopted to alleviate the overfitting problem. After the training is completed, save the trained model to disk for subsequent inference or further tuning. At the same time, a function to load the model also needs to be provided to reload and use the model when needed.

[0053] Exemplarily, step S1032 can be specifically implemented as steps one to six.

[0054] Step one, build an initial model and set the model parameters of the initial model as the model parameters of the current iteration process.

[0055] In some embodiments, the initial model includes a deep residual shrinkage module and a convolutional neural network module. The model parameters of the deep residual shrinkage module include attention parameters and soft threshold parameters; the model parameters of the convolutional neural network model include the number of layers, the number, and the weights of neurons.

[0056] Exemplarily, the deep residual shrinkage module combines the attention mechanism and the soft thresholding method to extract key features from vibration signals. The attention parameters are used to control the attention of the model to different frequency components, and the soft threshold parameters are used to remove noise and redundant information. The convolutional neural network module consists of multiple convolutional layers, pooling layers, and fully connected layers, and is used to further extract and classify features. Parameters such as the number of layers, the number, and the weights of neurons determine the complexity and learning ability of the network. Model parameter initialization: Use methods such as random initialization or pre-trained weights to assign values to the parameters of the initial model. These parameters will be updated through the backpropagation algorithm in subsequent iteration processes.

[0057] Step 2: Based on the two-dimensional time-frequency image of the current training sample, as well as the deep residual shrinkage module and model parameters in the current iteration process, obtain the key features.

[0058] Exemplarily, use the two-dimensional time-frequency image of the current training sample as the input of the deep residual shrinkage module. Through the attention mechanism and soft thresholding method of the deep residual shrinkage module, extract the key features in the image. These features reflect the changes of the vibration signal at different frequencies and time scales.

[0059] Exemplarily, Step 2 can be specifically implemented as A1 - A4.

[0060] A1: Based on the two-dimensional time-frequency image of the current training sample and the residual block of the deep residual shrinkage module, perform preliminary shrinkage on the two-dimensional time-frequency image to obtain the high-order features of the current training sample.

[0061] Among them, in the embodiments of the present invention, the two-dimensional time-frequency image of the current training sample can be used as the input of the deep residual shrinkage module. Ensure that the image data has been properly preprocessed, such as normalization, denoising, etc. The residual block in the deep residual shrinkage module includes components such as a convolutional layer, a batch normalization layer, and an activation function. The convolutional layer extracts local features in the image, the batch normalization layer accelerates the training process and improves the model stability, and the activation function (such as ReLU) introduces non-linearity. These components act together on the input image to generate preliminary high-order features. In the residual block, there is usually also a skip connection that directly adds the input features to the output features to achieve feature fusion and direct gradient transmission. This helps to alleviate the gradient vanishing problem in deep networks.

[0062] A2: Based on the attention parameters in the current iteration process, determine the weight coefficients of the features of each channel in the high-order features of the current training sample.

[0063] Exemplarily, an attention mechanism is introduced in the deep residual shrinkage module to dynamically adjust the weights of the features of each channel in the high-order features. This is usually achieved through components such as global average pooling, a fully connected layer, and a sigmoid activation function. The global average pooling layer compresses the spatial dimension of the high-order features into a single channel to obtain a compact feature representation. Then, through the fully connected layer, learn the correlation between channels, and use the sigmoid activation function to map the output to the interval [0, 1] as the weight coefficients of the features of each channel.

[0064] A3: Based on the high-order features of the current training sample and the weight coefficients of the features of each channel, generate attention features.

[0065] Exemplarily, the weight coefficients calculated in step two are multiplied element-wise with the high-order features to achieve feature weighting. This helps to emphasize important features and suppress irrelevant or noisy features. Attention feature generation: The weighted features are the attention features. They reflect the importance of each channel in the original high-order features and provide more valuable information for subsequent feature extraction and classification.

[0066] A4. Based on the soft threshold parameter in the current iteration process, perform deep shrinkage on the attention features to obtain key features.

[0067] Exemplarily, the soft thresholding operation in the deep residual shrinkage module is used to further remove noise and redundant information. It processes the attention features element-wise based on the soft threshold parameter in the current iteration process. The soft thresholding function is usually defined as f(x) = sign(x) * max(|x| - λ, 0), where x is an element in the attention features, λ is the soft threshold parameter, and sign(x) is the sign function. Through the soft thresholding operation, small values in the attention features (usually corresponding to noise or redundant information) are set to zero or values close to zero, while larger values (corresponding to important features) are retained. In this way, more refined and robust key features are obtained for subsequent classification tasks.

[0068] Step three, based on the convolutional neural network model and model parameters in the current iteration process, and the key features, output the predicted fault type.

[0069] Exemplarily, in the embodiment of the present invention, the extracted key features can be input into the convolutional neural network module, and through operations such as convolution, pooling, and fully connected layers, the features are mapped to the fault type space. According to the mapped features, a classifier such as the softmax function is used to output the probability distribution of the predicted fault type.

[0070] Step four, compare the deviation between the fault type of the current training sample and the predicted fault type.

[0071] Exemplarily, in the embodiment of the present invention, a metric method such as the cross-entropy loss function can be used to calculate the deviation between the true fault type and the predicted fault type of the current training sample. According to the value of the loss function, the prediction performance of the model is evaluated. The smaller the loss value, the better the prediction performance of the model.

[0072] Step five, if the deviation meets the preset condition, then exit the iteration and execute step six; if the deviation does not meet the preset condition, then change the model parameters of the deep residual shrinkage module and the convolutional neural network module, and repeat steps two to five until the deviation meets the preset condition.

[0073] Exemplarily, the preset conditions usually include the number of iterations, the loss value threshold, or the performance of the validation set, etc. When these conditions are met, the iterative process will terminate. Using the backpropagation algorithm and optimizers (such as Adam, SGD, etc.), update the model parameters of the deep residual shrinkage module and the convolutional neural network module according to the gradient of the loss function. Repeat steps two to five until the deviation meets the preset conditions or the maximum number of iterations is reached.

[0074] Step six, determine whether the training set in the training samples is trained. If it is trained, construct an artificial intelligence model based on the model parameters of the current deep residual shrinkage module and convolutional neural network module; if it is not trained, change the training samples and repeat steps two to six until the training set is trained.

[0075] Exemplarily, ensure that all samples in the training set have been used to train the model. Use the model parameters of the trained deep residual shrinkage module and convolutional neural network module to construct the final artificial intelligence model. Evaluate the performance of the model on the validation set to ensure its good generalization ability. If the performance of the validation set is not good, it may be necessary to adjust the model structure, optimizer parameters, or data augmentation strategy, etc. If the training set is not trained, replace the new training samples and repeat steps two to six until the training set is completely trained.

[0076] S104. Based on the artificial intelligence model, monitor the working state of the shape-integrated gearbox in real time to obtain a fault detection result.

[0077] Exemplarily, in the embodiment of the present invention, during the operation of the gearbox, the vibration signal can be collected in real time and subjected to wavelet transform processing to obtain a two-dimensional time-frequency image. Input the processed two-dimensional time-frequency image into the trained artificial intelligence model for inference to obtain a fault detection result. Output the fault detection result in a visual form, such as the fault type, fault degree, occurrence location, etc. At the same time, the result can be sent to the remote monitoring center or trigger the alarm system according to the need. According to the fault detection result, give corresponding maintenance suggestions, such as replacing damaged parts, adjusting operating parameters, etc., to ensure the safe and stable operation of the gearbox.

[0078] As a possible implementation manner, step S104 can be specifically implemented as steps S1041 - S1044.

[0079] S1041. Obtain the real-time vibration signal of the shape-integrated gearbox in the current period.

[0080] Exemplarily, in the embodiments of the present invention, acceleration sensors or vibration sensors can be installed at key parts of the formability integrated gearbox to capture the vibration signals of the gearbox in real time. Through the data acquisition system, the vibration signals of the gearbox are collected in real time at a certain sampling rate and resolution. Ensure that the sampling rate and resolution are high enough to capture the subtle changes in the vibration of the gearbox. The collected vibration signals are preprocessed, such as denoising, filtering, and normalization, etc., to improve the quality and reliability of the signals.

[0081] S1042. Perform wavelet transform processing based on the real-time vibration signals to obtain the two-dimensional time-frequency image of the current period.

[0082] Exemplarily, in the embodiments of the present invention, appropriate wavelet basis functions and decomposition levels can be selected for wavelet transform according to the characteristics of the gearbox vibration signals. Commonly used wavelet basis functions include Haar wavelet, Daubechies wavelet, etc. Through wavelet transform, the vibration signals are transformed from the time domain to the time-frequency domain, and the time-frequency images of different frequency components changing with time are obtained. This helps to capture the frequency changes and fault characteristics in the vibration of the gearbox. Visualize the time-frequency data obtained by wavelet transform to generate the two-dimensional time-frequency image of the current period. Ensure that the image is clear and easy to identify.

[0083] S1043. Input the two-dimensional time-frequency image of the current period into the deep residual shrinkage module of the artificial intelligence model to obtain the key features of the current period.

[0084] Exemplarily, in the embodiments of the present invention, a trained artificial intelligence model can be loaded, including a deep residual shrinkage module and a convolutional neural network module. The two-dimensional time-frequency image of the current period is used as the input of the deep residual shrinkage module. Ensure that the image format is consistent with the model input requirements. Through components such as residual blocks, attention mechanisms, and soft thresholding in the deep residual shrinkage module, feature extraction and noise suppression are performed on the input image to obtain the key features of the current period. These features reflect the key information in the gearbox vibration signals, such as fault characteristics, frequency changes, etc.

[0085] S1044. Output the fault detection result of the current period based on the key features of the current period and the convolutional neural network module of the artificial intelligence model.

[0086] Exemplarily, embodiments of the present invention can input the extracted key features into a convolutional neural network module. Through operations such as convolutional layers, pooling layers, and fully connected layers, the features are mapped into the fault type space. At the output layer of the convolutional neural network module, a classifier such as the softmax function is used to classify the mapped features, and the fault detection result for the current period is output. The detection result is usually given in the form of a probability distribution, indicating the possibilities of different fault types. According to the output result, the current working state of the gearbox and the possible fault causes are explained. If a fault is detected, an alarm is issued in a timely manner or corresponding measures are taken for maintenance.

[0087] The present invention provides a method for real-time monitoring of the working state of a gearbox. All feature information in scale spaces such as time domain and frequency domain is integrated through wavelet transform. Then, during the training process of the artificial intelligence model, methods such as attention mechanism and soft thresholding are adopted to screen the time domain information and frequency domain information for feature selection, achieve denoising, obtain key features related to the fault type, and train an artificial intelligence model based on the key features and fault types. By using the thus obtained artificial intelligence model to perform real-time monitoring on the gearbox, the noise in the vibration signal can be effectively reduced, the influence of noise on the gearbox fault prediction can be weakened, and the accuracy and reliability of the integrated shape and performance gearbox fault diagnosis can be improved.

[0088] Optionally, for the method for real-time monitoring of the working state of the gearbox provided by embodiments of the present invention, after step S104, steps S201 - S202 are further included.

[0089] S201. If the fault detection result is a minor fault, a fault monitoring instruction is generated.

[0090] In some embodiments, the fault monitoring instruction is used to instruct to monitor the gear system of the integrated shape and performance gearbox to operate with a fault; minor faults include wear, pitting, and scratches.

[0091] Exemplarily, if the fault types are wear, pitting, and scratches, the gear system will not fail immediately, but long-term existence may affect the performance and life of the gear system. The gear system can be monitored and operated with a fault.

[0092] Exemplarily, after step S201, steps B1 - B5 are further included.

[0093] B1. Record the infrared data after the gear system fails.

[0094] Exemplarily, after the gear system fails, first confirm the existence of the fault through visual inspection, vibration analysis, or other diagnostic means. Use an infrared thermal imager to record the infrared images or video data of the faulty gear system, ensuring coverage of the entire fault time period. Convert the infrared data into a digital format for subsequent processing and analysis.

[0095] B2. Divide the infrared data after the gear system fails into sliding time windows to obtain the infrared data of multiple sliding time windows.

[0096] Exemplarily, in the embodiments of the present invention, the size of the sliding time window (i.e., the time length included in each window) can be set according to the time range of the occurrence of the failure and the sampling frequency of the infrared data. The sliding time window should be small enough to capture the dynamic changes of the failure, and at the same time large enough to contain sufficient failure information. Determine the sliding step, that is, the time overlapping part between two adjacent windows. Divide the recorded infrared data according to the set sliding time window to obtain multiple continuous and non-overlapping (or with a certain overlap) time window data. The infrared data within each time window should contain sufficient temperature distribution information for subsequent stress analysis and fault detection.

[0097] B3. Based on the infrared data of multiple sliding time windows and the stress analysis model, determine the analysis results of multiple sliding time windows.

[0098] In some embodiments, the stress analysis model is used to analyze the temperature and stress changes in various states.

[0099] In some embodiments, the analysis results include the fault type, the fault probability of each fault type, and the fault location.

[0100] Exemplarily, for the infrared data of each sliding time window, use the constructed stress analysis model for processing. The model should be able to predict the stress distribution and possible fault types of the gear system based on the temperature distribution information in the infrared data. Extract information such as the fault type, the fault probability of each fault type, and the fault location from the output of the model. The fault type may include wear, pitting, tooth breakage, etc.; the fault probability indicates the possibility of each fault type within the current time window; the fault location indicates the specific location of the fault in the gear system.

[0101] B4. Based on the analysis results of multiple sliding time windows, analyze the faults at the fault location in the fault detection results to determine the fault trend.

[0102] Exemplarily, based on the analysis results of multiple sliding time windows, observe the changing trends of the fault type, the fault probability, and the fault location over time. Use methods such as statistical analysis, machine learning, or data mining to identify the development trend of the fault, such as whether the fault intensifies and whether it spreads to other components.

[0103] B5. Based on the fault detection results, the analysis results of multiple sliding time windows, and the fault trend, determine whether the gear system stops.

[0104] Exemplarily, combining the original fault detection results (such as the preliminary fault judgment obtained through the stress analysis model), the analysis results of multiple sliding time windows, and the fault trend analysis, comprehensively evaluate the overall state of the gear system. Considering factors such as the fault type, fault probability, fault location, fault development trend, as well as the importance of the gear system and the shutdown cost, make a decision on whether to shut down the machine.

[0105] If the evaluation result indicates that the gear system has a serious fault or the fault trend is uncontrollable, generate a shutdown instruction. The shutdown instruction should include information such as the reason for shutdown, shutdown time, safety measures after shutdown, and maintenance plan. Notify relevant personnel to perform the shutdown operation and take necessary safety measures and maintenance measures according to the requirements in the shutdown instruction.

[0106] S202. If the fault detection result is a serious fault, generate a shutdown instruction.

[0107] In some embodiments, the shutdown instruction is used to indicate the shutdown of the gear system, and the serious faults include broken teeth.

[0108] Exemplarily, if the fault type is a broken tooth, the gear system may fail immediately or pose a serious safety hazard, and it should be instructed to shut down the gear system immediately for maintenance.

[0109] In this way, the present invention can classify and process faults according to the severity of the gear system faults, reduce the shutdown time of the gear system, and improve the accuracy, convenience, and detection efficiency of gear system fault detection.

[0110] Optionally, the stress analysis model can be constructed through steps C1 - C9.

[0111] C1. Obtain the infrared data and stress data of the gear system in the gearbox with integrated shape and performance under various states.

[0112] Among them, various states include normal, worn, pitted, scratched, and broken teeth.

[0113] In some embodiments, the stress data includes the stress at each key point at each moment within a set time period under various states.

[0114] Exemplarily, the embodiments of the present invention can use a high-precision infrared thermal imager and stress sensors. The infrared thermal imager is used to capture the temperature distribution on the gear surface, and the stress sensors are arranged at key positions of the gear to measure stress changes. According to the operating speed and fault characteristics of the gear system, set an appropriate data sampling rate to ensure that sufficient fault information can be captured.

[0115] C2. Based on the infrared data within the gear system under various states, perform time window partitioning to generate multiple first input features.

[0116] C3. Generate stress features based on stress data in various states.

[0117] In some embodiments, the embodiments of the present invention can organize stress data in various states to form a stress time series or a stress distribution map. Preprocess the stress data, such as removing outliers, smoothing, etc. Extract features of the stress data, such as stress peaks, stress means, stress distribution ranges, etc., as stress features.

[0118] Exemplarily, step C3 can be specifically implemented as steps C31 - C34.

[0119] C31. Based on the three - dimensional twin model of the pre - constructed gear system, determine the stress - related positions of each key point.

[0120] In some embodiments, the stress - related positions include direct contact nodes and indirect contact nodes.

[0121] C32. Based on the stress data in various states, extract the stress data of each key point and the stress data of the stress - related positions of each key point.

[0122] C33. For any state, based on the stress data of each key point at each moment and the stress data of the stress - related positions of each key point in this state, construct the stress time - series sequence of each key point and the stress time - series sequence of the stress - related positions of each key point.

[0123] C34. Based on the stress time - series sequence of each key point and the stress time - series sequence of the stress - related positions of each key point, perform spatial sorting to obtain the stress features in this state.

[0124] C4. Based on the state types, divide the first input features and the stress features to obtain the first input features and stress features corresponding to each state type.

[0125] Exemplarily, the embodiments of the present invention can divide the first input features and the stress features according to the state types of the gear system (such as normal, broken tooth, wear, pitting, scratch, etc.) to obtain the first input features and stress features corresponding to each state type.

[0126] C5. Use the first input features corresponding to each state type as inputs and the stress features corresponding to each state type as outputs to obtain the first training samples.

[0127] C6. Use the stress features corresponding to each state type as inputs and the state types as outputs to obtain the second training samples.

[0128] C7. Based on the first training samples, perform neural network training to obtain the first model.

[0129] Exemplarily, embodiments of the present invention may select a suitable neural network architecture (such as a multi-layer perceptron, a convolutional neural network, etc.) and set corresponding hyperparameters (such as a learning rate, the number of iterations, etc.). The neural network is trained using the first training sample, and the network weights are adjusted through the backpropagation algorithm so that the network can accurately predict the stress characteristics under a given temperature change. After the training is completed, a first model is obtained, which can receive the temperature change as an input and output the predicted stress characteristics.

[0130] C8. Based on the second training sample, neural network training is performed to obtain a second model.

[0131] Exemplarily, embodiments of the present invention may select a suitable neural network architecture and hyperparameters. The neural network is trained using the second training sample so that the network can identify the state type of the gear system according to the stress characteristics. After the training is completed, a second model is obtained, which can receive the stress characteristics as an input and output the identified state type.

[0132] C9. Based on the first model and the second model, a stress analysis model is constructed.

[0133] Exemplarily, embodiments of the present invention may integrate the first model and the second model to construct a complete stress analysis model. This model can receive the temperature change of the gear system as an input, first predict the stress characteristics through the first model, and then identify the state type through the second model. The model can also be further optimized and adjusted as needed to improve the accuracy and robustness of the model.

[0134] Optionally, for the method for real-time monitoring of the working state of the gearbox provided by embodiments of the present invention, after step S104, steps S301 - S304 are further included.

[0135] S301. If the fault detection result is a gearbox fault, the real-time working condition parameters of the integrated gearbox at the current time period are obtained.

[0136] In some embodiments, the real-time working condition parameters include rotational speed, temperature, pressure, and load.

[0137] Exemplarily, embodiments of the present invention may install a variety of sensors on the gearbox, such as a rotational speed sensor, a temperature sensor, a pressure sensor, and a load sensor, for capturing the working condition parameters of the gearbox in real time. Through a data acquisition system, the data of these sensors are collected in real time, and the synchronization of the data is ensured. That is, the data of all sensors are collected at the same time point for subsequent analysis and processing. The collected working condition parameter data is preprocessed, including denoising, filtering, outlier processing, etc., to improve the accuracy and reliability of the data.

[0138] S302. Generate the operating condition features of the current period based on the real-time operating condition parameters of the current period and the two-dimensional time-frequency image of the current period.

[0139] Exemplarily, in the embodiments of the present invention, the real-time operating condition parameters (such as rotational speed, temperature, pressure, and load) can be fused with the features in the two-dimensional time-frequency image. This can be achieved by adding the operating condition parameters as additional feature dimensions to the feature vector of the time-frequency image. Using machine learning or deep learning techniques, key features that can reflect the current operating condition of the gearbox are extracted from the fused feature vector. These features may include the vibration mode of the gearbox, the temperature change trend, the pressure fluctuation, etc.

[0140] S303. Based on the operating condition features of the current period and the preset Gaussian process regression model, perform stress prediction to obtain the stress features of the shape-property integrated gearbox.

[0141] Exemplarily, in the embodiments of the present invention, before fault detection, it is necessary to use historical data and operating condition features to train the Gaussian process regression model. This model can learn the relationship between the operating condition features and the stress of the gearbox. Model input: Use the operating condition features of the current period as the input of the Gaussian process regression model. Stress prediction: Use the trained Gaussian process regression model to perform stress prediction on the operating condition features of the current period. The prediction results may include the stress distribution and stress level of each component of the gearbox. Stress feature extraction: Extract key features from the prediction results that can reflect the stress state of the gearbox, such as the maximum stress value and the stress change trend.

[0142] S304. Based on the stress features, check the fault detection results to determine the fault type of the shape-property integrated gearbox.

[0143] Exemplarily, in the embodiments of the present invention, a fault type library can be established: Establish a database containing various gearbox fault types, and each fault type corresponds to specific stress features. Match the predicted stress features with the features in the fault type library. Determine which fault type the current fault is closest to by calculating the similarity or distance metric. According to the matching result, determine the fault type of the shape-property integrated gearbox. If the match is successful, output the corresponding fault type; if the match is unsuccessful, it may indicate the occurrence of a new fault type or further diagnosis is required. According to the determined fault type, take corresponding maintenance measures or repair plans to ensure the normal operation of the gearbox and extend its service life.

[0144] In this way, the embodiments of the present invention can comprehensively use the stress features to check the fault detection results, ensure the accuracy of the fault detection results, and further improve the accuracy and reliability of the fault diagnosis of the shape-property integrated gearbox.

[0145] Optionally, before step S303, the real-time monitoring method for the working state of the gearbox provided by the embodiments of the present invention further includes steps S401 - S403.

[0146] S401. Obtain the vibration signals, operating condition parameters, and stress characteristics of the shape - property integrated gearbox in the normal state and various fault states.

[0147] Exemplarily, in order to obtain the data of the gearbox in the normal state and various fault states, a series of experiments need to be designed. These experiments should cover various operating conditions and fault types that the gearbox may encounter. During the experiments, sensors are used to collect the vibration signals of the gearbox in real - time (such as acceleration, displacement, etc.), operating condition parameters (such as rotational speed, temperature, pressure, load, etc.), and stress characteristics obtained through other means (such as strain gauges, finite element analysis, etc.). The collected data is labeled to clarify the gearbox state (normal or a certain fault type) corresponding to each group of data. This helps subsequent data analysis and model training. The collected data is stored in a safe and reliable storage medium for subsequent data processing and model training.

[0148] S402. Generate multiple operating condition features based on the vibration signals and operating condition parameters.

[0149] Exemplarily, the embodiments of the present invention can pre - process the collected vibration signals, such as denoising, filtering, standardization, etc., to improve the quality and consistency of the signals. Signal processing techniques (such as time - domain analysis, frequency - domain analysis, time - frequency analysis, etc.) and machine learning algorithms (such as principal component analysis, clustering analysis, etc.) are used to extract the key features from the vibration signals that can reflect the operating conditions of the gearbox. The extracted vibration features are integrated with the operating condition parameters (such as rotational speed, temperature, pressure, load, etc.) to form vectors containing multiple operating condition features. These feature vectors will be used for subsequent model training and stress prediction.

[0150] S403. Perform Gaussian regression fitting based on the multiple operating condition features and the stress characteristics corresponding to each operating condition feature to obtain a Gaussian process regression model.

[0151] Exemplarily, an embodiment of the present invention can use the working condition feature vector generated in step S402 as input and the corresponding stress feature as output to construct a training data set. Gaussian Process Regression (GPR) is selected as the modeling method. GPR is a non-parametric Bayesian regression method that can handle high-dimensional input data and non-linear relationships. The Gaussian process regression model is trained using the training data set. During the training process, the hyperparameters of the model (such as kernel function, kernel parameters, etc.) need to be set, and these parameters can be optimized by maximizing the marginal likelihood function of the model. The trained model is verified using cross-validation or other verification methods to ensure the accuracy and generalization ability of the model. The trained Gaussian process regression model is stored in an appropriate location for subsequent stress prediction and fault diagnosis.

[0152] In this way, before verifying the fault detection result through the stress feature, an embodiment of the present invention can perform data fitting on the working condition feature and the stress feature using the Gaussian process regression algorithm to obtain a Gaussian process regression model, so as to more comprehensively consider the fault detection result and further improve the accuracy and reliability of the integrated shape and performance gearbox fault diagnosis.

[0153] It should be understood that the magnitudes of the sequence numbers of the steps in the above embodiments do not mean the order of execution. The order of execution of each process should be determined by its function and internal logic, and should not constitute any limitation to the implementation process of the embodiments of the present invention.

[0154] The following is an apparatus embodiment of the present invention. For details not described in detail, reference can be made to the corresponding method embodiments above.

[0155] Figure 2 The structural schematic diagram of a real-time monitoring device for the working state of a gearbox provided by an embodiment of the present invention is shown. The monitoring device 500 includes a communication module 501 and a processing module 502.

[0156] The communication module 501 is used to obtain the vibration signals of the integrated shape and performance gearbox under various fault states and record the fault types of the vibration signals.

[0157] The processing module 502 is used to perform wavelet transform processing on the vibration signals to obtain a two-dimensional time-frequency image; the two-dimensional time-frequency image is used to reflect the time-domain information and frequency-domain information of the vibration signals; based on the two-dimensional time-frequency image and the fault types of the vibration signals, neural network training is performed to obtain an artificial intelligence model. During the training process, the attention mechanism and the soft thresholding method are used to denoise and screen the two-dimensional time-frequency image, and key features are extracted for fault type recognition; based on the artificial intelligence model, the working state of the integrated shape and performance gearbox is monitored in real time to obtain a fault detection result.

[0158] Figure 3It is a schematic structural diagram of an electronic device provided by an embodiment of the present invention. As Figure 3 shown, the electronic device 600 of this embodiment includes: a processor 601, a memory 602, and a computer program 603 stored in the memory 602 and executable on the processor 601. When the processor 601 executes the computer program 603, the steps in the above method embodiments are implemented, for example Figure 1 the steps S101 - S104 shown. Or, when the processor 601 executes the computer program 603, the functions of each module / unit in the above device embodiments are implemented. For example, Figure 2 the functions of the communication module 501 and the processing module 502 shown.

[0159] Exemplarily, the computer program 603 can be divided into one or more modules / units. The one or more modules / units are stored in the memory 602 and executed by the processor 601 to complete the present invention. The one or more modules / units can be a series of computer program instruction segments capable of performing specific functions, and these instruction segments are used to describe the execution process of the computer program 603 in the electronic device 600. For example, the computer program 603 can be divided into Figure 2 the communication module 501 and the processing module 502 shown.

[0160] The so-called processor 601 can be a central processing unit (CPU), or can also be other general-purpose processors, digital signal processors (DSPs), application specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor, etc.

[0161] The memory 602 may be an internal storage unit of the electronic device 600, such as a hard disk or memory of the electronic device 600. The memory 602 may also be an external storage device of the electronic device 600, such as a plug-in hard disk, a Smart Media Card (SMC), a Secure Digital (SD) card, a Flash Card, etc. equipped on the electronic device 600. Further, the memory 602 may also include both the internal storage unit of the electronic device 600 and an external storage device. The memory 602 is used to store the computer program and other programs and data required by the terminal. The memory 602 may also be used to temporarily store data that has been output or is to be output.

[0162] The above-described embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements on some of the technical features; and these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included in the protection scope of the present invention.

Claims

1. A real-time monitoring method for the working state of a gearbox, characterized in that, Including: Obtain the vibration signals of the formability integrated gearbox under various fault conditions, and record the fault types of each vibration signal; Perform wavelet transform processing on the vibration signals to obtain a two-dimensional time-frequency image; the two-dimensional time-frequency image is used to reflect the time-domain information and frequency-domain information of the vibration signals; Based on the two-dimensional time-frequency image and the fault types of each vibration signal, conduct neural network training to obtain an artificial intelligence model. During the training process, use the attention mechanism and soft thresholding method to denoise and screen the two-dimensional time-frequency image, and extract key features for fault type identification; Based on the artificial intelligence model, monitor the working state of the formability integrated gearbox in real time to obtain a fault detection result; After obtaining the fault detection result by monitoring the working state of the formability integrated gearbox in real time based on the artificial intelligence model, it further includes: if the fault detection result is a minor fault, generate a fault monitoring instruction, and the fault monitoring instruction is used to indicate monitoring the formability integrated gearbox with the gear system running with faults; the minor faults include wear, pitting, and scratches; if the fault detection result is a serious fault, generate a shutdown instruction, and the shutdown instruction is used to indicate the shutdown of the gear system, and the serious faults include broken teeth; After generating the fault monitoring instruction when the fault detection result is a minor fault, it further includes: record the infrared data after the gear system fails; divide the infrared data after the gear system fails into sliding time windows to obtain infrared data of multiple sliding time windows; based on the infrared data of multiple sliding time windows and the stress analysis model, determine the analysis results of multiple sliding time windows, and the analysis results include fault types, the fault probabilities of each fault type, and the fault positions; based on the analysis results of the multiple sliding time windows, analyze the faults at the fault positions in the fault detection result to determine the fault trend; based on the fault detection result, the analysis results of the multiple sliding time windows, and the fault trend, determine whether the gear system shuts down.

2. The real-time monitoring method for the working state of the gearbox according to claim 1, characterized in that, The performing wavelet transform processing on the vibration signals to obtain a two-dimensional time-frequency image includes: Analyze the frequency range, time-domain characteristics, and noise composition of the vibration signals; Based on the frequency range, time-domain characteristics, and noise composition, determine the wavelet basis function; Decompose the vibration signals based on the wavelet basis function to obtain wavelet coefficients of multiple scales; Reconstruct the wavelet coefficients of multiple scales to obtain the two-dimensional time-frequency image.

3. The real-time monitoring method for the working state of the gearbox according to claim 1, characterized in that, The conducting neural network training based on the two-dimensional time-frequency image and the fault types of each vibration signal to obtain an artificial intelligence model includes: Generate multiple training samples based on the two-dimensional time-frequency image and the fault types of each vibration signal; Based on the multiple training samples, conduct neural network training to obtain the artificial intelligence model.

4. The real-time monitoring method for the working state of the gearbox according to claim 3, characterized in that, The conducting neural network training based on the multiple training samples to obtain the artificial intelligence model includes: Step 1: Build an initial model and set the model parameters of the initial model as the model parameters for the current iteration process; the initial model includes a deep residual shrinkage module and a convolutional neural network module, and the model parameters of the deep residual shrinkage module include attention parameters and soft threshold parameters; the model parameters of the convolutional neural network model include the number of layers, quantity, and weights of neurons. Step 2: Based on the two-dimensional time-frequency image of the current training sample, as well as the deep residual shrinkage module and model parameters in the current iteration process, obtain key features. Step 3: Based on the convolutional neural network model and model parameters in the current iteration process, as well as the key features, output the predicted fault type. Step 4: Compare the deviation between the fault type of the current training sample and the predicted fault type. Step 5: If the deviation meets the preset condition, exit the iteration and execute Step 6; if the deviation does not meet the preset condition, change the model parameters of the deep residual shrinkage module and the convolutional neural network module, and repeat Steps 2 to 5 until the deviation meets the preset condition. Step 6: Determine whether the training set in the training samples is trained. If it is trained, build the artificial intelligence model based on the model parameters of the current deep residual shrinkage module and convolutional neural network module; if the training is not completed, change the training samples and repeat Steps 2 to 6 until the training set is trained.

5. The real-time monitoring method for the working state of the gearbox according to claim 4, characterized in that, The obtaining of key features based on the two-dimensional time-frequency image of the current training sample, as well as the deep residual shrinkage module and model parameters in the current iteration process, includes: Based on the two-dimensional time-frequency image of the current training sample and the residual blocks of the deep residual shrinkage module, perform preliminary shrinkage on the two-dimensional time-frequency image to obtain the high-order features of the current training sample. Based on the attention parameters in the current iteration process, determine the weight coefficients of the channel features in the high-order features of the current training sample. Based on the high-order features of the current training sample and the weight coefficients of the channel features, generate attention features. Based on the soft threshold parameters in the current iteration process, perform deep shrinkage on the attention features to obtain the key features.

6. The real-time monitoring method for the working state of the gearbox according to claim 1, characterized in that, The obtaining of the fault detection result by performing real-time monitoring on the working state of the shape-property integrated gearbox based on the artificial intelligence model includes: Obtain the real-time vibration signal of the shape-property integrated gearbox at the current time period. Based on the real-time vibration signal, perform wavelet transform processing to obtain the two-dimensional time-frequency image at the current time period. Input the two-dimensional time-frequency image at the current time period into the deep residual shrinkage module of the artificial intelligence model to obtain the key features at the current time period. Based on the key features at the current time period and the convolutional neural network module of the artificial intelligence model, output the fault detection result at the current time period.

7. The real-time monitoring method for the working state of the gearbox according to claim 1, characterized in that After obtaining the fault detection result by performing real-time monitoring on the working state of the shape-property integrated gearbox based on the artificial intelligence model, it further includes: If the fault detection result is a gearbox fault, obtain the real-time working condition parameters of the shape-property integrated gearbox at the current time period; the real-time working condition parameters include rotational speed, temperature, pressure, and load. Generate the working condition characteristics of the current period based on the real-time working condition parameters of the current period and the two-dimensional time-frequency image of the current period. Based on the working condition characteristics of the current period and a preset Gaussian process regression model, conduct stress prediction to obtain the stress characteristics of the shape-property integrated gearbox. Based on the stress characteristics, check the fault detection result to determine the fault type of the shape-property integrated gearbox.

8. A real-time monitoring device for the working state of a gearbox, characterized in that, It includes: A communication module for acquiring the vibration signals of the shape-property integrated gearbox under various fault states and recording the fault types of the vibration signals. A processing module for performing wavelet transform processing on the vibration signals to obtain a two-dimensional time-frequency image; the two-dimensional time-frequency image is used to reflect the time-domain information and frequency-domain information of the vibration signals; based on the two-dimensional time-frequency image and the fault types of the vibration signals, conduct neural network training to obtain an artificial intelligence model, and use the attention mechanism and soft thresholding method to denoise and screen the two-dimensional time-frequency image during the training process to extract key features for fault type recognition. Based on the artificial intelligence model, conduct real-time monitoring of the working state of the shape-property integrated gearbox to obtain a fault detection result. The processing module is further configured to generate a fault monitoring instruction if the fault detection result is a minor fault, and the fault monitoring instruction is used to indicate monitoring the shape-property integrated gearbox with the gear system operating with a fault; the minor faults include wear, pitting, and scratches. If the fault detection result is a serious fault, generate a shutdown instruction, and the shutdown instruction is used to indicate shutting down the gear system, and the serious fault includes broken teeth; record the infrared data after the gear system fails; divide the infrared data after the gear system fails into sliding time windows to obtain the infrared data of multiple sliding time windows. Based on the infrared data of multiple sliding time windows and a stress analysis model, determine the analysis results of multiple sliding time windows, and the analysis results include the fault type, the fault probability of each fault type, and the fault location. Based on the analysis results of the multiple sliding time windows, analyze the fault at the fault location in the fault detection result to determine the fault trend. Based on the fault detection result, the analysis results of the multiple sliding time windows, and the fault trend, determine whether to shut down the gear system.

Images