Method and device for detecting remaining service life of transmission component of construction waste disposal equipment

Abstract

The present application relates to the field of remaining service life detection, and particularly relates to a construction waste disposal equipment transmission component remaining service life detection method and device. The method comprises: obtaining a plurality of equipment transmission component data samples containing a remaining service life label for a full life cycle; performing data enhancement on the plurality of data samples to obtain a data set; constructing a detection model, the detection model comprising a multi-scale attention feature extraction module and a predictor, the multi-scale attention feature extraction module comprising a plurality of different scale dense connection blocks fused with an attention mechanism, and the predictor being configured to input multi-scale life state features to predict the remaining service life; training the detection model using the data set to obtain a trained detection model; and detecting the degradation vibration signal of the equipment transmission component using the trained detection model to obtain the remaining service life. The present application solves the problems of poor generalization performance and greater influence from noise interference in the prior art.

Description

Technical Field

[0001] This invention relates to the field of remaining service life testing, and specifically to a method and apparatus for testing the remaining service life of transmission components in construction waste disposal equipment. Background Technology

[0002] In construction waste disposal, excavators, feeders, crushers, and other engineering equipment and machinery are widely used. Load-bearing transmission components, such as rolling bearings, are key components of construction waste disposal equipment. Because they typically bear heavy loads and operate continuously, they are among the most vulnerable to damage. The lifespan of these transmission components significantly impacts the operation of construction waste disposal equipment and severely affects normal production schedules. Real-time status monitoring of transmission components, using the monitoring information to assess the equipment's operating status and degree of degradation, can accurately determine the remaining service life of the equipment. This facilitates timely repairs before failures occur and allows maintenance personnel to perform necessary upkeep, thereby maximizing equipment lifespan, significantly reducing downtime, and lowering the human and material costs associated with over-maintenance.

[0003] Current methods for detecting the remaining service life of transmission components are mainly divided into model-based methods and data-driven methods. Model-based methods simulate the degradation process of transmission components by establishing physical or mathematical models, but are limited by model complexity and have significant limitations under complex operating conditions. Data-driven methods extract degradation features by analyzing large amounts of historical data of transmission components, thereby revealing the correlation between monitoring data and remaining service life. With the rapid development of artificial intelligence and big data methods in recent years, especially the superior capabilities of neural networks in nonlinear fitting and big data processing, deep learning methods can automatically extract data features and determine the mapping relationships between data without human intervention or prior knowledge. Therefore, research on using deep learning for remaining service life prediction has become a hot area in academia and industry.

[0004] While deep learning methods perform well in detecting the remaining life of transmission components, the operating conditions of these components vary greatly in real-world industrial scenarios. Deep learning-based remaining life prediction models trained under a single operating condition exhibit poor generalization ability and significant performance degradation when applied to different conditions. Furthermore, their prediction accuracy is low when significantly affected by ambient noise. Therefore, there is a need for a method to detect the remaining life of transmission components in construction waste disposal equipment that possesses strong generalization capabilities and resistance to noise interference. Summary of the Invention

[0005] The technical problem to be solved by the present invention is to overcome the defects of the prior art and provide a method for detecting the remaining service life of transmission components of construction waste disposal equipment. It solves the problems of poor generalization performance and great influence of noise interference in the remaining service life detection of the prior art, and can realize accurate prediction of the remaining service life of transmission components of construction waste disposal equipment.

[0006] To solve the above-mentioned technical problems, the technical solution of the present invention is: a method for detecting the remaining service life of transmission components in construction waste disposal equipment, comprising:

[0007] S1, acquire full life cycle data samples of multiple construction waste disposal equipment transmission components, including remaining service life tags;

[0008] S2, perform data augmentation on multiple data samples to obtain a dataset;

[0009] S3. Construct a detection model. The detection model includes a multi-scale attention feature extraction module and a predictor. The multi-scale attention feature extraction module includes several densely connected blocks with attention mechanisms fused at different scales. The predictor is used to predict the remaining lifetime using multi-scale lifetime state features as input. The multi-scale lifetime state features are obtained by adaptively weighting and fusing the lifetime state features output by all densely connected blocks.

[0010] S4. Use the dataset to train the detection model to obtain a trained detection model;

[0011] S5 uses a trained detection model to detect the degradation vibration signals of the transmission components of the construction waste disposal equipment to obtain the remaining service life.

[0012] Furthermore, S1 includes:

[0013] S11, collect vibration signals of the transmission components of multiple construction waste disposal equipment throughout their entire life cycle under different working conditions; where the same speed and load represent the same working condition;

[0014] S12 divides the full life cycle vibration signal into data samples of equal length, labels them with the remaining lifespan, and normalizes them.

[0015] S13 divides the data samples into training sample set and test sample set according to different working conditions.

[0016] Furthermore, in S2, data augmentation is performed on the data samples, including adaptive scaling and transformation, as shown in the formula:

[0017] ;

[0018] In the formula, This represents the i-th data sample. For learnable affine parameters, and Corresponding to the input data samples respectively The mean and variance of.

[0019] Furthermore, in S3, each densely connected block includes several convolutional attention layers and one average pooling layer; wherein,

[0020] The input to the Kth convolutional attention layer is the input features of the densely connected block. and the output features of the first to the (K-1)th convolutional attention layers Connections along the channel dimension;

[0021] The output features of the Kth convolutional attention layer The calculation formula is:

[0022] ;

[0023] In the formula, Input features representing densely connected blocks and the output features of the first to the (K-1)th convolutional attention layers Connections along the channel dimension Let K be the nonlinear transformation function of the Kth layer, which includes BN, ReLU, Conv, and an attention layer;

[0024] The average pooling layer is used to flatten the output features of the last convolutional attention layer into a one-dimensional vector.

[0025] Furthermore, the attention layer is calculated as follows:

[0026] First, max pooling is performed on each channel of the input feature F to obtain the pooling vector;

[0027] Then, a one-dimensional convolution is performed on the pooling vector and activated by the Sigmoid function to obtain the channel weight vector. The formula is:

[0028] ;

[0029] In the formula, σ is the Sigmoid activation function. It is a convolution operation with a kernel size of k. Indicates max pooling;

[0030] Finally, the channel weight vector Multiplying the feature F by the attention-weighted feature yields the attention-weighted feature, which serves as the output of this attention layer.

[0031] Furthermore, in S3, the formula for adaptive weighted fusion of features output by several densely connected blocks is:

[0032] ;

[0033] In the formula, These are the lifetime state features extracted at the 1st to mth scales. These are learnable weight parameters with the same initial value, normalized by the Softmax function, and their sum of probabilities is 1. It represents the multi-scale lifetime state characteristics.

[0034] Furthermore, in S3, the predictor includes several fully connected layers.

[0035] Furthermore, in S4, during training, the model is optimized by minimizing the root mean square damage function (MSE).

[0036] ;

[0037] In the formula, This represents the actual remaining useful life. This represents the predicted remaining useful life, and N represents the number of input samples.

[0038] The present invention also provides a device for detecting the remaining service life of transmission components in construction waste disposal equipment, comprising:

[0039] The data sample acquisition module is used to acquire full life cycle data samples of multiple construction waste disposal equipment transmission components, including remaining service life tags.

[0040] The data augmentation module is used to augment multiple data samples to obtain a dataset;

[0041] The detection model construction module is used to build the detection model. The detection model includes a multi-scale attention feature extraction module and a predictor. The multi-scale attention feature extraction module includes several dense connected blocks with attention mechanisms fused at different scales. The predictor is used to predict the remaining lifetime using multi-scale lifetime state features as input. The multi-scale lifetime state features are obtained by adaptively weighting and fusing the lifetime state features output by all dense connected blocks.

[0042] The detection model training module is used to train the detection model using the dataset to obtain a trained detection model.

[0043] The detection module is used to detect the degradation vibration signals of the transmission components of the construction waste disposal equipment under test using a trained detection model, and to obtain the remaining service life.

[0044] The present invention also provides a device for detecting the remaining service life of transmission components of construction waste disposal equipment, including a device for detecting the remaining service life of transmission components of construction waste disposal equipment; or including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement a method for detecting the remaining service life of transmission components of construction waste disposal equipment.

[0045] By adopting the above technical solution, this invention overcomes the uniformity of samples in the training sample set through data augmentation, obtains samples with diverse distributions for training the detection model, and obtains the lifetime state features of samples at different scales by establishing a multi-scale attention feature extraction module. The attention mechanism is integrated in the dense connection block to enhance the model's ability to learn state information and reduce noise interference to focus on key state information, which significantly improves the model's generalization ability and prediction accuracy. It can be effectively applied to the detection of the remaining service life of transmission components of construction waste disposal equipment under different working conditions. Attached Figure Description

[0046] Figure 1 This is a flowchart of the method for detecting the remaining service life of transmission components in construction waste disposal equipment according to the present invention;

[0047] Figure 2 This is a schematic diagram illustrating the training process of the detection model of the present invention;

[0048] Figure 3 This is a schematic diagram of the structure of the dense connecting block of the present invention;

[0049] Figure 4 This is a schematic diagram of the detection process of the detection model of the present invention;

[0050] Figure 5 A graph showing the predicted remaining useful life of the detection model on the test sample set;

[0051] Figure 6 This is a schematic diagram of the remaining service life detection device for the transmission components of the construction waste disposal equipment of the present invention. Detailed Implementation

[0052] To make the content of this invention easier to understand, the invention will be further described in detail below with reference to specific embodiments and accompanying drawings.

[0053] Example 1: Figures 1 to 4 As shown, a method for detecting the remaining service life of transmission components in construction waste disposal equipment includes:

[0054] S1, acquire full life cycle data samples of multiple construction waste disposal equipment transmission components, including remaining service life tags;

[0055] S2, perform data augmentation on multiple data samples to obtain a dataset;

[0056] S3. Construct a detection model. The detection model includes a multi-scale attention feature extraction module and a predictor. The multi-scale attention feature extraction module includes several densely connected blocks with attention mechanisms fused at different scales. The predictor is used to predict the remaining lifetime using multi-scale lifetime state features as input. The multi-scale lifetime state features are obtained by adaptively weighting and fusing the lifetime state features output by all densely connected blocks.

[0057] S4. Use the dataset to train the detection model to obtain a trained detection model;

[0058] S5 uses a trained detection model to detect the degradation vibration signals of the transmission components of the construction waste disposal equipment to obtain the remaining service life.

[0059] Specifically, this detection method overcomes the uniformity of samples in the training sample set through data augmentation, obtains samples with diverse distributions for training the detection model, and obtains the lifetime state features of samples at different scales by establishing a multi-scale attention feature extraction module. The attention mechanism is integrated into the dense connection block to enhance the model's ability to learn state information and reduce noise interference to focus on key state information, which significantly improves the model's generalization ability and prediction accuracy. It can be effectively applied to the detection of the remaining service life of transmission components of construction waste disposal equipment under different working conditions.

[0060] In this embodiment, S1 includes:

[0061] S11, collect vibration signals of the transmission components of multiple construction waste disposal equipment throughout their entire life cycle under different working conditions; where the same speed and load represent the same working condition;

[0062] S12 divides the full life cycle vibration signal into data samples of equal length, labels them with the remaining lifespan, and normalizes them.

[0063] S13. The data samples are divided into training sample set and test sample set according to different working conditions. The remaining service life label contained in the training sample set can be used for model training, while the remaining service life label contained in the test sample set does not participate in model training and is only used to verify the accuracy of the model's prediction of the remaining service life of the test sample set.

[0064] As an example, the publicly available XJTU-SY rolling bearing accelerated life test dataset is used. Rolling bearings are one of the important transmission components of equipment. In the experiment, the sampling frequency was set to 25.6 kHz, the sampling interval was 1 min, and each sampling duration was 1.28 s. Bearing2_1, Bearing2_2, Bearing2_3, Bearing2_4, and Bearing2_5 under the second operating condition (speed 2250 r / min, radial force 11 kN) were used as the training sample set to train the constructed detection model. Bearing1_3 under the first operating condition (speed 2100 r / min, radial force 12 kN) was used as the test sample set to validate the constructed life detection model. The datasets used are described in Table 1.

[0065] .

[0066] It is important to note that the actual lifespan here refers to the actual service life of the bearing in the experiment, not the remaining service life value of a single sample.

[0067] For example, Bearing1_3 has a total of 158 samples and an actual lifetime of 2 hours and 38 minutes (158 minutes). Samples were collected every minute, resulting in 158 samples collected within the 158 minutes of operation. This means that the remaining lifetime for the first sample is 158 minutes, and for the last sample, it is 1 minute. The publicly available datasets all have corresponding labels.

[0068] In this embodiment, in S2, data augmentation is performed on the data samples, including adaptive scaling and transformation, to obtain diverse samples as input for model training. The formula is as follows:

[0069] ;

[0070] In the formula, This represents the i-th data sample. For learnable affine parameters, and Corresponding to the input data samples respectively The mean and variance of.

[0071] It should be noted that for the training sample set, each sample has a probability of 0.5 of being augmented to ensure that there are a sufficient number of original samples and augmented samples to train the model.

[0072] In this embodiment, as Figure 3 As shown, in S3, each densely connected block includes several convolutional attention layers and one average pooling layer; where,

[0073] The input to the Kth convolutional attention layer is the input features of the densely connected block. and the output features of the first to the (K-1)th convolutional attention layers Connections along the channel dimension;

[0074] The output features of the Kth convolutional attention layer The calculation formula is:

[0075] ;

[0076] In the formula, Input features representing densely connected blocks and the output features of the first to the (K-1)th convolutional attention layers Connections along the channel dimension Let K be the nonlinear transformation function of the Kth layer, which includes BN, ReLU, Conv, and an attention layer;

[0077] The average pooling layer is used to flatten the output features of the last convolutional attention layer into a one-dimensional vector.

[0078] Specifically, the feature transfer method of densely connected blocks is to directly connect the features of all previous layers in the channel dimension and then input them into the next layer, so that the features can be transferred more effectively, thereby enhancing the model's ability to learn state information.

[0079] In this embodiment, the attention layer is calculated as follows:

[0080] First, max pooling is performed on each channel of the input feature F to obtain the pooling vector;

[0081] Then, a one-dimensional convolution is performed on the pooling vector and activated by the Sigmoid function to obtain the channel weight vector. The formula is:

[0082] ;

[0083] In the formula, σ is the Sigmoid activation function. It is a convolution operation with a kernel size of k. Indicates max pooling;

[0084] Finally, the channel weight vector Multiplying the feature F by the attention-weighted feature yields the attention-weighted feature, which serves as the output of this attention layer.

[0085] Specifically, by assigning higher weights to key state information through an attention layer, the interference of noise in vibration signal samples is reduced, thereby focusing on more important lifetime state characteristics.

[0086] In this embodiment, as Figure 2 As shown, in S3, the formula for adaptive weighted fusion of features output by several densely connected blocks is:

[0087] ;

[0088] In the formula, These are the lifetime state features extracted at the 1st to mth scales. These are learnable weight parameters with the same initial value, normalized by the Softmax function, and their sum of probabilities is 1. It represents the multi-scale lifetime state characteristics.

[0089] Specifically, a multi-scale attention feature extraction module is used to extract rich state information at multiple scales, alleviating the problem of difficulty in extracting state information caused by changes in operating conditions.

[0090] As an example, such as Figure 2 As shown, three dense connect blocks are configured, each employing a different convolutional kernel size across the three scales: 1×11, 1×15, and 1×19. Each dense connect block is followed by an average pooling layer to flatten the extracted features into a one-dimensional vector. Then, adaptive weighted fusion of these one-dimensional vector features at different scales is performed to dynamically measure the importance of state features at different scales, ultimately yielding multi-scale lifetime state features. Among them, the weight parameters The parameters added to the detection model's parameter list can be trained and optimized during backpropagation optimization, therefore, These can be learnable parameters.

[0091] In this embodiment, in S3, the predictor includes several fully connected layers.

[0092] As an example, the predictor consists of two fully connected layers with 512 and 1 nodes respectively.

[0093] In this embodiment, during S4, the model is optimized by minimizing the root mean square damage function (MES) during training.

[0094] ;

[0095] In the formula, This represents the actual remaining useful life. This represents the predicted remaining useful life, and N represents the number of input samples.

[0096] In this embodiment, the root mean square error (RMSE), mean absolute error (MAE), and coefficient of determination (R²) are used as performance evaluation metrics for the model. RMSE, MAE, and R² are only used to evaluate the model's performance and do not participate in model parameter updates.

[0097] ;

[0098] ;

[0099] ;

[0100] In the formula, for The average value.

[0101] In this embodiment, the detection model undergoes 200 iterations of training, and an adaptive moment estimation algorithm is used as the optimization algorithm. Figure 2 The training model shown undergoes backpropagation optimization, and training stops when the number of iterations reaches a set value.

[0102] In this embodiment, the trained model is as follows: Figure 4 As shown, it includes a multi-scale attention feature extraction module and a predictor.

[0103] The evaluation results of the accuracy of the methods involved in the above embodiments for detecting the remaining useful life of the test sample set are shown in Table 2:

[0104] .

[0105] Among them, the root mean square error (RMSE) is more sensitive to large deviations, while the mean absolute error (MAE) provides an intuitive measure of error; the smaller the value, the smaller the model's prediction error. The coefficient of determination (R²) is also important. 2 The value reflects how well the model fits the data; the higher the value, the better the model fits the data.

[0106] It can be seen that this method uses the dataset under the second working condition as the training sample set and the dataset under the first working condition as the test sample set. The RMSE of the validation results is 0.0351, and the MAE of the validation results is 0.0289, reflecting that the difference between the predicted remaining useful life and the actual value is relatively small. 2 The result of 0.9852 also demonstrates that the model of this method fits the data well and the prediction results of the remaining useful life of the test sample set are relatively accurate.

[0107] The method described in the above embodiments detects the remaining useful life of the test sample set as follows: Figure 5As shown, the predicted remaining service life and the actual remaining service life have the same trend and a high degree of fit, which also reflects that the method of the present invention has good generalization ability and can accurately predict the remaining service life of the equipment.

[0108] Example 2: Figure 6 As shown, a device for detecting the remaining service life of transmission components in construction waste disposal equipment includes:

[0109] The data sample acquisition module is used to acquire full life cycle data samples of multiple construction waste disposal equipment transmission components, including remaining service life tags.

[0110] The data augmentation module is used to augment multiple data samples to obtain a dataset;

[0111] The detection model construction module is used to build the detection model. The detection model includes a multi-scale attention feature extraction module and a predictor. The multi-scale attention feature extraction module includes several dense connected blocks with attention mechanisms fused at different scales. The predictor is used to predict the remaining lifetime using multi-scale lifetime state features as input. The multi-scale lifetime state features are obtained by adaptively weighting and fusing the lifetime state features output by all dense connected blocks.

[0112] The detection model training module is used to train the detection model using the dataset to obtain a trained detection model.

[0113] The detection module is used to detect the degradation vibration signals of the transmission components of the construction waste disposal equipment under test using a trained detection model, and to obtain the remaining service life.

[0114] Example 3: A device for detecting the remaining service life of transmission components of construction waste disposal equipment, including the device for detecting the remaining service life of transmission components of construction waste disposal equipment as described in Example 2; or including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the method for detecting the remaining service life of transmission components of construction waste disposal equipment as described in Example 1.

[0115] Based on the above-described preferred embodiments of the present invention, and through the foregoing description, those skilled in the art can make various changes and modifications without departing from the inventive concept. The technical scope of this invention is not limited to the contents of the specification, but must be determined according to the scope of the claims.

Claims

1. A method for detecting the remaining service life of transmission components in construction waste disposal equipment, characterized in that, include: S1, acquire full life cycle data samples of multiple construction waste disposal equipment transmission components, including remaining service life tags, with each data sample having the same length; S2, perform data augmentation on multiple data samples to obtain a dataset; S3. Construct a detection model. The detection model includes a multi-scale attention feature extraction module and a predictor. The multi-scale attention feature extraction module includes several densely connected blocks set in parallel with convolutional kernels of different sizes. The predictor is used to predict the remaining lifetime using multi-scale lifetime state features as input. The multi-scale lifetime state features are obtained by adaptively weighting and fusing the lifetime state features output by all densely connected blocks. S4. Use the dataset to train the detection model to obtain a trained detection model; S5. The trained detection model is used to detect the degradation vibration signal of the transmission components of the construction waste disposal equipment to be tested, and the remaining service life is obtained. In S3, each densely connected block includes several convolutional attention layers and average pooling layers; among them, The input to the Kth convolutional attention layer is the input features of the densely connected block. and the output features of the first to the (K-1)th convolutional attention layers Connections along the channel dimension; The output features of the Kth convolutional attention layer The calculation formula is: ； In the formula, Input features representing densely connected blocks and the output features of the first to the (K-1)th convolutional attention layers Connections along the channel dimension Let K be the nonlinear transformation function of the Kth layer, which includes BN, ReLU, Conv, and an attention layer; The average pooling layer is used to flatten the output features of the last convolutional attention layer into a one-dimensional vector; The attention layer is calculated as follows: First, max pooling is performed on each channel of the input feature F to obtain the pooling vector; Then, a one-dimensional convolution is performed on the pooling vector and activated by the Sigmoid function to obtain the channel weight vector. The formula is: ； In the formula, σ is the Sigmoid activation function. It is a convolution operation with a kernel size of k. Indicates max pooling; Finally, the channel weight vector Multiplying the feature F by the attention-weighted feature yields the attention-weighted feature, which serves as the output of this attention layer. In S3, the formula for adaptive weighted fusion of features output by several densely connected blocks is: ； In the formula, , … This represents the lifetime state feature vectors extracted at the 1st to mth scales. These are learnable weight parameters with the same initial value, normalized by the Softmax function, and their sum of probabilities is 1. It represents the multi-scale lifetime state characteristics.

2. The method for detecting the remaining service life of transmission components in construction waste disposal equipment according to claim 1, characterized in that, S1 includes: S11, collect vibration signals of the transmission components of multiple construction waste disposal equipment throughout their entire life cycle under different working conditions; where the same speed and load represent the same working condition; S12 divides the full-life-cycle vibration signal into data samples of equal length, labels them with the remaining service life, and normalizes them. S13 divides the data samples into training sample set and test sample set according to different working conditions.

3. The method for detecting the remaining service life of transmission components in construction waste disposal equipment according to claim 1, characterized in that, In S2, data augmentation is performed on the data samples, including adaptive scaling and transformation, as shown in the formula: ； In the formula, Let i represent the i-th sample. For learnable affine parameters, and Corresponding to the input samples respectively The mean and variance of.

4. The method for detecting the remaining service life of transmission components in construction waste disposal equipment according to claim 1, characterized in that, In S3, the predictor consists of several fully connected layers.

5. The method for detecting the remaining service life of transmission components in construction waste disposal equipment according to claim 1, characterized in that, In S4, during training, the model is optimized using the Minimum Mean Square Damage Function (MES). ； In the formula, This represents the actual remaining useful life. This represents the predicted remaining useful life, and N represents the number of input samples.

6. A device for detecting the remaining service life of transmission components in construction waste disposal equipment, characterized in that, include: The data sample acquisition module is used to acquire data samples of the entire life cycle of multiple transmission components of construction waste disposal equipment, including tags of remaining service life. Each data sample has the same length. The data augmentation module is used to augment multiple data samples to obtain a dataset; The detection model construction module is used to build the detection model. The detection model includes a multi-scale attention feature extraction module and a predictor. The multi-scale attention feature extraction module includes several densely connected blocks with convolutional kernels of different sizes set in parallel. The predictor is used to predict the remaining lifetime using multi-scale lifetime state features as input. The multi-scale lifetime state features are obtained by adaptively weighting and fusing the lifetime state features output by all densely connected blocks. The detection model training module is used to train the detection model using the dataset to obtain a trained detection model. The detection module is used to detect the degradation vibration signals of the transmission components of the construction waste disposal equipment under test using a trained detection model, and to obtain the remaining service life. in, Each densely connected block includes several convolutional attention layers and average pooling layers; among them, The input to the Kth convolutional attention layer is the input features of the densely connected block. and the output features of the first to the (K-1)th convolutional attention layers Connections along the channel dimension; The output features of the Kth convolutional attention layer The calculation formula is: ； In the formula, Input features representing densely connected blocks and the output features of the first to the (K-1)th convolutional attention layers Connections along the channel dimension Let K be the nonlinear transformation function of the Kth layer, which includes BN, ReLU, Conv, and an attention layer; The average pooling layer is used to flatten the output features of the last convolutional attention layer into a one-dimensional vector; The attention layer is calculated as follows: First, max pooling is performed on each channel of the input feature F to obtain the pooling vector; Then, a one-dimensional convolution is performed on the pooling vector and activated by the Sigmoid function to obtain the channel weight vector. The formula is: ； In the formula, σ is the Sigmoid activation function. It is a convolution operation with a kernel size of k. This indicates max pooling; Finally, the channel weight vector Multiplying the feature F by the attention-weighted feature yields the attention-weighted feature, which serves as the output of this attention layer. In S3, the formula for adaptive weighted fusion of features output by several densely connected blocks is: ； In the formula, , … This represents the lifetime state feature vector extracted at the 1st to mth scales. These are learnable weight parameters with the same initial value, normalized by the Softmax function, and their sum of probabilities is 1. It represents the multi-scale lifetime state characteristics.

7. A device for detecting the remaining service life of transmission components in construction waste disposal equipment, characterized in that, Includes the device for detecting the remaining service life of transmission components in construction waste disposal equipment as described in claim 6; It may include a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the method for detecting the remaining service life of transmission components of construction waste disposal equipment as described in any one of claims 1-5.

Images