Method for predicting residual service life of scraper conveyor, medium and equipment

Abstract

The invention discloses a scraper conveyor remaining service life prediction method, medium and equipment, and relates to the technical field of remaining service life prediction.The scraper conveyor remaining service life prediction method comprises the steps that multiple pieces of original graph data are constructed for obtained different sensor data of a scraper conveyor; taking graph data constructed after preprocessing as the input of a model through a convolution layer, extracting features, obtaining a dynamic dependency relationship between sensors through a graph structure learning layer, and fusing the spatial correlation of the sensors; spatial features of aggregation nodes are updated through a graph attention network in a space-time fusion layer, then a time attention mechanism and a bidirectional gating circulation unit are utilized to extract a dependency relationship between sequences, and finally fused space-time features are input into a full connection layer to determine the final remaining service life, so that life prediction of the scraper conveyor is realized.

Description

Technical Field

[0001] The invention relates to the technical field of remaining service life prediction, and in particular to a method, medium and equipment for predicting the remaining service life of a scraper conveyor. Background Art

[0002] Scraper conveyor is an important transportation equipment for coal mining working face in coal mines. Its reliability and service life directly affect the efficiency and safety of coal mine production. Due to the harsh environment of coal mines, scraper conveyors are prone to various failures in long-term operation. Once a failure or maintenance problem occurs, it will cause huge economic losses. Therefore, it is of great significance to predict its remaining service life.

[0003] There are two main types of deep learning networks currently used for remaining useful life (RUL) prediction: convolutional neural networks (CNNs) and recurrent neural networks (RNNs). The CNN-based RUL prediction model can extract local and global features of the original data and achieve feature fusion through multi-layer convolution and pooling operations. Yang et al. proposed a method based on a dual CNN model architecture, using the first CNN to identify early failure points and the second CNN as a regression model to predict RUL. However, the size of the convolution kernel limits the ability to extract long-term relevant information in the time series. The RNN-based RUL prediction model can learn features on the time scale and use internal states to store historical information. However, the RNN-based method ignores the relationship between data variables. In addition, some scholars combine CNN and RNN to form a hybrid prediction model. Ma et al. designed a convolution-based long short-term memory (CLSTM) network to predict the RUL of rotating machinery by mining vibration data.

[0004] However, the above RUL prediction models ignore the potential spatial interaction between sensors. Some scholars try to obtain spatial correlation. Liu et al. introduced an attention mechanism to adaptively weight the features of the sensor at each time point to improve the prediction performance. Zhang et al. stacked sensors and then used 2D-CNN to capture spatial dependencies. Song et al. designed a RUL prediction model based on a distributed attention mechanism to perform weighted processing on different sensors and time steps respectively. The above DL models have achieved good results in processing Euclidean data, but the data in many practical application scenarios are generated from non-Euclidean space. Therefore, when faced with more complex industrial equipment, existing technologies may not be able to accurately predict RUL, resulting in inaccurate predictions of the life degradation trend of scraper conveyors. Summary of the invention

[0005] The present invention provides a method, medium and device for predicting the remaining service life of a scraper conveyor, which are used to solve the above-mentioned problem existing in the prior art, that is, how to accurately predict the service life degradation trend of a scraper conveyor in the prior art. The present invention provides a method for predicting the remaining service life of a scraper conveyor, which comprises:

[0006] For different sensor data of the scraper conveyor obtained, multiple original graph data are constructed; multi-scale convolution operation is used to extract features of the original graph data to obtain feature graphs of different scales, and the feature graphs of different scales are fused to obtain fused graph data;

[0007] By obtaining the spatial installation layout of the sensor, the prior adjacency matrix is ​​determined. Based on the fused graph data, the initial adjacency matrix is ​​constructed by using the KNN algorithm, and the initial adjacency matrix is ​​reconstructed by the graph autoencoder to obtain the similarity adjacency matrix. The prior adjacency matrix is ​​fused with the similarity adjacency matrix through an adaptive mechanism to obtain an adaptive adjacency matrix.

[0008] By adopting the graph attention mechanism to fuse the graph data and the adaptive adjacency matrix, the updated node spatial features are obtained. The updated node spatial features are weighted by the time attention mechanism to determine the global time features. The global time features are fused with the original time series to obtain new time series features.

[0009] A bidirectional gated recurrent unit is used to encode the fused new time series features to obtain the fused spatiotemporal features. The fused spatiotemporal features are input into the fully connected layer to determine the final remaining service life and predict the life of the scraper conveyor.

[0010] Optionally, the obtained different sensor data of the scraper conveyor are used to construct a plurality of original image data, specifically including:

[0011] The obtained data of different sensors of the scraper conveyor are normalized, and the normalized data are obtained using the following formula:

[0012]

[0013] Among them, x i,j is the i-th raw data of the j-th sensor, is the normalized data, and are the minimum and maximum values ​​of the raw data of the jth sensor, respectively;

[0014] The sliding window technology is used to divide the normalized data to obtain the original image data of the sensor.

[0015] Optionally, the method of obtaining the spatial installation layout of the sensor, determining the prior adjacency matrix, constructing an initial adjacency matrix based on the fused graph data by adopting the KNN algorithm, and reconstructing the initial adjacency matrix by using a graph autoencoder to obtain a dynamic similarity adjacency matrix specifically includes:

[0016] The prior adjacency matrix is ​​determined using the following formula:

[0017]

[0018] Among them, A′ is the prior adjacency matrix, Represents node X i and node X j There is interaction;

[0019] Based on the fused graph data, the KNN algorithm is used to determine the distance between each node and all nodes, and the k nodes closest to each other are selected to determine the neighbor node set N. i , the initial adjacency matrix is ​​obtained using the following formula:

[0020]

[0021] Among them, A t Represents the node X in the tth graph data i Is it related to node X? n adjacent;

[0022] The initial adjacency matrix is ​​reconstructed using the following formula to obtain a similar adjacency matrix:

[0023] Z=GCN(h,A t ),

[0024] A t ″=σ(ZZ T ),

[0025] Where GCN(·) represents the update process of GCN, h is the node feature, σ is the sigmoid activation function, Z is the embedding vector of the node, and A″ t is the t-th similarity adjacency matrix.

[0026] Optionally, the graph data and the adjacency matrix are fused by adopting a graph attention mechanism to obtain updated spatial node features, specifically including:

[0027] The graph data and the adjacency matrix are fused by adopting the graph attention mechanism. The similarity coefficient between adjacent nodes is obtained by the nonlinear activation function LeakyReLU, and then the normalized attention coefficient is obtained after activation by the Softmax function. The updated node spatial features are obtained through the multi-head attention weights and output vectors.

[0028] Optionally, the nonlinear activation function LeakyReLU obtains the similarity coefficient between adjacent nodes, and then obtains the normalized attention coefficient after activation by the Softmax function, specifically including:

[0029] The similarity coefficient between adjacent nodes and the normalized attention coefficient are obtained using the following formula:

[0030]

[0031]

[0032] in, is the feature of the i-th node, is the feature of its adjacent nodes, F is the original feature dimension of each node, e ij is the similarity coefficient between adjacent nodes, a ij is the normalized attention coefficient, a∈R 2F′ is a learnable weight vector, W∈R F×F′ is a learnable shared weight matrix, and F′ represents the feature dimension of the node output.

[0033] Optionally, the method adopts a time attention mechanism to weight the updated node spatial features to determine the global time features, and concatenates and fuses the global time features with the original time series to obtain new time series features, specifically including:

[0034] The time attention mechanism is used to weight the updated spatial node features, and the original weight is determined by the following formula:

[0035] ω t =o T tanh(h′ t ),

[0036] The normalized weight is obtained using the following formula:

[0037]

[0038] The global time feature is determined according to the normalized weight. The specific formula is as follows:

[0039]

[0040] The global time feature is fused with the original time series, and the new time series feature μ is obtained using the following formula t :

[0041] μ t =[h′ t ||S],

[0042] Among them, h′ t is the updated spatial node feature, o∈R N×F′ is a learnable vector, F′ is the feature dimension of the node output, ω t is the original weight, θ t is the normalized weight, S∈R T It is the representation of global time characteristics.

[0043] Optionally, the bidirectional gated recurrent unit is used to encode the fused new time series features to obtain fused spatiotemporal features, specifically including:

[0044] The forward hidden state and the reverse hidden state are obtained using the following formula:

[0045]

[0046]

[0047] The following formula is used to concatenate the forward hidden state and the reverse hidden state to obtain the fused spatiotemporal features:

[0048]

[0049] Among them, μ t is the new time series feature after fusion, and Represent the forward and reverse hidden states at time t, Y t is the output after fusing the bidirectional hidden values, GRU(·) represents the GRU update process, u t , v t , b t are the corresponding learnable parameters.

[0050] Optionally, the sensor data specifically includes:

[0051] Material level sensor data, zero speed sensor data and speed sensor data for speed regulation.

[0052] The present invention provides a computer-readable storage medium, wherein the storage medium stores a computer program, and when the computer program is executed by a processor, the method for predicting the remaining service life of a scraper conveyor is implemented.

[0053] The present invention provides a computer device, comprising a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor implements the above-mentioned method for predicting the remaining service life of a scraper conveyor when executing the program.

[0054] Compared with the prior art, the beneficial effects of the present invention are as follows: the present invention provides a method for predicting the remaining service life of a scraper conveyor, which mines the hidden features of raw data at multiple scales by using one-dimensional convolution kernels of different sizes, and more comprehensive features provide support for capturing the correlation between sensors; a new association graph construction method is designed to realize the combination of a priori static graph structure and dynamic similarity graph structure, wherein a new dynamic similarity graph structure construction method KNN-GAE is proposed, which dynamically adjusts the graph structure based on the degradation characteristics of the data, and an adaptive fusion mechanism is proposed to adaptively update the spatial coupling relationship between sensors from two aspects; a new RUL prediction model based on GAT and BiGRU is proposed, which uses GAT to aggregate the spatial features between sensors, uses the attention mechanism to realize the fusion of local and global features between sequences, and uses BiGRU to capture the temporal dependency between sequences, and the fused spatiotemporal features improve the RUL prediction performance. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the invention and, together with the description, serve to explain the principles of the invention.

[0056] Figure 1 An overall framework diagram of a method for predicting the remaining service life of a scraper conveyor provided by an embodiment of the present invention;

[0057] Figure 2 A data diagram of a sliding window construction diagram provided by an embodiment of the present invention;

[0058] Figure 3 A multi-scale convolution graph provided by an embodiment of the present invention;

[0059] Figure 4 A dynamic similarity graph provided by an embodiment of the present invention;

[0060] Figure 5 A graph attention mechanism network diagram provided for an embodiment of the present invention;

[0061] Figure 6 A fusion diagram of local and global features of a sequence provided by an embodiment of the present invention;

[0062] Figure 7 A schematic diagram of a computer device for predicting the remaining useful life of a scraper conveyor provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0063] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the drawings of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0064] During the operation of the scraper conveyor, multiple sensors are needed to monitor the operating status of the scraper conveyor, and there may be potential dependencies between sensors with similar positions.

[0065] However, existing technologies cannot fully exploit the spatial interaction relationship between sensors. The association graph constructed using prior knowledge or feature similarity ignores the dynamic evolution relationship between sensors over time, and the association graph from a single angle is not sufficient to fully reflect the spatial interaction relationship between sensors, which limits their RUL prediction performance and leads to inaccurate prediction of the life degradation trend of scraper conveyors.

[0066] In response to the above problems, the present invention provides a method for predicting the remaining service life of a scraper conveyor. The method fully captures the dynamic coupling relationship between sensors by making full use of the spatial structural characteristics of the scraper conveyor and the dynamic interaction relationship between sensors during the degradation process, thereby accurately predicting the life degradation trend of the scraper conveyor.

[0067] The technical solution of the present invention and how the technical solution of the present invention solves the above-mentioned technical problems are described in detail below with specific embodiments. The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be repeated in some embodiments. The embodiments of the present invention will be described below in conjunction with the accompanying drawings.

[0068] Example 1

[0069] Figure 1 This is an overall framework diagram of a method for predicting the remaining service life of a scraper conveyor provided by an embodiment of the present invention. The sensors are mapped as nodes in the diagram, including a piece of feature data. Each edge represents whether there is interaction between sensors, and the set of edges is used as an adjacency matrix. Figure 1 As shown, this embodiment shows a method for predicting the remaining service life of a scraper conveyor, comprising:

[0070] S101: construct multiple original graph data based on different sensor data of the scraper conveyor obtained.

[0071] Optionally, the acquired data of different sensors of the scraper conveyor are normalized, and the normalized data is obtained using the following formula:

[0072]

[0073] Among them, x i,j is the i-th raw data of the j-th sensor, is the normalized data, and are the minimum and maximum values ​​of the raw data of the jth sensor, respectively.

[0074] Exemplarily, the non-sensor data of the plate conveyor is obtained, which may include, for example, material level sensor data, zero speed sensor data, and speed sensor data for speed regulation.

[0075] For example, Figure 2 As shown, the sliding window technology can be used to divide the normalized data to obtain the original image data of the sensor.

[0076] S102: extract features from the original graph data using a multi-scale convolution operation to obtain feature graphs of different scales, and fuse the feature graphs of different scales to obtain fused graph data.

[0077] For example, one-dimensional convolution kernels of different sizes are used to perform multiple convolution operations on the input image data, and then a 1×1 convolution kernel is used to fuse the features in different channel directions to form a more complete feature representation. The convolution process is as follows: Figure 3 shown.

[0078] In this step, convolutional layers are used to extract features at different scales and used as the initial features of the nodes in order to obtain information at different levels.

[0079] S103: By obtaining the spatial installation layout of the sensor, the prior adjacency matrix is ​​determined, based on the fused graph data, the initial adjacency matrix is ​​constructed by adopting the KNN algorithm, and the initial adjacency matrix is ​​reconstructed by the graph autoencoder to obtain the similarity adjacency matrix, and the prior adjacency matrix is ​​fused with the similarity adjacency matrix by an adaptive mechanism to obtain an adaptive adjacency matrix.

[0080] For example, physically adjacent or functionally similar sensors can be assigned a higher probability of interaction. For example, the transition slot is physically connected to the middle slot and the head frame, so it is considered that the sensors monitoring them have an interactive relationship; at the same time, when the scraper chain speed increases, the temperature of the middle plate in the middle slot will rise, so it is considered that the sensors monitoring them also have an interactive relationship.

[0081] Optionally, the prior adjacency matrix is ​​determined using the following formula:

[0082]

[0083] Among them, A′ is the prior adjacency matrix, Represents node X i and node X j There is interaction.

[0084] Based on the fused graph data, such as Figure 4 As shown, the KNN algorithm is used to determine the distance between each node and all nodes, and the distance vector D(i,n)∈R is obtained. N , select the k nodes closest to each other and determine the neighbor node set N i , the initial adjacency matrix A t ∈R N×N It is expressed as:

[0085]

[0086] Among them, A t Represents the node X in the tth graph data i Is it related to node X? n Adjacent.

[0087] The graph autoencoder is used to reconstruct the initial adjacency matrix to obtain a similar adjacency matrix. The specific formula is as follows:

[0088] Z=GCN(h,A t ).

[0089] A t ″=σ(ZZ T ).

[0090] Where GCN(·) represents the update process of GCN, h is the node feature, σ is the sigmoid activation function, Z is the embedding vector of the node, and A″ t is the t-th similarity adjacency matrix.

[0091] The gating mechanism process is as follows:

[0092] gate=σ(A′W g1 +A″ t W g2 +b g ).

[0093] A″′ t =gate#A′+(1-gate)A″ t .

[0094] Among them, A″′ t is the graph structure after adaptive fusion, W g1 , W g2 ∈R N×N is the learnable weight, b g ∈R Nis a learnable bias.

[0095] The gating mechanism can adaptively adjust the weight of the graph structure according to different inputs, thereby obtaining richer interactions and dependencies between features.

[0096] S104: The graph data and the adaptive adjacency matrix are fused by adopting the graph attention mechanism to obtain updated node spatial features, the updated node spatial features are weighted by adopting the time attention mechanism to determine the global time features, and the global time features are fused with the original time series to obtain new time series features.

[0097] Optionally, the graph data and the adjacency matrix are fused by adopting the graph attention mechanism, and the nonlinear activation function LeakyReLU obtains the similarity coefficient between adjacent nodes, which is then activated by the Softmax function to obtain the normalized attention coefficient. The updated spatial node features are obtained through the multi-head attention weights and output vectors.

[0098] Exemplarily, the specific formula of the similarity coefficient between adjacent nodes is as follows:

[0099]

[0100] The specific formula of the normalized attention coefficient is as follows:

[0101]

[0102] in, is the feature of the ith node, is the feature of its adjacent nodes, F is the original feature dimension of each node, e ij is the similarity coefficient between adjacent nodes, such as Figure 5 As shown in (a), a ij is the normalized attention coefficient, a∈R 2F′ is a learnable weight vector, W∈R F×F′ is a learnable shared weight matrix, and F′ represents the feature dimension of the node output.

[0103] After randomly discarding the attention coefficient, the sparse attention coefficient is obtained. The specific calculation formula is as follows:

[0104] a i ' j =Droupout(a ij ).

[0105] The feature representation of node i after aggregating neighboring node information is:

[0106]

[0107] like Figure 5 As shown in (b), the output multi-channel feature map is averaged and the new node feature is obtained using the following formula:

[0108]

[0109]

[0110] Among them, K is the number of attention mechanisms introduced, and || represents the vector concatenation operation.

[0111] After GAT fusion, a new set of node features will be generated It is the new feature after the node is updated, and the potential spatial dependency between sensors is integrated into the feature h′.

[0112] Optional, such as Figure 6 As shown in Figure 1, the time attention mechanism is used to weight the updated node spatial feature sequence, and the original weight is determined by the following formula:

[0113] ω t =o T tanh(h′ t ).

[0114] The normalized weight is obtained using the following formula:

[0115]

[0116] The global time feature is determined according to the normalized weight. The specific formula is as follows:

[0117] S=[θ1||θ2…||θ T ].

[0118] The global time feature is fused with the original time series, and the new time series feature μ is obtained using the following formula t :

[0119] μ t =[h′ t ||S].

[0120] Among them, h′ t is the updated node spatial feature, o∈R N×F′ is a learnable vector, F′ is the feature dimension of the node output, ω t is the original weight, θ t is the normalized weight, S∈R T It is the representation of global time characteristics.

[0121] S105: A bidirectional gated recurrent unit is used to encode the fused new time series features to obtain fused spatiotemporal features, and the fused spatiotemporal features are input into the fully connected layer to determine the final remaining service life, so as to predict the life of the scraper conveyor.

[0122] Optionally, the forward hidden state and the reverse hidden state are obtained using the following formula:

[0123]

[0124]

[0125] The following formula is used to concatenate the forward hidden state and the reverse hidden state to obtain the fused spatiotemporal features:

[0126]

[0127] Among them, μ t is the new time series feature after fusion, and Represent the forward and reverse hidden states at time t, Y t is the output after fusing the bidirectional hidden values, GRU(·) represents the GRU update process, u t , v t , b t are the corresponding learnable parameters.

[0128] In this step, a bidirectional gated recurrent unit is used to encode the fused new time series features in order to better capture the dynamic changes and long-term dependencies of the time series. Finally, the output of the integrated spatial and temporal features is passed to the fully connected layer to achieve the final remaining life prediction task.

[0129] This embodiment provides a method for predicting the remaining service life of a scraper conveyor. The convolution layer uses the graph data constructed after preprocessing as the input of the model, which can extract hidden features at different scales and use them as the initial features of the nodes. Then, the graph structure learning layer uses the device structure characteristics to construct and convolute features to capture the dynamic dependencies between sensors, thereby fully integrating the correlation of the graph data. The GAT network is used to update the spatial features of the aggregated nodes through the spatiotemporal fusion layer. On this basis, the temporal attention module and BiGRU are used to extract the dependencies between sequences. Finally, the fused spatiotemporal features are input into the fully connected layer to obtain the final RUL, thereby accurately predicting the service life degradation trend of the scraper conveyor.

[0130] Example 2

[0131] This example verifies the effectiveness of this method by selecting the full life cycle degradation data of four different scraper conveyors in a coal mine.

[0132] Optionally, the root mean square error RMSE and score function Score are used to evaluate the performance of the model. The specific formula is as follows:

[0133]

[0134]

[0135] in, and i are the predicted and true RUL, respectively, and N is the number of samples in the test set. The RMSE function measures the deviation between the predicted value and the true value, treating early and late predictions equally.

[0136] In contrast, the Score function has a greater penalty for late predictions to avoid more serious consequences caused by lagging predictions. The lower the values ​​of these two indicators, the better the prediction performance of the model. In order to eliminate the influence of randomness, the model was tested ten times on each scraper conveyor data, and the average value was taken as the final evaluation result.

[0137] As shown in Table 1, the scoring comparison of the scraper conveyor by the method used in the present invention and other methods is shown.

[0138] Table 1 Comparison of different methods for scoring scraper conveyors:

[0139]

[0140] As can be seen from Table 1, the machine learning-based methods ELM, RF, and DBN methods perform poorly, which is due to the lack of nonlinear expression capabilities of traditional machine learning methods when processing high-dimensional and massive data. The deep learning-based methods DCNN and MS-DCNN both extract local features of sensor data, but MS-DCNN enhances the prediction performance by fusing multi-scale local features. However, the CNN-based model ignores the temporal dependency of the data. The deep learning-based methods Bi-LSTM and ATT-LSTM use recursive networks to extract the temporal dependency of data, and ATT-LSTM improves the prediction performance to a certain extent by introducing an attention mechanism. However, the RNN-based model ignores the relationship between variables. The hybrid CNN-RNN method BLCNN is generally better than the single type method, while the deep learning-based methods 2D-CNN and AGCNN that consider the correlation of sensors perform better overall; although the transformer-based method is powerful in capturing global dependencies in the data, it may ignore the order of time series data.

[0141] The GNN-based method integrates temporal and spatial features and achieves more significant results. In terms of graph structure construction, STFA relies on the prior structural knowledge of the device, but it ignores the dynamic relationship between sensors in degradation. GGCN uses the cosine similarity between sensor features to construct, but does not fully utilize the prior knowledge of the device. CDSG constructs the prior graph structure and the designed recursive graph structure using the Hadamard product, but it is difficult to achieve perfect fusion of the graph structures of the two angles directly through the Hadamard product. In response to the above problems, the present invention proposes a new adaptive fusion method to construct the graph structure, that is, adding more weights to important graph structures, making up for the one-sidedness of the single association graph structure, and more richly reflecting the interactive relationship between sensors. The experimental results verify the prediction performance of the method of the present invention.

[0142] The present invention also provides a computer-readable storage medium, which stores a computer program. The computer program can be used to execute the method for predicting the remaining service life of a scraper conveyor provided above.

[0143] The present invention also provides Figure 7 The structural diagram of the computer device shown in FIG. Figure 7 As shown, at the hardware level, the computer device includes a processor, an internal bus, a network interface, a memory, and a non-volatile memory, and may also include hardware required for other services. The processor reads the corresponding computer program from the non-volatile memory into the memory and then runs it to implement the method for predicting the remaining service life of a scraper conveyor provided in the above embodiment.

[0144] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided by the present invention can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory or optical memory, etc. Volatile memory can include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM).

[0145] The technical features of the above embodiments may be arbitrarily combined. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of the present invention.

Claims

1. A method for predicting the remaining service life of a scraper conveyor, characterized in that: include: For different sensor data of the scraper conveyor obtained, multiple original graph data are constructed; multi-scale convolution operation is used to extract features of the original graph data to obtain feature graphs of different scales, and the feature graphs of different scales are fused to obtain fused graph data; By obtaining the spatial installation layout of the sensor, the prior adjacency matrix is ​​determined. Based on the fused graph data, the initial adjacency matrix is ​​constructed by using the KNN algorithm, and the initial adjacency matrix is ​​reconstructed by the graph autoencoder to obtain the similarity adjacency matrix. The prior adjacency matrix is ​​fused with the similarity adjacency matrix through an adaptive mechanism to obtain an adaptive adjacency matrix. By adopting the graph attention mechanism to fuse the graph data and the adaptive adjacency matrix, the updated node spatial features are obtained. The updated node spatial features are weighted by the time attention mechanism to determine the global time features. The global time features are spliced ​​and fused with the original time series to obtain new time series features. A bidirectional gated recurrent unit is used to encode the fused new time series features to obtain the fused spatiotemporal features. The fused spatiotemporal features are input into the fully connected layer to determine the final remaining service life and predict the life of the scraper conveyor.

2. The method for predicting the remaining service life of a scraper conveyor according to claim 1, characterized in that: The different sensor data of the scraper conveyor are obtained to construct a plurality of original image data, specifically including: The obtained data of different sensors of the scraper conveyor are normalized, and the normalized data are obtained using the following formula: Among them, x i,j is the i-th raw data of the j-th sensor, is the normalized data, and are the minimum and maximum values ​​of the raw data of the jth sensor, respectively; The sliding window technology is used to divide the normalized data to obtain the original image data of the sensor.

3. The method for predicting the remaining service life of a scraper conveyor according to claim 1, characterized in that: The method obtains the spatial installation layout of the sensor, determines the prior adjacency matrix, constructs the initial adjacency matrix based on the fused graph data by adopting the KNN algorithm, and reconstructs the initial adjacency matrix by using the graph autoencoder to obtain the similarity adjacency matrix, specifically including: The prior adjacency matrix is ​​determined using the following formula: Among them, A′ is the prior adjacency matrix, Represents node X i and node X j There is interaction; Based on the fused graph data, the KNN algorithm is used to determine the distance between each node and all nodes, and the k nodes closest to each other are selected to determine the neighbor node set N. i , the initial adjacency matrix is ​​obtained using the following formula: Among them, A t Represents the node X in the tth graph data i Is it related to node X? n adjacent; The initial adjacency matrix is ​​reconstructed using the following formula to obtain a similar adjacency matrix: Z=GCN(h,A t ), A″ t =σ(ZZ T ), Where GCN(·) represents the update process of GCN, h is the node feature, σ is the sigmoid activation function, Z is the embedding vector of the node, and A″ t is the t-th similarity adjacency matrix.

4. The method for predicting the remaining service life of a scraper conveyor according to claim 1, characterized in that: The updated node spatial features are obtained by fusing the graph data and the adjacency matrix using the graph attention mechanism, which specifically includes: The graph data and the adjacency matrix are fused by adopting the graph attention mechanism. The similarity coefficient between adjacent nodes is obtained by the nonlinear activation function LeakyReLU, and then the normalized attention coefficient is obtained after activation by the Softmax function. The updated node spatial features are obtained through the multi-head attention weights and output vectors.

5. The method for predicting the remaining service life of a scraper conveyor as claimed in claim 4, characterized in that: The nonlinear activation function LeakyReLU obtains the similarity coefficient between adjacent nodes, and then obtains the normalized attention coefficient after activation by the Softmax function, specifically including: The similarity coefficient between adjacent nodes and the normalized attention coefficient are obtained using the following formula: in, is the feature of the ith node, is the feature of its adjacent nodes, F is the original feature dimension of each node, e ij is the similarity coefficient between adjacent nodes, a ij is the normalized attention coefficient, a∈R 2F′ is a learnable weight vector, W∈R F×F′ is a learnable shared weight matrix, and F′ represents the feature dimension of the node output.

6. The method for predicting the remaining service life of a scraper conveyor according to claim 1, characterized in that: The time attention mechanism is used to weight the updated spatial node features to determine the global time features, and the global time features are spliced ​​and fused with the original time series to obtain new time series features, specifically including: The time attention mechanism is used to weight the updated node spatial features, and the original weight is determined by the following formula: ω t =o T tanh(h t ′), The normalized weight is obtained using the following formula: The global time feature is determined according to the normalized weight. The specific formula is as follows: The global time feature is fused with the original time series, and the new time series feature μ is obtained using the following formula t : m t =[h t ′||S], Among them, h t ′ is the updated spatial node feature, o∈R N×F′ is a learnable vector, F′ is the feature dimension of the node output, ω t is the original weight, θ t is the normalized weight, S∈R T It is the representation of global time characteristics.

7. The method for predicting the remaining service life of a scraper conveyor according to claim 1, characterized in that: The bidirectional gated recurrent unit is used to encode the fused new time series features to obtain fused spatiotemporal features, which specifically include: The forward hidden state and the reverse hidden state are obtained using the following formula: The following formula is used to concatenate the forward hidden state and the reverse hidden state to obtain the fused spatiotemporal features: Among them, μ t is the new time series feature after fusion, and Represent the forward and reverse hidden states at time t, Y t is the output after fusing the bidirectional hidden values, GRU(·) represents the GRU update process, u t , v t , b t are the corresponding learnable parameters.

8. The method for predicting the remaining service life of a scraper conveyor according to claim 1, characterized in that: The sensor data specifically includes: Material level sensor data, zero speed sensor data and speed sensor data for speed regulation.

9. A computer-readable storage medium, characterized in that: The storage medium stores a computer program, and when the computer program is executed by a processor, the method for predicting the remaining service life of a scraper conveyor as described in any one of claims 1 to 8 is implemented.

10. A computer device, characterized in that: It comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the method for predicting the remaining service life of a scraper conveyor as described in any one of claims 1 to 8 is implemented.

Images