Bearing Fault Diagnosis Method and System Based on Dual Attention Mechanism Reinforced Hierarchical Decision Network

Abstract

The present invention discloses a bearing fault diagnosis method and system based on a dual attention mechanism to enhance a hierarchical decision-making network. The method steps include: collecting bearing vibration signals under different health states; constructing a hierarchical multi-class fault diagnosis model based on the bearing vibration signals; and using the hierarchical multi-class fault diagnosis model to determine the fault location and size of the bearing. The present invention can not only accurately identify the bearing health state in an unbalanced and strong noise environment, but also the decision-making process of the sample state type is interpretable. At the same time, the decision-making idea of the present invention, which is to "locate the fault position first and then quantify the fault size", is different from the previous practice of "mixed decision-making at the same level for fault position and size", and is more in line with the common cognition of humans.

Description

Technical Field

[0001] The present invention relates to the field of bearing fault detection, and specifically to a bearing fault diagnosis method and system based on a dual attention mechanism enhanced hierarchical decision network. Background Art

[0002] In the existing detection of bearing faults, the fault diagnosis work of bearing health state classification is carried out under a relatively balanced data set; however, in the actual fault diagnosis work, on the one hand, the data distributions of different types of bearing faults are uneven, making the training and testing of the model more complex. On the other hand, the inevitable noise environment may further interfere with the feature extraction and diagnosis process of bearing signals. Therefore, how to achieve good bearing fault diagnosis in an environment of unbalanced data and strong noise interference is still a very challenging task. In terms of related technologies, although the attention mechanism-assisted CNN shows powerful capabilities in feature extraction and has achieved encouraging results in bearing fault diagnosis tasks, they have two main disadvantages: (1) They only focus on the input signal and output result, while ignoring the effective reasoning in the intermediate process, reducing the credibility of the diagnosis result; (2) Classifying the bearing fault location and size at the same level for mixing deviates from the conventional human thinking process. Summary of the Invention

[0003] To solve the technical problems in the above background, the present invention proposes a hierarchical multi-class fault diagnosis model of a dual attention-guided tree-shaped hierarchical decision network, aiming to achieve the fault diagnosis work of bearing health state classification in an environment of unbalanced strong noise.

[0004] To achieve the above object, the present invention provides a bearing fault diagnosis method based on a dual attention mechanism enhanced hierarchical decision network, and the steps include:

[0005] Collect bearing vibration signals under different health states;

[0006] Based on the bearing vibration signals, construct a hierarchical multi-class fault diagnosis model;

[0007] Use the hierarchical multi-class fault diagnosis model to determine the fault location and size of the bearing.

[0008] Preferably, the hierarchical multi-class fault diagnosis model includes a dual attention guidance mechanism and a tree heuristic hierarchical decision network;

[0009] The dual attention guidance mechanism is used to enhance the information closely related to the fault information in the bearing fault signal and weaken the interference information that has little relevance to the fault information;

[0010] The tree heuristic hierarchical decision network is used to make decisions on the position and size of the bearing fault step by step.

[0011] Preferably, the construction method of the dual attention guidance mechanism includes: integrating a triple attention mechanism and a multi-head convolutional attention mechanism into a CNN model to form the dual attention guidance mechanism;

[0012] Among them, the triple attention mechanism introduces a convolutional block attention module through the concept of cross-dimensional interaction, making the interaction between the channel and spatial dimensions more compact and comprehensive;

[0013] Among them, the multi-head convolutional self-attention mechanism is used to reduce the memory usage and computational cost during training or inference, while maintaining the interactivity and diversity between multiple heads.

[0014] Preferably, the tree heuristic hierarchical decision network is designed based on the basic logic of the fault diagnosis task, including: a fault type layer and a fault size layer;

[0015] The fault type layer is used to determine the fault type of the input sample;

[0016] The fault size layer is used to determine the fault size of the input sample.

[0017] Preferably, the dual attention guidance mechanism is used as the backbone network of the hierarchical multi-class fault diagnosis model; at the same time, a tree heuristic hierarchical decision network with a two-layer structure is built using two fully connected layers, and the weight information generated by the fully connected layer in the backbone network is inherited by the threshold of the seed node of the tree heuristic hierarchical decision network and further determines the threshold of the leaf node by the embedding decision rule between the seed node and the leaf node.

[0018] The present invention also provides a bearing fault diagnosis system based on a dual attention mechanism enhanced hierarchical decision network, which is used to implement the above method, including: a collection module, a construction module and a detection module;

[0019] The collection module is used to collect bearing vibration signals under different health states;

[0020] The construction module is used to construct a hierarchical multi-class fault diagnosis model based on the bearing vibration signal;

[0021] The detection module is used to determine the fault location and size of the bearing by using the hierarchical multi-class fault diagnosis model.

[0022] Preferably, the hierarchical multi-class fault diagnosis model includes a dual attention guidance mechanism and a tree heuristic hierarchical decision network;

[0023] The dual attention guidance mechanism is used to enhance the information closely related to the fault information in the bearing fault signal and weaken the interference information with little relevance to the fault information;

[0024] The tree-inspired hierarchical decision-making network is used to make decisions on the location and size of bearing faults step by step.

[0025] Preferably, the workflow of the building block includes integrating a triple attention mechanism and a multi-head convolutional attention mechanism into a CNN model to form the dual attention guidance mechanism;

[0026] Among them, the triple attention mechanism introduces a convolutional block attention module through the concept of cross-dimensional interaction, making the interaction between the channel and spatial dimensions more compact and comprehensive;

[0027] Among them, the multi-head convolutional self-attention mechanism is used to reduce the memory usage and computational cost during training or inference, while maintaining the interactivity and diversity among the multi-heads.

[0028] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0029] The present invention can not only accurately identify the bearing health state in an unbalanced strong noise environment, but also the decision-making process of the sample state type is interpretable. At the same time, the decision-making idea of the present invention, which is to "locate the fault position first and then quantify the fault size", is different from the previous practice of "mixed decision-making of the same level for fault position and size", and is more in line with the conventional human cognition. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] In order to more clearly illustrate the technical solutions of the present invention, the following briefly introduces the drawings required for the embodiments. Obviously, the drawings in the following description 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.

[0031] Figure 1 It is a schematic flowchart of the method according to the embodiment of the present invention;

[0032] Figure 2 It is a schematic diagram of the input part of the model structure according to the embodiment of the present invention;

[0033] Figure 3 It is a schematic diagram of the dual attention guidance mechanism of the model structure according to the embodiment of the present invention;

[0034] Figure 4 It is a schematic diagram of the tree-inspired hierarchical decision-making layer of the model structure according to the embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0035] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0036] To make the above objects, features, and advantages of the present invention more obvious and understandable, the present invention will be further described in detail below in conjunction with the accompanying drawings and specific embodiments.

[0037] Embodiment 1

[0038] As Figure 1 shown, it is a schematic diagram of the method flow of the present invention, and the steps include:

[0039] S1. Collect bearing vibration signals under different health states.

[0040] In this embodiment, the collected vibration signal samples are unbalanced bearing sample vibration signals with various health states under noise background collected from a fault test bench using a dedicated acceleration sensor.

[0041] S2. Based on the bearing vibration signals, construct a hierarchical multi-class fault diagnosis model.

[0042] To achieve the fault diagnosis work of bearing health state classification in an environment of strong unbalanced noise, this embodiment proposes a hierarchical multi-class fault diagnosis model of a Dual Attention guided Tree-inspired Grade Decision Network (DATGDN), as Figure 2 、 3 、4 shown; use the hierarchical multi-class fault diagnosis model to diagnose the bearing vibration signals and determine the fault location and size. It not only reduces the complexity related to the diagnosis task but also enhances the accuracy of fault diagnosis.

[0043] The hierarchical multi-class fault diagnosis model includes a dual attention guidance mechanism and a tree-inspired hierarchical decision network, and its construction method is as follows: First, integrate the triple attention mechanism and the multi-head convolutional attention mechanism into a new CNN model to form a dual attention guidance mechanism (as Figure 3 shown) to enhance the information closely related to the fault information in the bearing fault signal and weaken the interference information with little relevance to the fault information; secondly, design a hierarchical decision network with a two-layer tree structure to form a tree-inspired hierarchical decision layer (as Figure 4As shown in the figure, it is used to make hierarchical decisions on the location and size of bearing faults; finally, the tree-inspired hierarchical decision-making layer is arranged behind the double-attention-guided CNN model to seamlessly integrate the hierarchical multi-class fault diagnosis model.

[0044] To more clearly describe the proposed triple-attention mechanism, this embodiment first introduces cross-dimensional interaction and Z-pooling operations.

[0045] (1) Cross-dimensional interaction

[0046] Cross-dimensional interaction involves solving the problem of insufficient mutual dependence between the channel and spatial dimensions in traditional methods. Usually, traditional methods use singular weights when calculating channel attention, resulting in an obvious loss of spatial features in signal processing. This leads to a lack of a tight association between the channel dimension and the spatial dimension. To overcome this problem, the Convolutional Block Attention Module (CBAM) is introduced, which uses spatial attention as a supplementary module for channel attention. Specifically, CBAM makes up for the deficiency in terms of where to focus within the channel by introducing spatial attention, enabling the model to more comprehensively understand the positions that should be focused on within the channel. At the same time, channel attention still plays the role of specifying which channel to focus on, thus maintaining the attention to channel information.

[0047] (2) Z-pool

[0048] The task of Z-pool is to reduce the zeroth dimension of the tensor to two by concatenating the features of global average pooling (GMP) and global max pooling (GAP) in the dimension. The purpose of this operation is to improve the efficiency of subsequent calculations by reducing the depth while maintaining the comprehensive representation of the original tensor. The introduction of the Z-pool layer enables the model to more effectively process the input data while still maintaining a comprehensive grasp of the overall features. Its mathematical representation is shown in Equation 1. The Z-Pool operation generates a tensor of shape (2×H×W) from a tensor of shape (C×H×W).

[0049] Z-pool(χ) = [GMP 0d (χ), GAP 0d (χ)] (1)

[0050] In the formula, 0d represents the 0th dimension of the tensor on which the GMP and GAP operations are performed; x represents the input tensor.

[0051] As Figure 3 shown, for a given input tensor x ∈ R C×H×W , it is respectively passed into the three branches of the proposed triple-attention module.

[0052] In the first branch, an interaction is constructed between the height dimension and the channel dimension. To achieve this function, first, the input tensor x is rotated counterclockwise by 90° along the H axis, generating a rotated tensor x with the shape of (W×H×C). 0H ; Second, x 0H is further processed through the Z-Pool operation to obtain x with the shape and size of (2×H×C). 0HZ ; Third, x 0HZ is convolved with a convolutional kernel of size k×k and batch-normalized, generating an intermediate output tensor with the dimension of (1×H×C); Next, the generated intermediate output tensor is processed through a sigmoid activation layer (σ) to generate attention weights and applied to x 0H ; Finally, it is rotated clockwise by 90° along the H axis to generate an attention weight tensor with the same shape as the original input tensor x.

[0053] Similarly, within the second branch, the original signal is rotated counterclockwise by 90° along the W axis to generate a rotated tensor x with the shape and size of (H×C×W). 1W ; Second, this tensor is processed through the Z-pool operation to obtain x with the shape and size of (2×C×W). 1WZ ; Third, x 1WZ is convolved with a convolutional kernel of size k×k and batch-normalized, generating an intermediate output tensor with the dimension of (1×H×C); Next, the generated intermediate output tensor is processed through a sigmoid activation layer (σ) to generate attention weights and applied to x 1W ; Finally, it is rotated clockwise by 90° along the W axis to generate an attention weight tensor with the same shape as the original input tensor x.

[0054] For the third branch, first, x with the shape of (C×H×W) of the input tensor is directly processed through the Z-pool operation to generate a tensor x with the shape and size of (2×H×W). 2Z ; Second, x 2Z is convolved with a convolutional kernel of size k×k and batch-normalized to generate an intermediate output tensor with the dimension of (1×H×C); Finally, the obtained intermediate output tensor is processed through a sigmoid activation layer (σ) to generate an attention weight tensor with the shape and size of (1×H×W).

[0055] Finally, the tensors with the shape and size of (C×H×W) generated by the three branches are first merged and then averaged to obtain a refined attention tensor y. Its calculation process is shown in Equation 2:

[0056]

[0057] In the formula, ψ1, ψ2, and ψ3 all represent standard convolution operations.

[0058] Equation 2 can be simplified to Equation 3:

[0059]

[0060] Where ω1, ω2, and ω3 respectively represent three cross-dimensional attention weights calculated in the three branches of the ternary attention; y1, y2, and y3 respectively represent the attention weight tensors generated by the first branch, the second branch, and the third branch.

[0061] In short, the ternary attention mechanism introduces CBAM through the concept of cross-dimensional interaction, making the interaction between the channel and spatial dimensions more compact and comprehensive. Further, the ternary attention can simultaneously consider the information of channels, space, and channel-space cross-dimensions, improving the model's perception ability of key features.

[0062] As Figure 3 shown, this embodiment also proposes a multi-head convolutional self-attention (MCSA) based on single-head convolutional attention (SCA). The aim is to reduce memory usage and computational costs during training or inference while maintaining the interactivity and diversity between multiple heads. MCSA overcomes the memory and computational limitations of the standard MSA to more effectively handle the processing requirements of large-scale data and high-dimensional data. This enables MCSA to handle complex tasks while maintaining the advantages of the multi-head self-attention mechanism. The specific method is as follows:

[0063] First, the one-dimensional original signal is used to generate a two-dimensional input tensor by means of embedded encoding where n and d m respectively represent the spatial dimension and the channel dimension of x1.

[0064] Second, similar to MSA, MCSA first uses a set of projections to obtain queries. At the same time, to effectively compress memory, the two-dimensional input tensor is reshaped into a three-dimensional tensor along the spatial dimension

[0065] Third, the three-dimensional input tensor is subjected to depthwise separable convolution (DWConv) and layer normalization techniques to obtain a new three-dimensional tensor In the depthwise separable convolution operation, the three-dimensional input tensor is reduced in its height and width dimensions by a scaling factor s (determined according to the size of the feature map) to obtain a new tensor Among them, the convolution kernel size, stride, and padding are s + 1, s, and s / 2 respectively.

[0066] Next, the new three-dimensional input tensor is reshaped into a new two-dimensional input tensor and two sets of fully connected feature maps are obtained for Key and Value respectively, where n = (h / s) × (w / s).

[0067] Furthermore, use Equation 5 to replace Equation 4 to calculate Query (Q), Key (K), and Value (V) in the attention function, multiply Query and Key and normalize them, then input them into the convolution operation with Softmax and perform instance normalization (IN) operation.

[0068]

[0069] In the formula, Conv(·) represents the standard 1*1 convolution operation, which is used to construct the interaction of information between different heads in the multi-head attention; d k represents the channel dimension of the input data; CSA represents the convolutional self-attention mechanism; SA represents the self-attention mechanism; T represents the transpose.

[0070] Finally, the output values of each head are concatenated and passed through a linear projection to form the final output.

[0071] MCSA(Q, K, V) = Concat(head i )W o (6)

[0072] In the formula, W o represents the weight matrix generated by the linear projection operation; MCSA represents the multi-head convolutional self-attention mechanism; Concat represents the concatenation operation; head i represents the detection head.

[0073] After the multi-head convolutional attention mechanism is completed, it is integrated with the triple attention mechanism into the CNN model to form a Double Attention-guided Convolutional Neural Network (DACNN), which is used as the backbone network of the overall multi-class fault diagnosis model. In the constructed DACNN, the two attentions are applied as two branches to enhance the CNN respectively, ensuring that the bearing fault features extracted by the two attention mechanisms do not interfere with each other. Finally, the fault features extracted by the two attention-enhanced CNNs are aggregated. This operation ensures that the fault features extracted by each individual attention-enhanced CNN retain their specific meanings and guarantees that these features strengthen each other rather than weaken each other.

[0074] There is an inherent logical relationship between fault location localization and the determination of fault size severity, and most deep learning models do not reflect this internal connection. In this embodiment, a novel tree-inspired hierarchical decision network is designed, as Figure 4 shown, to establish the diagnostic logic between the fault location and the fault size. The tree-inspired hierarchical decision network is designed based on the basic logic of the fault diagnosis task. In this embodiment, a two-layer hierarchical fault diagnosis architecture is deployed, corresponding to the fault type layer and the fault size layer of the bearing respectively. The first layer focuses on determining the fault type of the input sample to obtain the corresponding superclass attributes, while the second layer focuses on determining the fault size of the input sample to obtain the corresponding subclass attributes.

[0075] To better understand the distribution of the embedding features generated by this model, a two-layer tree-inspired hierarchical decision network is built using two fully connected layers. In this process, the weight information generated by the fully connected layer in the backbone network is inherited by the threshold of the seed node of the tree-inspired hierarchical decision network, and the threshold of the leaf node is further determined by the embedding decision rule between the seed node and the leaf node. The weight of the seed node directly inherits the probability distribution of the pre-trained fully connected layer, which ensures that the recognition ability of the subclass is similar to that of the pre-trained hierarchical multi-class fault diagnosis model.

[0076] The bearing vibration signals with known states are input into the multi-class fault diagnosis model for training to complete the construction of the hierarchical multi-class fault diagnosis model.

[0077] S3. Use the hierarchical multi-class fault diagnosis model to determine the fault location and size of the bearing.

[0078] The bearing vibration signals with unknown states are input into the trained multi-class fault diagnosis model to detect the health condition of the bearings to be tested.

[0079] Embodiment 2

[0080] This embodiment also provides a bearing fault diagnosis system based on a dual attention mechanism enhanced hierarchical decision-making network, including: an acquisition module, a construction module, and a detection module; the acquisition module is used to acquire bearing vibration signals under different health states; the construction module is used to construct a hierarchical multi-class fault diagnosis model based on the bearing vibration signals; the detection module is used to determine the fault location and size of the bearing by using the hierarchical multi-class fault diagnosis model.

[0081] The hierarchical multi-class fault diagnosis model includes a dual attention guidance mechanism and a tree heuristic hierarchical decision-making network; the dual attention guidance mechanism is used to enhance the information closely related to the fault information in the bearing fault signal and weaken the interference information with little relevance to the fault information; the tree heuristic hierarchical decision-making network is used to make decisions on the location and size of the bearing fault step by step. The working process of the construction module includes: integrating the triple attention mechanism and the multi-head convolutional attention mechanism into a CNN model to form a dual attention guidance mechanism; among them, the triple attention mechanism introduces a convolutional block attention module through the cross-dimensional interaction concept to make the interaction between the channel and spatial dimensions more compact and comprehensive; constructing a multi-head convolutional self-attention mechanism based on the single-head convolutional attention.

[0082] This embodiment designs a tree heuristic hierarchical decision-making network based on the basic logic of the fault diagnosis task, including: a fault type layer and a fault size layer; the fault type layer is used to determine the fault type of the input sample; the fault size layer is used to determine the fault size of the input sample. The dual attention guidance mechanism is used as the backbone network of the hierarchical multi-class fault diagnosis model; at the same time, a tree heuristic hierarchical decision-making network with a two-layer structure is built by using two fully connected layers, and the weight information generated by the fully connected layer in the backbone network is inherited by the threshold of the seed node of the tree heuristic hierarchical decision-making network and the threshold of the leaf node is further determined by the embedding decision rule between the seed node and the leaf node.

[0083] The above-described embodiments are only descriptions of the preferred embodiments of the present invention, and do not limit the scope of the present invention. Without departing from the design spirit of the present invention, various deformations and improvements made by those of ordinary skill in the art to the technical solutions of the present invention shall fall within the protection scope determined by the claims of the present invention.

Claims

1. A bearing fault diagnosis method based on a dual attention mechanism to strengthen the hierarchical decision network, characterized in that the steps include: Collect bearing vibration signals under different health conditions; Based on the bearing vibration signal, a hierarchical multi-category fault diagnosis model is constructed; The hierarchical multi-class fault diagnosis model includes a dual attention guidance mechanism and a tree-inspired hierarchical decision network; The method for constructing the dual attention guidance mechanism includes: integrating the ternary attention mechanism and the multi-head convolutional attention mechanism into a CNN model to form the dual attention guidance mechanism; the multi-head convolutional self-attention is constructed based on the single-head convolutional attention, and the steps include: First, the one-dimensional original signal is converted into a two-dimensional input tensor using an embedded coding method. where n and d m Represent the spatial dimension and channel dimension of x1 respectively; Second, a set of projections is used to obtain the query query. At the same time, the two-dimensional input tensor Reshape into a 3D tensor along the spatial dimension Again, the three-dimensional input tensor Perform depth-separable convolution and layer normalization techniques to obtain a new three-dimensional tensor In a depth-wise separable convolution operation, the three-dimensional input tensor The new tensor is obtained by reducing its height and width dimensions by scaling factor s Among them, the convolution kernel size, step size and padding are s+1, s and s / 2 respectively; Next, the new 3D input tensor is reshaped into a new 2D input tensor And perform two sets of fully connected feature mapping to obtain Key and Value, where n = (h / s) × (w / s); Furthermore, the Query, Key, and Value in the attention function are calculated and the Query and Key are multiplied and normalized, and then input into the convolution operation with Softmax and instance normalization operation is performed. The calculation formula is as follows: Where Q represents Query, K represents Key, V represents Value, Conv(·) represents a standard 1*1 convolution operation, which is used to construct the interaction of information between different heads in multi-head attention. k Represents the channel dimension of the input data; SA represents the self-attention mechanism; T represents transposition; Finally, the output values ​​of each head are concatenated and linearly projected to form the final output: MCSA(Q,K,V)=Concat(head i )W O Where W O represents the weight matrix generated by the linear projection operation; MCSA represents the multi-head convolution self-attention mechanism; Concat represents the connection operation; head i Indicates the detection head; The hierarchical multi-class fault diagnosis model is used to determine the fault location and size of the bearing.

2. The bearing fault diagnosis method based on the dual attention mechanism enhanced hierarchical decision network according to claim 1 is characterized in that: The dual attention guidance mechanism is used to enhance the information in the bearing fault signal that is closely related to the fault information and weaken the interference information that is not closely related to the fault information; The tree-inspired hierarchical decision network is used to determine the location and size of bearing faults step by step.

3. The bearing fault diagnosis method based on the dual attention mechanism enhanced hierarchical decision network according to claim 2 is characterized in that: The ternary attention mechanism introduces the convolutional block attention module through the concept of cross-dimensional interaction, making the interaction between channels and spatial dimensions more compact and comprehensive; Among them, the multi-head convolutional self-attention mechanism is used to reduce memory usage and computational costs during training or inference while maintaining interactivity and diversity among multiple heads.

4. The bearing fault diagnosis method based on dual attention mechanism enhanced hierarchical decision network according to claim 2 is characterized in that: The tree-inspired hierarchical decision network is designed based on the basic logic of the fault diagnosis task, including: a fault type layer and a fault size layer; The fault type layer is used to determine the fault type of the input sample; The fault size layer is used to determine the fault size of input samples.

5. The bearing fault diagnosis method based on dual attention mechanism enhanced hierarchical decision network according to claim 2 is characterized in that: The dual attention guidance mechanism is used as the backbone network of the hierarchical multi-category fault diagnosis model; at the same time, a tree-inspired hierarchical decision network with a two-layer structure is constructed using two fully connected layers. The weight information generated by the fully connected layer in the backbone network is inherited by the threshold of the seed node of the tree-inspired hierarchical decision network, and the threshold of the leaf node is further determined by the embedded decision rule of the seed node and the leaf node.

6. A bearing fault diagnosis system based on a dual attention mechanism to strengthen a hierarchical decision network, the system being used to implement the method described in any one of claims 1 to 5, characterized in that: include: Acquisition module, construction module and detection module; The acquisition module is used to collect bearing vibration signals in different health states; The building module is used to build a hierarchical multi-category fault diagnosis model based on the bearing vibration signal; The detection module is used to determine the fault location and size of the bearing using the hierarchical multi-category fault diagnosis model.

7. The bearing fault diagnosis system based on the dual attention mechanism enhanced hierarchical decision network according to claim 6 is characterized in that: The hierarchical multi-class fault diagnosis model includes a dual attention guidance mechanism and a tree-inspired hierarchical decision network; The dual attention guidance mechanism is used to enhance the information in the bearing fault signal that is closely related to the fault information and weaken the interference information that is not closely related to the fault information; The tree-inspired hierarchical decision network is used to determine the location and size of bearing faults step by step.

8. The bearing fault diagnosis system based on the dual attention mechanism enhanced hierarchical decision network according to claim 7 is characterized in that: The workflow of the building module includes: integrating the ternary attention mechanism and the multi-head convolutional attention mechanism into a CNN model to form the dual attention guidance mechanism; Among them, the ternary attention mechanism introduces the convolutional block attention module through the concept of cross-dimensional interaction, making the interaction between channels and spatial dimensions more compact and comprehensive; Among them, the multi-head convolutional self-attention mechanism is used to reduce memory usage and computational costs during training or inference while maintaining interactivity and diversity among multiple heads.

Images