Fault diagnosis method and device for vibrating screen spring

Abstract

The invention provides a vibration screen spring fault diagnosis method and device. The method comprises the steps that an original vibration signal of a vibration screen spring is collected through a vibration sensor; extracting a time-frequency characteristic signal from the original vibration signal; utilizing a pre-trained feature fusion neural network model to extract spatial features and time features from the time-frequency feature signals; fusing the spatial features and the time features to obtain fused features; and inputting the fusion features into a pre-trained fault classifier, and outputting a fault diagnosis result. The time features and the space features of the signals are extracted through the feature fusion neural network model, finally, the features are fused to achieve spring fault diagnosis, vibration screen spring faults can be diagnosed timely and accurately, and the method has great significance in guaranteeing safe and stable operation of a coal dressing production system.

Description

Technical Field

[0001] This invention relates to the field of mechanical fault diagnosis, and in particular to a fault diagnosis method and apparatus for a vibrating screen spring. Background Technology

[0002] As a key piece of equipment in coal preparation plants, the stable operation of vibrating screen springs plays a crucial role in production efficiency and product quality. However, vibrating screens operate under high-speed rotation and strong vibration environments, facing frequent changes in alternating loads and unexpected impacts. These factors can easily lead to failures in components such as the springs. Fault diagnosis of vibrating screen springs is of great significance for ensuring the safe and stable operation of coal preparation production systems. Currently, research on vibrating screen spring fault diagnosis technology mainly focuses on using intelligent diagnostic methods to improve the accuracy of diagnostic results, especially under conditions of limited fault samples. Fault diagnosis technology for vibrating screen springs is developing towards intelligence and precision, which has significant practical implications for improving manufacturing levels and optimizing production processes.

[0003] However, current technologies do not fully consider the combined impact of multiple factors on the diagnosis of vibrating screen spring faults, especially the combined noise interference generated by multiple vibrating screens operating simultaneously. Such interference may mask the true fault characteristic signals, leading to misjudgment or omission by the diagnostic model and reducing the accuracy of fault identification. Summary of the Invention

[0004] This invention provides a fault diagnosis method and device for vibrating screen springs, which is used to solve the problem of nonlinear and non-stationary fault signals caused by external noise and other factors during the coal preparation process, and to achieve effective extraction and accurate diagnosis of uncertain fault characteristics.

[0005] In a first aspect, the present invention provides a method for diagnosing spring faults in a vibrating screen, comprising: The original vibration signal of the vibrating screen spring is collected by a vibration sensor; Extract time-frequency feature signals from the original vibration signal; Using a pre-trained feature fusion neural network model, spatial features and temporal features are extracted from the time-frequency feature signal, respectively. The spatial features and the temporal features are fused to obtain the fused features; The fused features are input into a pre-trained fault classifier, which outputs fault diagnosis results.

[0006] Optionally, extracting time-frequency feature signals from the original vibration signal includes: Perform continuous wavelet transform on the original vibration signal to extract the corresponding first time-frequency features; The original vibration signal is subjected to eigenmode decomposition to obtain a series of intrinsic mode functions as the second time-frequency features; The original vibration signal is subjected to a short-time Fourier transform to obtain a third time-frequency feature; the time-frequency feature signal includes at least one of the first time-frequency feature, the second time-frequency feature and the third time-frequency feature.

[0007] Optionally, the feature fusion neural network model includes: a dynamic graph convolutional network and a long short-term memory network; using the pre-trained feature fusion neural network model, spatial features and temporal features are extracted from the time-frequency feature signal, respectively, including: The time-frequency characteristic signal is plotted as a spatiotemporal graph data; The spatial features are extracted from the spatiotemporal graph data using the dynamic graph convolutional network. The time features are extracted from the time-frequency feature signal using the long short-term memory network.

[0008] Optionally, the spatial features are extracted from the spatiotemporal graph data using the dynamic graph convolutional network, including: The spatiotemporal graph data is processed by the dynamic graph convolutional network, and the EdgeConv operation is performed to dynamically update the graph structure and capture local geometric features. A channel attention mechanism is introduced to adaptively adjust the weights of different channel features in the local geometric features; The weighted local geometric features are reduced in dimensionality by using a max pooling layer and a global average pooling layer, and global spatial information is integrated to obtain the spatial features.

[0009] Optionally, the time feature is extracted from the time-frequency feature signal using the long short-term memory network, including: Perform one-dimensional convolution and batch normalization on the time series; The processed features are input into the Long Short-Term Memory network, and long-term dependencies are captured through its gating mechanism. The time feature is formed based on the long-term dependency relationship.

[0010] Optionally, the feature fusion neural network model further includes a feature fusion sub-network; fusing the spatial features and the temporal features to obtain fused features, including: The spatial features and the temporal features are concatenated along the feature dimension; The spliced ​​features are input into the fully connected layer of the feature fusion subnetwork for nonlinear combination; Regularization is performed on the features combined by the fully connected layer to generate the fused features.

[0011] Secondly, the present invention provides a vibrating screen spring fault diagnosis device, comprising: The acquisition module is used to acquire the original vibration signal of the vibrating screen spring through a vibration sensor; The signal extraction module is used to extract time-frequency feature signals from the original vibration signal; The feature extraction module is used to extract spatial features and temporal features from the time-frequency feature signal using a pre-trained feature fusion neural network model. The fusion module is used to fuse the spatial features and the temporal features to obtain fused features; The prediction module is used to input the fused features into a pre-trained fault classifier and output fault diagnosis results.

[0012] Thirdly, the present invention provides an electronic device including a processor and a memory, the memory storing computer-readable instructions that, when executed by the processor, perform the steps of the method provided in the first aspect above.

[0013] Fourthly, the present invention provides a storage medium having a computer program stored thereon, which, when executed by a processor, performs the steps of the method provided in the first aspect above.

[0014] Fifthly, the present invention provides a computer program product comprising a computer program that, when executed by a processor, performs the steps of the method provided in the first aspect above.

[0015] As can be seen from the above technical solutions, the present invention has the following advantages: This invention provides a method and apparatus for diagnosing spring faults in vibrating screens. The method includes: acquiring raw vibration signals of the vibrating screen springs using a vibration sensor; extracting time-frequency feature signals from the raw vibration signals; extracting spatial and temporal features from the time-frequency feature signals using a pre-trained feature fusion neural network model; fusing the spatial and temporal features to obtain fused features; and inputting the fused features into a pre-trained fault classifier to output fault diagnosis results. By using a feature fusion neural network model to extract the temporal and spatial features of the signal respectively, and finally fusing the features to achieve spring fault diagnosis, this method can diagnose vibrating screen spring faults in a timely and accurate manner, which is of great significance for ensuring the safe and stable operation of coal preparation production systems. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 This is a flowchart illustrating the steps of a method for diagnosing spring faults in a vibrating screen according to the present invention. Figure 2 This is a flowchart illustrating the steps of a second embodiment of the method for diagnosing spring faults in a vibrating screen according to the present invention. Figure 3 This is one of the characteristic mode decomposition results of a second embodiment of the vibrating screen spring fault diagnosis method of the present invention; Figure 4 This is the second characteristic mode decomposition result diagram of a second embodiment of the vibrating screen spring fault diagnosis method of the present invention; Figure 5 The third characteristic mode decomposition result diagram is shown in Embodiment 2 of the vibrating screen spring fault diagnosis method of the present invention. Figure 6 The fourth figure shows the characteristic mode decomposition results of a second embodiment of the vibrating screen spring fault diagnosis method of the present invention. Figure 7 The fifth figure shows the characteristic mode decomposition results of Embodiment 2 of the vibrating screen spring fault diagnosis method of the present invention; Figure 8 This is a schematic diagram of the dynamic convolutional network structure of a second embodiment of the vibrating screen spring fault diagnosis method of the present invention; Figure 9 This is a flowchart illustrating the fault diagnosis process of a second embodiment of the vibrating screen spring fault diagnosis method of the present invention. Figure 10 This is a structural block diagram of an embodiment of a vibrating screen spring fault diagnosis device according to the present invention. Detailed Implementation

[0018] This invention provides a fault diagnosis method and device for vibrating screen springs, which is used to solve the problem of nonlinear and non-stationary fault signals caused by external noise and other factors during the coal preparation process, and to achieve effective extraction and accurate diagnosis of uncertain fault characteristics.

[0019] To make the objectives, features, and advantages of this invention more apparent and understandable, the technical solutions of the embodiments of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the embodiments described below are only some embodiments of this invention, and not all embodiments. Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this invention.

[0020] Example 1 Please see Figure 1 , Figure 1 This is a flowchart illustrating the steps of a fault diagnosis method for a vibrating screen spring according to an embodiment of the present invention. The method includes: Step S101: Acquire the original vibration signal of the vibrating screen spring through a vibration sensor; In this embodiment, a vibration sensor is used to collect real-time signals from the springs of a vibrating screen in a coal preparation plant, obtaining the original vibration signals under its working condition.

[0021] Step S102: Extract time-frequency feature signals from the original vibration signal; In the embodiments of this application, the original vibration signal is preprocessed and multi-dimensional features are extracted to better characterize nonlinear and non-stationary fault information.

[0022] In practical implementation, the first time-frequency feature of the signal is extracted through continuous wavelet transform; the signal is decomposed into a series of intrinsic mode functions (EMFs) through eigenmode decomposition, which serve as the second time-frequency feature; and the signal is transformed into the frequency domain through short-time Fourier transform to obtain the third time-frequency feature. The extracted time-frequency feature signal contains a combination of one or more of the above features.

[0023] Step S103: Using a pre-trained feature fusion neural network model, spatial features and temporal features are extracted from the time-frequency feature signal, respectively. In an optional embodiment, the feature fusion neural network model includes: a dynamic graph convolutional network and a long short-term memory network; using the pre-trained feature fusion neural network model, spatial features and temporal features are extracted from the time-frequency feature signal, respectively, including: The time-frequency characteristic signal is plotted as a spatiotemporal graph data; The spatial features are extracted from the spatiotemporal graph data using the dynamic graph convolutional network. The time features are extracted from the time-frequency feature signal using the long short-term memory network.

[0024] In this embodiment of the application, the time-frequency characteristic signal is plotted as spatiotemporal graph data; In this embodiment, the time-frequency feature signal extracted in step S102 is converted into graph structure data to simultaneously express the spatial relationships and temporal evolution between data points. Then, the dynamic graph convolutional network is used to extract the spatial features from the spatiotemporal graph data; DGCNN dynamically updates the graph structure through its core EdgeConv operation to capture local geometric features in the spatiotemporal graph data. To further enhance feature representation capabilities, a channel attention mechanism is introduced to adaptively adjust the weights of different channel features, highlighting key information. Subsequently, the features are dimensionality-reduced and global spatial information is integrated through max pooling and global average pooling layers, ultimately forming fused spatial features. The long short-term memory network is used to extract the temporal features from the time-frequency feature signal, and simultaneously, the time-frequency feature signal is processed as a time series. First, it undergoes one-dimensional convolution and batch normalization to initially extract temporal patterns and stabilize the training. Then, the processed features are input into the long short-term memory network, utilizing its gating mechanism to capture long-term dependencies in the time series, ultimately forming temporal features.

[0025] Step S104: The spatial features and the temporal features are fused to obtain fused features; In this embodiment, the spatial features extracted by DGCNN and the temporal features extracted by LSTM are concatenated along the feature dimension to form a comprehensive feature vector. This concatenated feature is then input into the fully connected layer of the feature fusion subnetwork for nonlinear combination to learn the deep correlation between spatiotemporal features. Finally, regularization is performed on the features output by the fully connected layer to generate the final, robust fused feature, which simultaneously contains key information about the fault in both the spatial and temporal dimensions.

[0026] Step S105: Input the fused features into a pre-trained fault classifier and output the fault diagnosis result.

[0027] In this embodiment of the application, the fused features are input to a pre-trained fault classifier. The classifier calculates the probability of different fault types (such as spring torsion deformation, spring wire crack, jamming, insufficient spring force, etc.) based on the fused features and outputs the final fault diagnosis result, thereby realizing accurate and automated diagnosis of the condition of the vibrating screen spring.

[0028] This invention provides a method for diagnosing spring faults in a vibrating screen, comprising: acquiring raw vibration signals of the vibrating screen spring using a vibration sensor; extracting time-frequency feature signals from the raw vibration signals; extracting spatial and temporal features from the time-frequency feature signals using a pre-trained feature fusion neural network model; fusing the spatial and temporal features to obtain fused features; and inputting the fused features into a pre-trained fault classifier to output a fault diagnosis result. By using a feature fusion neural network model to extract the temporal and spatial features of the signal respectively, and finally fusing the features to achieve spring fault diagnosis, this method can diagnose vibrating screen spring faults in a timely and accurate manner, which is of great significance for ensuring the safe and stable operation of coal preparation production systems.

[0029] Example 2 Please see Figure 2 , Figure 2 This is a flowchart illustrating a second embodiment of the vibrating screen spring fault diagnosis method of the present invention. The steps include: Step S201: Collect the original vibration signal of the vibrating screen spring using a vibration sensor; In this embodiment of the application, a vibration sensor is used to collect signals from the vibrating screen spring to obtain the original vibration signal.

[0030] Step S202: Extract time-frequency feature signals from the original vibration signal; In this embodiment, the original vibration signal is subjected to continuous wavelet transform to extract the corresponding first time-frequency feature; the original vibration signal is subjected to eigenmode decomposition to obtain a series of intrinsic mode functions as the second time-frequency feature; the original vibration signal is subjected to short-time Fourier transform to obtain the third time-frequency feature; the time-frequency feature signal includes at least one of the first time-frequency feature, the second time-frequency feature and the third time-frequency feature.

[0031] In practical implementation, the Eigenmode Decomposition (EMD) algorithm is used to decompose the acquired raw vibration signal to improve the effectiveness of subsequent feature extraction. The theoretical formula for Eigenmode Decomposition can be expressed as:

[0032] in Indicates the input signal. Indicates the first One characteristic mode function, This represents the residual. The Eigenmode Decomposition process requires extracting multiple... The specific steps are as follows: (1) The input signal I(n) is first preprocessed, such as denoising or normalization, to improve the accuracy of decomposition; (2) Perform a fast Fourier transform on the preprocessed signal to obtain its frequency domain representation; (3) In the frequency domain, the signal is filtered to separate different frequency components; (4) Perform inverse fast Fourier transform on the filtered frequency domain signal to obtain an approximate signal in the time domain; (5) Subtract the approximate signal from the original signal to obtain the first intrinsic mode function IMF1(n). Repeat the above steps to decompose the remaining signal until all IMF components are obtained. (6) The remaining signal after the last IMF component is the residual ResM(n); (7) Add all IMF components and residuals together to reconstruct the original signal and verify the accuracy of the decomposition; Vibration signals processed by the above FMD can be better used for subsequent feature extraction and fault diagnosis, improving the accuracy of diagnosis.

[0033] Please see Figures 3-7 , Figures 3-7 This paper presents typical results obtained after processing the original vibration signal using the Eigenmode Decomposition (EMD) algorithm. The figure visualizes how EMD decomposes a complex, non-stationary original signal into a series of relatively stationary intrinsic mode functions (EMFs) with different center frequencies and a residual component. The decomposition results demonstrate that the EMD algorithm can effectively remove noise and redundant information from the original signal, separating complex fault features into different EMF components. This significantly improves the efficiency and accuracy of feature extraction using subsequent dynamic graph convolutional networks and long short-term memory (LSTM) networks, laying a solid foundation for high-precision fault diagnosis.

[0034] Step S203: Construct the time-frequency characteristic signal into a spatiotemporal graph data; In this embodiment, the extracted time-frequency feature signals are converted into graph-structured data to form spatiotemporal graph data. This step aims to capture the spatial dependencies and temporal evolution patterns in the vibration signals, providing input for subsequent dynamic graph convolutional network processing.

[0035] Step S204: Extract the spatial features from the spatiotemporal graph data using the dynamic graph convolutional network in the pre-trained fusion neural network model; the fusion neural network model includes a long short-term memory network and a feature fusion sub-network in addition to the dynamic graph convolutional network. In this embodiment, the spatiotemporal graph data is processed by the dynamic graph convolutional network, and the EdgeConv operation is performed to dynamically update the graph structure and capture local geometric features. A channel attention mechanism is introduced to adaptively adjust the weights of different channel features in the local geometric features. The weighted local geometric features are reduced in dimensionality and integrated with global spatial information by max pooling layer and global average pooling layer to obtain the spatial features.

[0036] Please see Figure 8 , Figure 8 This is a schematic diagram of the dynamic graph convolutional network structure of a second embodiment of the vibrating screen spring fault diagnosis method of the present invention. The structure includes: (1) Input layer: receiving vibration signal data represented by a graph structure, where nodes represent data points and edges represent the relationships between points. (2) EdgeConv operation: aggregating the local neighborhood information of nodes in each layer to capture local geometric features, and dynamically updating the connection relationship of the graph according to the learned new features, thereby non-locally spreading information, so that the network can adaptively learn the graph structure most conducive to fault diagnosis. (3) Channel attention mechanism: a channel attention mechanism is introduced into the feature sequence. This mechanism can adaptively learn and calibrate the weights of different feature channels, thereby highlighting the key features related to the fault and suppressing unimportant features. (4) Pooling layer: the network finally reduces the dimensionality of the features processed by the multi-layer cascaded EdgeConv and attention mechanism through operations such as max pooling and global average pooling, and integrates global spatial information, and finally outputs a fixed-length, highly abstract spatial feature vector.

[0037] Step S205: Using the long short-term memory network, extract the time features from the time-frequency feature signal; Long Short-Term Memory (LSTM) networks are a typical deep learning method. The collected vibration signals are stored in time series. The storage units and gating mechanisms in LTM networks can solve the gradient vanishing or gradient exploding problems and can handle data with long-term dependencies very well.

[0038] In this embodiment, a one-dimensional convolution operation and batch normalization are performed on the time series; the processed features are input into the Long Short-Term Memory network, and long-term dependencies are captured through its gating mechanism; the time features are formed based on the long-term dependencies.

[0039] In the specific implementation, a one-dimensional convolution operation is performed on the time series to extract time series features. Then, the features are normalized through a batch normalization layer to improve the stability of model training. Finally, the processed features are input into a long short-term memory network, which captures long-term dependencies through its gating mechanism, and forms the time features based on the long-term dependencies.

[0040] Step S206: Concatenate the spatial features and the temporal features along the feature dimension; In this embodiment, the spatial features extracted by the dynamic graph convolutional network and the temporal features extracted by the long short-term memory network are concatenated along the feature dimension to form a preliminary fused feature vector, which provides input for subsequent nonlinear combination and classification.

[0041] Step S207: Input the spliced ​​features into the fully connected layer of the feature fusion subnetwork for nonlinear combination; In this embodiment, the spliced ​​features are input into the fully connected layer of the feature fusion subnetwork to perform nonlinear combination and transformation of the features, thereby further enhancing the expressive power and discriminative power of the features.

[0042] Step S208: Perform regularization processing on the features combined by the fully connected layer to generate the fused features; In this embodiment, regularization is performed on the features output by the fully connected layer to prevent overfitting and generate the final fused features, thereby improving the model's generalization ability.

[0043] Step S209: Input the fused features into a pre-trained fault classifier and output the fault diagnosis result.

[0044] In this embodiment, the fused features are input to a fault classifier, which outputs the fault type diagnosis results of the vibrating screen spring, thereby achieving real-time monitoring and accurate diagnosis of spring faults.

[0045] Please see Figure 9 , Figure 9This is a flowchart illustrating the fault diagnosis process of a second embodiment of the vibrating screen spring fault diagnosis method of the present invention. The process begins with the acquisition of vibration signals from the physical world and ends with the output of fault diagnosis results and decision support. The prediction process is roughly divided into three stages: signal preprocessing, spatial information fusion module, and temporal information fusion module. In the signal preprocessing stage, the original vibration signal first undergoes continuous wavelet transform to extract the time-frequency features of the signal. Then, the signal is further decomposed through fast mode decomposition to obtain vibration features of different modes. Finally, the signal is converted to the frequency domain through short-time Fourier transform, providing a basis for subsequent feature extraction. Next, the spatial information fusion module is entered. This module first performs multiple two-dimensional convolution operations on the preprocessed signal to extract local spatial features. Subsequently, a max pooling layer is used to reduce the feature dimension, retaining the most important feature information. To further enhance the expressive power of the features, a channel attention mechanism is introduced, which can adaptively adjust the weights of different channel features, thereby highlighting key features. In addition, a global average pooling layer is used to integrate global spatial information and further extract global features. Finally, these features are concatenated to form a fused spatial feature representation. Finally, the temporal information fusion module is entered. This module first performs a one-dimensional convolution operation on the concatenated features to extract time-series features. Next, a batch normalization layer normalizes the features, improving the model's training stability. Then, a max-pooling layer further reduces the feature dimensionality. Afterward, the features are input into a fully connected layer for non-linear combination. To capture long-short-term dependencies, a Long Short-Term Memory (LSTM) network is introduced, which effectively models long-term dependencies in time series data. Finally, a regularization layer and a fault diagnosis layer output the final fault diagnosis results, such as "normal," "spring torsion," or "spring wire crack." This enables timely and accurate diagnosis of vibrating screen spring faults, which is crucial for ensuring the safe and stable operation of the coal preparation production system. In some optional embodiments, corresponding maintenance strategies or alarm information can be generated based on the fault diagnosis results, providing decision-making support for equipment maintenance personnel, thereby achieving predictive maintenance and ensuring production safety.

[0046] Example 3 Please see Figure 10 , Figure 10 This is a structural block diagram of an embodiment of a vibrating screen spring fault diagnosis device according to the present invention. The device includes: The acquisition module 301 is used to acquire the original vibration signal of the vibrating screen spring through a vibration sensor; The signal extraction module 302 is used to extract time-frequency feature signals from the original vibration signal; Feature extraction module 303 is used to extract spatial features and temporal features from the time-frequency feature signal using a pre-trained feature fusion neural network model. The fusion module 304 is used to fuse the spatial features and the temporal features to obtain fused features; The prediction module 305 is used to input the fused features into a pre-trained fault classifier and output the fault diagnosis result.

[0047] In an optional embodiment, the signal extraction module 302 includes: The first extraction submodule is used to perform continuous wavelet transform on the original vibration signal and extract the corresponding first time-frequency feature. The second lifting submodule is used to perform eigenmode decomposition on the original vibration signal to obtain a series of intrinsic mode functions as the second time-frequency features; The third extraction submodule is used to perform a short-time Fourier transform on the original vibration signal to obtain a third time-frequency feature; the time-frequency feature signal includes at least one of the first time-frequency feature, the second time-frequency feature and the third time-frequency feature.

[0048] In an optional embodiment, the feature fusion neural network model includes: a dynamic graph convolutional network and a long short-term memory network; the feature extraction module 303 includes: The mapping submodule is used to map the time-frequency characteristic signal into spatiotemporal graph data; The first feature extraction submodule is used to extract the spatial features from the spatiotemporal graph data using the dynamic graph convolutional network; The second feature extraction submodule is used to extract the time features from the time-frequency feature signal using the long short-term memory network.

[0049] In an optional embodiment, the first feature extraction submodule includes: The capture unit is used to process the spatiotemporal graph data through the dynamic graph convolutional network, perform the EdgeConv operation to dynamically update the graph structure and capture local geometric features; An adjustment unit is used to introduce a channel attention mechanism to adaptively adjust the weights of different channel features in the local geometric features; The integration unit is used to reduce the dimensionality of the weighted local geometric features and integrate global spatial information through a max pooling layer and a global average pooling layer to obtain the spatial features.

[0050] In an optional embodiment, the second feature extraction submodule includes: A standardization processing unit is used to perform one-dimensional convolution operations and batch standardization on the time series. The dependency capture unit is used to input the processed features into the long short-term memory network and capture long-term dependencies through its gating mechanism. A time feature forming unit is used to form the time feature based on the long-term dependency relationship.

[0051] In an optional embodiment, the feature fusion neural network model further includes a feature fusion subnetwork; the fusion module 304 includes: The splicing submodule is used to splice the spatial features and the temporal features along the feature dimension; The nonlinear combination submodule is used to input the spliced ​​features into the fully connected layer of the feature fusion subnetwork for nonlinear combination; The regularization processing submodule is used to perform regularization processing on the features combined by the fully connected layer to generate the fused features.

[0052] Example 4 This invention also provides an electronic device, including a memory and a processor. The memory stores a computer program, which, when executed by the processor, causes the processor to perform the steps of a fault diagnosis method for a vibrating screen spring according to any embodiment.

[0053] Example 5 This invention also provides a computer storage medium storing a computer program thereon, which, when executed by the processor, implements the steps of a fault diagnosis method for a vibrating screen spring according to any embodiment.

[0054] Example 6 This invention also provides a computer program product having a computer program stored thereon, wherein when the computer program is executed by the processor, it implements the steps of a fault diagnosis method for a vibrating screen spring according to any embodiment.

[0055] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0056] In the several embodiments provided in this application, it should be understood that the methods, apparatuses, electronic devices, and storage media disclosed in this invention can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative. For instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the displayed or discussed mutual couplings, direct couplings, or communication connections may be through some interfaces; indirect couplings or communication connections between devices or units may be electrical, mechanical, or other forms.

[0057] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0058] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0059] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a readable storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned readable storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0060] The above-described embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for diagnosing spring faults in a vibrating screen, characterized in that, include: The original vibration signal of the vibrating screen spring is collected by a vibration sensor; Extract time-frequency feature signals from the original vibration signal; Using a pre-trained feature fusion neural network model, spatial features and temporal features are extracted from the time-frequency feature signal, respectively. The spatial features and the temporal features are fused to obtain the fused features; The fused features are input into a pre-trained fault classifier, which outputs fault diagnosis results.

2. The method for diagnosing spring faults in a vibrating screen according to claim 1, characterized in that, Extracting time-frequency feature signals from the original vibration signal includes: Perform continuous wavelet transform on the original vibration signal to extract the corresponding first time-frequency features; The original vibration signal is subjected to eigenmode decomposition to obtain a series of intrinsic mode functions as the second time-frequency features; The original vibration signal is subjected to a short-time Fourier transform to obtain a third time-frequency feature; the time-frequency feature signal includes at least one of the first time-frequency feature, the second time-frequency feature and the third time-frequency feature.

3. The method for diagnosing spring faults in a vibrating screen according to claim 1, characterized in that, The feature fusion neural network model includes a dynamic graph convolutional network and a long short-term memory network; using the pre-trained feature fusion neural network model, spatial features and temporal features are extracted from the time-frequency feature signal, including: The time-frequency characteristic signal is plotted as a spatiotemporal graph data; The spatial features are extracted from the spatiotemporal graph data using the dynamic graph convolutional network. The time features are extracted from the time-frequency feature signal using the long short-term memory network.

4. The method for diagnosing spring faults in a vibrating screen according to claim 3, characterized in that, Using the dynamic graph convolutional network, the spatial features are extracted from the spatiotemporal graph data, including: The spatiotemporal graph data is processed by the dynamic graph convolutional network, and the EdgeConv operation is performed to dynamically update the graph structure and capture local geometric features. A channel attention mechanism is introduced to adaptively adjust the weights of different channel features in the local geometric features; The weighted local geometric features are reduced in dimensionality by using a max pooling layer and a global average pooling layer, and global spatial information is integrated to obtain the spatial features.

5. The method for diagnosing spring faults in a vibrating screen according to claim 3, characterized in that, Using the Long Short-Term Memory network, the time features are extracted from the time-frequency feature signal, including: Perform one-dimensional convolution and batch normalization on the time series; The processed features are input into the Long Short-Term Memory network, and long-term dependencies are captured through its gating mechanism. The time feature is formed based on the long-term dependency relationship.

6. The method for diagnosing spring faults in a vibrating screen according to claim 3, characterized in that, The feature fusion neural network model further includes a feature fusion sub-network; The spatial features and the temporal features are fused to obtain fused features, including: The spatial features and the temporal features are concatenated along the feature dimension; The spliced ​​features are input into the fully connected layer of the feature fusion subnetwork for nonlinear combination; Regularization is performed on the features combined by the fully connected layer to generate the fused features.

7. A device for diagnosing spring faults in a vibrating screen, characterized in that, include: The acquisition module is used to acquire the original vibration signal of the vibrating screen spring through a vibration sensor; The signal extraction module is used to extract time-frequency feature signals from the original vibration signal; The feature extraction module is used to extract spatial features and temporal features from the time-frequency feature signal using a pre-trained feature fusion neural network model. The fusion module is used to fuse the spatial features and the temporal features to obtain fused features; The prediction module is used to input the fused features into a pre-trained fault classifier and output fault diagnosis results.

8. An electronic device, characterized in that, It includes a processor and a memory, the memory storing computer-readable instructions that, when executed by the processor, perform the method as described in any one of claims 1-6.

9. A storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it performs the method as described in any one of claims 1-6.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the method of any one of claims 1-6.

Images