Fault monitoring methods, devices, equipment, storage media and software products

Abstract

This application discloses a fault monitoring method, apparatus, equipment, storage medium, and program product, relating to the field of equipment monitoring technology. The method includes: acquiring audio signals from various monitoring points of a belt conveyor via optical fibers arranged along the belt's movement direction; encoding the audio signal of each monitoring point based on a channel attention mechanism to obtain the encoding features of each monitoring point; for each monitoring point, performing fault identification based on the encoding features to obtain the identification result for that monitoring point; the identification result characterizes whether a fault exists in the vicinity of the monitoring point and the type of fault if it exists. This achieves the purpose of automatically monitoring faults in the belt conveyor, reducing monitoring costs and improving monitoring efficiency while ensuring the safety of personnel.

Description

Technical Field

[0001] This application relates to the field of equipment monitoring technology, and in particular to a fault monitoring method, apparatus, equipment, storage medium, and program product. Background Technology

[0002] Currently, to ensure the smooth operation of industrial production, it is necessary to monitor the malfunctions of industrial equipment to prevent production accidents. However, the primary method for equipment malfunction monitoring is manual, periodic inspections. Inspectors rely on visual inspection, listening, and experience to determine if the equipment is abnormal. The accuracy of this manual inspection method largely depends on the professional knowledge of the staff and the frequency of inspections, and it is also costly. Furthermore, some equipment is sometimes located in dangerous positions, making it impossible to completely guarantee the safety of staff during inspections, thus highlighting the limitations of manual inspections.

[0003] Therefore, it is necessary to provide an automatic fault monitoring solution for industrial equipment. Summary of the Invention

[0004] In view of the above problems, this application provides a fault monitoring method, apparatus, device, storage medium, and program product to achieve automatic fault monitoring of industrial equipment. The specific solution is as follows:

[0005] The first aspect of this application provides a fault monitoring method, the method comprising:

[0006] Audio signals from various monitoring points of the belt conveyor are collected using optical fibers arranged along the belt's direction of movement.

[0007] The audio signal at each monitoring point is encoded using a channel attention mechanism to obtain the encoded features of each monitoring point;

[0008] For each monitoring point, fault identification is performed on the monitoring point based on its coding characteristics to obtain the identification result of the monitoring point. The identification result of the monitoring point indicates whether there is a fault in the belt conveyor near the monitoring point, and the type of fault if there is a fault.

[0009] In one possible implementation, the optical fiber is arranged on a bracket on the belt conveyor parallel to the direction of belt movement, and the optical fiber is located below or to the side of the idler roller that carries the belt.

[0010] In one possible implementation, the acquisition of audio signals from various monitoring points of the belt conveyor via optical fibers arranged along the belt movement direction includes:

[0011] Audio signals from various monitoring points on the belt conveyor are collected using two parallel optical fibers set on both sides of the belt.

[0012] The monitoring points set on both sides of the belt are located at different positions in the direction of belt movement.

[0013] In one possible implementation, encoding the audio signal at each monitoring point based on a channel attention mechanism to obtain the encoded features of each monitoring point includes:

[0014] For each monitoring point, an adaptive filter is applied to the audio signal at that monitoring point to obtain a filtered audio signal; the background noise of the filtered audio signal is less than the background noise of the audio signal at that monitoring point.

[0015] The filtered audio signal is encoded using a channel attention mechanism to obtain the encoded features of the monitoring point.

[0016] In one possible implementation, the audio signal at each monitoring point is encoded using a channel attention mechanism. For each monitoring point, the process of fault identification based on the encoded features of that monitoring point includes:

[0017] The audio signal of each monitoring point is encoded using a channel attention mechanism based on a fault identification model. Fault identification is then performed on the monitoring point based on the encoding features of each monitoring point to obtain the identification result for each monitoring point.

[0018] The fault identification model is obtained by pre-training using a first audio dataset and then fine-tuning the pre-trained fault identification model using a second audio dataset.

[0019] The audio data in the first audio dataset is audio data in a non-belt conveyor scenario. The labels of the audio data in the first audio dataset represent whether there is noise or fault in the audio data, and the noise category when there is noise or the fault category when there is fault.

[0020] The audio data in the second audio dataset is audio data in the context of a belt conveyor. The labels of the audio data in the second audio dataset indicate whether there is a fault in the audio data, and the type of fault when a fault exists.

[0021] In one possible implementation, the process of pre-training the fault identification model using the first audio dataset and fine-tuning the pre-trained fault identification model using the second audio dataset includes:

[0022] The acoustic features of the audio data in the first audio dataset are input into the fault identification model to obtain the first identification result of each audio data in the first audio dataset; the parameters of the fault identification model are updated with the goal of making the first identification result of each audio data in the first audio dataset approach the label of the audio data in the first audio dataset, so as to obtain the pre-trained fault identification model.

[0023] The acoustic features of the audio data in the second audio dataset are input into the pre-trained fault identification model to obtain the second identification result of each audio data in the second audio dataset. With the goal of the second identification result of each audio data in the second audio dataset approaching the label of the audio data in the second audio dataset, the parameters of the target level of the encoding module and the parameters of the non-encoding module of the pre-trained fault identification model are updated to obtain the trained fault identification model.

[0024] The output of the non-target level of the encoding module is connected to the input of the target level; the input of the non-target level is the input of the fault identification model, the non-target level does not include the channel attention layer, and the target level includes the channel attention layer.

[0025] A second aspect of this application provides a fault monitoring device, comprising:

[0026] The acquisition module is used to acquire audio signals from various monitoring points of the belt conveyor through optical fibers arranged along the belt movement direction.

[0027] The encoding module is used to encode the audio signal of each monitoring point based on the channel attention mechanism to obtain the encoding features of each monitoring point;

[0028] The identification module is used to identify faults at each monitoring point based on the coding characteristics of that monitoring point, and obtain the identification result of that monitoring point. The identification result of the monitoring point indicates whether there is a fault in the belt conveyor near the monitoring point, and the type of fault if there is a fault.

[0029] A third aspect of this application provides a computer program product including computer-readable instructions that, when executed on an electronic device, cause the electronic device to implement the fault monitoring method of the first aspect or any implementation thereof.

[0030] A fourth aspect of this application provides an electronic device, including at least one processor and a memory connected to the processor, wherein:

[0031] The memory is used to store computer programs;

[0032] The processor is used to execute the computer program so that the electronic device can implement the fault monitoring method of the first aspect or any implementation thereof.

[0033] The fifth aspect of this application provides a computer storage medium carrying one or more computer programs, which, when executed by an electronic device, enable the electronic device to implement the fault monitoring method described in the first aspect or any implementation thereof.

[0034] By employing the above technical solutions, the fault monitoring method, apparatus, equipment, storage medium, and program products provided in this application acquire audio signals from various monitoring points of the belt conveyor through optical fibers arranged along the belt movement direction. The audio signals from each monitoring point are encoded based on a channel attention mechanism to obtain the encoding features of each monitoring point. For each monitoring point, fault identification is performed based on its encoding features to obtain the identification result. This identification result characterizes whether a fault exists in the vicinity of the monitoring point and, if so, the type of fault. This achieves the purpose of automatically monitoring faults in the belt conveyor, ensuring worker safety while reducing monitoring costs and improving monitoring efficiency. Attached Figure Description

[0035] The above and other features, advantages, and aspects of the embodiments of this disclosure will become more apparent from the accompanying drawings and the following detailed description. Throughout the drawings, the same or similar reference numerals denote the same or similar elements. It should be understood that the drawings are schematic, and the originals and elements are not necessarily drawn to scale.

[0036] Figure 1 A flowchart illustrating an implementation of the fault monitoring method provided in this application;

[0037] Figure 2a A distribution example diagram of monitoring points set on one side of the belt, provided in this application;

[0038] Figure 2b A distribution example diagram of monitoring points set on both sides of the belt provided in this application;

[0039] Figure 3 A flowchart is provided for implementing the encoding of the audio signal of each monitoring point based on the channel attention mechanism to obtain the encoding features of each monitoring point, as provided in this application.

[0040] Figure 4 A schematic diagram of the fault monitoring device provided in this application;

[0041] Figure 5 A schematic diagram of the structure of the electronic device provided in this application. Detailed Implementation

[0042] The embodiments of this application are described below with reference to the accompanying drawings. The terminology used in the implementation section of this application is for explaining specific embodiments only and is not intended to limit the scope of this application.

[0043] The embodiments of this application will now be described with reference to the accompanying drawings. Those skilled in the art will recognize that, with technological advancements and the emergence of new scenarios, the technical solutions provided in the embodiments of this application are equally applicable to similar technical problems.

[0044] The terms "first," "second," etc., used in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such terms are interchangeable where appropriate; this is merely a way of distinguishing objects with the same attributes in the embodiments of this application. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion, so that a process, method, system, product, or apparatus that comprises a series of elements is not necessarily limited to those elements, but may include other elements not explicitly listed or inherent to those processes, methods, products, or apparatuses.

[0045] Taking belt conveyors for transporting coal as an example, their working environment is relatively harsh. Key equipment is constantly subjected to collisions, wear, and other unstable conditions, which can easily lead to equipment failures and damage, thus affecting the efficiency of the entire transportation chain and causing numerous inconveniences and direct economic losses to production. Idler rollers, as important components of belt conveyors for transporting coal, play a role in supporting the conveyor belt and the transported goods. If idler roller failures are not monitored, they can sometimes even cause serious production accidents.

[0046] Currently, monitoring such faults mainly relies on regular manual inspections. Inspectors visually assess whether idlers are stuck or broken, and combine this with the sounds emitted by the conveyor belt and past experience to determine if there are any abnormalities such as slippage or wear. However, the accuracy of this method largely depends on the professional knowledge of the staff and the frequency of inspections. Furthermore, some equipment is sometimes installed in relatively dangerous locations, making it impossible to fully guarantee the safety of staff during inspections, thus presenting certain limitations. Therefore, developing a highly efficient and accurate fault detection system is particularly important.

[0047] This application proposes a solution to achieve automatic fault monitoring of belt conveyors.

[0048] like Figure 1 The diagram shown is a flowchart of one implementation of the fault monitoring method provided in this application, which may include:

[0049] Step S101: Acquire audio signals from each monitoring point of the belt conveyor using optical fibers arranged along the belt movement direction of the belt conveyor.

[0050] In this application, the optical fiber in the distributed fiber acoustic sensing (DAS) system extends along the direction of movement of the conveyor belt. At each monitoring point, the optical fiber is coiled at least once before extending backward. The principle of sound acquisition in the distributed fiber acoustic sensing system can be found in existing solutions and will not be detailed here.

[0051] Optical fibers can be placed on one side of the belt or on both sides. If optical fibers are placed on both sides of the belt, two distributed fiber optic acoustic sensing systems are required, with different distributed fiber optic acoustic sensing systems placed on each side of the belt.

[0052] like Figure 2a The diagram shown illustrates a distribution example of monitoring points set on one side of a belt according to an embodiment of this application. In this example diagram, arrows indicate the direction of belt movement, dashed lines represent optical fibers deployed on supports below idlers (not shown), and each black dot represents a monitoring point. The idlers are located below and in contact with the belt.

[0053] Multiple monitoring points can be evenly distributed or unevenly distributed. For example, more monitoring points can be set up in areas prone to failure, and fewer monitoring points can be set up in areas less prone to failure.

[0054] Compared with setting up microphones at monitoring points, using fiber optic audio reception has advantages such as quick deployment (simply run the cable, while microphones need to be deployed individually), low cost, good sound reception, and high accuracy.

[0055] Step S102: Encode the audio signal of each monitoring point based on the channel attention mechanism to obtain the coding features of each monitoring point.

[0056] For each monitoring point, the acoustic features of the audio signal at that monitoring point can be extracted, and the acoustic features can be encoded based on the channel attention mechanism to obtain the encoded features of that monitoring point.

[0057] Among them, acoustic features may include, but are not limited to, any of the following: Mel frequency cepstral coefficients (MFCC), Fbank features, etc.

[0058] Step S103: For each monitoring point, fault identification is performed on the monitoring point based on the coding characteristics of the monitoring point to obtain the identification result of the monitoring point. The identification result of the monitoring point indicates whether there is a fault in the belt conveyor near the monitoring point, and the type of fault when there is a fault.

[0059] The audio signal of each monitoring point can be encoded using a deep neural network based on the channel attention mechanism. Based on the encoding features of each monitoring point, the fault of the monitoring point can be identified, and the identification result of the monitoring point can be obtained.

[0060] The fault monitoring method provided in this application collects audio signals from various monitoring points of the conveyor belt using optical fibers arranged along the belt movement direction. The audio signal of each monitoring point is encoded based on a channel attention mechanism to obtain the encoding features of each monitoring point. For each monitoring point, fault identification is performed based on the encoding features to obtain the identification result. This identification result indicates whether a fault exists in the vicinity of the monitoring point and, if so, the type of fault. This achieves the purpose of automatically monitoring faults in the conveyor belt, reducing monitoring costs and improving monitoring efficiency while ensuring the safety of personnel.

[0061] In an optional embodiment, the belt of the conveyor is supported by a plurality of idlers distributed along the direction of belt movement, and these idlers are fixed on a bracket parallel to the direction of belt movement. Accordingly, each optical fiber is arranged on the bracket parallel to the direction of belt movement on the conveyor, with the optical fiber located below or to the side of the idler supporting the belt.

[0062] In other words, the belt conveyor has brackets fixed below or to the side of the idler rollers along the direction of belt movement for deploying optical fibers.

[0063] In an optional embodiment, one way to collect audio signals from various monitoring points of the belt conveyor using optical fibers arranged along the belt's direction of movement is as follows:

[0064] Audio signals from various monitoring points on the belt conveyor are collected using two parallel optical fibers set on both sides of the belt.

[0065] The monitoring points set on both sides of the belt are located at different positions along the direction of belt movement.

[0066] In other words, two optical fibers can be set on both sides of the belt to collect audio signals from each monitoring point. At each monitoring point, only that monitoring point is monitored in the direction perpendicular to the optical fiber.

[0067] like Figure 2bThe diagram shown is an example of the distribution of monitoring points set on both sides of a belt according to an embodiment of this application. In this example diagram, the arrows indicate the direction of belt movement, each dashed line represents an optical fiber deployed on a bracket below the idler roller (not shown in the diagram), and each black dot represents a monitoring point. The idler roller is located below the belt and in contact with the belt.

[0068] In an optional embodiment, a flowchart illustrating one approach to encoding the audio signal at each monitoring point using a channel attention mechanism to obtain the encoded features for each monitoring point is shown below. Figure 3 As shown, it may include:

[0069] Step S301: For each monitoring point, perform adaptive filtering on the audio signal of that monitoring point (referred to as the original audio signal for ease of description and distinction) to obtain the filtered audio signal; the background noise of the filtered audio signal is less than the background noise of the original audio signal.

[0070] Optionally, the audio signal at each monitoring point can be subjected to least mean square (LMS) adaptive filtering. Alternatively, the audio signal at each monitoring point can be subjected to recursive least squares (RLS) adaptive filtering.

[0071] Adaptive filtering can remove background noise around the belt conveyor and accurately recover the true signal emitted by the belt conveyor.

[0072] Step S302: Encode the filtered audio signal based on the channel attention mechanism to obtain the encoded features of the monitoring point.

[0073] For each monitoring point, the audio frames of the filtered audio signal at that monitoring point can be encoded based on a channel attention mechanism.

[0074] Optionally, acoustic features of each audio frame can be extracted, and the acoustic features of each audio frame can be encoded using a channel attention mechanism to obtain the encoded features of each audio frame. Specifically, for the acoustic features of each audio frame, the acoustic features of the audio frame can be encoded at multiple levels. These multiple levels can include multiple convolutional layers and multiple channel attention layers, which are interleaved with the multiple convolutional layers.

[0075] Optionally, the multiple channel attention layers can be the same or different.

[0076] As an example, each channel attention layer can be any of the following channel attention layers: SE-Nets (Squeeze-and-Excitation Networks) attention layer, ECA (Efficient Channel Attention) attention layer, etc.

[0077] In an optional embodiment, the above-described encoding of the audio signal at each monitoring point based on a channel attention mechanism, and the fault identification of the monitoring point based on the encoding characteristics of that monitoring point, can be implemented as follows:

[0078] The audio signal of each monitoring point is encoded using a channel attention mechanism based on a fault identification model. Fault identification is then performed on the monitoring point based on the encoding features of each monitoring point, resulting in the identification result for each monitoring point.

[0079] Optionally, the fault identification model can adopt any of the following network structures: deep convolutional neural network, time delay neural network (TDNN), Efficient network, etc.

[0080] The fault identification model includes an encoding module.

[0081] As an example, in a deep convolutional neural network, the encoding module may include multiple convolutional layers and multiple channel attention layers; the channel attention layers and convolutional layers are interspersed. For example, the layers of the encoding module may be arranged as follows: the first 5 layers are convolutional layers, the 6th layer is a channel attention layer, the 7th and 8th layers are convolutional layers, the 9th layer is a channel attention layer, the 10th and 11th layers are convolutional layers, and the 12th layer is a channel attention layer. The structures of the various convolutional layers may be the same or different.

[0082] As an example, in a time-delay neural network, the encoding module may include multiple feedforward network layers and multiple channel attention layers, which are interspersed.

[0083] As an example, in an Efficient network, the encoding module may include multiple MBConv modules and multiple channel attention modules (i.e., multiple channel attention layers). Each MBConv module consists of at least one ordinary convolutional layer and at least one depthwise separable convolutional layer. Each channel attention layer may be deployed in one MBConv module or between two MBConv modules.

[0084] The fault identification model is obtained by pre-training using the first audio dataset and then fine-tuning the pre-trained fault identification model using the second audio dataset.

[0085] The audio data in the first audio dataset is audio data in non-belt conveyor scenarios. The labels of the audio data in the first audio dataset are whether there is noise or fault, and the noise category when there is noise or the fault category when there is fault.

[0086] Optionally, the first audio dataset includes audio data from general scenarios, i.e., audio data containing environmental noise in typical scenarios. Environmental noise may include, but is not limited to, bird calls, human voices, rain sounds, wind sounds, and thunder. The corresponding labels for the audio data from general scenarios characterize whether noise exists in the audio data and, if so, the type of noise. The labels for the audio data can be manually assigned or automatically assigned by a labeling system.

[0087] The first audio dataset can also include audio data from other industrial scenarios, such as equipment sounds from wind power generation sites, hydropower generation sites, and manufacturing processes. The corresponding labels for the industrial scenario audio data indicate whether the equipment is malfunctioning and, if so, the type of malfunction. The labels for the audio data can be manually added or automatically added by a labeling system.

[0088] As an example, for each audio data in the first audio dataset, each audio frame of the audio data can be labeled separately. The label of each audio frame indicates whether the audio frame has noise or fault, and the type of noise when noise exists or the type of fault when fault exists.

[0089] The audio data in the second audio dataset is audio data in the context of belt conveyor scenarios. The labels of the audio data indicate whether the belt conveyor has a fault, and if so, the type of fault.

[0090] The labels on the audio data in the second audio dataset can be manually labeled or automatically labeled by a labeling system.

[0091] As an example, for each audio data in the second audio dataset, each audio frame of the audio data can be labeled separately. The label of each audio frame indicates whether the audio frame has a fault, and if a fault exists, the type of fault.

[0092] Because labeled data is not readily available in belt conveyor scenarios, and the second audio dataset contains a relatively small amount of data, the fault identification model trained solely on the second audio dataset exhibits poor generalization ability and is prone to overfitting. To enhance the generalization ability of the fault identification model and mitigate overfitting, the fault identification model can be pre-trained using the first audio dataset. This allows the model to learn and identify features related to belt conveyor faults, such as visual features, on the first audio dataset.

[0093] In an optional embodiment, the process of pre-training the fault identification model using a first audio dataset and fine-tuning the pre-trained fault identification model using a second audio dataset may include:

[0094] The acoustic features of the audio data in the first audio dataset are input into the fault identification model to obtain the identification results of each audio data in the first audio dataset (for ease of description and distinction, they are referred to as the first identification results). The parameters of the fault identification model are updated with the goal of making the first identification results of each audio data in the first audio dataset as close as possible to the labels of the audio data in the first audio dataset, so as to obtain the pre-trained fault identification model.

[0095] The first identification result of each audio data represents whether there is noise or fault in the audio data, and the type of noise when noise is present or the type of fault when fault is present.

[0096] For each audio data in the first audio data, the audio data can be divided into frames to obtain multiple audio frames. The acoustic features of each audio frame are extracted and input into the fault identification model. The fault identification model encodes the acoustic features of each audio frame to obtain the encoded features of each audio frame. Based on the encoded features of each audio frame, the first identification result of each audio frame is determined. The first identification result of each audio frame indicates whether there is noise or fault in the audio frame, and the type of noise or fault when noise or fault exists.

[0097] Calculate the loss between the first recognition result and the label for each audio frame; with the goal of minimizing the loss of each audio frame (i.e., the loss between the first recognition result and the label obtained by processing the audio data using the fault recognition model after parameter update is smaller than the first recognition result obtained by processing the audio data using the fault recognition model before parameter update), update the parameters of the fault recognition model to obtain the pre-trained fault recognition model.

[0098] The acoustic features of the audio data in the second audio dataset are input into the pre-trained fault recognition model to obtain the recognition results of each audio data in the second audio dataset (referred to as the second recognition results for ease of description and distinction). With the goal of the second recognition results of each audio data in the second audio dataset approximating the labels of the audio data in the second audio dataset, the parameters of the target layer of the encoding module and the parameters of the non-encoding module of the pre-trained fault recognition model are updated to obtain the trained fault recognition model. The output of the non-target layer of the encoding module is connected to the input of the target layer. The input of the non-target layer is the input of the fault recognition model. The non-target layer does not include a channel attention layer, while the target layer does.

[0099] The second identification result for each audio data point indicates whether a fault exists in the audio data, and if so, the type of fault.

[0100] For each audio data in the second audio data, the audio data can be divided into frames to obtain multiple audio frames. The acoustic features of each audio frame are extracted and input into a pre-trained fault recognition model. The pre-trained fault recognition model encodes the acoustic features of each audio frame to obtain the encoded features of each audio frame. Based on the encoded features of each audio frame, the second recognition result of each audio frame is determined. The second recognition result of each audio frame represents whether there is a fault in the audio frame and the type of fault if there is a fault.

[0101] Calculate the loss between the second recognition result and the label for each audio frame; update some parameters of the pre-trained fault recognition model with the goal of minimizing the loss for each audio frame. That is, when fine-tuning the pre-trained fault recognition model, freeze some parameters and only update the parameters that do not need to be frozen.

[0102] Taking the aforementioned encoding module with the following layers in sequence: the first 5 layers are convolutional, the 6th layer is a channel attention layer, the 7th and 8th layers are convolutional layers, the 9th layer is a channel attention layer, the 10th and 11th layers are convolutional layers, and the 12th layer is a channel attention layer, as an example, this application updates the parameters of both the encoding and non-encoding modules of the fault identification model during pre-training. During fine-tuning the pre-trained fault identification model, the parameters of the first 5 layers of the encoding module are frozen, and only the parameters after the 5th layer (including the parameters of layers 6-12 of the encoding module and the parameters of the non-encoding modules of the fault identification model) are updated. In this example, the first 5 layers of the encoding module constitute the non-target layers of the encoding module, and layers 6-12 constitute the target layers of the encoding module.

[0103] As an example, a fault identification model may include an encoding module and a prediction module. The encoding module encodes each audio frame of the audio data using a channel attention mechanism to obtain the encoded features of each audio frame. The prediction module determines the identification result of each audio frame based on its encoded features. During pre-training of the fault identification model, the parameters of both the encoding and prediction modules are updated. During fine-tuning of the pre-trained fault identification model, the parameters of the non-target level of the encoding module are frozen, and only the parameters of the target level of the encoding module and the parameters of the prediction module are updated.

[0104] Corresponding to the method embodiments, this application also provides a fault monitoring device. A schematic diagram of the structure of the fault monitoring device provided in the embodiments of this application is shown below. Figure 4 As shown, it may include:

[0105] Acquisition module 401, encoding module 402 and recognition module 403;

[0106] The acquisition module 401 is used to acquire audio signals from various monitoring points of the belt conveyor through an optical fiber arranged along the belt movement direction of the belt conveyor.

[0107] The encoding module 402 is used to encode the audio signal of each monitoring point based on the channel attention mechanism to obtain the encoding features of each monitoring point;

[0108] The identification module 403 is used to identify faults at each monitoring point based on the coding characteristics of the monitoring point, and obtain the identification result of the monitoring point. The identification result of the monitoring point indicates whether there is a fault in the belt conveyor near the monitoring point, and the type of fault when there is a fault.

[0109] The fault monitoring device provided in this application collects audio signals from various monitoring points of the conveyor belt through optical fibers arranged along the belt movement direction. The audio signal of each monitoring point is encoded based on a channel attention mechanism to obtain the encoding features of each monitoring point. For each monitoring point, fault identification is performed based on the encoding features to obtain the identification result. This identification result indicates whether a fault exists in the vicinity of the monitoring point and, if so, the type of fault. This achieves the purpose of automatically monitoring faults in the conveyor belt, reducing monitoring costs and improving monitoring efficiency while ensuring the safety of personnel.

[0110] In an optional embodiment, the optical fiber is arranged on a bracket on the belt conveyor parallel to the direction of belt movement, and the optical fiber is located below or to the side of the idler roller that carries the belt.

[0111] In an optional embodiment, when the acquisition module 401 acquires audio signals from various monitoring points of the belt conveyor via optical fibers arranged along the belt movement direction, it is used for:

[0112] Audio signals from various monitoring points on the belt conveyor are collected using two parallel optical fibers set on both sides of the belt.

[0113] The monitoring points set on both sides of the belt are located at different positions in the direction of belt movement.

[0114] In an optional embodiment, when the encoding module 402 encodes the audio signal of each monitoring point based on a channel attention mechanism to obtain the encoded features of each monitoring point, it is used for:

[0115] For each monitoring point, an adaptive filter is applied to the audio signal at that monitoring point to obtain a filtered audio signal; the background noise of the filtered audio signal is less than the background noise of the audio signal at that monitoring point.

[0116] The filtered audio signal is encoded using a channel attention mechanism to obtain the encoded features of the monitoring point.

[0117] In an optional embodiment, the encoding module 402 encodes the audio signal of each monitoring point based on a channel attention mechanism, and the identification module 403, corresponding to each monitoring point, performs fault identification on that monitoring point based on its encoding features, for the following purposes:

[0118] The audio signal of each monitoring point is encoded using a channel attention mechanism based on a fault identification model. Fault identification is then performed on the monitoring point based on the encoding features of each monitoring point to obtain the identification result for each monitoring point.

[0119] The fault identification model is obtained by pre-training using a first audio dataset and then fine-tuning the pre-trained fault identification model using a second audio dataset.

[0120] The audio data in the first audio dataset is audio data in a non-belt conveyor scenario. The labels of the audio data in the first audio dataset represent whether there is noise or fault in the audio data, and the noise category when there is noise or the fault category when there is fault.

[0121] The audio data in the second audio dataset is audio data in the context of a belt conveyor. The labels of the audio data in the second audio dataset indicate whether there is a fault in the audio data, and the type of fault when a fault exists.

[0122] In an optional embodiment, the fault monitoring device further includes a training module for pre-training the fault identification model using the first audio dataset and fine-tuning the pre-trained fault identification model using the second audio dataset, specifically including:

[0123] The acoustic features of the audio data in the first audio dataset are input into the fault identification model to obtain the first identification result of each audio data in the first audio dataset; the parameters of the fault identification model are updated with the goal of making the first identification result of each audio data in the first audio dataset approach the label of the audio data in the first audio dataset, so as to obtain the pre-trained fault identification model.

[0124] The acoustic features of the audio data in the second audio dataset are input into the pre-trained fault identification model to obtain the second identification result of each audio data in the second audio dataset. With the goal of the second identification result of each audio data in the second audio dataset approaching the label of the audio data in the second audio dataset, the parameters of the target level of the encoding module and the parameters of the non-encoding module of the pre-trained fault identification model are updated to obtain the trained fault identification model.

[0125] The output of the non-target level of the encoding module is connected to the input of the target level; the input of the non-target level is the input of the fault identification model, the non-target level does not include the channel attention layer, and the target level includes the channel attention layer.

[0126] This application also provides an electronic device in its embodiments. (See reference...) Figure 5 As shown, it illustrates a structural schematic diagram of an electronic device suitable for implementing the fault monitoring method in the embodiments of this application. The electronic device in the embodiments of this application can be a terminal device (e.g., a mobile phone, tablet computer, laptop computer, desktop computer, etc.) or a server (which can be a single server, a server cluster, or a cloud server, etc.). Figure 5 The electronic device shown is merely an example and should not impose any limitation on the functionality and scope of use of the embodiments of this application.

[0127] like Figure 5As shown, the electronic device may include a processing unit (e.g., a central processing unit, a graphics processing unit, etc.) 501, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 502 or a program loaded from a storage device 508 into a random access memory (RAM) 503. When the electronic device is powered on, the RAM 503 also stores various programs and data required for the operation of the electronic device. The processing unit 501, ROM 502, and RAM 503 are interconnected via a bus 504. An input / output (I / O) interface 505 is also connected to the bus 504.

[0128] Typically, the following devices can be connected to I / O interface 505: input devices 506 including, for example, touchscreens, touchpads, keyboards, mice, cameras, microphones, accelerometers, gyroscopes, etc.; output devices 507 including, for example, liquid crystal displays (LCDs), speakers, vibrators, etc.; storage devices 508 including, for example, memory cards, hard drives, etc.; and communication devices 509. Communication device 509 allows electronic devices to communicate wirelessly or wiredly with other devices to exchange data. Although Figure 5 Electronic devices with various devices are shown, but it should be understood that it is not required to implement or have all of the devices shown. More or fewer devices may be implemented or have alternatively.

[0129] This application also provides a computer program product including computer-readable instructions, which, when executed on an electronic device, cause the electronic device to implement any of the fault monitoring methods provided in this application.

[0130] This application also provides a computer-readable storage medium that carries one or more computer programs. When the one or more computer programs are executed by an electronic device, the electronic device can implement any of the fault monitoring methods provided in this application.

[0131] It should be noted that the device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and 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 modules can be selected to achieve the purpose of this embodiment according to actual needs. In addition, in the device embodiment drawings provided in this application, the connection relationship between modules indicates that they have a communication connection, which can be implemented as one or more communication buses or signal lines.

[0132] Through the above description of the embodiments, those skilled in the art can clearly understand that this application can be implemented by means of software plus necessary general-purpose hardware, or it can be implemented by special-purpose hardware including application-specific integrated circuits, special-purpose CPUs, special-purpose memory, special-purpose components, etc. Generally, any function performed by a computer program can be easily implemented by corresponding hardware, and the specific hardware structure used to implement the same function can also be diverse, such as analog circuits, digital circuits, or special-purpose circuits. However, for this application, software program implementation is more often a better implementation method. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a readable storage medium, such as a computer floppy disk, USB flash drive, mobile hard disk, ROM, RAM, magnetic disk, or optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, training equipment, or network device, etc.) to execute the methods described in the various embodiments of this application.

[0133] In the above embodiments, the functionality can be implemented entirely or partially through software, hardware, firmware, or any combination thereof. When implemented in software, it can be implemented entirely or partially as a computer program product. Those skilled in the art can use different methods to implement the described functions for each specific solution, but such implementation should not be considered beyond the scope of this application.

[0134] The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions may be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions may be transmitted from one website, computer, training device, or data center to another website, computer, training device, or data center via wired (e.g., coaxial cable, fiber optic, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium may be any available medium that a computer can store or a data storage device such as a training device or data center that integrates one or more available media. The available media may be magnetic media (e.g., floppy disks, hard disks, magnetic tapes), optical media (e.g., DVDs), or semiconductor media (e.g., solid-state drives (SSDs)).

[0135] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. The same or similar parts between the various embodiments can be referred to each other.

[0136] The above description of the disclosed embodiments enables those skilled in the art to make or use this application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A fault monitoring method, characterized in that, The method includes: Audio signals from various monitoring points of the belt conveyor are collected using optical fibers arranged along the belt's direction of movement. The audio signal of each monitoring point is encoded using a channel attention mechanism through a fault identification model to obtain the encoding features of each monitoring point. For each monitoring point, fault identification is performed based on the encoding features of that monitoring point to obtain the identification result of that monitoring point. The identification result of that monitoring point indicates whether the belt conveyor has a fault near the monitoring point, and the type of fault if a fault exists. The fault identification model is obtained by pre-training using a first audio dataset and then fine-tuning the pre-trained model using a second audio dataset. The fault identification model includes an encoding module and a prediction module. During pre-training, the parameters of both the encoding and prediction modules are updated. During fine-tuning, the parameters of the non-target layers of the encoding module are frozen, while the parameters of the target layers and the parameters of the prediction module are updated. The non-target layers do not include channel attention layers, while the target layers include channel attention layers and convolutional layers. The output of the non-target layers of the encoding module is connected to the input of the target layers. The input of the non-target layers serves as the input to the fault identification model. The audio data in the first audio dataset is audio data in non-belt conveyor scenarios. The labels of the audio data in the first audio dataset represent whether there is environmental noise or fault in the audio data, and the noise category when there is environmental noise or the fault category when there is fault. The first audio dataset includes audio data including ambient sound in general scenarios, as well as audio data from other industrial scenarios other than belt conveyor scenarios. The audio data in the second audio dataset is audio data in the context of a belt conveyor. The labels of the audio data in the second audio dataset indicate whether there is a fault in the audio data, and the type of fault when a fault exists.

2. The method according to claim 1, characterized in that, The optical fiber is arranged on a bracket on the belt conveyor parallel to the direction of belt movement, and the optical fiber is located below or to the side of the idler roller that carries the belt.

3. The method according to claim 1, characterized in that, The method of acquiring audio signals from various monitoring points of the belt conveyor via optical fibers arranged along the belt's direction of movement includes: Audio signals from various monitoring points on the belt conveyor are collected using two parallel optical fibers set on both sides of the belt. The monitoring points set on both sides of the belt are located at different positions in the direction of belt movement.

4. The method according to claim 1, characterized in that, The process of encoding the audio signal at each monitoring point based on a channel attention mechanism to obtain the encoded features of each monitoring point includes: For each monitoring point, an adaptive filter is applied to the audio signal at that monitoring point to obtain a filtered audio signal; the background noise of the filtered audio signal is less than the background noise of the audio signal at that monitoring point. The filtered audio signal is encoded using a channel attention mechanism to obtain the encoded features of the monitoring point.

5. The method according to claim 1, characterized in that, The process of pre-training the fault identification model using the first audio dataset and fine-tuning the pre-trained fault identification model using the second audio dataset includes: The acoustic features of the audio data in the first audio dataset are input into the fault identification model to obtain the first identification result of each audio data in the first audio dataset; the parameters of the fault identification model are updated with the goal of making the first identification result of each audio data in the first audio dataset approach the label of the audio data in the first audio dataset, so as to obtain the pre-trained fault identification model. The acoustic features of the audio data in the second audio dataset are input into the pre-trained fault identification model to obtain the second identification result of each audio data in the second audio dataset. With the goal of the second identification result of each audio data in the second audio dataset approaching the label of the audio data in the second audio dataset, the parameters of the target level of the encoding module and the parameters of the non-encoding module of the pre-trained fault identification model are updated to obtain the trained fault identification model.

6. A fault monitoring device, characterized in that, include: The acquisition module is used to acquire audio signals from various monitoring points of the belt conveyor through optical fibers arranged along the belt movement direction of the belt conveyor. The encoding module is used to encode the audio signal of each monitoring point based on the channel attention mechanism through the fault identification model to obtain the encoding features of each monitoring point; The identification module is used to identify faults at each monitoring point based on the coding features of the monitoring point through the fault identification model, and obtain the identification result of the monitoring point. The identification result of the monitoring point indicates whether there is a fault in the belt conveyor near the monitoring point, and the type of fault when there is a fault. The fault identification model is obtained by pre-training using a first audio dataset and then fine-tuning the pre-trained model using a second audio dataset. The fault identification model includes an encoding module and a prediction module. During pre-training, the parameters of both the encoding and prediction modules are updated. During fine-tuning, the parameters of the non-target layers of the encoding module are frozen, while the parameters of the target layers and the parameters of the prediction module are updated. The non-target layers do not include channel attention layers, while the target layers include channel attention layers and convolutional layers. The output of the non-target layers of the encoding module is connected to the input of the target layers. The input of the non-target layers serves as the input to the fault identification model. The audio data in the first audio dataset is audio data in non-belt conveyor scenarios. The labels of the audio data in the first audio dataset represent whether there is environmental noise or fault in the audio data, and the noise category when there is environmental noise or the fault category when there is fault. The first audio dataset includes audio data including ambient sound in general scenarios, as well as audio data from other industrial scenarios other than belt conveyor scenarios. The audio data in the second audio dataset is audio data in the context of a belt conveyor. The labels of the audio data in the second audio dataset indicate whether there is a fault in the audio data, and the type of fault when a fault exists.

7. A computer program product, characterized in that, It includes computer-readable instructions that, when executed on an electronic device, cause the electronic device to implement the fault monitoring method as described in any one of claims 1 to 5.

8. An electronic device, characterized in that, The electronic device includes at least one processor and a memory connected to the processor, wherein: The memory is used to store computer programs; The processor is used to execute the computer program to enable the electronic device to implement the fault monitoring method as described in any one of claims 1 to 5.

9. A computer storage medium, characterized in that, The storage medium carries one or more computer programs that, when executed by an electronic device, enable the electronic device to implement the fault monitoring method as described in any one of claims 1 to 5.

Images