Conveyor carrier roller online monitoring method and device and electronic equipment

Abstract

The invention discloses a conveyor carrier roller online monitoring method and device and electronic equipment. The method comprises the following steps of: firstly, acquiring a sound wave vibration signal generated in real time when the carrier roller of the conveyor moves; wherein the sound wave vibration signal is obtained based on reverse Rayleigh scattering light in the optical fiber; performing feature extraction from the sound wave vibration signal to obtain an energy feature value of each frame; performing normalization processing on the energy characteristic value to obtain a target characteristic value; judging whether the target characteristic value exceeds a preset threshold value or not: if the target characteristic value exceeds the preset threshold value, inputting the target characteristic value into a pre-constructed target neural network model, so that the target neural network model outputs a fault type corresponding to the target characteristic value; therefore, online monitoring of the conveyor carrier roller based on optical fiber sensing multi-feature fusion is completed. According to the method, the fault type can be automatically and accurately identified and output, so that the overall performance and safety of the conveyor carrier roller are enhanced.

Description

Technical Field

[0001] The present invention belongs to the technical field of monitoring, and particularly relates to an online monitoring method, device and electronic equipment for conveyor rollers. Background Art

[0002] Conveyor rollers are crucial components in a conveyor system, and their operating status directly affects the overall performance and safety. However, due to the large number, wide distribution and complex working conditions of the rollers, the existing monitoring methods mainly rely on manual inspections and vibration sensors, and there are some obvious deficiencies. First, in complex environments such as coal mines, it is very inconvenient to install vibration sensors, and the fault location is not accurate enough, and the maintenance frequency is high, which makes it difficult to meet the requirements of real-time monitoring in long-distance and high-noise environments; second, manual detection not only has high costs, but also has low monitoring accuracy, slow alarm response, lack of an effective early warning mechanism, and is even unable to accurately identify the type of fault. These problems limit the effectiveness of the existing monitoring means, so there is an urgent need for a more efficient and reliable monitoring method to improve the overall performance and safety of conveyor rollers. Summary of the Invention

[0003] The technical problem to be solved by the present invention is to propose an online monitoring method, device and electronic equipment for conveyor rollers based on multi-feature fusion of fiber optic sensing in view of the above deficiencies of the prior art. This method can automatically and accurately identify and output the type of fault, thereby enhancing the overall performance and safety of conveyor rollers.

[0004] In a first aspect, the present invention provides an online monitoring method for conveyor rollers based on multi-feature fusion of fiber optic sensing, and the method includes the following steps:

[0005] Step S1: Obtain the acoustic vibration signal generated in real time when the conveyor roller is running;

[0006] Wherein, the acoustic vibration signal is obtained based on the backward Rayleigh scattered light in the optical fiber;

[0007] Step S2: Extract the target feature value based on the acoustic vibration signal;

[0008] Step S3: Determine whether the target feature value exceeds a preset threshold:

[0009] If the target feature value exceeds the preset threshold, go to step S4; if the target feature value does not exceed the preset threshold, end the process;

[0010] Step S4: Input the target feature value into a pre-constructed target neural network model so that the target neural network model outputs the type of fault corresponding to the target feature value, thereby completing the online monitoring of the conveyor roller based on multi-feature fusion of fiber optic sensing.

[0011] Further, step S1 specifically includes the following steps:

[0012] Transmit the pulsed laser signal through the vibration measurement optical fiber to the conveyor idler;

[0013] Receive the backward Rayleigh scattered light generated in the optical fiber;

[0014] Among them, the backward Rayleigh scattered light is obtained according to the acoustic wave vibration generated in real time when the conveyor idler runs;

[0015] Perform optoelectronic signal conversion on the received backward Rayleigh scattered light signal to obtain the corresponding electrical signal;

[0016] Convert the electrical signal into a vibration signal, so as to obtain the acoustic wave vibration signal generated in real time when the conveyor idler runs.

[0017] Further, step S2 specifically includes the following steps:

[0018] Step S21: Extract features from the vibration signal to obtain the energy eigenvalue of each frame;

[0019] Step S22: Normalize the energy eigenvalue to obtain the target eigenvalue.

[0020] Further, step S21 specifically includes the following steps:

[0021] Preprocess the vibration signal to obtain the preprocessed vibration signal;

[0022] Among them, the preprocessing includes increasing the high-frequency components of the vibration signal;

[0023] Frame the preprocessed vibration signal according to a time window to obtain each frame of signal including a continuous fixed duration;

[0024] Perform windowing processing on each frame of signal to obtain each frame of windowed signal;

[0025] Convert the time-domain signal in each frame of windowed signal into a frequency-domain signal through fast Fourier transform to obtain the spectrum information of the signal;

[0026] Use a Mel filter to extract the band energy value from the spectrum information to obtain the output value of the Mel filter;

[0027] Sum the output values of all Mel filters in each frame to obtain the energy eigenvalue of each frame.

[0028] Further, step S22 specifically includes the following steps:

[0029] Traverse all the energy eigenvalues to obtain the minimum and maximum values among the energy eigenvalues;

[0030] Calculate the target eigenvalue according to the minimum and maximum values among the energy eigenvalues;

[0031] The calculation formula of the target eigenvalue is as follows:

[0032]

[0033] wherein,

[0034] x′ represents the target eigenvalue;

[0035] min(x) represents the minimum value among the energy eigenvalues;

[0036] max(x) represents the maximum value among the energy eigenvalues.

[0037] Furthermore, before step S4, the method further includes step S0;

[0038] Step S0: Construct a target neural network model;

[0039] Step S0 specifically includes the following steps:

[0040] Step S01: Collect an annotated sample set; the annotated samples include the operation data of the conveyor idler for normal samples and different fault type samples;

[0041] Step S02: Divide the annotated sample set into a training set and a validation set;

[0042] Step S03: Construct a residual neural network with the multi-feature fusion value in the annotated sample as the input and the fault type as the output, and use the training set to train the residual neural network to obtain a residual neural network prediction model;

[0043] wherein, the multi-feature fusion value includes time-domain feature data, frequency-domain feature data, envelope feature data, and video feature data;

[0044] Step S04: Verify, optimize, and test the residual neural network prediction model through the validation set to obtain a target neural network model.

[0045] In a second aspect, the present invention provides an on-line monitoring device for a conveyor idler based on fiber optic sensing multi-feature fusion, and the device includes:

[0046] An acquisition unit for acquiring the acoustic vibration signal generated in real time during the operation of the conveyor idler;

[0047] wherein, the acoustic vibration signal is obtained based on the backward Rayleigh scattered light in the optical fiber;

[0048] An extraction unit, connected to the acquisition unit, for extracting target feature values based on the acoustic wave vibration signal;

[0049] A determination unit, connected to the extraction unit, for determining whether the target feature value exceeds a preset threshold;

[0050] An input unit, connected to the determination unit, for inputting the target feature value into a pre-constructed target neural network model when the determination unit determines that the target feature value exceeds the preset threshold, so that the target neural network model outputs the fault type corresponding to the target feature value, thereby completing the online monitoring of the conveyor idler based on fiber optic sensing multi-feature fusion.

[0051] Further, the acquisition unit includes:

[0052] A transmission module for transmitting the emitted pulsed laser signal to the conveyor idler through the vibration measuring optical fiber;

[0053] A receiving module, connected to the transmission module, for receiving the backward Rayleigh scattered light generated in the optical fiber;

[0054] Wherein, the backward Rayleigh scattered light is obtained according to the acoustic wave vibration generated in real time when the conveyor idler operates;

[0055] A first conversion module, connected to the receiving module, for converting the received backward Rayleigh scattered light signal into an optoelectronic signal to obtain the corresponding electrical signal;

[0056] A conversion module, connected to the first conversion module, for converting the electrical signal into a vibration signal, thereby obtaining the acoustic wave vibration signal generated in real time when the conveyor idler operates.

[0057] Further, the extraction unit includes:

[0058] An extraction subunit for performing feature extraction from the vibration signal to obtain the energy feature value of each frame;

[0059] A calculation subunit, connected to the extraction subunit, for normalizing the energy feature value to obtain the target feature value.

[0060] Further, the extraction subunit includes:

[0061] A preprocessing module for preprocessing the vibration signal to obtain the preprocessed vibration signal;

[0062] Wherein, the preprocessing includes increasing the high-frequency components of the vibration signal;

[0063] The frame processing module, connected to the preprocessing module, is used to frame the preprocessed vibration signal according to a time window to obtain each frame of signal including a continuous fixed duration;

[0064] The windowing processing module, connected to the frame processing module, is used to perform windowing processing on each frame of signal to obtain each frame of windowed signal;

[0065] The second conversion module, connected to the windowing processing module, is used to convert the time-domain signal in each frame of windowed signal into a frequency-domain signal through fast Fourier transform to obtain the spectral information of the signal;

[0066] The extraction module, connected to the second conversion module, is used to extract the band energy value from the spectral information using a Mel filter to obtain the output value of the Mel filter;

[0067] The first calculation module, connected to the extraction module, is used to sum the output values of all Mel filters in each frame to obtain the energy eigenvalue of each frame.

[0068] Further, the calculation sub-unit includes:

[0069] The acquisition module is used to traverse all the energy eigenvalues to obtain the minimum value and the maximum value in the energy eigenvalues;

[0070] The second calculation module, connected to the acquisition module, is used to calculate the target eigenvalue according to the minimum value and the maximum value in the energy eigenvalues;

[0071] Wherein, the following calculation formula of the target eigenvalue is stored in the second calculation module:

[0072]

[0073] Wherein,

[0074] x′ represents the target eigenvalue;

[0075] min(x) represents the minimum value in the energy eigenvalues;

[0076] max(x) represents the maximum value in the energy eigenvalues.

[0077] Further, the device further includes a construction unit;

[0078] The construction unit is connected to the input unit and is used to construct a target neural network model so that the input unit inputs the target eigenvalue into the pre-trained target neural network model;

[0079] The construction unit includes:

[0080] A collection module for collecting an annotated sample set; the annotated samples include the operation data of normal samples and conveyor idler samples of different fault types;

[0081] A partitioning module, connected to the collection module, for partitioning the annotated sample set into a training set and a validation set;

[0082] A construction module, connected to the partitioning module, for constructing a residual neural network with the multi-feature fusion value in the annotated samples as the input and the fault type as the output;

[0083] Wherein, the multi-feature fusion value includes time-domain feature data, frequency-domain feature data, envelope feature data, and video feature data;

[0084] A training module, respectively connected to the construction module and the partitioning module, for training the residual neural network with the training set to obtain a residual neural network prediction model;

[0085] A tuning module, respectively connected to the training module and the partitioning module, for validating, tuning, and testing the residual neural network prediction model through the validation set to obtain a target neural network model.

[0086] In a third aspect, the present invention provides an electronic device for online monitoring of a conveyor idler based on fiber optic sensing multi-feature fusion. The electronic device includes a memory and a processor. A computer program is stored in the memory. When the processor runs the computer program stored in the memory, the processor executes the method for online monitoring of a conveyor idler based on fiber optic sensing multi-feature fusion according to the first aspect.

[0087] Based on fiber optic sensing multi-feature fusion, the present invention can automatically and accurately identify and output the fault type, thereby enhancing the overall performance and safety of the conveyor idler. The specific effective effects are as follows:

[0088] 1. Real-time monitoring ability: Based on the backscattered Rayleigh light, the present invention can obtain and analyze vibration signals in real time, and timely identify the abnormal operation of the idler, thereby reducing the response time after a fault occurs.

[0089] 2. Improved fault diagnosis accuracy: Through the normalization processing of energy eigenvalue and the intelligent analysis of the target neural network, the present invention significantly improves the recognition accuracy of fault types and reduces the misjudgment rate.

[0090] 3. Intelligent management: The present invention realizes the automation of fault detection, reduces the dependence on manual detection, reduces the possibility of human errors, and further improves the monitoring efficiency.

[0091] 4. Cost savings: By reducing the need for frequent manual inspections and maintenance, the present invention significantly reduces operating costs and improves the utilization efficiency of resources.

[0092] 5. Adaptability to complex environments: The present invention utilizes fiber optic sensing technology and can operate stably in high-noise environments, ensuring the reliability of monitoring data.

[0093] 6. Enhanced early warning capability: Through real-time feature monitoring and threshold judgment, the present invention can provide early warnings of potential faults, prevent serious damage to equipment, and thus avoid losses caused by production stoppages. Description of the Drawings

[0094] Figure 1 Schematic diagram of the online monitoring method for conveyor idlers based on fiber optic sensing multi-feature fusion in an embodiment of the present invention;

[0095] Figure 2 Framework diagram of the online monitoring method for conveyor idlers based on fiber optic sensing multi-feature fusion in an embodiment of the present invention;

[0096] Figure 3 Overall flowchart of the online monitoring of conveyor idlers based on fiber optic sensing multi-feature fusion in an embodiment of the present invention;

[0097] Figure 4 Another overall flowchart of the online monitoring of conveyor idlers based on fiber optic sensing multi-feature fusion in an embodiment of the present invention;

[0098] Figure 5 Schematic diagram of the online monitoring device for conveyor idlers based on fiber optic sensing multi-feature fusion in an embodiment of the present invention;

[0099] Figure 6 Architecture diagram of the electronic device for online monitoring of conveyor idlers based on fiber optic sensing multi-feature fusion in an embodiment of the present invention.

[0100] Reference numerals: 1, belt conveyor; 2, voiceprint detector; 3, vibration measurement optical fiber; 4, personal computer; 10, acquisition unit; 20, extraction unit; 30, determination unit; 40, input unit; 100, memory; 200, processor. Detailed Embodiments

[0101] To enable those skilled in the art to better understand the technical solutions of the present invention, the embodiments of the present invention will be further described in detail below with reference to the drawings.

[0102] It can be understood that the specific embodiments and drawings described herein are only for explaining the present invention and are not intended to limit the present invention.

[0103] It is understood that, without conflict, the embodiments in the present invention and the features in the embodiments may be combined with each other.

[0104] It is understood that, for the convenience of description, only the parts related to the present invention are shown in the drawings of the present invention, while the parts unrelated to the present invention are not shown in the drawings.

[0105] It is understood that each unit and module involved in the embodiments of the present invention may correspond to only one entity structure, or may be composed of multiple entity structures. Alternatively, multiple units and modules may also be integrated into one entity structure.

[0106] It is understood that, without conflict, the functions and steps marked in the flowcharts and block diagrams of the present invention may occur in an order different from that marked in the drawings.

[0107] It is understood that in the flowcharts and block diagrams of the present invention, the possible architectures, functions, and operations of the systems, devices, equipment, and methods according to the embodiments of the present invention are shown. Among them, each block in the flowchart or block diagram may represent a unit, module, program segment, or code, which contains executable instructions for implementing the specified function. Moreover, each block or combination of blocks in the block diagram and flowchart may be implemented by a hardware-based system for implementing the specified function, or may be implemented by a combination of hardware and computer instructions.

[0108] It is understood that the units and modules involved in the embodiments of the present invention may be implemented in software or in hardware. For example, the units and modules may be located in the processor.

[0109] Embodiment 1:

[0110] This embodiment provides an online monitoring method for conveyor rollers based on multi-feature fusion of optical fiber sensing. This method is applicable to long-distance conveyor systems in complex industrial environments such as coal mines, ports, power plants, and mines. This method can collect multi-dimensional data such as vibration, temperature, and strain of the rollers in real time through distributed optical fiber sensors, and combined with multi-feature fusion technology, it can accurately identify early faults (such as wear, jamming, eccentricity, etc.) and achieve precise positioning of the fault points. It has the characteristics of explosion-proof, anti-electromagnetic interference, and corrosion resistance, enabling it to operate stably under harsh conditions of high dust, high noise, and strong electromagnetic interference. In addition, through real-time monitoring and intelligent early warning, this method significantly reduces the cost of manual inspection, improves the accuracy and timeliness of fault detection, provides a reliable guarantee for the safe and efficient operation of the conveyor system, thereby enhancing the operation safety and reliability of the equipment, and ensuring the continuity and efficiency of the production process. As Figure 1 shown, the online monitoring method includes the following steps:

[0111] Step S0: Construct a target neural network model.

[0112] Specifically, step S0 specifically includes the following steps:

[0113] Step S01: Collect an annotated sample set; the annotated samples include the operation data of the conveyor idler for normal samples and different fault type samples;

[0114] Step S02: Divide the annotated sample set into a training set and a validation set;

[0115] Step S03: Construct a residual neural network with the multi-feature fusion value in the annotated samples as the input and the fault type as the output, and use the training set to train the residual neural network to obtain a residual neural network prediction model;

[0116] Among them, the multi-feature fusion value includes time-domain feature data, frequency-domain feature data, envelope feature data, and video feature data;

[0117] Step S04: Verify, optimize, and test the residual neural network prediction model through the validation set to obtain a target neural network model.

[0118] Step S1: Obtain the acoustic vibration signal generated in real time during the operation of the conveyor idler; the acoustic vibration signal is obtained based on the backward Rayleigh scattered light in the optical fiber;

[0119] As a specific implementation manner, step S1 specifically includes the following steps:

[0120] Transmit the emitted pulsed laser signal to the conveyor idler through the vibration measurement optical fiber;

[0121] Receive the backward Rayleigh scattered light generated in the optical fiber;

[0122] Among them, the backward Rayleigh scattered light is obtained according to the acoustic vibration generated in real time during the operation of the conveyor idler;

[0123] Perform optoelectronic signal conversion on the received backward Rayleigh scattered light signal to obtain the corresponding electrical signal;

[0124] Convert the electrical signal into a vibration signal, thereby obtaining the acoustic vibration signal generated in real time during the operation of the conveyor idler.

[0125] Step S2: Extract target feature values based on the acoustic vibration signal.

[0126] Specifically, this step S2 specifically includes the following steps:

[0127] Step S21: Perform feature extraction from the vibration signal to obtain the energy feature value of each frame.

[0128] As a specific implementation, step S21 specifically includes the following steps:

[0129] Preprocess the vibration signal to obtain a preprocessed vibration signal;

[0130] Among them, the preprocessing includes increasing the high-frequency components of the vibration signal;

[0131] Frame the preprocessed vibration signal according to a time window to obtain each frame of signal including a continuous fixed duration;

[0132] Perform windowing processing on each frame of signal to obtain each frame of windowed signal;

[0133] Convert the time-domain signal in each frame of windowed signal into a frequency-domain signal through fast Fourier transform to obtain the spectral information of the signal;

[0134] Use a Mel filter to extract the band energy value from the spectral information to obtain the output value of the Mel filter;

[0135] Sum the output values of all Mel filters in each frame to obtain the energy eigenvalue of each frame.

[0136] Step S22: Normalize the energy eigenvalue to obtain a target eigenvalue.

[0137] As a specific implementation, step S22 specifically includes the following steps:

[0138] Traverse all the energy eigenvalues to obtain the minimum and maximum values among the energy eigenvalues;

[0139] Calculate the target eigenvalue according to the minimum and maximum values among the energy eigenvalues;

[0140] The calculation formula of the target eigenvalue is as follows:

[0141]

[0142] Among them,

[0143] x′ represents the target eigenvalue;

[0144] min(x) represents the minimum value among the energy eigenvalues;

[0145] max(x) represents the maximum value among the energy eigenvalues.

[0146] Step S3: Determine whether the target eigenvalue exceeds a preset threshold: If the target eigenvalue exceeds the preset threshold, go to step S4; if the target eigenvalue does not exceed the preset threshold, end the process.

[0147] Step S4: Input the target feature value into a pre-constructed target neural network model, so that the target neural network model outputs the fault type corresponding to the target feature value, thereby completing the online monitoring of the conveyor idler based on fiber optic sensing multi-feature fusion.

[0148] The online monitoring method provides a systematic process for monitoring the operating status of the conveyor idler through acoustic vibration signals. The following is a detailed elaboration of this process.

[0149] First, a target neural network model is constructed in step S0. In step S01, an annotated sample set is collected. It includes vibration data of the idler in normal operating conditions and different fault types (such as slight wear, severe wear, or insufficient lubrication, etc.). In the annotated sample set, the acoustic vibration signals of the idler in each state and the corresponding fault type labels are included. The collection of these data is usually carried out within a specific time period to ensure the diversity and representativeness of the samples. Then, in step S02, the collected annotated sample set is divided into a training set and a validation set. Generally, 80% of the samples are allocated for model training, while 20% are used for subsequent validation and testing. This division helps to evaluate the generalization ability of the model during the training process and ensures that the model can perform well on unknown data. In step S03, a residual neural network is constructed based on the annotated samples. Researchers use multi-feature fusion values such as time-domain features, frequency-domain features, envelope features, and video features as inputs and the fault type as the output for network training. During the training process, cross-validation is used to ensure that the model does not overfit, and the hyperparameters are continuously adjusted to optimize the model performance. After multiple rounds of training, the obtained residual neural network prediction model can effectively identify different fault types. Subsequently, in step S04, the validation set is used to test, validate, and optimize the model to ensure that the finally obtained target neural network model has high accuracy and robustness.

[0150] Enter step S1 to obtain the acoustic vibration signal generated during the operation of the conveyor idler in real time. The specific implementation process is to emit a pulsed laser signal and transmit it through the vibration measurement optical fiber to the idler, and the backscattered Rayleigh light generated in the optical fiber is received. The backscattered Rayleigh light is generated according to the actual vibration state of the idler operation, so it can reflect the operation of the idler in real time. After photoelectric signal conversion, the corresponding electrical signal is obtained, and the electrical signal is converted into a vibration signal, thereby obtaining the acoustic vibration signal of the real-time operation of the idler.

[0151] Figure 3Shown is the online monitoring process of conveyor idlers based on multi - feature fusion of fiber - optic sensing and machine learning. In this process, first, a fiber - optic sensor is used to collect the vibration signals of the idlers, and high - sensitivity, distributed monitoring is achieved through the reverse Rayleigh scattering light technology. In the signal - processing stage, steps such as pre - emphasis, framing, and windowing are used to pre - process the original signal: pre - emphasis is used to enhance high - frequency signals, framing divides the continuous signal into multiple short time intervals for frame - by - frame analysis, and windowing reduces spectral leakage through a window function. Then, the signal is mapped from the frequency domain to the Mel scale through a Mel filter bank, and band - energy features are extracted. Combining time - domain features (such as mean, variance), frequency - domain features (such as spectral energy), envelope - diagram features (extracted through Hilbert transform), and time - frequency features (such as short - time Fourier transform or wavelet transform) for multi - feature fusion to comprehensively describe the operating state of the idler. In the fault - diagnosis stage, a ResNet neural network is used to classify and predict the fused features. ResNet solves the problem of gradient disappearance in the training of deep networks by introducing residual connections and can efficiently identify various fault types of idlers, such as bearing abnormal noise, bearing wear, rubber ring wear flat, shaft fracture, etc. At the output end of the system, the operating state of the idler is judged according to the analysis result. If a fault is detected, an alarm is triggered to achieve precise positioning and early warning of the fault point. This method is applicable to complex industrial environments such as coal mines, mines, and ports, and has the characteristics of anti - electromagnetic interference, corrosion resistance, high precision, and strong real - time performance, significantly reducing the cost of manual inspection and improving the accuracy and timeliness of fault detection, providing a reliable guarantee for the safe and efficient operation of the conveyor system.

[0152] As Figure 4 shown, in addition to faults caused by sound (such as bearing abnormal noise, bearing wear, rubber ring wear flat, shaft fracture, etc.), dust and high temperature may also cause equipment failures. Therefore, we use dust concentration and temperature as inputs to the learning model. This approach helps to comprehensively evaluate the working environment of the conveyor idler, thus more accurately identifying and predicting potential faults. By integrating information from multiple dimensions such as sound, dust, and temperature, the learning model can better learn different fault patterns and improve the accuracy and reliability of fault detection.

[0153] Embodiment 2:

[0154] As Figure 5 shown, this embodiment provides an online monitoring device for conveyor idlers based on multi - feature fusion of fiber - optic sensing, and the device includes:

[0155] An acquisition unit 10 for acquiring vibration signals generated by the real - time audio of the conveyor idler;

[0156] wherein, the vibration signal is obtained based on the reverse Rayleigh scattering light in the optical fiber;

[0157] The extraction unit 20, connected to the acquisition unit 10, is configured to extract target feature values based on the acoustic wave vibration signal;

[0158] The determination unit 30, connected to the extraction unit 20, is configured to determine whether the target feature value exceeds a preset threshold;

[0159] The input unit 40, connected to the determination unit 30, is configured to input the target feature value into a pre-constructed target neural network model when the determination unit 30 determines that the target feature value exceeds the preset threshold, so that the target neural network model outputs the fault type corresponding to the target feature value, thereby completing the online monitoring of the conveyor idler based on fiber optic sensing multi-feature fusion.

[0160] As a specific implementation manner, the acquisition unit 10 includes:

[0161] The transmission module is configured to transmit the emitted pulsed laser signal to the conveyor idler through the vibration measuring optical fiber;

[0162] The receiving module, connected to the transmission module, is configured to receive the backward Rayleigh scattered light generated in the optical fiber;

[0163] Wherein, the backward Rayleigh scattered light is obtained according to the fluctuation of the real-time audio of the conveyor idler;

[0164] The first conversion module, connected to the receiving module, is configured to perform optoelectronic signal conversion on the received backward Rayleigh scattered light signal to obtain the corresponding electrical signal;

[0165] The conversion module, connected to the first conversion module, is configured to convert the electrical signal into a vibration signal, thereby obtaining the vibration signal generated by the real-time audio of the conveyor idler.

[0166] As a specific implementation manner, the extraction unit 20 includes:

[0167] The extraction subunit is configured to perform feature extraction from the vibration signal to obtain the energy feature value of each frame;

[0168] The calculation subunit, connected to the extraction subunit, is configured to perform normalization processing on the energy feature value to obtain the target feature value.

[0169] As a more specific implementation manner, the extraction subunit includes:

[0170] The preprocessing module is configured to preprocess the vibration signal to obtain the preprocessed vibration signal;

[0171] Wherein, the preprocessing includes increasing the high-frequency components of the vibration signal;

[0172] The frame processing module, connected to the preprocessing module, is configured to frame the preprocessed vibration signal according to a time window to obtain each frame of signal including a continuous fixed duration;

[0173] The windowing processing module, connected to the frame processing module, is configured to perform windowing processing on each frame of signal to obtain each frame of windowed signal;

[0174] The second conversion module, connected to the windowing processing module, is configured to convert the time-domain signal in each frame of windowed signal into a frequency-domain signal through fast Fourier transform to obtain the spectral information of the signal;

[0175] The extraction module, connected to the second conversion module, is configured to extract the band energy value from the spectral information using a Mel filter to obtain the output value of the Mel filter;

[0176] The first calculation module, connected to the extraction module, is configured to sum the output values of all Mel filters in each frame to obtain the energy eigenvalue of each frame.

[0177] As a specific implementation manner, the calculation sub-unit includes:

[0178] The acquisition module is configured to traverse all the energy eigenvalues to acquire the minimum value and the maximum value in the energy eigenvalues;

[0179] The second calculation module, connected to the acquisition module, is configured to calculate the target eigenvalue according to the minimum value and the maximum value in the energy eigenvalues;

[0180] Wherein, the following calculation formula of the target eigenvalue is stored in the second calculation module:

[0181]

[0182] Wherein,

[0183] x′ represents the target eigenvalue;

[0184] min(x) represents the minimum value in the energy eigenvalues;

[0185] max(x) represents the maximum value in the energy eigenvalues.

[0186] As a specific implementation manner, the device further includes a construction unit;

[0187] The construction unit is connected to the input unit 40 and is configured to construct a target neural network model, so that the input unit inputs the target eigenvalue into the pre-trained target neural network model;

[0188] The construction unit includes:

[0189] A collection module for collecting an annotated sample set; the annotated samples include the operation data of normal samples and conveyor idler samples of different fault types.

[0190] A division module, connected to the collection module, for dividing the annotated sample set into a training set and a validation set.

[0191] A construction module, connected to the division module, for constructing a residual neural network with the multi-feature fusion value in the annotated samples as the input and the fault type as the output.

[0192] Wherein, the multi-feature fusion value includes time-domain feature data, frequency-domain feature data, envelope feature data, and video feature data.

[0193] A training module, respectively connected to the construction module and the division module, for training the residual neural network with the training set to obtain a residual neural network prediction model.

[0194] An optimization module, respectively connected to the training module and the division module, for validating, optimizing, and testing the residual neural network prediction model through the validation set to obtain a target neural network model.

[0195] The device in this embodiment can execute the method in Embodiment 1.

[0196] Embodiment 3:

[0197] As Figure 6 shown, this embodiment provides an electronic device for on-line monitoring of conveyor idlers based on optical fiber sensing multi-feature fusion. The electronic device includes a memory 100 and a processor 200. A computer program is stored in the memory 100. When the processor 200 runs the computer program stored in the memory 100, the processor 200 executes the on-line monitoring method of conveyor idlers based on optical fiber sensing multi-feature fusion according to the first aspect.

[0198] As a specific implementation manner of this embodiment, the electronic device is a personal computer 4 or a mobile terminal.

[0199] As Figure 2 shown, the electronic device is a personal computer 4. The core advantage of this electronic device lies in the combination of the voiceprint detector and the vibration measurement optical fiber, enabling it to comprehensively and accurately monitor the operation status of the idler. Due to its explosion-proof, anti-electromagnetic interference, and corrosion-resistant characteristics, even in harsh environments such as high noise and high dust, the system can still work stably, ensuring the safe operation of the conveyor. In addition, the real-time monitoring and intelligent early warning functions significantly improve the accuracy and timeliness of fault detection, reduce the manual inspection cost, and provide a reliable basis for equipment maintenance and production optimization.

[0200] Specifically, the belt conveyor 1 is an important device of this monitoring system and is responsible for material transportation. It consists of a conveyor belt, idlers, and a driving device, etc. Among them, the operating state of the idlers is the key monitoring object. Through the real-time data collection of sensors, effective fault diagnosis of the idlers can be carried out. The acoustic detector 2 is installed on the external channel steel of the belt conveyor and is fixed to the device with bolts. Its main function is to collect and amplify the sound signals from the idlers. After amplification processing, it helps to better restore the sound, providing support for subsequent fault analysis and identification. These acoustic signals will help the monitoring personnel identify abnormal sounds during the operation of the device, thereby inferring the types of faults. The vibration measurement optical fiber 3 is led out from the personal computer 4, passes through the acoustic detector and is fixed on the side channel steel of the conveyor, making close contact with the channel steel, and uses optical fiber sensing technology to collect vibration signals in real time. The optical fiber has extremely high anti-interference ability, ensuring its stable operation in a complex industrial environment. The collected vibration signals will be transmitted to the personal computer 4 for data analysis and fault diagnosis. The personal computer 4 is the core processing device of the entire monitoring system. It is connected to the vibration measurement optical fiber and the acoustic detector, receives and processes the collected vibration and sound signals. The personal computer 4 is built-in with a laser emission and reception module and a signal processing and analysis module, responsible for further processing and analysis of the data. According to the preset algorithm, the personal computer 4 can identify different fault modes and feedback the analysis results to the operator, thus facilitating effective maintenance decisions.

[0201] It can be understood that the above embodiments are only exemplary embodiments adopted to illustrate the principle of the present invention. However, the present invention is not limited thereto. For those of ordinary skill in the art, various modifications and improvements can be made without departing from the spirit and essence of the present invention, and these modifications and improvements are also regarded as the protection scope of the present invention.

Claims

1. A conveyor roller online monitoring method based on optical fiber sensing multi-feature fusion, characterized in that: The method comprises the following steps: Step S1: Acquire the acoustic vibration signal generated in real time when the conveyor roller is running; Wherein, the acoustic vibration signal is obtained based on the reverse Rayleigh scattered light in the optical fiber; Step S2: extracting target feature values ​​based on the acoustic vibration signal; Step S3: Determine whether the target feature value exceeds a preset threshold: If the target feature value exceeds the preset threshold, the process proceeds to step S4; if the target feature value does not exceed the preset threshold, the process ends; Step S4: input the target feature value into a pre-built target neural network model, so that the target neural network model outputs the fault type corresponding to the target feature value, thereby completing the online monitoring of conveyor rollers based on optical fiber sensing multi-feature fusion.

2. The conveyor roller online monitoring method based on optical fiber sensing multi-feature fusion according to claim 1 is characterized in that: The step S1 specifically includes the following steps: The emitted pulse laser signal is transmitted to the conveyor roller via the vibration measuring optical fiber; Receiving the reverse Rayleigh scattered light generated in the optical fiber; Wherein, the reverse Rayleigh scattered light is obtained according to the real-time acoustic vibration generated when the conveyor roller is running; Convert the received reverse Rayleigh scattered light signal into a photoelectric signal to obtain an electrical signal corresponding thereto; The electrical signal is converted into a vibration signal, thereby obtaining a sonic vibration signal generated in real time when the conveyor roller is running.

3. The conveyor roller online monitoring method based on optical fiber sensing multi-feature fusion according to claim 1 is characterized in that: The step S2 specifically includes the following steps: Step S21: extracting features from the vibration signal to obtain energy feature values ​​for each frame; Step S22: normalize the energy eigenvalue to obtain a target eigenvalue.

4. The conveyor roller online monitoring method based on optical fiber sensing multi-feature fusion according to claim 3 is characterized in that: The step S21 specifically includes the following steps: Preprocessing the vibration signal to obtain a preprocessed vibration signal; Wherein, the preprocessing includes increasing the high-frequency components of the vibration signal; The preprocessed vibration signal is divided into frames according to the time window to obtain each frame signal including a continuous fixed time length; Perform windowing processing on each frame signal to obtain each frame signal after windowing; The time domain signal in each frame of the windowed signal is converted into a frequency domain signal by fast Fourier transform to obtain the spectrum information of the signal; Extracting a frequency band energy value from the frequency spectrum information using a Mel filter to obtain an output value of the Mel filter; The output values ​​of all Mel filters in each frame are summed to obtain the energy feature value of each frame.

5. The conveyor roller online monitoring method based on optical fiber sensing multi-feature fusion according to claim 3 is characterized in that: The step S22 specifically includes the following steps: Traverse all energy characteristic values ​​and obtain the minimum and maximum values ​​among the energy characteristic values; Calculate the target eigenvalue according to the minimum and maximum values ​​of the energy eigenvalues; The calculation formula of the target characteristic value is as follows: in, x′ represents the target feature value; min(x) represents the minimum value among the energy eigenvalues; max(x) represents the maximum value among the energy eigenvalues.

6. The conveyor roller online monitoring method based on optical fiber sensing multi-feature fusion according to any one of claims 1 to 5 is characterized in that: Before step S4, the method further includes step S0; Step S0: construct a target neural network model; The step S0 specifically includes the following steps: Step S01: Collecting a set of labeled samples; the labeled samples include normal samples and conveyor roller operation data of samples of different fault types; Step S02: dividing the labeled sample set into a training set and a validation set; Step S03: construct a residual neural network with the multi-feature fusion value in the labeled sample as input and the fault type as output, and train the residual neural network with the training set to obtain a residual neural network prediction model; Wherein, the multi-feature fusion value includes time domain feature data, frequency domain feature data, envelope feature data, and video feature data; Step S04: verify, tune and test the residual neural network prediction model through a verification set to obtain a target neural network model.

7. An online monitoring device for conveyor rollers based on multi-feature fusion of optical fiber sensing, characterized in that: include: An acquisition unit, used for acquiring the acoustic vibration signal generated in real time when the conveyor roller is running; Wherein, the acoustic vibration signal is obtained based on the reverse Rayleigh scattered light in the optical fiber; An extraction unit, connected to the acquisition unit, for extracting a target characteristic value based on the acoustic vibration signal; A determination unit, connected to the extraction unit, for determining whether the target feature value exceeds a preset threshold; An input unit is connected to the determination unit and is used for inputting the target characteristic value into a pre-built target neural network model when the determination unit determines that the target characteristic value exceeds a preset threshold value, so that the target neural network model outputs the fault type corresponding to the target characteristic value, thereby completing the online monitoring of conveyor rollers based on optical fiber sensing multi-feature fusion.

8. The conveyor roller online monitoring device based on optical fiber sensing multi-feature fusion according to claim 7 is characterized in that: The acquisition unit comprises: A transmission module, used for transmitting the emitted pulse laser signal to the conveyor roller via the vibration measuring optical fiber; A receiving module, connected to the transmission module, for receiving reverse Rayleigh scattered light generated in the optical fiber; Wherein, the reverse Rayleigh scattered light is obtained according to the real-time acoustic vibration generated when the conveyor roller is running; A first conversion module, connected to the receiving module, for converting the received reverse Rayleigh scattered light signal into an optical-electrical signal to obtain an electrical signal corresponding thereto; The conversion module is connected to the first conversion module and is used to convert the electrical signal into a vibration signal, thereby obtaining a sound wave vibration signal generated in real time when the conveyor roller is running.

9. The conveyor roller online monitoring device based on optical fiber sensing multi-feature fusion according to claim 7 is characterized in that: The extraction unit comprises: An extraction subunit, used to extract features from the vibration signal to obtain an energy feature value of each frame; The calculation subunit is connected to the extraction subunit and is used to normalize the energy eigenvalue to obtain a target eigenvalue.

10. The conveyor roller online monitoring device based on optical fiber sensing multi-feature fusion according to claim 9 is characterized in that: The extraction subunit comprises: A preprocessing module, used for preprocessing the vibration signal to obtain a preprocessed vibration signal; Wherein, the preprocessing includes increasing the high-frequency components of the vibration signal; A frame processing module, connected to the preprocessing module, for framing the preprocessed vibration signal according to the time window to obtain each frame signal including a continuous fixed time length; A windowing processing module, connected to the frame processing module, is used to perform windowing processing on each frame signal to obtain each frame signal after windowing; A second conversion module is connected to the windowing processing module and is used to convert the time domain signal in each frame signal after windowing into a frequency domain signal through fast Fourier transform to obtain the spectrum information of the signal; an extraction module, connected to the second conversion module, and configured to extract a frequency band energy value from the frequency spectrum information using a Mel filter to obtain an output value of the Mel filter; The first calculation module is connected to the extraction module and is used to sum the output values ​​of all Mel filters in each frame to obtain the energy characteristic value of each frame.

11. The conveyor roller online monitoring device based on optical fiber sensing multi-feature fusion according to claim 9 is characterized in that: The computing subunit comprises: An acquisition module is used to traverse all energy characteristic values ​​and obtain the minimum and maximum values ​​among the energy characteristic values; A second calculation module, connected to the acquisition module, is used to calculate the target characteristic value according to the minimum value and the maximum value of the energy characteristic value; The second calculation module stores the following calculation formula for the target characteristic value: in, x′ represents the target feature value; min(x) represents the minimum value among the energy eigenvalues; max(x) represents the maximum value among the energy eigenvalues.

12. The conveyor roller online monitoring device based on optical fiber sensing multi-feature fusion according to any one of claims 7 to 11, characterized in that: The device also includes a construction unit; The construction unit is connected to the input unit and is used to construct a target neural network model, so that the input unit inputs the target feature value into the pre-trained target neural network model; The building block comprises: A collection module, used to collect a set of labeled samples; the labeled samples include normal samples and conveyor roller operation data of samples of different fault types; A division module, connected to the collection module, for dividing the labeled sample set into a training set and a validation set; A construction module, connected to the division module, for constructing a residual neural network with a multi-feature fusion value in the labeled sample as input and a fault type as output; Wherein, the multi-feature fusion value includes time domain feature data, frequency domain feature data, envelope feature data, and video feature data; A training module, connected to the construction module and the partitioning module respectively, for training the residual neural network using a training set to obtain a residual neural network prediction model; The tuning module is connected to the training module and the partitioning module respectively, and is used to verify, tune and test the residual neural network prediction model through a verification set to obtain a target neural network model.

13. An electronic device for online monitoring of conveyor rollers based on multi-feature fusion of optical fiber sensing, characterized in that: It includes a memory and a processor, wherein a computer program is stored in the memory. When the processor runs the computer program stored in the memory, the processor executes the conveyor roller online monitoring method based on optical fiber sensing multi-feature fusion according to any one of claims 1 to 6.

Images