A method, equipment, medium, and product for online safety monitoring of belt conveyors.

Abstract

This application discloses a method, equipment, medium, and product for online safety monitoring of belt conveyors, relating to the field of belt conveyor safety monitoring. The method includes using a vibration-damping and sensitivity-enhancing acoustic-optical cable device to collect multi-source vibration signals and temperature signals from the belt conveyor. The vibration-damping and sensitivity-enhancing acoustic-optical cable device adopts a multi-core fiber spiral winding structure with an integrated gradient vibration-damping layer on the outer layer. The multi-source vibration signals are sequentially subjected to wavelet packet decomposition, blind source separation, and deep learning optimization to determine the single-source vibration signal corresponding to each part. The single-source vibration signal and its corresponding temperature signal are synchronized in time using a dual-fiber redundant link and a fiber optic gyroscope. Based on the synchronized single-source vibration signal and its corresponding temperature signal, combined with the sound signal, multi-source data fusion diagnosis is performed using D-S evidence theory to generate a fault warning signal. This application can improve the accuracy and stability of distributed optical fiber monitoring of belt conveyors.

Description

Technical Field

[0001] This application relates to the field of safety monitoring of belt conveyors, and in particular to a method, equipment, medium and product for online safety monitoring of belt conveyors. Background Technology

[0002] Belt conveyors, as core material transport equipment in industries such as mining, ports, and metallurgy, directly impact production efficiency and operational safety. Related safety monitoring technologies fall at the intersection of industrial equipment condition monitoring and fiber optic sensing applications. This field integrates fiber optic sensing, signal processing, and intelligent algorithms to achieve real-time detection and early warning of faults such as belt conveyor roller imbalance, idler abnormalities, and belt tears. Among these, distributed fiber optic monitoring technology, due to its characteristics of resistance to electromagnetic interference, long-distance distributed measurement, and tolerance to harsh environments, has become the mainstream direction for replacing traditional electrical sensor monitoring. It primarily relies on temperature-sensing and acoustic-vibration fiber optic cables to construct a comprehensive monitoring network.

[0003] Traditional distributed fiber optic monitoring systems for belt conveyors are implemented in a relatively simple way: at the sensing level, single-core optical fibers are laid in a straight line, and there is no special vibration reduction design on the outer layer of the optical cable; physical protection is only achieved through the basic sheath. Signal processing relies on Fourier transform or simple filtering algorithms, which have limited effect on separating aliased signals generated by multiple vibration sources such as motors, drums, and idlers. Model training relies entirely on field-measured fault samples, which is difficult to cover all fault types under complex working conditions. Data transmission uses a single fiber optic link, and the synchronization accuracy can only reach the millisecond level. Fault diagnosis is mostly based on a single signal threshold (such as temperature exceeding 80℃ or vibration amplitude exceeding a preset value) to determine the equipment status.

[0004] The existing distributed optical fiber monitoring system for belt conveyors has the following shortcomings: (1) In environments with strong background noise, the demodulation dynamic range of the distributed optical fiber acoustic vibration signal is limited, making it difficult to capture signals such as 0.1 mm / s. 2 (1) Early weak fault signals such as micro-vibration; (2) Traditional detection optical cables lack special vibration reduction and sensitivity enhancement design, resulting in low coupling efficiency of acoustic and vibration signals, weak anti-interference ability, short service life and high maintenance cost; (3) The mixed signals generated by multiple vibration sources (motor, roller, idler, etc.) are difficult to detect and separate effectively, and a single signal processing algorithm is prone to fault misjudgment or missed judgment; at the same time, it also includes derivative problems such as insufficient model generalization ability caused by the scarcity of fault samples and low data transmission synchronization accuracy affecting diagnostic reliability.

[0005] Therefore, in order to address the above problems, there is an urgent need to provide a new method or system for online safety monitoring of belt conveyors. Summary of the Invention

[0006] The purpose of this application is to provide a method, equipment, medium, and product for online safety monitoring of belt conveyors, which can improve the accuracy and stability of distributed optical fiber monitoring of belt conveyors.

[0007] To achieve the above objectives, this application provides the following solution:

[0008] In a first aspect, this application provides a method for online safety monitoring of a belt conveyor, the method comprising:

[0009] A vibration-damping and sensitivity-enhancing acoustic-optical cable device is used to collect multi-source vibration signals and temperature signals from a belt conveyor. The multi-source vibration signals include single-source vibration signals from multiple locations within the belt conveyor. The vibration-damping and sensitivity-enhancing acoustic-optical cable device employs a multi-core fiber spiral winding structure and integrates a gradient vibration-damping layer on its outer layer. The gradient vibration-damping layer consists of a metal corrugated tube layer, a rubber buffer layer, and a carbon fiber reinforcement layer, arranged sequentially from the inside out.

[0010] The multi-source signal is sequentially subjected to wavelet packet decomposition, blind source separation and deep learning optimization to determine the single-source signal corresponding to each part.

[0011] For single-source vibration signals and corresponding temperature signals, time synchronization is achieved using dual-fiber redundant links and fiber optic gyroscopes.

[0012] Based on the synchronized single-source vibration signal and the corresponding temperature signal, combined with the sound signal, the DS evidence theory is used to perform multi-source data fusion diagnosis and generate a fault warning signal.

[0013] Optionally, the vibration-damping and sensitivity-enhancing acoustic optical cable device includes: four parallel optical fibers; each optical fiber is wrapped with a highly elastic polymer layer, and damping material is filled between adjacent optical fibers; the optical fiber winding pitch is harmonicly related to the operating frequency of the belt conveyor.

[0014] Optionally, the step of sequentially performing wavelet packet decomposition, blind source separation, and deep learning optimization on the multi-source signal to determine the single-source signal corresponding to each part specifically includes:

[0015] The db8 wavelet was used to perform 5-level wavelet packet decomposition on the multi-source signal to obtain multiple sub-frequency bands;

[0016] The feature matrix X is constructed based on the sub-components whose energy percentage within a sub-band of a set frequency is greater than the energy threshold.

[0017] Blind source separation is performed on the feature matrix X based on the improved FastICA algorithm to obtain the separation matrix W;

[0018] Based on the separation matrix W, the signal is classified using the objective function that maximizes non-Gaussianity by introducing regularization coefficients, and the preliminary separated source signals are obtained.

[0019] Based on the initially separated vibration source signals, a single vibration source signal is obtained using a CNN-LSTM model; the CNN-LSTM model includes a convolutional neural network and a long short-term memory network.

[0020] Optionally, the step of classifying signals based on the separation matrix W using a non-Gaussian objective function with introduced regularization coefficients to obtain preliminarily separated source signals specifically includes:

[0021] Using formula Determine the objective function to maximize non-Gaussianity ;

[0022] Using formula Perform iterative updates;

[0023] in, For the separation matrix, This refers to the number of sub-components within a given frequency band whose energy percentage exceeds an energy threshold. For the number of vibration sources, The characteristic matrix, The number of sampling points. The regularization coefficient is . Let be the separation matrix at time t+1. Let be the separation matrix at time t. The iteration step size is T, and the superscript T is the transpose. Stop iteration when, via The initial separated vibration source signals were obtained. It is an identity matrix.

[0024] Optionally, the step of using the synchronized single-source vibration signal and the corresponding temperature signal, combined with the sound signal, to perform multi-source data fusion diagnosis using DS evidence theory to generate a fault warning signal specifically includes:

[0025] Kurtosis analysis was performed on the vibration signal from a single vibration source to obtain a first diagnostic result; the first diagnostic result is a mechanical fault of roller imbalance.

[0026] The temperature signal is classified using a support vector machine to obtain a second diagnostic result; the second diagnostic result is either normal temperature rise or abnormal heating.

[0027] Mel frequency cepstral coefficients are extracted from the sound signal in the single-source signal and combined with a hidden Markov model to obtain a third diagnostic result; the third diagnostic result is the belt slippage or tearing state.

[0028] Based on the first, second, and third diagnostic results, decision-level fusion is performed using DS evidence theory to generate fault warning signals.

[0029] Optionally, the step of generating a fault warning signal by performing multi-source data fusion diagnosis based on the synchronized single-source signal and the corresponding temperature signal, combined with the sound signal, using DS evidence theory, further includes:

[0030] The iteration step size of blind source separation is dynamically adjusted based on the diagnostic results of multi-source data fusion.

[0031] Adjust the damping coefficient of the vibration-damping and sensitivity-enhancing optical cable device based on real-time monitoring of environmental noise.

[0032] Secondly, this application provides an online safety monitoring device for a belt conveyor, the online safety monitoring device for the belt conveyor comprising:

[0033] The vibration-damping and sensitivity-enhancing acoustic-optical cable module is used to collect multi-source vibration signals and temperature signals from a belt conveyor using a vibration-damping and sensitivity-enhancing acoustic-optical cable device. The multi-source vibration signals include single-source vibration signals from multiple locations within the belt conveyor. The vibration-damping and sensitivity-enhancing acoustic-optical cable device adopts a multi-core optical fiber spiral winding structure and integrates a gradient vibration damping layer on its outer layer. The gradient vibration damping layer consists of a metal corrugated tube layer, a rubber buffer layer, and a carbon fiber reinforcement layer, from the inside out.

[0034] The multi-source signal analysis module is used to sequentially perform wavelet packet decomposition, blind source separation and deep learning optimization on multi-source signals to determine the single-source signal corresponding to each part.

[0035] The data transmission and synchronization module is used to synchronize the single-source signal with the corresponding temperature signal using a dual-fiber redundant link and a fiber optic gyroscope.

[0036] The fault diagnosis and early warning module is used to perform multi-source data fusion diagnosis based on the synchronized single vibration source signal and the corresponding temperature signal, combined with the sound signal, using DS evidence theory, and generate a fault early warning signal.

[0037] Thirdly, this application provides a computer device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the aforementioned online safety monitoring method for belt conveyors.

[0038] Fourthly, this application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the aforementioned online safety monitoring method for belt conveyors.

[0039] Fifthly, this application provides a computer program product, including a computer program that, when executed by a processor, implements the aforementioned online safety monitoring method for belt conveyors.

[0040] According to the specific embodiments provided in this application, this application has the following technical effects:

[0041] This application provides a method, equipment, medium, and product for online safety monitoring of belt conveyors. It utilizes a vibration-damping and sensitivity-enhancing acoustic-optical cable device to collect multi-source vibration signals and temperature signals from the belt conveyor, and analyzes the multi-source vibration signals to achieve a speed of 0.1 mm / s. 2 The system captures micro-vibration signals and accurately separates signals from multiple vibration sources. Furthermore, the multi-core spiral winding structure and gradient damping layer design of the vibration-damping and sensitivity-enhancing acoustic-vibration optical cable device significantly enhance the coupling efficiency of acoustic-vibration signals and suppress environmental noise. Microsecond-level time synchronization is achieved through fiber optic gyroscopes, and dual-fiber redundant links ensure the continuity and reliability of data transmission. When the main link bit error rate exceeds the threshold, a switching mechanism is triggered to ensure stable transmission of monitoring data. Attached Figure Description

[0042] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0043] Figure 1 This is a schematic flowchart of an online safety monitoring method for a belt conveyor according to one embodiment of this application;

[0044] Figure 2 This is a schematic diagram of the overall process of an online safety monitoring method for a belt conveyor according to one embodiment of this application;

[0045] Figure 3 A schematic diagram illustrating the process of determining a single-source signal;

[0046] Figure 4 This is a schematic diagram of the structure of an online safety monitoring device for a belt conveyor according to one embodiment of this application. Detailed Implementation

[0047] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0048] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0049] In one exemplary embodiment, such as Figure 1 and Figure 2 As shown, a method for online safety monitoring of a belt conveyor is provided, comprising the following steps S101 to S104: Wherein:

[0050] S101, a vibration-damping and sensitivity-enhancing acoustic-optical cable device is used to collect multi-source vibration signals and temperature signals of a belt conveyor; the multi-source vibration signals include single-source vibration signals from multiple parts of the belt conveyor; the vibration-damping and sensitivity-enhancing acoustic-optical cable device adopts a multi-core optical fiber spiral winding structure, and a gradient damping layer is integrated on the outer layer of the vibration-damping and sensitivity-enhancing acoustic-optical cable device to enhance the coupling efficiency of acoustic signals through elastic body deformation; the gradient damping layer consists of a metal corrugated tube layer, a rubber buffer layer, and a carbon fiber reinforcement layer from the inside to the outside, and the three-layer structure works together to absorb vibration energy;

[0051] The vibration-damping and sensitivity-enhancing acoustic-vibration optical cable device includes: four parallel-arranged optical fibers; each optical fiber is wrapped with a highly elastic polymer layer, and damping material is filled between adjacent optical fibers to suppress environmental interference through noise correlation between multi-core optical fiber channels; the optical fiber winding pitch is harmonicly related to the operating frequency of the belt conveyor, enhancing the signal response in a specific frequency band. Specifically, the parallel-arranged four optical fibers include a temperature-sensing optical cable and an acoustic-vibration optical fiber, with the other two optical fibers serving as redundant backup fibers. The temperature-sensing optical cable is used to acquire temperature signals, and the acoustic-vibration optical fiber is used to acquire acoustic signals.

[0052] The harmonic relationship between fiber optic winding pitch and the operating frequency of a belt conveyor refers to the spiral winding of optical fibers at a specific physical spacing, such that the winding period (pitch) corresponds to an integer multiple of the characteristic vibration frequency of the belt conveyor. Specifically, when the belt conveyor operates at its natural frequency... During vibration, if the fiber winding pitch... satisfy ( If the integer is positive, then a mechanical resonance enhancement effect will occur at that frequency.

[0053] For example, when the rotation frequency of the belt conveyor roller is 10Hz, winding the optical fiber with a pitch of 5cm (corresponding to n=2) or 2.5cm (corresponding to n=4) can make the vibration signal of that frequency generate a stronger strain response in the optical fiber.

[0054] By harmonic matching, the characteristic fault frequencies of the belt conveyor (such as the 20Hz side frequency caused by roller imbalance and the 120Hz impact frequency caused by idler damage) are resonantly amplified, thereby increasing the signal strength of the above key frequency bands by 30%-50% and significantly improving the signal-to-noise ratio.

[0055] For environmental noise with non-harmonic relationships (such as high-frequency electromagnetic interference from motors and random external vibrations), the signal transmission efficiency is greatly reduced because the resonance condition is not met, and the equivalent noise suppression ratio can reach more than 15dB.

[0056] The enhanced feature frequency signal is easier to identify and separate in subsequent wavelet packet decomposition and FastICA blind source separation, reducing the algorithm's dependence on complex preprocessing and enabling the CNN-LSTM model to converge to the fault feature pattern faster. The training time of the CNN-LSTM model is shortened by about 25%.

[0057] S102, perform wavelet packet decomposition, blind source separation and deep learning optimization on the multi-source signal in sequence to determine the single-source signal corresponding to each part;

[0058] like Figure 3 As shown, S102 specifically includes:

[0059] S21, the db8 wavelet is used to perform 5-level wavelet packet decomposition on the multi-source signal to obtain multiple sub-frequency bands;

[0060] Specifically, the multi-source signal is divided into 32 sub-bands covering the range of 0kHz-20kHz.

[0061] S22, The feature matrix X is formed by the sub-components whose energy percentage in the sub-band of the set frequency is greater than the energy threshold;

[0062] For typical fault characteristics of belt conveyors (iron roller noise 8kHz-12kHz and belt friction 12kHz-16kHz), sub-components in the 8kHz-16kHz frequency band are extracted and their energy proportions are calculated. The calculation formula is:

[0063] ;

[0064] in, For the first Signal sequences of each sub-frequency band This is a sequence of multi-source vibration signals (original acoustic and vibration signals). Filtering. ≥18% of the sub-components constitute the characteristic matrix ,in The number of sampling points. This refers to the number of sub-components within a given frequency band whose energy percentage exceeds an energy threshold; that is, the number of filtered sub-components. The above formula accurately retains high-energy characteristic components, providing effective input for subsequent separation.

[0065] S23, Blind source separation is performed on the feature matrix X based on the improved FastICA algorithm to obtain the separation matrix W;

[0066] S24. Based on the separation matrix W, the signal is classified using the objective function that maximizes non-Gaussianity by introducing regularization coefficients, and the preliminary separated source signals are obtained.

[0067] Using formula Determine the objective function to maximize non-Gaussianity ;

[0068] Using formula Perform iterative updates;

[0069] in, For the separation matrix, For example, the number of vibration sources. Let's set it to 3, corresponding to the motor, drum, and idler roller respectively. The characteristic matrix, The number of sampling points. This is the regularization coefficient (value 0.01). Let be the separation matrix at time t+1. Let be the separation matrix at time t. The iteration step size (dynamically adjustable from 0.005 to 0.02) is given by the superscript T, which stands for transpose. Stop iteration when, via The initial separated vibration source signals were obtained. Regularization constraints are used to avoid ill-conditioned separation matrices and improve the separation accuracy of aliased signals. I is the identity matrix, which is used to ensure the orthogonality of the separation matrix W through regularization constraints, thereby avoiding ill-conditioned separation matrices and improving the separation accuracy of aliased signals.

[0070] S25, Based on the initially separated vibration source signals, a single vibration source signal is obtained using a CNN-LSTM model; the CNN-LSTM model includes a convolutional neural network and a long short-term memory network. The convolutional neural network (CNN) includes four convolutional layers (kernel sizes of 7×1, 5×1, 3×1, and 3×1, with numbers of 64, 128, 256, and 256 respectively), outputting a feature vector: The Long Short-Term Memory (LSTM) network consists of two hidden layers (64 and 32 neurons respectively), and the attention mechanism is implemented through a weight matrix. Adjusting the source weight The calculation formula is:

[0071] ;

[0072] in, The feature vector (256 dimensions) is a learnable vector. The score is calculated by the inner product of the feature vector and the source-specific vector, and then normalized using softmax to highlight effective source features. The feature vector refers to the 1×256 feature vector F output by the CNN part, and the source-specific vector refers to a 256-dimensional learnable vector corresponding to different vibration sources such as motors, rollers, and idlers. The LSTM outputs a clean single-source signal through a fully connected layer, providing a high signal-to-noise ratio signal source for the data transmission and synchronization modules.

[0073] Training the CNN-LSTM model also includes the following steps:

[0074] (1) Generative Adversarial Networks (GANs) are used to synthesize virtual signals containing different fault types to expand the training dataset;

[0075] (2) Using transfer learning technology, the vibration signal model parameters of hydropower units are transferred to the monitoring scenario of belt conveyors, reducing the need for training samples.

[0076] The model takes a belt conveyor vibration / sound signal dataset (including normal and fault state samples) and pre-trained model parameters (such as a hydropower unit vibration signal model) as input, and a trained CNN-LSTM model (including weight parameters and feature extraction logic) as output. The training process of the CNN-LSTM model is as follows:

[0077] (1) Data preprocessing: The original signal is standardized (the signal amplitude is normalized to the [0,1] interval), and the training set and validation set are divided in an 8:2 ratio. The training set includes measured fault signals (such as roller imbalance and belt tear) and normal operation signals.

[0078] (2) Dataset expansion: Start the training of Generative Adversarial Network (GAN), take the measured fault signal as input, and synthesize virtual fault samples (including gradually changing signals of different fault degrees) that match the features of the measured signal through the generator. Then, merge the virtual samples with the measured samples to form an expanded training set (expansion ratio is 1:1).

[0079] (3) Model initialization: Load the basic parameters of transfer learning (extract the initial weights of the convolutional layer and LSTM layer from the vibration signal model of the hydropower unit), and construct the CNN-LSTM model structure (the CNN part contains 3 convolutional layers and 2 pooling layers, the LSTM part contains 2 hidden layers, and the output layer uses the softmax activation function).

[0080] (4) Training iteration: Set the initial learning rate to 0.001, the batch size to 32, and the number of iterations to 100. Perform the following operations in each iteration:

[0081] Forward propagation: Input the training set samples into the CNN-LSTM model. The CNN layer extracts the frequency domain features of the signal (such as the energy distribution in the 8-16kHz frequency band), and the LSTM layer captures the temporal changes of the features (such as the duration and amplitude fluctuations of the fault signal). Loss calculation: The cross-entropy loss function is used to calculate the deviation between the predicted fault type and the actual label. The loss value is denoted as L.

[0082] Backpropagation: The model weights are updated through the Adam optimizer, prioritizing the adjustment of convolution kernel parameters related to belt conveyor-specific faults (such as roller noise);

[0083] (5) Validation and evaluation: The model accuracy is calculated every 10 rounds using the validation set. If the accuracy improvement is less than 0.5% for 3 consecutive rounds, the learning rate is reduced to 1 / 10 of the current value.

[0084] (6) Model optimization: An attention mechanism is introduced to assign dynamic weights to the temporal features of the LSTM output (the weight values ​​are calculated by the importance of the fault features of the training set samples) to enhance the feature contribution of signals in key time periods (such as the initial stage of a fault).

[0085] (7) Generalization verification: Use an independent test set (including novel fault samples not involved in training) to evaluate the model. If the accuracy is ≥95%, terminate the training; otherwise, return to step (2) to add 50 iterations and adjust the feature distribution of the GAN synthetic samples.

[0086] (8) Model export: Save the final model parameters and network structure, and generate a deployable lightweight model file (the parameter precision is compressed from 32-bit floating point to 16-bit through model quantization).

[0087] S103 uses a dual-fiber redundant link and fiber optic gyroscope for time synchronization between the single-source acoustic vibration signal and the corresponding temperature signal. The main link transmits the single-source acoustic vibration signal and the synchronized temperature signal, while the backup link transmits calibration parameters and control commands. Microsecond-level time synchronization is achieved via the fiber optic gyroscope, and the link automatically switches to the backup link when the main link's bit error rate exceeds a threshold. The calibration parameters include the iterative step size calibration parameters for blind source separation and the damping coefficient calibration parameters for the vibration-damping and sensitivity-enhancing acoustic optical cable device. The control commands include commands for adjusting the iterative step size for blind source separation and commands for adjusting the damping coefficient.

[0088] Specifically, multi-channel signal time-division multiplexing transmission is implemented based on the ARINC 818 protocol.

[0089] S104 uses the synchronized single-source signal and corresponding temperature signal, combined with the sound signal, to perform multi-source data fusion diagnosis using DS evidence theory, and generates a fault warning signal.

[0090] S104 specifically includes:

[0091] S41, perform kurtosis analysis on the vibration signal in the single vibration source signal to obtain the first diagnostic result; the first diagnostic result is a mechanical fault of roller imbalance.

[0092] S42, the temperature signal is classified using a support vector machine (SVM) to obtain a second diagnostic result; the second diagnostic result is normal temperature rise or abnormal heating;

[0093] S43, Mel frequency cepstral coefficients (MFCC) are extracted from the sound signal in the single-source signal, and combined with a hidden Markov model (HMM) to obtain a third diagnostic result; the third diagnostic result is the belt slippage or tearing state;

[0094] S44. Based on the first, second, and third diagnostic results, decision-level fusion is performed using DS evidence theory to generate a fault warning signal.

[0095] Following S104 are:

[0096] S1, dynamically adjust the iteration step size of blind source separation based on the multi-source data fusion diagnostic results;

[0097] S2 adjusts the damping coefficient of the vibration-damping and sensitivity-enhancing optical cable device based on real-time monitoring of environmental noise.

[0098] Adaptive optimization is achieved through closed-loop control via S1 and S2;

[0099] The vibration reduction and sensitivity enhancement optical cable device includes a multi-core fiber spiral winding structure. A gradient vibration reduction layer composed of a metal corrugated tube layer, a rubber buffer layer, and a carbon fiber reinforcement layer is integrated on the outer layer of the optical cable. Four-core optical fibers are arranged in parallel, each wrapped with a highly elastic polymer layer, and damping material is filled between adjacent fibers to suppress crosstalk. Simultaneously, the fiber winding pitch is harmonicly related to the operating frequency of the belt conveyor, thereby enhancing the coupling efficiency of the acoustic signal and suppressing environmental interference, improving the sensitivity and lifespan of the optical cable. The original acoustic signal undergoes wavelet packet decomposition to extract 8-16kHz frequency band feature components. In the blind source separation stage, the FastICA algorithm is used to estimate the mixing matrix to separate the main vibration source signals such as motors, rollers, and idlers. In the deep learning optimization stage, the separated signal is input into a pre-trained CNN-LSTM model, and the feature weights of each vibration source are dynamically adjusted through an attention mechanism, ultimately outputting a pure single-source signal. These steps together constitute the key protection points for signal analysis.

[0100] The key points and areas for protection in CNN-LSTM model training lie in the combination of dataset expansion and transfer learning. Specifically, generative adversarial networks are used to synthesize virtual signals containing different fault types to expand the training dataset. Simultaneously, transfer learning techniques are employed to transfer the vibration signal model parameters of hydropower units to the monitoring scenario of belt conveyors, reducing the need for training samples and thus improving model training efficiency and adaptability. The closed-loop control adaptive optimization mechanism in this application feeds back the fused diagnostic results to the multi-source signal analysis to dynamically adjust the iteration step size of the blind source separation algorithm. Simultaneously, the damping coefficient of the vibration-damping and sensitivity-enhancing optical cable device is automatically adjusted based on the real-time monitored environmental noise level, forming a perception-analysis-optimization closed-loop system to enhance monitoring stability. Dual-fiber redundant links are used to ensure the continuity and reliability of data transmission. When the main link bit error rate exceeds a threshold, a switching mechanism is triggered to ensure stable transmission of monitoring data. Kurtosis analysis of vibration signals is used to identify mechanical faults such as roller imbalance. Support vector machine is used to classify temperature signals to distinguish between normal temperature rise and abnormal heat generation. Mel frequency cepstral coefficients are extracted from sound signals and combined with hidden Markov models to determine belt slippage or tearing. Multi-dimensional data fusion is used to improve the accuracy and comprehensiveness of fault identification.

[0101] Based on the same inventive concept, this application also provides a belt conveyor safety online monitoring device for implementing the above-mentioned belt conveyor safety online monitoring method. The solution provided by this device is similar to the implementation described in the above method; therefore, the specific limitations in one or more embodiments of the belt conveyor safety online monitoring device provided below can be found in the limitations of the belt conveyor safety online monitoring method described above, and will not be repeated here.

[0102] In one exemplary embodiment, such as Figure 4 As shown, a safety online monitoring device for a belt conveyor is provided, comprising:

[0103] The vibration-damping and sensitivity-enhancing acoustic-optical cable module is used to collect multi-source vibration signals and temperature signals from a belt conveyor using a vibration-damping and sensitivity-enhancing acoustic-optical cable device. The multi-source vibration signals include single-source vibration signals from multiple locations within the belt conveyor. The vibration-damping and sensitivity-enhancing acoustic-optical cable device adopts a multi-core optical fiber spiral winding structure and integrates a gradient vibration damping layer on its outer layer. The gradient vibration damping layer consists of a metal corrugated tube layer, a rubber buffer layer, and a carbon fiber reinforcement layer, from the inside out.

[0104] The multi-source signal analysis module is used to sequentially perform wavelet packet decomposition, blind source separation and deep learning optimization on multi-source signals to determine the single-source signal corresponding to each part.

[0105] The data transmission and synchronization module is used to synchronize the single-source signal with the corresponding temperature signal using a dual-fiber redundant link and a fiber optic gyroscope.

[0106] The fault diagnosis and early warning module is used to perform multi-source data fusion diagnosis based on the synchronized single vibration source signal and the corresponding temperature signal, combined with the sound signal, using DS evidence theory, and generate a fault early warning signal.

[0107] Among them, the vibration reduction and sensitivity enhancement acoustic optical cable module and the multi-source signal analysis module are connected through an optical fiber link. The decoupled signal output by the multi-source signal analysis module is transmitted to the fault diagnosis and early warning module through the data transmission and synchronization module.

[0108] The vibration-damping and sensitivity-enhancing acoustic optical cable module has more than 3 times the sensitivity of traditional single-core optical cables, reduces the equivalent noise sound pressure by 9dB, and can detect speeds as low as 0.1mm / s. 2 Micro-vibrations.

[0109] The fault diagnosis and early warning module feeds back the diagnosis results to the multi-source signal analysis module and dynamically adjusts the iteration step size of the blind source separation algorithm;

[0110] The damping coefficient of the vibration-damping and amplifying optical fiber cable module is automatically adjusted based on the real-time monitored ambient noise level.

[0111] This application supports access to remote management platforms, enabling local real-time diagnosis through edge computing nodes, while simultaneously uploading raw data and diagnostic results to the cloud for long-term trend analysis.

[0112] The lifespan of the vibration-damping and sensitivity-enhancing acoustic optical cable module is more than twice that of traditional optical cables, and the maintenance cycle is extended from 3 months to 1 year, significantly reducing operation and maintenance costs.

[0113] Specifically, the multi-source signal analysis module includes: a data processing unit, a blind source separation unit, and a deep learning optimization unit;

[0114] This application utilizes a multi-core helical winding structure and gradient damping layer design in its vibration-damping and sensitivity-enhancing acoustic-vibration optical cable module to significantly improve the coupling efficiency of acoustic-vibration signals and suppress environmental noise. Combined with wavelet packet decomposition, FastICA algorithm, and CNN-LSTM model in the multi-source signal analysis module, it can achieve accurate separation and feature extraction of multi-source aliasing signals under strong background noise. This means the system can capture weak fault signals (such as 0.1 mm / s) that are difficult to identify using traditional monitoring methods. 2The micro-vibrations of the optical fiber cable module form a complete high-sensitivity monitoring link from the signal source to the processing end, significantly improving the accuracy of fault identification and early warning capabilities. The closed-loop adaptive control mechanism and intelligent training strategy of this application further enhance its adaptability. By dynamically adjusting the iteration step size of the blind source separation algorithm based on fault diagnosis results and adjusting the optical cable damping coefficient according to real-time noise levels, the system can maintain stable performance under complex working conditions. The application of GAN-synthesized virtual samples and transfer learning technology solves the problem of scarce fault samples in industrial scenarios, allowing the model to achieve efficient training and generalization even with limited measured data, ensuring reliable monitoring accuracy under different operating states and environmental interference. From an engineering application perspective, the vibration-damping and sensitivity-enhancing optical fiber cable module has a lifespan more than twice that of traditional optical cables, and the maintenance cycle is extended from 3 months to 1 year, significantly reducing long-term operation and maintenance costs. Simultaneously, the edge computing node combines local real-time diagnosis with long-term trend analysis in the cloud, meeting both the need for immediate on-site response and providing data support for the full lifecycle management of equipment. This dual advantage of technical performance and practical value makes this invention highly applicable and competitive in the field of industrial monitoring.

[0115] In one exemplary embodiment, a computer device is provided, which may be a server or a terminal. The computer device includes a processor, memory, input / output interfaces (I / O), and a communication interface. The processor, memory, and I / O interfaces are connected via a system bus, and the communication interface is connected to the system bus via the I / O interfaces. The processor of the computer device provides computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores an operating system, computer programs, and a database. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The I / O interfaces of the computer device are used for exchanging information between the processor and external devices. The communication interface of the computer device is used for communication with external terminals via a network connection. When the computer program is executed by the processor, it implements a method for online safety monitoring of a belt conveyor.

[0116] In one exemplary embodiment, a computer device is provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps in the above-described method embodiments.

[0117] In one exemplary embodiment, a computer-readable storage medium is provided storing a computer program that, when executed by a processor, implements the steps in the above-described method embodiments.

[0118] In one exemplary embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps in the above-described method embodiments.

[0119] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of the relevant data must comply with relevant regulations.

[0120] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments described above. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM).

[0121] The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.

[0122] In this application, all actions to acquire signals, information, or data are carried out in compliance with the relevant data protection laws and policies of the country where the location is situated, and with the authorization granted by the owner of the relevant device.

[0123] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0124] This document uses specific examples to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. Furthermore, those skilled in the art will recognize that, based on the ideas of this application, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A method of safety on-line monitoring of a belt conveyor, characterized in that The belt conveyor safety online monitoring method comprises: Collecting multi-vibration source signals and temperature signals of the belt conveyor by using a damping and sensitivity enhancing acoustic vibration optical cable device; the multi-vibration source signals comprise single vibration source signals of multiple parts in the belt conveyor; the damping and sensitivity enhancing acoustic vibration optical cable device adopts a multi-core optical fiber spiral winding structure, and a gradient damping layer is integrated on the outer layer of the damping and sensitivity enhancing acoustic vibration optical cable device; the gradient damping layer comprises, from inside to outside, a metal bellows layer, a rubber buffer layer and a carbon fiber reinforced layer; Performing wavelet packet decomposition, blind source separation and deep learning optimization on the multi-vibration source signals in sequence to determine single vibration source signals corresponding to each part; Performing time synchronization on the single vibration source signals and corresponding temperature signals by using a double optical fiber redundant link and an optical fiber gyroscope; Performing multi-source data fusion diagnosis on the synchronized single vibration source signals and corresponding temperature signals by using D-S evidence theory in combination with sound signals to generate a fault warning signal.

2. The belt conveyor safety online monitoring method according to claim 1, characterized in that, The damping and sensitivity enhancing acoustic vibration optical cable device comprises four-core optical fibers arranged in parallel; each core optical fiber is wrapped with a high-elasticity polymer layer, and a damping material is filled between adjacent optical fibers; the optical fiber winding pitch is in a harmonic relationship with the running frequency of the belt conveyor.

3. The belt conveyor safety online monitoring method according to claim 1, characterized in that, The wavelet packet decomposition, blind source separation and deep learning optimization on the multi-vibration source signals in sequence to determine single vibration source signals corresponding to each part specifically comprise: Performing 5-layer wavelet packet decomposition on the multi-vibration source signals by using db8 wavelet to obtain multiple sub-frequency bands; Constructing a feature matrix X according to sub-components with energy proportion greater than an energy threshold in a set frequency sub-band; Performing blind source separation on the feature matrix X based on an improved FastICA algorithm to obtain a separation matrix W; Performing signal classification by using a maximum non-Gaussianity objective function with a regularization coefficient based on the separation matrix W to obtain preliminarily separated vibration source signals; Obtaining single vibration source signals based on a CNN-LSTM model according to the preliminarily separated vibration source signals; the CNN-LSTM model comprises a convolutional neural network and a long short-term memory network.

4. The belt conveyor safety online monitoring method according to claim 3, characterized in that, The signal classification by using a maximum non-Gaussianity objective function with a regularization coefficient based on the separation matrix W to obtain preliminarily separated vibration source signals specifically comprises: Using the formula determining the maximum non-Gaussianity objective function ; Equation (1) is iteratively updated using the formula Equation (1) is iteratively updated using the formula wherein, is a separation matrix, is the number of sub-components whose energy proportion in the frequency sub-band is greater than the energy threshold, is the number of sources, is a feature matrix, is the number of sampling points, is a regularization coefficient, is the separation matrix at time t+1, is the separation matrix at time t, is an iteration step, the superscript T is a transpose, and when the iteration is stopped, and the preliminary separated source signals are obtained by is an identity matrix.​ 5. The belt conveyor safety online monitoring method according to claim 1, characterized in that, The multi-source data fusion diagnosis by using D-S evidence theory on the synchronized single vibration source signals and corresponding temperature signals in combination with sound signals to generate a fault warning signal specifically comprises: Performing kurtosis analysis on vibration signals in the single vibration source signals to obtain a first diagnostic result; the first diagnostic result is a drum imbalance mechanical fault; Performing support vector machine classification on the temperature signals to obtain a second diagnostic result; the second diagnostic result is normal temperature rise or abnormal heating; Performing Mel-frequency cepstral coefficient extraction on sound signals in the single vibration source signals and obtaining a third diagnostic result in combination with a hidden Markov model; the third diagnostic result is a belt slip or tearing state; Performing decision-level fusion based on D-S evidence theory according to the first diagnostic result, the second diagnostic result and the third diagnostic result to generate a fault warning signal.

6. The belt conveyor safety online monitoring method of claim 1, wherein, The D-S evidence theory is used for multi-source data fusion diagnosis according to the synchronized single-vibration-source signal and the corresponding temperature signal in combination with the sound signal to generate a fault warning signal, and then further includes: According to the multi-source data fusion diagnosis result, the iteration step of blind source separation is dynamically adjusted. According to the real-time monitored environmental noise, the damping coefficient of the vibration-damping and sensitivity-enhancing sound-vibration-light cable device is adjusted.

7. A belt conveyor safety online monitoring apparatus, characterized by The belt conveyor safety online monitoring device includes: A vibration-damping and sensitivity-enhancing sound-vibration-light cable module is used to collect multi-vibration-source signals and temperature signals of the belt conveyor by using a vibration-damping and sensitivity-enhancing sound-vibration-light cable device; the multi-vibration-source signals include single-vibration-source signals of multiple parts of the belt conveyor; the vibration-damping and sensitivity-enhancing sound-vibration-light cable device adopts a multi-core optical fiber spiral winding structure, and a gradient damping layer is integrated on the outer layer of the vibration-damping and sensitivity-enhancing sound-vibration-light cable device; the gradient damping layer includes, from inside to outside, a metal bellows layer, a rubber buffer layer, and a carbon fiber reinforced layer; A multi-vibration-source signal analysis module is used to sequentially perform wavelet packet decomposition, blind source separation, and deep learning optimization on the multi-vibration-source signals to determine the single-vibration-source signal corresponding to each part; A data transmission and synchronization module is used to synchronize the single-vibration-source signals and the corresponding temperature signals in time by using a double-fiber redundant link and an optical fiber gyroscope; A fault diagnosis and warning module is used to perform multi-source data fusion diagnosis according to the synchronized single-vibration-source signals and the corresponding temperature signals in combination with the sound signal by using the D-S evidence theory to generate a fault warning signal.

8. A computer device comprising: A memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that the processor executes the computer program to implement the belt conveyor safety online monitoring method of any one of claims 1-6.

9. A computer-readable storage medium having stored thereon a computer program, characterized in that, The computer program is executed by the processor to implement the belt conveyor safety online monitoring method of any one of claims 1-6.

10. A computer program product comprising a computer program, characterized in that, The computer program is executed by the processor to implement the belt conveyor safety online monitoring method of any one of claims 1-6.

Images