Power overhead line fitting fault monitoring method and device based on grating array

Through the combination of grating array and spectrum image recognition model, the problems of large workload and low recognition efficiency in power overhead line metal tools are solved, efficient and accurate monitoring and alarm of metal tools are achieved, and the risk of grid faults is reduced.

CN118914917AInactive Publication Date: 2024-11-08WUHAN FENGLI OPTOELECTRONICS TECH CO LTD +1
View PDF 4 Cites 0 Cited by

Patent Information

Application Number
CN202411066075.X
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-05
Publication Date
2024-11-08
Estimated Expiration
Not applicable · inactive patent

AI Technical Summary

Technical Problem

The existing overhead wire metal monitoring methods have a large workload and low identification efficiency. They cannot predict the stress status of cable metal in a timely and accurate manner, which can easily cause power grid failure.

Method used

The fault monitoring method based on grating array is adopted to obtain the grating array monitoring data near the metal tool for preprocessing, and the spectrum image recognition model based on residual neural network, Transformer network and cross-channel attention network are used to predict the global characteristics of the monitoring data. Extract local features to determine whether there is a fault in the instrument.

Benefits of technology

It reduces the workload of the fault monitoring process of overhead wire metal tools, improves the accuracy and efficiency of fault monitoring, reduces costs, and realizes timely identification and alarm of metal tools.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118914917A_ABST
    Figure CN118914917A_ABST
Patent Text Reader

Abstract

The invention relates to a power overhead line fitting fault monitoring method and device based on a grating array, and belongs to the technical field of optical fiber monitoring, the power overhead line fitting fault monitoring method based on the grating array comprises the following steps: obtaining monitoring data of the grating array in a preset range near a fitting, the monitoring data of the grating array are preprocessed; and inputting the preprocessed monitoring data of the grating array into a frequency spectrum image recognition model, and determining whether the hardware fitting has a fault or not based on an output result of the frequency spectrum image recognition model. According to the invention, while the fault monitoring of the electric power overhead line fitting is realized, the workload of the fault monitoring process of the electric power overhead line fitting is reduced, so that the cost of fault monitoring of the electric power overhead line fitting is reduced, and the accuracy of fault monitoring of the electric power overhead line fitting is improved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention relates to the technical field of optical fiber monitoring, and in particular to a method and a device for monitoring power overhead line hardware faults based on a grating array. Background Art

[0002] my country's smart grid is a new type of grid based on the coordinated development of UHV grid and various voltage levels. It integrates modern advanced sensor measurement technology, communication technology, information technology, computer technology and control technology with the physical grid. The most important front-end asset of the grid is the transmission cable, including optical fiber composite overhead line.

[0003] Fiber-optic composite overhead lines have a unique structure of optical fiber bundle tubes installed inside the conductors, so pre-twisted wire fittings must be used to protect the optical fiber during installation. During the actual use of fiber-optic composite overhead lines on transmission lines, if the fittings loosen and slip and the optical cable twisted wire fittings loose strands, it will seriously affect the safe and stable operation of the power system. Timely monitoring of the firmness of the fittings throughout the line is of great significance to disaster warning, troubleshooting and status assessment of the power grid infrastructure.

[0004] The existing monitoring of power overhead line hardware mainly relies on manual and drone line patrols, which has a large workload and low identification efficiency. It cannot timely and accurately predict the stress state of cable hardware in advance, which can easily cause serious power grid failures. Summary of the invention

[0005] In view of this, it is necessary to provide a method and device for monitoring power overhead line fittings faults based on a grating array, so as to solve the problems of large workload and low recognition efficiency in the existing power overhead line fittings monitoring methods.

[0006] In order to solve the above problems, the present invention provides a method for monitoring power overhead line fittings faults based on a grating array, comprising: Acquire monitoring data of the grating array within a preset range near the hardware, and pre-process the monitoring data of the grating array; The pre-processed monitoring data of the grating array is input into the spectrum image recognition model, and based on the output result of the spectrum image recognition model, it is determined whether the hardware has a fault; The spectral image recognition model is based on a residual neural network, a Transformer network and a cross-channel attention network, and the spectral image recognition model is trained based on historical monitoring data of labeled grating arrays.

[0007] In a possible implementation, the preprocessing of the monitoring data of the grating array includes: Performing high-pass filtering on the monitoring data of the grating array based on a preset frequency, and converting the monitoring data of the grating array after the high-pass filtering into time spectrum image data based on a short-time Fourier transform to obtain first spectrum image data; Performing spectrum image enhancement on the first spectrum image data to obtain second spectrum image data; performing data normalization processing on the second spectrum image data to obtain third spectrum image data; performing data expansion on the third spectrum image data to obtain fourth spectrum image data; Performing Gamma filtering correction on the fourth spectrum image data to obtain fifth spectrum image data; The fifth spectrum image data is subjected to data standardization to obtain preprocessed monitoring data of the grating array.

[0008] In a possible implementation, performing spectrum image enhancement on the first spectrum image data to obtain second spectrum image data includes: The first spectrum image data is enhanced based on the following formula to obtain the second spectrum image data:

[0009] in, represents the spectrum value of the data point in the second spectrum image data, Indicates the spectrum value of the data point in the first spectrum image data.

[0010] In a possible implementation manner, performing data expansion on the third spectrum image data to obtain fourth spectrum image data includes: Based on the bilinear interpolation method, data expansion is performed on the third spectrum image data to obtain fourth spectrum image data.

[0011] In a possible implementation, performing data normalization on the fifth spectrum image data to obtain preprocessed monitoring data of the grating array includes: The fifth spectrum image data is normalized based on the following formula to obtain the preprocessed monitoring data of the grating array:

[0012] in, represents the spectrum value of the data point in the monitoring data of the preprocessed grating array, represents the spectrum value of the data point in the fifth spectrum image data, represents the average value of all data points in the fifth spectrum image data, represents the standard deviation of all data points in the fifth spectrum image data.

[0013] In a possible implementation, the training of the spectrum image recognition model includes: Adjusting a first ratio of monitoring data in historical monitoring data of the labeled grating array so that the first ratio is greater than or equal to a preset ratio, and obtaining a training data set, wherein the first ratio is a ratio between monitoring data labeled as fault and monitoring data labeled as normal in the historical monitoring data of the labeled grating array; The training data set is input into the spectral image recognition model for multiple iterations, forward propagation and back propagation are performed in each iteration, and the model parameters of the spectral image recognition model are updated based on the Adam optimization algorithm.

[0014] In a possible implementation, determining whether a hardware has a fault based on an output result of the spectrum image recognition model includes: Obtain the hardware fault monitoring result of the drone inspection. When the hardware fault monitoring result of the drone inspection indicates that the hardware is faulty and the output result of the spectrum image recognition model indicates that the hardware is faulty, determine that the hardware is faulty.

[0015] The present invention also provides a power overhead line hardware fault monitoring device based on a grating array, comprising: An acquisition module is used to acquire monitoring data of the grating array within a preset range near the hardware, and pre-process the monitoring data of the grating array; A determination module, used to input the pre-processed monitoring data of the grating array into the spectrum image recognition model, and determine whether the hardware has a fault based on the output result of the spectrum image recognition model; The spectral image recognition model is based on a residual neural network, a Transformer network and a cross-channel attention network, and the spectral image recognition model is trained based on historical monitoring data of labeled grating arrays.

[0016] The present invention also provides an electronic device, including a memory and a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, the method for monitoring faults of power overhead line fittings based on a grating array as described above is implemented.

[0017] The present invention also provides a computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the method for monitoring faults of power overhead line fittings based on a grating array as described above is implemented.

[0018] The beneficial effects of the present invention are as follows: the method and device for monitoring the fault of power overhead line fittings based on a grating array provided by the present invention first obtain the monitoring data of the grating array within a preset range near the fittings to ensure the correlation between the monitoring data and the fittings, then pre-process the monitoring data to make the fault characteristics of the fittings easier to identify, and then extract the global and local features of the monitoring data through a spectrum image recognition model to improve the recognition accuracy and obtain the output result of the spectrum image recognition model, and finally determine whether the fittings have a fault based on the output result of the spectrum image recognition model to achieve fault monitoring of the power overhead line fittings. While achieving fault monitoring of the power overhead line fittings, the present invention reduces the workload of the power overhead line fitting fault monitoring process, thereby reducing the cost of power overhead line fitting fault monitoring and improving the accuracy of power overhead line fitting fault monitoring. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 A schematic diagram of a flow chart of an embodiment of a method for monitoring power overhead line fittings faults based on a grating array provided by the present invention; Figure 2 A schematic structural diagram of an embodiment of an optical fiber composite power overhead line and a double-layer hardware fixing structure with a built-in grating array sensing optical fiber provided by the present invention; Figure 3 A schematic diagram of a signal waveform of an embodiment of a typical characteristic signal of a sudden loosening of a hardware provided by the present invention; Figure 4 A schematic structural diagram of an embodiment of a power overhead line fittings fault monitoring device based on a grating array provided by the present invention; Figure 5 A schematic structural diagram of an electronic device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0020] The preferred embodiments of the present invention are described in detail below in conjunction with the accompanying drawings, wherein the accompanying drawings constitute a part of this application and are used together with the embodiments of the present invention to illustrate the principles of the present invention, but are not used to limit the scope of the present invention.

[0021] In the description of the present invention, the terms "first" and "second" are used for descriptive purposes only and cannot be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first" and "second" may explicitly or implicitly include at least one of the features. In addition, the meaning of "plurality" is at least two, such as two, three, etc., unless otherwise clearly and specifically defined.

[0022] In the description of the present invention, reference to an "embodiment" means that a particular feature, structure, or characteristic described in conjunction with the embodiment may be included in at least one embodiment of the present invention. The appearance of the phrase in various places in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive of other embodiments. It is explicitly and implicitly understood by those skilled in the art that the described embodiments may be combined with other embodiments.

[0023] my country's smart grid is a new type of grid based on a strong grid with UHV grid as the backbone and coordinated development of grids of various voltage levels. It integrates modern advanced sensor measurement technology, communication technology, information technology, computer technology and control technology with the physical grid. The most important front-end asset of the grid is the transmission cable, including the optical fiber composite overhead line. The optical fiber composite overhead line mainly includes two types: optical power ground wire (OPGW) and optical phase conductor (OPPC). OPGW optical cable is a special power optical cable that integrates overhead ground wire and communication optical cable. It is installed at the same time as the transmission line and is one of the pillars for the safe and stable operation of the power communication system. OPPC is a new type of special power optical cable, which is an optical cable that combines optical fiber units in the phase line. It has the dual functions of phase line and communication. It is mainly used for voltage levels below 110kV, suburban distribution networks, and rural power grids. The key to the intelligent construction of the power grid is the intelligence of power cables, especially the power overhead lines implanted with sensors, which further reflects the advanced level of power grid intelligence.

[0024] Fiber-optic composite overhead lines have a unique structure of optical fiber bundle tubes installed inside the conductors, so pre-twisted wire fittings must be used to protect the optical fiber during installation. The use of pre-twisted wire fittings has three advantages: first, the construction is simple and quick, and there is no need to drag heavy compressors and crimping pliers to the site, which improves labor efficiency and reduces physical labor; and pre-twisted wire fittings are good conductors with good electrical conductivity and significant energy-saving effects; third, the pre-twisted wire fittings are installed on the line, the contact surface between the line and the conductor is increased, the length is increased, and the force is evenly distributed, which reduces the fatigue of the conductor, extends the life of the conductor, and improves the shockproof ability. In the actual use of fiber-optic composite overhead lines on transmission lines, if the fittings are loose and slipped and the optical cable twisted wire fittings are scattered, it will seriously affect the safe and stable operation of the power system. Timely monitoring of the firmness of the fittings across the line can be of great significance to disaster warning, troubleshooting and status assessment of the power grid infrastructure.

[0025] With the development of fiber optic sensing technology and the advancement of fiber optic composite overhead line technology, the application of new fiber optic sensing technology to power overhead lines is inevitable. Since it can not only utilize existing fiber optic communication links, but also avoid the interference of strong electromagnetic fields on general electromagnetic sensors to achieve distributed passive monitoring, fiber optic sensors should be the best choice for power transmission and communication line maintenance and fault alarm. The use of fiber optic sensors to achieve online cable monitoring is gradually becoming an important research direction for technological development. Although the Brillouin optical time domain reflectometry (BOTDR) technology in existing fiber optic sensing technology can achieve simultaneous measurement of temperature and strain, the existing communication optical fibers in the fiber optic composite overhead line cables have excess length problems, making it impossible for BOTDR technology to effectively monitor strain and effectively evaluate the stress state of hardware.

[0026] In response to the need for identifying loose and slipping strands of optical fiber composite power overhead line hardware, traditional methods mainly rely on manual inspections, drone video inspections, etc. The above methods mainly rely on manual and drone line inspections, which are labor-intensive and have low recognition efficiency. They cannot timely and accurately predict the stress state of cable hardware in advance, which can easily cause serious power grid failures. In response to this problem, the present invention proposes a method for identifying power overhead line hardware faults based on a grating array. The method uses a grating array sensor unit embedded in an optical unit to perceive the temperature, vibration, and strain information of the overhead line throughout the entire process, and can achieve all-weather cable status multi-parameter monitoring. For the hardware clamping position, it can extract temperature gradients, stress mutations, vibration frequencies and other features for the identification of abnormal behavior of hardware fixation. Compared with traditional methods, it has the technical advantages of accurate fault event positioning, fast response speed, and wide coverage, which meets the needs of smart grid construction.

[0027] The specific embodiments are described in detail below: A specific embodiment of the present invention discloses a method for monitoring power overhead line hardware faults based on a grating array, Figure 1 Come and see, Figure 1 The flowchart of an embodiment of a method for monitoring power overhead line fittings fault based on a grating array provided by the present invention includes step S101 and step S102, wherein: In step S101, the monitoring data of the grating array within a preset range near the hardware is acquired, and the monitoring data of the grating array is preprocessed; In step S102, the pre-processed monitoring data of the grating array is input into the spectrum image recognition model, and based on the output result of the spectrum image recognition model, it is determined whether the hardware has a fault; The spectral image recognition model is based on a residual neural network, a Transformer network and a cross-channel attention network, and the spectral image recognition model is trained based on historical monitoring data of labeled grating arrays.

[0028] During implementation, the monitoring data of the grating array within a preset range (for example, within 50 cm) near the hardware can be first obtained to ensure the correlation between the monitoring data and the hardware.

[0029] The monitoring data of the grating array can then be preprocessed to make the characteristics of the monitoring data more obvious and the characteristics of the hardware failure easier to identify.

[0030] After preprocessing the monitoring data of the grating array, the preprocessed monitoring data of the grating array can be input into the spectrum image recognition model, and then it is determined whether the hardware has a fault according to the output result of the spectrum image recognition model.

[0031] The spectrum image recognition model can be composed of a residual neural network, a Transformer network and a cross-channel attention network. The residual neural network and the Transformer network can extract global features, and the cross-channel attention network can extract local features. The spectrum image recognition model can extract global features and local features to improve recognition accuracy. In addition, the spectrum image recognition model can be trained based on historical monitoring data of labeled grating arrays.

[0032] The power overhead line fittings fault monitoring method based on grating array provided by the present invention can be applied to power overhead line fittings fault monitoring scenarios, and can also be applied to other overhead line fittings fault monitoring scenarios, and the present invention does not make specific limitations on this.

[0033] Compared with the prior art, the method for monitoring power overhead line fittings fault based on grating array provided in this embodiment first obtains monitoring data of the grating array within a preset range near the fittings to ensure the correlation between the monitoring data and the fittings, then pre-processes the monitoring data to make the fitting fault characteristics easier to identify, and then extracts the global and local features of the monitoring data through a spectrum image recognition model to improve the recognition accuracy and obtain the output result of the spectrum image recognition model, and finally determines whether the fittings have faults based on the output result of the spectrum image recognition model to realize fault monitoring of power overhead line fittings. While realizing fault monitoring of power overhead line fittings, the present invention reduces the workload of the power overhead line fitting fault monitoring process, thereby reducing the cost of power overhead line fitting fault monitoring and improving the accuracy of power overhead line fitting fault monitoring.

[0034] Exemplarily, the preprocessing of the monitoring data of the grating array includes: Performing high-pass filtering on the monitoring data of the grating array based on a preset frequency, and converting the monitoring data of the grating array after the high-pass filtering into time spectrum image data based on a short-time Fourier transform to obtain first spectrum image data; Performing spectrum image enhancement on the first spectrum image data to obtain second spectrum image data; performing data normalization processing on the second spectrum image data to obtain third spectrum image data; performing data expansion on the third spectrum image data to obtain fourth spectrum image data; Performing Gamma filtering correction on the fourth spectrum image data to obtain fifth spectrum image data; The fifth spectrum image data is subjected to data standardization to obtain preprocessed monitoring data of the grating array.

[0035] Specifically, when preprocessing the monitoring data of the grating array, the monitoring data of the grating array can first be high-pass filtered according to a preset frequency (for example, 1 Hz), and then the high-pass filtered monitoring data of the grating array can be converted into time spectrum image data through short-time Fourier transform to obtain first spectrum image data.

[0036] Then, the first spectrum image data may be subjected to spectrum image enhancement, data normalization, data expansion, Gamma filter correction and data standardization operations in sequence, and finally the pre-processed monitoring data of the grating array may be obtained.

[0037] When performing data normalization processing, a maximum and minimum value normalization algorithm may be used, or other normalization algorithms may be used, which is not specifically limited in the present invention.

[0038] Exemplarily, performing spectrum image enhancement on the first spectrum image data to obtain second spectrum image data includes: The first spectrum image data is enhanced based on the following formula to obtain the second spectrum image data:

[0039] in, represents the spectrum value of the data point in the second spectrum image data, Indicates the spectrum value of the data point in the first spectrum image data.

[0040] Specifically, when spectrum image enhancement is performed on the first spectrum image data, the spectrum image enhancement may be performed according to the above formula.

[0041] Exemplarily, performing data expansion on the third spectrum image data to obtain fourth spectrum image data includes: Based on the bilinear interpolation method, data expansion is performed on the third spectrum image data to obtain fourth spectrum image data.

[0042] Specifically, when data expansion is performed on the third spectrum image data, the third spectrum image data can be expanded by bilinear interpolation to obtain fourth spectrum image data. Data expansion is beneficial to convolution operations in the network, thereby reducing the amount of calculation.

[0043] Exemplarily, the step of normalizing the fifth spectrum image data to obtain pre-processed monitoring data of the grating array includes: The fifth spectrum image data is normalized based on the following formula to obtain the preprocessed monitoring data of the grating array:

[0044] in, represents the spectrum value of the data point in the monitoring data of the preprocessed grating array, represents the spectrum value of the data point in the fifth spectrum image data, represents the average value of all data points in the fifth spectrum image data, represents the standard deviation of all data points in the fifth spectrum image data.

[0045] Specifically, when data normalization is performed on the fifth spectrum image data, the data normalization may be performed using the above formula.

[0046] Exemplarily, the training of the spectrum image recognition model includes: Adjusting a first ratio of monitoring data in historical monitoring data of the labeled grating array so that the first ratio is greater than or equal to a preset ratio, and obtaining a training data set, wherein the first ratio is a ratio between monitoring data labeled as fault and monitoring data labeled as normal in the historical monitoring data of the labeled grating array; The training data set is input into the spectral image recognition model for multiple iterations, forward propagation and back propagation are performed in each iteration, and the model parameters of the spectral image recognition model are updated based on the Adam optimization algorithm.

[0047] Specifically, when training the spectral image recognition model, since the sample distribution in the original data is extremely uneven, the number of data under normal working conditions is much larger than the number of data under fault conditions. In order to improve the training effect of the model, the ratio between the number of data under fault conditions and the number of data under normal conditions in the historical monitoring data of the labeled grating array can be adjusted to be greater than the preset ratio.

[0048] Then the training data set can be input into the spectral image recognition model for multiple iterations, forward propagation and back propagation are performed in each iteration, and the model parameters of the spectral image recognition model are updated based on the Adam optimization algorithm until the model achieves the expected effect.

[0049] Exemplarily, determining whether a hardware has a fault based on the output result of the spectrum image recognition model includes: Obtain the hardware fault monitoring result of the drone inspection. When the hardware fault monitoring result of the drone inspection indicates that the hardware is faulty and the output result of the spectrum image recognition model indicates that the hardware is faulty, determine that the hardware is faulty.

[0050] Specifically, when determining whether a hardware failure exists based on the output results of the spectrum image recognition model, the hardware failure monitoring results of the drone inspection (or the hardware failure monitoring results of the manual inspection) can also be obtained, and then when the hardware failure monitoring results of the drone inspection indicate that the hardware failure exists, and the output results of the spectrum image recognition model indicate that the hardware failure exists, it is determined that the hardware failure exists. After determining that the hardware failure exists, an alarm prompt can also be issued.

[0051] The following is a specific application scenario to better illustrate the technical solution of the present invention: The present invention combines traditional methods and deep learning methods to monitor overhead line hardware failure events in real time, and alarms and locates points that are determined to be hardware failure conditions by both traditional methods and deep learning. This can effectively monitor hardware failure events that affect the safety of overhead lines in real time, stop them in time, and ensure the safety of overhead power lines. The monitoring system with the addition of deep learning methods effectively removes the alarm status of points that are often misreported by traditional methods, and improves the accuracy of alarms.

[0052] Combination Figure 2 Come and see, Figure 2 The schematic diagram of the structure of an embodiment of the optical fiber composite power overhead line and double-layer hardware fixing structure with built-in grating array sensing optical fiber provided by the present invention is as follows: Figure 2 As shown, a grating array and a sensing optical fiber can be set inside the power overhead line to monitor the operation of the hardware.

[0053] The specific process of power overhead line fittings fault monitoring is as follows: 1. Acquisition of multi-parameter sensing data of grating array near hardware.

[0054] The sensing data near the hardware collected by the fiber Bragg grating multi-parameter sensing fiber is obtained. The data is filtered by a 1Hz high-pass filter, and then the raw signal data is converted into time spectrum image data using short-time Fourier transform.

[0055] Split the spectrum data into several independent data according to a certain time length and sliding window size, and retain the frequency range with loose hardware characteristics. Label the data set and divide all data into two categories: normal working conditions and loose hardware. The time spectrum signal characteristics of normal working conditions are: there is no obvious convex signal in frequency; the characteristics of loose hardware conditions are: there is a step mutation signal on the stable signal. Figure 3 Come and see, Figure 3 A schematic diagram of a signal waveform of an embodiment of a typical characteristic signal of a sudden change in hardware loosening provided by the present invention.

[0056] 2. Spectrum image data preprocessing.

[0057] In order to better identify the straight line and curve features in the frequency spectrum of the hardware loosening signal, it is necessary to preprocess the data to remove the influence of noise and enhance useful information. The specific preprocessing steps are as follows: Spectrum enhancement processing: The original signal is enhanced by logarithmic transformation, so that the signals under different working conditions can show the characteristics of straight lines and curves on the image, which is conducive to recognition. The specific function is:

[0058] Data normalization: Normalize the data to between 0 and 1. The importance of each feature of the normalized data is more clearly compared, and the influence of each feature is relatively uniform, which is conducive to improving the model convergence speed, improving the generalization ability of the model and the stability of calculation. The normalization method is maximum and minimum value normalization, and the calculation formula is:

[0059] Data size expansion: The bilinear interpolation method is used for the data, expanding it from the original 144×310 to 256×256, which is beneficial to the convolution operation in the network, can reduce the amount of calculation and accelerate model training.

[0060] Gamma filter correction: Gamma filter correction is an image processing technology used to improve the contrast and brightness of an image and improve the quality of the image. Gamma filtering can reduce noise in the data and enhance the frequency signal in the data. It works by adjusting the gamma value of the image. To change the brightness and contrast of the image: when the gamma value is 1, the image is linear, when the gamma value is less than 1, the image will become darker, and when the gamma value is greater than 1, the image will become brighter. The specific calculation formula is:

[0061] Data standardization: Assign a minimum value to the 0 value in the data, and standardize the data to convert the data into a standard Gaussian distribution. The data value is between -1 and 1, with a mean of 0 and a standard deviation of 1. This can make the data better respond to the activation function, improve the expressiveness of the data, reduce gradient explosion and gradient disappearance, and facilitate model training. The standardization formula is:

[0062] 3. Spectral image recognition model training.

[0063] Model training includes the steps of preparing spectrum data, building a network model for spectrum image recognition, model training, and model testing.

[0064] Spectrum data preparation: Split the labeled hardware loose sample data, randomly shuffle the data, and split the data set into training set, validation set, and test set in a ratio of 8:1:1. Then encapsulate the original data and the labels corresponding to each data according to the above split. Since the original signal data sample distribution is extremely uneven, the number of normal working conditions is far greater than the number of alarm conditions such as hardware loosening. This unbalanced number of samples will greatly affect the training effect of the network model. Therefore, the training samples were re-selected to form a relatively balanced sub-data set for training to ensure that the amount of data for different working conditions is more balanced, thereby reducing the impact of unbalanced data on training results.

[0065] Network model construction for spectrum image recognition: The original task requirement is transformed from the recognition of raw vibration signals to the recognition of spectrum image data. Therefore, a deep neural network model combining multi-layer Transformer and cross-channel attention mechanism can effectively recognize the spectrum graph. It contains 4 modules, namely: multi-layer residual neural network, multi-scale Transformer module, cross-channel attention module and classification probability fusion module.

[0066] The multi-layer residual neural network can extract convolutional features at different levels of the spectrogram, such as low-level features such as color and texture, as well as higher-level semantic features such as frequency distribution, which helps the model extract multi-dimensional feature information of the spectrogram. The multi-scale Transformer module can learn the attention relationship between different positions and subspaces of the spectrogram, such as the correlation between different frequency bands, which can enhance the model's ability to learn the whole picture. The cross-channel attention module cross-fuses the features extracted by different ResNet and Transformer layers to form a composite feature representation containing contextual relationships, which helps to extract the overall structural information of the spectrogram. The classification probability fusion module combines global visual features (ResNet features) and local features (cross-channel attention features) to generate the final classification decision features, which is conducive to considering the overall and local detail information of the spectrogram at the same time and improving the recognition accuracy.

[0067] Model training: Input the training sample data set into the network model for iterative training. Each training iteration includes two steps: forward propagation and back propagation. Forward propagation: Input a batch of spectrum samples into the network model and obtain the output results through forward calculation. Back propagation: Calculate the classification cross entropy loss function and calculate the gradient of the weight parameters according to the loss. Parameter update: Use the Adam optimization algorithm to update the parameters according to the gradient of the loss function to achieve small batch stochastic gradient descent training.

[0068] The following adjustment strategies were adopted during the training process: the number of iterations was increased from 100 to 300 to ensure that the model converged in identifying spectral features; the learning rate was adjusted to adopt a gradually decreasing or increasing strategy to find the best learning efficiency of the network in learning spectral data features; the batch size was tuned to balance the computational efficiency and convergence performance of spectral samples; and the dropout layer was added to improve the generalization ability of the model. Through continuous iteration and parameter updates, a network model with high classification accuracy was finally obtained.

[0069] Model testing: The optimal network model obtained through training was applied to the test set of spectrum data for evaluation. The overall classification accuracy of the test results was 97.69%, the false alarm rate was 1.85%, and the missed alarm rate was 1.06%, indicating that the model can effectively identify most alarm events with high recognition accuracy.

[0070] 4. Identification of abnormal events of loosening of hardware.

[0071] The model is deployed on the online monitoring platform, and the traditional method and deep learning method are combined to monitor and alarm the abnormal events of loose overhead line hardware. The specific strategy is: the traditional monitoring algorithm is jointly operated with the deep learning algorithm, and the judgment results of the two methods are obtained at the same time; when the traditional algorithm detects the alarm level, the deep learning algorithm is triggered to classify the data in the signal window; only when both the traditional algorithm and the deep learning algorithm are judged as the alarm category, the event is considered to be a real alarm and an alarm command is sent.

[0072] The present invention utilizes the grating array multi-parameter sensing optical fiber contained in the optical unit in the overhead line near the hardware to obtain characteristic information such as temperature gradient, stress mutation and vibration frequency. Combined with the deep learning method, it can accurately monitor the loosening, slipping and stranding events of the hardware in real time and send an alarm.

[0073] The embodiment of the present invention also provides a power overhead line hardware fault monitoring device based on a grating array, Figure 4 Come and see, Figure 4 The schematic diagram of the structure of an embodiment of a power overhead line hardware fault monitoring device based on a grating array provided by the present invention is as follows. The power overhead line hardware fault monitoring device based on a grating array 400 comprises: An acquisition module is used to acquire monitoring data of the grating array within a preset range near the hardware, and pre-process the monitoring data of the grating array; A determination module, used to input the pre-processed monitoring data of the grating array into the spectrum image recognition model, and determine whether the hardware has a fault based on the output result of the spectrum image recognition model; The spectral image recognition model is based on a residual neural network, a Transformer network and a cross-channel attention network, and the spectral image recognition model is trained based on historical monitoring data of labeled grating arrays.

[0074] The specific implementation methods of each module of the power overhead line fittings fault monitoring device based on grating array can be found in the description of the above-mentioned power overhead line fittings fault monitoring method based on grating array, and have similar beneficial effects, which will not be repeated here.

[0075] It should be noted that the power overhead line fittings fault monitoring device based on the grating array can be set in a power dispatching control center or in power monitoring stations in various places, and the present invention does not make specific limitations on this.

[0076] The embodiment of the present invention further provides an electronic device, Figure 5 Come and see, Figure 5This is a structural diagram of an embodiment of an electronic device provided by the present invention. The electronic device 500 includes a processor 501, a memory 502, and a computer program stored in the memory 502 and executable on the processor 501. When the processor 501 executes the program, the method for monitoring power overhead line fittings faults based on a grating array as described above is implemented.

[0077] As a preferred embodiment, the electronic device 500 further includes a display 503 for displaying that the processor 501 executes the above-mentioned method for monitoring faults of power overhead line fittings based on grating array.

[0078] Exemplarily, the computer program may be divided into one or more modules / units, one or more modules / units are stored in the memory 502, and are executed by the processor 501 to complete the present invention. One or more modules / units may be a series of computer program instruction segments that can complete specific functions, and the instruction segments are used to describe the execution process of the computer program in the electronic device 500. For example, the computer program may be divided into the acquisition module 401 and the determination module 402 in the above embodiment, and the specific functions of each module are as described above, and are not repeated here one by one.

[0079] The electronic device 500 may be a desktop computer, a notebook, a PDA or a smart phone with an adjustable camera module.

[0080] Among them, the processor 501 may be an integrated circuit chip with signal processing capabilities. The above-mentioned processor 501 may be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it may also be a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field-programmable gate array (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components. The methods, steps and logic block diagrams disclosed in the embodiments of the present invention can be implemented or executed. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.

[0081] The memory 502 may be, but is not limited to, a random access memory (RAM), a read only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable read-only memory (EEPROM), etc. The memory 502 is used to store programs, and the processor 501 executes the program after receiving an execution instruction. The process definition method disclosed in any of the above embodiments of the present invention may be applied to the processor 501, or implemented by the processor 501.

[0082] The display 503 may be an LCD display screen or an LED display screen, for example, a display screen on a mobile phone.

[0083] Understandably, Figure 5 The structure shown is only a schematic diagram of the structure of the electronic device 500. The electronic device 500 may also include Figure 5 More or fewer components as shown. Figure 5 Each component shown in the figure can be implemented by hardware, software or a combination thereof.

[0084] The electronic device provided according to the above-mentioned embodiment of the present invention can be implemented with reference to the specific description of the method for monitoring power overhead line fittings faults based on a grating array as described above according to the present invention, and has similar beneficial effects as the method for monitoring power overhead line fittings faults based on a grating array as described above, which will not be repeated here.

[0085] An embodiment of the present invention further provides a computer-readable storage medium on which a computer program is stored. When the program is executed by a processor, the above-mentioned method for monitoring faults of power overhead line fittings based on a grating array is implemented.

[0086] Generally speaking, the computer instructions for implementing the method of the present invention can be carried by any combination of one or more computer-readable storage media. Non-transitory computer-readable storage media can include any computer-readable media except for the signal itself that is temporarily propagating.

[0087] Computer-readable storage media may be, for example, but not limited to, electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatuses, or devices, or any combination thereof. More specific examples of computer-readable storage media (a non-exhaustive list) include: an electrical connection having one or more conductors, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In this document, a computer-readable storage medium may be any tangible medium that contains or stores a program that may be used by or in conjunction with an instruction execution system, apparatus, or device.

[0088] Computer program codes for performing the operations of the present invention may be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages, such as Java, Smalltalk, C++, and conventional procedural programming languages, such as "C" or similar programming languages, in particular, Python suitable for neural network computing and platform frameworks based on TensorFlow, PyTorch, etc. may be used. The program code may be executed entirely on the user's computer, partially on the user's computer, as an independent software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer via any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0089] Those skilled in the art will appreciate that all or part of the processes of the above-mentioned embodiments can be implemented by instructing related hardware through a computer program, and the program can be stored in a computer-readable storage medium, wherein the computer-readable storage medium is a disk, an optical disk, a read-only storage memory, or a random access memory, etc.

[0090] The invention discloses a method and device for monitoring the fault of electric power overhead line hardware based on a grating array. First, monitoring data of the grating array within a preset range near the hardware is obtained to ensure the correlation between the monitoring data and the hardware. Then, the monitoring data is preprocessed to make the hardware fault characteristics easier to be identified. Then, the global characteristics and local characteristics of the monitoring data are extracted through a spectrum image recognition model to improve the recognition accuracy and obtain the output result of the spectrum image recognition model. Finally, it is determined whether the hardware is faulty according to the output result of the spectrum image recognition model to realize the fault monitoring of the electric power overhead line hardware. While realizing the fault monitoring of the electric power overhead line hardware, the invention reduces the workload of the electric power overhead line hardware fault monitoring process, thereby reducing the cost of the electric power overhead line hardware fault monitoring and improving the accuracy of the electric power overhead line hardware fault monitoring.

[0091] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by any technician familiar with the technical field within the technical scope disclosed by the present invention should be covered within the protection scope of the present invention.

Claims

1. A method for monitoring power overhead line hardware faults based on a grating array, characterized in that: include: Acquire monitoring data of the grating array within a preset range near the hardware, and pre-process the monitoring data of the grating array; The pre-processed monitoring data of the grating array is input into the spectrum image recognition model, and based on the output result of the spectrum image recognition model, it is determined whether the hardware has a fault; The spectral image recognition model is based on a residual neural network, a Transformer network and a cross-channel attention network, and the spectral image recognition model is trained based on historical monitoring data of labeled grating arrays.

2. The method for monitoring power overhead line fittings faults based on grating array according to claim 1, characterized in that: The preprocessing of the monitoring data of the grating array includes: Performing high-pass filtering on the monitoring data of the grating array based on a preset frequency, and converting the monitoring data of the grating array after the high-pass filtering into time spectrum image data based on a short-time Fourier transform to obtain first spectrum image data; Performing spectrum image enhancement on the first spectrum image data to obtain second spectrum image data; performing data normalization processing on the second spectrum image data to obtain third spectrum image data; performing data expansion on the third spectrum image data to obtain fourth spectrum image data; Performing Gamma filtering correction on the fourth spectrum image data to obtain fifth spectrum image data; The fifth spectrum image data is subjected to data standardization to obtain preprocessed monitoring data of the grating array.

3. The method for monitoring power overhead line fittings fault based on grating array according to claim 2, characterized in that: The step of performing spectrum image enhancement on the first spectrum image data to obtain second spectrum image data includes: The first spectrum image data is enhanced based on the following formula to obtain the second spectrum image data: in, represents the spectrum value of the data point in the second spectrum image data, Indicates the spectrum value of the data point in the first spectrum image data.

4. The method for monitoring power overhead line fittings fault based on grating array according to claim 2, characterized in that: The step of performing data expansion on the third spectrum image data to obtain fourth spectrum image data includes: Based on the bilinear interpolation method, data expansion is performed on the third spectrum image data to obtain fourth spectrum image data.

5. The method for monitoring power overhead line fittings fault based on grating array according to claim 2, characterized in that: The step of normalizing the fifth spectrum image data to obtain pre-processed monitoring data of the grating array includes: The fifth spectrum image data is normalized based on the following formula to obtain the preprocessed monitoring data of the grating array: in, represents the spectrum value of the data point in the monitoring data of the preprocessed grating array, represents the spectrum value of the data point in the fifth spectrum image data, represents the average value of all data points in the fifth spectrum image data, represents the standard deviation of all data points in the fifth spectrum image data.

6. The method for monitoring power overhead line fittings fault based on grating array according to claim 1, characterized in that: The training of the spectrum image recognition model includes: Adjusting a first ratio of monitoring data in historical monitoring data of the labeled grating array so that the first ratio is greater than or equal to a preset ratio, and obtaining a training data set, wherein the first ratio is a ratio between monitoring data labeled as fault and monitoring data labeled as normal in the historical monitoring data of the labeled grating array; The training data set is input into the spectral image recognition model for multiple iterations, forward propagation and back propagation are performed in each iteration, and the model parameters of the spectral image recognition model are updated based on the Adam optimization algorithm.

7. The method for monitoring power overhead line fittings faults based on a grating array according to any one of claims 1 to 6, characterized in that: Based on the output results of the spectrum image recognition model, determine whether the hardware is faulty, including: Obtain the hardware fault monitoring result of the drone inspection. When the hardware fault monitoring result of the drone inspection indicates that the hardware is faulty and the output result of the spectrum image recognition model indicates that the hardware is faulty, determine that the hardware is faulty.

8. A power overhead line hardware fault monitoring device based on a grating array, characterized in that: include: An acquisition module is used to acquire monitoring data of the grating array within a preset range near the hardware, and pre-process the monitoring data of the grating array; A determination module, used to input the pre-processed monitoring data of the grating array into the spectrum image recognition model, and determine whether the hardware has a fault based on the output result of the spectrum image recognition model; The spectral image recognition model is based on a residual neural network, a Transformer network and a cross-channel attention network, and the spectral image recognition model is trained based on historical monitoring data of labeled grating arrays.

9. An electronic device, characterized in that: The invention comprises a memory and a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, the method for monitoring faults of power overhead line fittings based on a grating array according to any one of claims 1 to 7 is implemented.

10. A computer-readable storage medium, characterized in that: A computer program is stored thereon, and when the computer program is executed by a processor, the method for monitoring power overhead line fittings faults based on a grating array as claimed in any one of claims 1 to 7 is implemented.

Citation Information

Patent Citations

  • Multi-parameter optical fiber sensing detection method and system of leading-down optical cable

    CN111024283A

  • Automatic alarm device and method for cracking of electric power fitting string

    CN116110205A

  • Pipeline leakage positioning method based on frequency spectrum weighting and residual convolutional neural network

    CN117633588A

  • Subway external intrusion event identification method and device, electronic equipment and storage medium

    CN117710949A