Optical Fiber Grating Sensor Fault Location Method and System

Abstract

This application relates to the technical field of fault location. Specifically, it discloses a method and system for fault location of fiber Bragg grating sensors. First, it obtains the first signal data of the fiber Bragg grating sensors. After performing empirical mode decomposition on the first signal data, it further introduces a signal processing technology based on deep learning to extract features from each sensor signal component, and conducts feature selection based on the feature distribution guidance of each sensor signal component to filter out noise and redundant information, realize the efficient expression of key features of the signal, and on this basis, conduct intelligent identification of the fault types of fiber Bragg grating sensors. In this way, noise interference can be effectively avoided, thereby improving the accuracy and stability of fault location of fiber Bragg grating sensors.

Description

Technical Field

[0001] This application relates to the technical field of fault location, and more specifically, to a method and system for fault location of fiber Bragg grating sensors. Background Art

[0002] As an advanced optical measurement device, fiber Bragg grating sensors have been widely used in the fields of structural health monitoring, communication networks, oil and gas pipeline monitoring, etc. due to their high precision, long-distance monitoring ability, and good adaptability to harsh environments. However, due to working in a complex and changeable environment for a long time, fiber Bragg grating sensors are easily affected by various factors, such as temperature fluctuations, mechanical vibrations, electromagnetic interference, etc., resulting in a decline in sensor performance or failures. Therefore, fault location is a key link to ensure the stable operation and timely maintenance of fiber Bragg grating sensors.

[0003] In the prior art, the invention patent with the publication number CN118035855A discloses a method for fault location of fiber Bragg grating sensors. It processes signal data through a preset signal decomposition algorithm to obtain sensor data components, and then encodes the fault categories based on the signal characteristics of each sensor data component to construct a feature sample set, so as to train a preset machine learning model. Thus, the target fault location model after training is used to analyze the data to be analyzed of the sensor, and the fault location result is obtained.

[0004] Although the above method improves the automation degree and accuracy of fiber Bragg grating sensor fault recognition to a certain extent, there are still some challenges in practical applications. For example, when the signal decomposition algorithm processes sensor signals, it will decompose the noise and interference components in the signal together, resulting in the sensor data components obtained containing more redundant information. And in the prior art, mainly relying on statistics or simple mathematical transformations for feature extraction, it is difficult to capture the deep correlation and subtle differences between different signal components, which may lead to the omission of key features or the interference of noise and redundant information, introducing unnecessary errors, and thus reducing the accuracy and stability of fault location.

[0005] Therefore, an optimized method and system for fault location of fiber Bragg grating sensors are expected. Summary of the Invention

[0006] To solve the above technical problems, the present application is proposed. Embodiments of the present application provide a method and system for fault location of fiber Bragg grating sensors. First, first signal data of the fiber Bragg grating sensors is acquired. After performing empirical mode decomposition on the first signal data, signal processing technology based on deep learning is further introduced to extract features of each sensor signal component, and feature selection is performed based on the feature distribution guidance of each sensor signal component to filter out noise and redundant information, realize efficient expression of key features of the signal, and on this basis, perform intelligent identification of the fault types of the fiber Bragg grating sensors. In this way, noise interference can be effectively avoided, thereby improving the accuracy and stability of the fault location of the fiber Bragg grating sensors.

[0007] According to one aspect of the present application, a method for fault location of fiber Bragg grating sensors is provided, which includes:

[0008] Acquire first signal data of the fiber Bragg grating sensors;

[0009] Perform signal decomposition on the first signal data based on the empirical mode decomposition algorithm to obtain a plurality of sensor data components;

[0010] Extract features of each sensor data component in the plurality of sensor data components to obtain a set of encoded feature vectors of the sensor signal components;

[0011] Perform feature selection based on feature distribution guidance on the set of encoded feature vectors of the sensor signal components to obtain a set of sparsified encoded feature vectors of the sensor signal components;

[0012] Based on the set of sparsified encoded feature vectors of the sensor signal components, determine the fault type of the fiber Bragg grating sensors.

[0013] According to another aspect of the present application, a system for fault location of fiber Bragg grating sensors is provided, which includes:

[0014] A first signal data acquisition module for acquiring first signal data of the fiber Bragg grating sensors;

[0015] A signal data decomposition module for performing signal decomposition on the first signal data based on the empirical mode decomposition algorithm to obtain a plurality of sensor data components;

[0016] A feature extraction module for extracting features of each sensor data component in the plurality of sensor data components to obtain a set of encoded feature vectors of the sensor signal components;

[0017] A feature selection module, configured to perform feature selection based on feature distribution guidance on a set of encoded feature vectors of the sensor signal components to obtain a set of sparsified encoded feature vectors of the sensor signal components;

[0018] A fault type identification module, configured to determine the fault type of the fiber Bragg grating sensor based on the set of sparsified encoded feature vectors of the sensor signal components.

[0019] Compared with the prior art, the fiber Bragg grating sensor fault location method and system provided by the present application first obtain the first signal data of the fiber Bragg grating sensor. After performing empirical mode decomposition on the first signal data, a signal processing technology based on deep learning is further introduced to extract features of each sensor signal component, and feature selection is performed based on the feature distribution guidance of each sensor signal component to filter out noise and redundant information, realize the efficient expression of key features of the signal, and on this basis, perform intelligent identification of the fault type of the fiber Bragg grating sensor. In this way, noise interference can be effectively avoided, thereby improving the accuracy and stability of fiber Bragg grating sensor fault location. Description of the Drawings

[0020] By describing the embodiments of the present application in more detail in conjunction with the drawings, the above and other objects, features, and advantages of the present application will become more obvious. The drawings are used to provide a further understanding of the embodiments of the present application, and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation to the present application. In the drawings, the same reference numerals generally represent the same components or steps.

[0021] Figure 1 It is a flowchart of the fiber Bragg grating sensor fault location method according to an embodiment of the present application.

[0022] Figure 2 It is a schematic diagram of data flow of the fiber Bragg grating sensor fault location method according to an embodiment of the present application.

[0023] Figure 3 It is a flowchart of sub-step S4 of the fiber Bragg grating sensor fault location method according to an embodiment of the present application.

[0024] Figure 4 It is a flowchart of sub-step S41 of the fiber Bragg grating sensor fault location method according to an embodiment of the present application.

[0025] Figure 5 It is a flowchart of sub-step S42 of the fiber Bragg grating sensor fault location method according to an embodiment of the present application.

[0026] Figure 6 It is a flowchart of sub-step S5 of the fiber Bragg grating sensor fault location method according to an embodiment of the present application.

[0027] Figure 7 It is a block diagram of a fiber Bragg grating sensor fault location system according to an embodiment of the present application. Detailed implementation manners

[0028] As shown in the present application and the claims, unless the context clearly indicates an exception, words such as "a", "an", "one", and / or "the" are not specifically singular and may also include the plural. Generally speaking, the terms "comprising" and "including" only indicate the inclusion of the clearly identified steps and elements, and these steps and elements do not constitute an exclusive list. The method or device may also include other steps or elements.

[0029] Although the present application makes various references to certain modules in the system according to the embodiments of the present application, however, any number of different modules can be used and run on the user terminal and / or the server. The modules are only illustrative, and different aspects of the system and method can use different modules.

[0030] Flowcharts are used in the present application to illustrate the operations performed by the system according to the embodiments of the present application. It should be understood that the operations before or below do not necessarily need to be executed precisely in sequence. On the contrary, according to needs, various steps can be executed in reverse order or simultaneously. At the same time, other operations can also be added to these processes, or one or several steps can be removed from these processes.

[0031] Next, exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments of the present application. It should be understood that the present application is not limited by the exemplary embodiments described here.

[0032] It is worth noting that in the present application, all actions of obtaining data are carried out on the premise of complying with the corresponding data protection regulations and policies of the country where it is located and obtaining the authorization given by the owner of the corresponding device.

[0033] As mentioned in the above background art, Patent CN118035855A proposes a fiber Bragg grating sensor fault location method, which processes signal data through a preset signal decomposition algorithm to obtain sensor data components, and then encodes the fault categories based on the signal characteristics of each sensor data component to construct a feature sample set, so as to train a preset machine learning model, and thus analyze the data to be analyzed of the sensor according to the trained target fault location model to obtain the fault location result.

[0034] Although the above methods have improved the automation and accuracy of fiber Bragg grating sensor fault identification to a certain extent, there are still some challenges in practical applications. For example, while signal decomposition techniques analyze sensing data, they also break down accompanying noise and interference factors, resulting in a large amount of redundant information in the final data components. Existing technologies mainly rely on statistical analysis or basic mathematical transformations to extract features, making it difficult to capture the complex relationships and subtle differences between individual signal components. This may lead to the neglect of key features or being affected by noise and redundant data, causing additional misjudgments and reducing the accuracy and reliability of fault location.

[0035] To address the above technical problems, this application proposes an optimized fiber Bragg grating sensor fault location method. First, it obtains the first signal data of the fiber Bragg grating sensor. After performing empirical mode decomposition on the first signal data, it further introduces a deep learning-based signal processing technology to extract features from each sensor signal component, and conducts feature selection based on the feature distribution orientation of each sensor signal component to filter out noise and redundant information, achieving an efficient expression of the key features of the signal. Based on this, it conducts intelligent identification of the fault types of the fiber Bragg grating sensor. In this way, noise interference can be effectively avoided, thereby improving the accuracy and stability of fiber Bragg grating sensor fault location.

[0036] Figure 1 It is a flowchart of the fiber Bragg grating sensor fault location method according to an embodiment of this application. Figure 2 It is a schematic diagram of data flow of the fiber Bragg grating sensor fault location method according to an embodiment of this application. As Figure 1 and Figure 2 shown, the fiber Bragg grating sensor fault location method includes the steps: S1, obtaining the first signal data of the fiber Bragg grating sensor; S2, performing signal decomposition on the first signal data based on the empirical mode decomposition algorithm to obtain multiple sensor data components; S3, extracting features from each of the multiple sensor data components to obtain a set of sensor signal component coding feature vectors; S4, performing feature selection based on the feature distribution orientation on the set of sensor signal component coding feature vectors to obtain a set of sparsified sensor signal component coding feature vectors; S5, determining the fault type of the fiber Bragg grating sensor based on the set of sparsified sensor signal component coding feature vectors.

[0037] In the above method for fault location of fiber Bragg grating sensors, in step S1, first signal data of the fiber Bragg grating sensors is obtained. It should be understood that fiber Bragg grating sensors work based on the grating principle. When the physical quantity to be measured externally changes, it will cause changes in the light propagation path or phase inside the optical fiber, and then change the light interference pattern. The fiber Bragg grating sensors convert the physical quantity information into changes in parameters such as the intensity and wavelength of the optical signal by analyzing the interference light, so as to realize the measurement of the physical quantity to be measured. In practical applications, fiber Bragg grating sensors will be affected by various factors, such as environmental changes, structural deformations, equipment aging, etc., which may cause problems such as chirping phenomena, damage to the fiber head, base solder joint detachment, and fiber optic cable breakage in the sensors, thus causing changes in the measurement data. Therefore, this application expects to capture abnormal changes in the physical quantity to be measured by analyzing the first signal data collected by the fiber Bragg grating sensors, so as to accurately identify the fault types of the fiber Bragg grating sensors.

[0038] Specifically, first in the hardware preparation stage, it must be ensured that the fiber Bragg grating sensors are correctly installed on the object to be monitored. For different application scenarios, the installation methods will also be different. For example, in structural health monitoring, the sensors may need to be embedded inside a bridge or a building; while in pipeline monitoring, they are laid along the pipeline. In either case, care must be taken to avoid applying external pulling or squeezing forces to the sensors, so as not to affect their sensitivity and accuracy. In addition, the fiber Bragg grating sensors need to be connected to a reading device, such as a wavelength demodulator or a spectral analyzer, which can analyze the optical signals reflected by the fiber Bragg gratings and convert them into data that can be used for subsequent processing. Before the first use, the entire system needs to be calibrated to ensure the accuracy of the data. The calibration process usually includes using a standard light source and a reference sample with a known reflectivity for comparison and adjustment, so as to ensure the measurement accuracy of the system.

[0039] After the completion of the hardware preparation work, the next thing to focus on is the setting of the measurement environment. In order to ensure the consistency and reliability of the data, the influence of external factors on the measurement results should be minimized as much as possible. This includes maintaining stable temperature and humidity conditions, because temperature changes may cause wavelength drift in the fiber Bragg grating sensors, thereby affecting the measurement results. At the same time, strong electromagnetic interference sources should be avoided, and mechanical vibrations should be prevented from being transmitted to the sensors, because these factors will introduce additional noise and reduce the quality of the data. In addition, when working in some special environments, appropriate safety precautions need to be taken. For example, when working near a high-voltage electric field, relevant safety regulations should be followed; in a flammable and explosive environment, explosion-proof requirements need to be noted to ensure the safety of personnel and equipment.

[0040] When entering the data acquisition parameter setting stage, it is necessary to select an appropriate sampling frequency according to the characteristics of the monitored phenomenon. For rapidly changing processes, such as vibration monitoring, a higher sampling rate is necessary to capture transient events; while for slower changes, such as long-term structural health monitoring, a lower sampling rate can be adopted to save storage space and reduce computational load. In addition to the sampling frequency, it is also important to define the time length of each measurement (i.e., the time window). A longer time window helps to capture periodic or trending changes, but increases the data volume; while a shorter time window is more suitable for real-time monitoring of transient events. The setting of trigger conditions cannot be ignored either. For example, sometimes it is desired to start recording data only under specific conditions, and then the trigger conditions need to be set. The trigger conditions can be based on time (timed trigger), external signals (such as switch status), or other physical quantities (such as stress or strain reaching a certain threshold) to improve the effectiveness and pertinence of the data.

[0041] Finally, after everything is ready, the reading device can be started and the predetermined parameter configuration file can be loaded to initialize all components, check the status indicator of the system, and confirm normal operation. Then, the measurement process is started by pressing the "Start" button or sending a command through the software interface. At this time, the reading device will collect the reflected spectral information from the fiber Bragg grating sensor according to the predetermined parameters and convert it into a digital signal. Since the originally acquired data usually contains some noise and other unnecessary components, preliminary data cleaning is often required before saving or further processing. This may involve removing outliers, filtering, or applying other data preprocessing techniques to improve the quality of subsequent analysis. In addition, considering the uncertainties and environmental changes in practical applications, an adaptive data acquisition strategy can also be designed so that the system can automatically adjust parameters to a certain extent to meet the measurement requirements in different situations.

[0042] In the above method for fault location of fiber Bragg grating sensors, in step S2, the first signal data is decomposed based on the empirical mode decomposition algorithm to obtain multiple sensor data components. Specifically, in this application, considering that the signal data of fiber Bragg grating sensors is usually non-linear and non-stationary, containing various frequency components, noise interference, and abnormal fluctuations caused by sensor failures. Therefore, in order to effectively extract the key features in the sensor signal, this application uses the empirical mode decomposition algorithm to decompose the first signal data, so as to adaptively decompose the complex signal into multiple intrinsic mode function (IMF) components and a residual component, thereby obtaining multiple sensor data components. It should be understood that the empirical mode decomposition algorithm is based on the local characteristic time scale of the signal. By continuously finding the local maximum points and minimum points of the signal, constructing the upper envelope and the lower envelope, and calculating their mean curve, then subtracting the mean curve from the original signal to obtain a new signal, and repeating this process until a specific stop criterion is met, so as to decompose the signal into a series of IMF components with different frequencies. Among them, each IMF component represents the oscillation mode of the signal in different frequency bands. Through this decomposition process, the first signal data can be separated from the complex mixed state, enabling subsequent feature extraction to focus on each component with independent physical meaning, more effectively mining the fault feature information, avoiding the mutual influence between different frequency components and interference factors, and improving the accuracy of fault diagnosis.

[0043] Specifically, for fiber Bragg grating (FBG) sensors, their reflection spectra change with the changes of external physical quantities such as temperature and stress. However, there are often various interference factors in the actual monitoring environment, making the first signal data collected complex and non-stationary. Therefore, the EMD algorithm has become one of the effective tools for extracting useful information from such complex signals. Fiber Bragg grating sensors have been widely used in the fields of structural health monitoring, communication networks, etc. due to their high sensitivity, strong anti-electromagnetic interference ability, etc. However, due to being exposed to a complex natural environment for a long time, the signals received by fiber Bragg grating sensors will inevitably be affected by various noises and interferences, such as environmental noise (including external factors such as mechanical vibration and wind disturbance), temperature fluctuations (temperature changes will cause the expansion and contraction of the optical fiber itself, thereby affecting the position of the FBG reflection wavelength), electromagnetic interference (although the optical fiber itself has good anti-electromagnetic interference performance, interference phenomena may still occur in some special occasions), and multi-physical field coupling effects (when multiple physical quantities act on the same FBG at the same time, there may be mutual influence between the physical quantities, increasing the difficulty of signal analysis). In order to overcome the above challenges, advanced signal processing technologies need to be adopted to improve the data quality and analysis accuracy. Among them, the EMD algorithm has become an ideal choice due to its ability to automatically adapt to the characteristics of the input signal and effectively separate different frequency components.

[0044] Before performing EMD decomposition, it is usually necessary to perform necessary preprocessing on the original signal to remove the DC bias and reduce the interference of irrelevant information through methods such as smoothing filtering. For fiber Bragg grating sensors, this means ensuring that the acquired first signal data is as pure as possible so that useful features can be captured more accurately in subsequent steps. Next, enter the core part of EMD - constructing the IMF sequence. First, identify all the maximum and minimum positions of the signal within a given time period, and then fit a smooth upper envelope line and a smooth lower envelope line based on these extreme points respectively. By calculating the average value between the upper and lower envelope lines and subtracting it from the original signal, a new residual signal is obtained. Then, check whether the newly obtained residual signal already meets the definition of IMF. If it does, output it as an IMF; otherwise, repeat the above steps until the condition is met. Finally, subtract the current IMF from the original signal to obtain a new remaining part as the object for the next decomposition. This process is repeated continuously until the remaining part no longer contains valid IMF components, that is, it becomes a trend term or approaches a constant.

[0045] As the EMD decomposition progresses, the original complex signal is gradually stripped layer by layer, forming multiple different IMF components. Each IMF represents the oscillation behavior of the original signal at a specific time scale. These IMFs not only retain the main features of the original signal, but also, since each IMF corresponds to a different frequency band, can more clearly display the hidden information inside the signal. Especially for fiber Bragg grating sensors, which are extremely sensitive to environmental changes, the IMFs obtained through EMD decomposition can help better understand the specific contributions of different physical quantities to the sensor response, thus providing a solid foundation for further feature extraction and fault diagnosis.

[0046] In the above-mentioned fiber Bragg grating sensor fault location method, in step S3, feature extraction is performed on each of the multiple sensor data components to obtain a set of sensor signal component coding feature vectors. In a specific example of the present application, step S3 includes: using a signal feature extractor based on a one-dimensional convolutional layer to perform feature extraction on each of the multiple sensor data components respectively to obtain the set of sensor signal component coding feature vectors. Here, considering that each sensor data component represents an oscillation mode in a different frequency band decomposed from the original signal and reflects different fault feature information. Therefore, in order to accurately capture and efficiently express the features of each sensor data component, the present application introduces a signal processing technology based on deep learning, and uses a signal feature extractor based on a one-dimensional convolutional layer to perform feature learning on each sensor data component to automatically extract the deep features of each sensor data component and obtain the set of sensor signal component coding feature vectors. It should be understood that the one-dimensional convolutional layer has a powerful local feature extraction ability, which extracts the feature patterns of local regions by sliding the convolutional kernel in the time dimension of the sensor data component, such as local frequency changes, amplitude mutations, etc. With the stacking of convolutional layers and the learning and update of parameters, the signal feature extractor can gradually extract higher-level feature representations, transform the original sensor data components into more abstract and representative feature vectors, thereby providing a more discriminative feature representation for subsequent fault diagnosis and improving the accuracy of fault location.

[0047] In the above-mentioned fiber Bragg grating sensor fault location method, in step S4, feature selection based on feature distribution guidance is performed on the set of sensor signal component coding feature vectors to obtain a sparsified set of sensor signal component coding feature vectors. It should be understood that since the number of sensor data components obtained through signal decomposition is large and contains a certain degree of redundancy and correlation, directly using them for fault identification may increase the computational complexity and introduce unnecessary noise. Therefore, in order to improve the robustness and computational efficiency of feature representation, the present application further performs feature selection on the set of sensor signal component coding feature vectors, screens the set of sensor signal component coding feature vectors based on the feature correlation distribution between each sensor signal component coding feature vector, retains the feature representations that contribute more to fault identification and have a lower redundancy between them, and removes redundant and noise features, thereby realizing the dimensionality reduction and sparsification of the feature space to reduce the computational burden and improve the accuracy and speed of fault identification. Among them, Figure 3 is a flowchart of sub-step S4 of the fiber Bragg grating sensor fault location method according to an embodiment of the present application. As Figure 3As shown, step S4 includes steps: S41. Based on the feature distribution hub of the set of sensor signal component encoding feature vectors, calculate the fine-grained ablation factor of each sensor signal component encoding feature vector in the set of sensor signal component encoding feature vectors to obtain a set of sensor signal component fine-grained ablation factors; S42. Based on the set of sensor signal component fine-grained ablation factors, perform feature sparsification processing on the set of sensor signal component encoding feature vectors to obtain the set of sparsified sensor signal component encoding feature vectors.

[0048] Figure 4 FIG. is a flowchart of sub-step S41 of the fiber grating sensor fault location method according to an embodiment of the present application. As Figure 4 shown, step S41 includes steps: S411. Perform feature distribution hub search on the set of sensor signal component encoding feature vectors to obtain a sensor signal component feature distribution hub encoding vector; S412. Calculate the fine-grained ablation factor of each sensor signal component encoding feature vector in the set of sensor signal component encoding feature vectors relative to the sensor signal component feature distribution hub encoding vector to obtain the set of sensor signal component fine-grained ablation factors.

[0049] More specifically, in a specific example of the present application, step S411 includes: First, calculate the feature difference factor of each sensor signal component encoding feature vector in the set of sensor signal component encoding feature vectors relative to the set of sensor signal component encoding feature vectors to obtain a set of sensor signal component feature difference factors, which is expressed by the formula:

[0050]

[0051]

[0052] where represents the set of sensor signal component encoding feature vectors, , , , and respectively represent the first, second, th, th, th and th sensor signal component encoding feature vectors in the set of sensor signal component encoding feature vectors, represents the transpose of the vector, is the and the Covariance matrix between is the feature difference factor of the sensor signal component.

[0053] That is, by performing a feature distribution hub search on the set of encoded feature vectors of the sensor signal components, key feature points located at the center of the feature distribution or with significant discrimination are identified to characterize the main feature distribution trend of the first signal data. Specifically, during the search process, first, by calculating the average distance of each encoded feature vector of the sensor signal component relative to the overall feature of the set, the cooperation mode and interaction effect between different component features are understood, thereby revealing the connection and its hierarchical structure between each sensor data component, which helps to preliminarily evaluate the importance of each sensor data component feature. In this way, the essential characteristics of the original signal can be more accurately reflected, providing a solid foundation for subsequent analysis, effectively screening out the most representative features, reducing redundant information, and improving the efficiency and accuracy of the fault location model.

[0054] Then, the negative of each sensor signal component feature difference factor in the set of sensor signal component feature difference factors is taken and normalized based on the Softmax function to obtain the set of sensor signal component hub feature correlation factors; finally, using the set of sensor signal component hub feature correlation factors as the weight distribution, the set of encoded feature vectors of the sensor signal components is weighted and aggregated to obtain the sensor signal component feature distribution hub encoded vector, which is expressed by the formula:

[0055]

[0056] where represents the exponential operation with the natural constant as the base, represents the sensor signal component feature distribution hub encoded vector.

[0057] That is, first through normalization, the weight distribution of each sensor data component feature is determined, and a weighted aggregation method is used to construct the feature distribution hub of the set of encoded feature vectors of the sensor signal components, that is, the sensor signal component feature distribution hub encoded vector, which is used as an overall description and general expression of the core feature distribution of the first signal data, providing a key reference standard for subsequent feature selection and sparsification processing. In this way, it helps to extract the most representative information, thus laying a foundation for the next data analysis step.

[0058] More specifically, step S412 is expressed by the formula:

[0059]

[0060] Among them, represents the norm of the vector, represents the maximum operation, is a very small positive number used to prevent the denominator from being zero, represents the fine-grained ablation factor of the sensor signal component.

[0061] That is, based on the hub coding vector of the sensor signal component feature distribution, using contrast learning technology, calculate the fine-grained ablation factor between each sensor signal component coding feature vector and the hub coding vector of the sensor signal component feature distribution, and evaluate the influence degree of removing each sensor signal component feature on the overall expression of the hub feature. In this way, the semantic association strength between each sensor data component and the core feature of the first signal data can be revealed, thereby quantifying the importance of each sensor signal component coding feature vector.

[0062] Figure 5 is a flowchart of sub-step S42 of the fiber grating sensor fault location method according to an embodiment of the present application. As Figure 5 shown, the step S42 includes steps: S421, input each sensor signal component fine-grained ablation factor in the set of sensor signal component fine-grained ablation factors into a sparsification module based on a gating function to obtain a set of sparsified sensor signal component fine-grained ablation factors; S422, use the set of sparsified sensor signal component fine-grained ablation factors as a weight distribution to perform weighted modulation on the set of sensor signal component coding feature vectors to obtain the set of sparsified sensor signal component coding feature vectors.

[0063] More specifically, the step S421 is expressed by the formula:

[0064]

[0065] Among them, represents the gating function, represents the gating threshold, represents the sparsified sensor signal component fine-grained ablation factor.

[0066] More specifically, the step S422 is expressed by the formula:

[0067]

[0068] Among them, represents the corresponding sparsified sensor signal component coding feature vector, Represents a set of encoded feature vectors of the sparsified sensor signal components.

[0069] That is, further using a gating mechanism to screen each fine-grained ablation factor, and regarding the sensor signal component feature vectors corresponding to the lower fine-grained ablation factors as redundant or noise features to be filtered and removed, while retaining the sensor signal component features that have an important impact on the overall expression of the core features of the first signal data, so as to achieve effective sparsification processing of the set of the encoded feature vectors of the sensor signal components. In this way, the subsequent fault identification process can focus more on key features, thereby reducing the risks of false alarms and missed alarms and enhancing the reliability and stability of diagnosis. In addition, by sparsifying the feature space, the consumption of computing resources can also be reduced, and the execution efficiency of the algorithm can be improved.

[0070] In the above optical fiber grating sensor fault location method, in step S5, based on the set of the encoded feature vectors of the sparsified sensor signal components, the fault type of the optical fiber grating sensor is determined. Among them, Figure 6 is a flowchart of sub-step S5 of the optical fiber grating sensor fault location method according to an embodiment of the present application. As Figure 6 shown, step S5 includes steps: S51, performing feature splicing on the set of the encoded feature vectors of the sparsified sensor signal components to obtain a global feature encoding vector of the sensor signal; S52, inputting the global feature encoding vector of the sensor signal into the trained target fault location model to obtain a fault location result, and the fault location result is a fault type label of the optical fiber grating sensor.

[0071] Specifically, in step S51, the set of the encoded feature vectors of the sparsified sensor signal components is subjected to feature splicing to obtain a global feature encoding vector of the sensor signal. It should be understood that although the encoded feature vectors of the sensor signal components after feature selection each carry the signal features in different frequency bands of the first signal data, they lack integrity. Therefore, in order to achieve the global feature description of the first signal data, the present application further performs feature splicing on the set of the encoded feature vectors of the sparsified sensor signal components to integrate the scattered local feature representations into a global feature encoding vector of the sensor signal, comprehensively reflecting the overall features and fault patterns of the first signal data.

[0072] Specifically, in step S52, the global feature encoding vector of the sensor signal is input into the trained target fault location model to obtain a fault location result, which is the fault type label of the fiber Bragg grating sensor. In a specific example of the present application, the target fault location model is constructed based on the support vector machine (SVM). In the training stage, the target fault location model learns through a large number of known fault types and corresponding sensor feature data, and continuously adjusts its own parameters and structure to optimize the classification ability for different fault types. During the fault location process, the model matches and compares the input global feature encoding vector of the sensor signal with various fault patterns learned during the training process, and determines the fault type label that best matches the input features according to the maximum margin classification principle in the SVM, so as to achieve the fault location of the fiber Bragg grating sensor.

[0073] Preferably, when each sensor signal component encoding feature vector in the set of sensor signal component encoding feature vectors represents the sensor signal component feature, during the process of performing feature sparsification based on the feature distribution hub, it may cause valid information to be erroneously deleted, which results in insufficient encoding representation of the long-distance valid information of the global feature encoding vector of the sensor signal, thereby reducing the expression effect of the global feature encoding vector of the sensor signal relative to the trained target fault location model and affecting the accuracy of the fault location result obtained by inputting it into the trained target fault location model.

[0074] Therefore, in one example, when the global feature encoding vector of the sensor signal is input into the trained target fault location model, the global feature encoding vector of the sensor signal is optimized, and the optimization includes the following steps:

[0075] Parse the implicit parameter topology framework from the target fault location model;

[0076] Embed the global feature encoding vector of the sensor signal into the micro-manifold of the implicit parameter topology framework to generate a sensor signal mimic criterion mapping encoding;

[0077]

[0078] Wherein, represents the global feature encoding vector of the sensor signal, represents the implicit parameter topology framework, represents the sensor signal mimic criterion mapping encoding;

[0079] Apply a non-linear gain operation to the sensor signal mimic criterion mapping encoding to output a sensor signal scale convergence mapping encoding;

[0080]

[0081] Among them, represents the sensor signal scale convergence mapping encoding;

[0082] Generate a sensor signal multi-scale convergence coding tensor based on the sensor signal scale convergence mapping encoding and the sensor signal mimicry criterion mapping encoding;

[0083]

[0084] Among them, represents the transpose operation, represents the vector multiplication, represents the length of the sensor signal scale convergence mapping encoding, represents the sensor signal multi-scale convergence coding tensor;

[0085] Perform entropy regularization on the sensor signal multi-scale convergence coding tensor to generate a sensor signal steady-state retrieval architecture;

[0086]

[0087] Among them, represents the sensor signal steady-state retrieval architecture;

[0088] Reparameterize the sensor signal global feature encoding vector to the Riemannian space of the sensor signal steady-state retrieval architecture to obtain a sensor signal criterion incremental characterization;

[0089]

[0090] Among them, represents the sensor signal criterion incremental characterization;

[0091] By coupling the sensor signal global feature encoding vector and the sensor signal criterion incremental characterization, output an adjusted sensor signal global feature encoding vector;

[0092]

[0093] Among them, and represent weighted hyperparameters, represents the adjusted sensor signal global feature encoding vector. Finally, input the adjusted sensor signal global feature encoding vector into the trained target fault location model to obtain a fault location result.

[0094] Here, relying on the set of orthogonal quantities of the separating hyperplane anchored by the model's implicit cognitive graph primitives, the original feature vector is driven to evolve into a high-order separability space, and the correlation dimension with the category criterion is significantly enhanced through a nonlinear gain mapper. Then, a multi-scale convergence retrieval field of classification features is constructed as an information distillation field to capture incremental representations that are helpful for criterion decision-making, and a parameter sensitivity adaptation mechanism is used to fuse the incremental representations of the criterion. The optimized classification features are more suitable for the elastic scaling convergence of the implicit parameter topology of the target fault location model, improving the accuracy of the fault location results obtained by the target fault location model after input training.

[0095] After initially determining the type of fault in the fiber grating sensor, a more in-depth fault analysis is required. This step includes a physical inspection of the sensor and its related components to identify any visible damage or abnormalities. For example, check whether the sensor has obvious mechanical damage, or whether the connecting cable is broken. At the same time, combine historical data with currently collected data to try to find the root cause of the fault. By comparing the differences between the data characteristics during normal operation and the fault state, the specific manifestations and scope of the fault can be better understood. In addition, simulation tools can be used to simulate the behavior of sensors under different conditions to help engineers locate the problem more accurately.

[0096] After the fault type is identified, the next step is to evaluate the impact of the fault on the overall system performance. For some key application areas, such as aerospace, oil and gas pipeline monitoring, any minor fault may lead to serious consequences. Therefore, it is necessary to quantify the negative impact of the fault on monitoring accuracy, response time, etc., and judge whether immediate action is needed based on this. In addition, it is necessary to consider whether the fault will cause other potential problems to provide a basis for subsequent decision-making. For example, if a sensor is found to have a problem of decreased temperature sensitivity, it may affect the accuracy of the data that depends on the sensor in the entire monitoring network, and then affect all decisions made based on this data. Therefore, it is necessary to fully evaluate the impact of the fault in order to make the most appropriate response.

[0097] Based on the above analysis results, the next step is to develop a specific repair plan. This process usually includes selecting an appropriate repair method (such as hardware replacement, software update, parameter adjustment, etc.) and planning a timetable for implementation. If the failure is caused by external factors, such as performance degradation caused by environmental changes, then in addition to repairing the sensor itself, measures need to be taken to improve the installation environment to prevent similar problems from happening again. For internal failures, such as aging or failure of sensor components, it may be necessary to replace the entire sensor unit or specific components. At this stage, the cost-benefit ratio should also be considered to select the optimal solution that can solve the problem without causing unnecessary waste of resources.

[0098] After the repair plan is determined, it enters the execution stage. The execution stage includes both actual repair operations and related coordination work. For on-site operations, it is very important to ensure that there are enough technical support personnel on site. Technical support personnel can quickly and effectively handle various emergencies encountered. At the same time, in order not to affect normal business operations, it is often necessary to arrange repairs during off-peak hours and try to shorten the downtime. In addition, if there is collaboration between multiple departments, it is also essential to communicate and plan in advance to ensure that all parties can clearly understand their respective tasks and time nodes.

[0099] Finally, after all repair work is completed, a comprehensive performance verification must be carried out to ensure that the system has returned to the expected working state. This step can be carried out by recalibrating the sensors, collecting new data and comparing it with the previous benchmarks. Performance verification is not only a test of the repair work, but also a guarantee of the future system stability. By setting a series of strict test criteria, it can effectively detect whether there are residual problems or new hidden dangers. Only when all indicators meet the requirements can it be officially announced that the fault has been completely resolved and the system is put back into normal use.

[0100] In summary, the fiber Bragg grating sensor fault location method based on the embodiments of the present application is elucidated. It first obtains the first signal data of the fiber Bragg grating sensor. After performing empirical mode decomposition on the first signal data, it further introduces a signal processing technology based on deep learning to extract features from each sensor signal component, and performs feature selection based on the feature distribution orientation of each sensor signal component to filter out noise and redundant information, realize the efficient expression of the key features of the signal, and on this basis, perform intelligent identification of the fiber Bragg grating sensor fault type. In this way, noise interference can be effectively avoided, thereby improving the accuracy and stability of the fiber Bragg grating sensor fault location.

[0101] Furthermore, a fiber Bragg grating sensor fault location system is also provided.

[0102] Figure 7 is a block diagram of the fiber Bragg grating sensor fault location system according to the embodiments of the present application. As Figure 7As shown, the fiber Bragg grating sensor fault location system 100 according to an embodiment of the present application includes: a first signal data acquisition module 110 for acquiring first signal data of the fiber Bragg grating sensor; a signal data decomposition module 120 for performing signal decomposition on the first signal data based on an empirical mode decomposition algorithm to obtain a plurality of sensor data components; a feature extraction module 130 for extracting features from each of the plurality of sensor data components to obtain a set of sensor signal component coding feature vectors; a feature selection module 140 for performing feature distribution-guided feature selection on the set of sensor signal component coding feature vectors to obtain a sparsified set of sensor signal component coding feature vectors; and a fault type identification module 150 for determining the fault type of the fiber Bragg grating sensor based on the sparsified set of sensor signal component coding feature vectors.

[0103] Here, those skilled in the art can understand that the specific operations of the various modules in the above fiber Bragg grating sensor fault location system have been described in detail above with reference to Figures 1 to 6 the description of the fiber Bragg grating sensor fault location method, and thus, the repeated description thereof will be omitted.

[0104] The basic principles of the present invention have been described above in conjunction with specific embodiments. However, it should be noted that the advantages, advantages, effects, etc. mentioned in the present invention are only examples and not limitations, and it cannot be considered that these advantages, advantages, effects, etc. are essential for each embodiment of the present invention. In addition, the specific details of the above embodiments are only for the purposes of illustration and easy understanding, rather than limitations, and the above details do not limit the present invention to necessarily adopt the above specific details to implement.

[0105] In the above embodiments, the descriptions of the various embodiments have their own emphases. For parts not described in detail or recorded in a certain embodiment, reference may be made to the relevant descriptions of other embodiments. In the several embodiments provided by the present invention, it should be understood that the disclosed systems and methods can be implemented in other ways. For example, the system embodiments described above are only illustrative. For example, the unit division is only a logical function division, and there may be other division methods in actual implementation. The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed to multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0106] For those skilled in the art, it is obvious that the present invention is not limited to the details of the above-described exemplary embodiments, and the present invention can be implemented in other specific forms without departing from the spirit or basic characteristics of the present invention. Therefore, from any point of view, the embodiments should be regarded as exemplary and non-limiting. The scope of the present invention is defined by the appended claims rather than the above description. Therefore, all changes falling within the meaning and scope of the equivalent elements of the claims are intended to be embraced by the present invention. Any reference signs in the claims should not be construed as limiting the claims involved.

[0107] In addition, it is obvious that the word "comprising" does not exclude other units or steps, and the singular does not exclude the plural. The multiple units stated in the system claims can also be implemented by one unit through software or hardware.

[0108] Finally, it should be noted that the above description has been given for purposes of illustration and description. In addition, the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A method for locating a fiber grating sensor fault, characterized in that: include: Acquiring first signal data of the fiber grating sensor; Performing signal decomposition based on an empirical mode decomposition algorithm on the first signal data to obtain a plurality of sensor data components; Performing feature extraction on each of the plurality of sensor data components to obtain a set of sensor signal component encoding feature vectors; Performing feature selection based on feature distribution guidance on the set of sensor signal component encoding feature vectors to obtain a sparse set of sensor signal component encoding feature vectors; Determining the fault type of the fiber grating sensor based on the set of sparsely encoded feature vectors of the sensor signal components; Performing feature selection based on feature distribution guidance on the set of sensor signal component encoding feature vectors to obtain a sparse set of sensor signal component encoding feature vectors, including: Calculating a characteristic difference factor of each sensor signal component encoding feature vector in the set of sensor signal component encoding feature vectors relative to the set of sensor signal component encoding feature vectors to obtain a set of sensor signal component characteristic difference factors; Taking the inverse of each sensor signal component feature difference factor in the set of sensor signal component feature difference factors and performing normalization processing based on a Softmax function to obtain a set of sensor signal component hub feature correlation factors; Using the set of sensor signal component hub feature correlation factors as weight distribution, weighted aggregation is performed on the set of sensor signal component encoding feature vectors to obtain the sensor signal component feature distribution hub encoding vector; Respectively calculating the fine-grained ablation factor of each sensor signal component encoding feature vector in the set of sensor signal component encoding feature vectors relative to the sensor signal component feature distribution hub encoding vector to obtain the set of sensor signal component fine-grained ablation factors; Inputting each sensor signal component fine-grained ablation factor in the set of sensor signal component fine-grained ablation factors into a sparse module based on a gating function to obtain a set of sparse sensor signal component fine-grained ablation factors; The set of the sparse sensor signal component fine-grained ablation factors is used as a weight distribution, and the set of the sensor signal component encoding feature vectors is weighted modulated to obtain the set of the sparse sensor signal component encoding feature vectors.

2. The fiber grating sensor fault location method according to claim 1, characterized in that: Extracting features from each of the plurality of sensor data components to obtain a set of sensor signal component encoding feature vectors includes: A signal feature extractor based on a one-dimensional convolutional layer is used to perform feature extraction on each of the multiple sensor data components to obtain a set of encoded feature vectors of the sensor signal components.

3. The fiber grating sensor fault location method according to claim 2, characterized in that: Determining the fault type of the fiber grating sensor based on the set of sparsely encoded feature vectors of the sensor signal components comprises: Performing feature concatenation on the set of the sparsely encoded feature vectors of the sensor signal components to obtain a sensor signal global feature encoding vector; The sensor signal global feature encoding vector is input into the trained target fault location model to obtain a fault location result, and the fault location result is a fault type label of the fiber grating sensor.

4. A fiber grating sensor fault location system, characterized in that: include: A first signal data acquisition module, used to acquire first signal data of the fiber grating sensor; A signal data decomposition module, configured to perform signal decomposition based on an empirical mode decomposition algorithm on the first signal data to obtain a plurality of sensor data components; A feature extraction module, configured to extract features from each of the plurality of sensor data components to obtain a set of sensor signal component encoding feature vectors; A feature selection module, configured to perform feature selection based on feature distribution guidance on the set of sensor signal component encoding feature vectors to obtain a sparse set of sensor signal component encoding feature vectors; A fault type identification module, used to determine the fault type of the fiber Bragg grating sensor based on the set of sparse sensor signal component encoding feature vectors; The feature selection module is used to: calculate the feature difference factor of each sensor signal component encoding feature vector in the set of sensor signal component encoding feature vectors relative to the set of sensor signal component encoding feature vectors to obtain a set of sensor signal component feature difference factors; Taking the inverse of each sensor signal component feature difference factor in the set of sensor signal component feature difference factors and performing normalization processing based on a Softmax function to obtain a set of sensor signal component hub feature correlation factors; Using the set of sensor signal component hub feature correlation factors as weight distribution, weighted aggregation is performed on the set of sensor signal component encoding feature vectors to obtain the sensor signal component feature distribution hub encoding vector; Respectively calculating the fine-grained ablation factor of each sensor signal component encoding feature vector in the set of sensor signal component encoding feature vectors relative to the sensor signal component feature distribution hub encoding vector to obtain the set of sensor signal component fine-grained ablation factors; Inputting each sensor signal component fine-grained ablation factor in the set of sensor signal component fine-grained ablation factors into a sparse module based on a gating function to obtain a set of sparse sensor signal component fine-grained ablation factors; The set of the sparse sensor signal component fine-grained ablation factors is used as a weight distribution, and the set of the sensor signal component encoding feature vectors is weighted modulated to obtain the set of the sparse sensor signal component encoding feature vectors.

Images