A high-voltage fuse

Abstract

The present invention relates to the field of fuse technology and discloses a high-voltage fuse comprising: a fuse mechanism and a control module, the control module being electrically connected to the fuse mechanism; the fuse mechanism comprising: a contact portion, an exhaust contact seat portion, an insulator, a fuse carrier portion, and a fiber grating (FBG) sensor, wherein the lower portion of the insulator is connected to the lower portion of the fuse carrier portion via the exhaust contact seat portion, the fiber grating (FBG) sensor is located within the fuse carrier portion, and the upper portion of the insulator is connected to the upper portion of the fuse carrier portion via the contact portion. This application improves the accuracy of sensing the operating status of the fuse.

Description

Technical Field

[0001] The present invention relates to the technical field of fuses, and in particular to a high-voltage fuse. Background Art

[0002] A high-voltage fuse is a protective device used in power systems. Its primary function is to protect electrical equipment from damage by shutting off the current by melting its critical component (the fuse) when an overload or short circuit occurs. Under normal current, the fuse remains conductive; when the current is abnormal, the fuse heats up and melts, triggering the arc-extinguishing medium to absorb energy and cool the arc, thus disconnecting the circuit. A drop-out type high-voltage fuse is a type of high-voltage fuse commonly used in power distribution networks. Its characteristic feature is that after melting, the fuse tube automatically drops to form a visible disconnect point, combining circuit protection and physical isolation. It is primarily used to protect distribution transformers, overhead lines, and branch lines, and is particularly widely used in outdoor environments due to its simple structure, low cost, and intuitive maintenance.

[0003] However, the design concept of existing drop-type high-voltage fuses is centered on "passive protection". Their function focuses on rapid melting and arc extinction when a fault occurs. However, whether the fuse tube has fallen and the fuse element has melted needs to be judged by manual visual inspection or regular inspections, which is inefficient and poses a safety hazard. Summary of the Invention

[0004] The purpose of the present invention is to provide a high-voltage fuse to solve the problems existing in the above-mentioned prior art.

[0005] To achieve the above object, the present invention provides the following solution: The present invention provides a high-voltage fuse, comprising:

[0006] A fuse mechanism and a control module, wherein the control module is electrically connected to the fuse mechanism and is used to control the fuse mechanism; the fuse mechanism includes:

[0007] A contact portion, an exhaust contact seat portion, an insulator, a melt carrier portion, and a fiber Bragg grating sensor, wherein the lower portion of the insulator is connected to the lower portion of the melt carrier portion via the exhaust contact seat portion, the fiber Bragg grating sensor is located within the melt carrier portion, and the fiber Bragg grating sensor is fixedly connected to the melt carrier portion via a ceramic sleeve, and the upper portion of the insulator is connected to the upper portion of the melt carrier portion via the contact portion;

[0008] The control module includes:

[0009] an acquisition unit configured to acquire multidimensional data of the fiber Bragg grating sensor and synchronously acquire a plurality of the multidimensional data based on wavelength division multiplexing, and further configured to acquire voltage data;

[0010] a processing unit configured to obtain temperature data and strain data of the melt carrier based on the multidimensional data, and further configured to preprocess and aggregate the temperature data, the strain data, and the voltage data to obtain a data set;

[0011] a judgment unit configured to train a condition monitoring model based on historical data, and further configured to input the data set into the condition monitoring model for comparison;

[0012] The early warning unit is configured to determine abnormal events based on the comparison results and send corresponding early warning information based on each abnormal event.

[0013] Furthermore, the control module includes:

[0014] an acquisition unit configured to acquire sensor data, store the sensor data at fixed time intervals, and perform denoising, the acquisition unit further configured to align timestamps of the sensor data and process missing values ​​using KNN interpolation;

[0015] a judgment unit configured to extract the sensor data and input the sensor data into a pre-trained state model for comparison, and the judgment unit is further configured to determine the current operating state of the oil level gauge assembly based on the comparison result of the state model;

[0016] The early warning unit is configured to determine an abnormal area based on the abnormal state and send corresponding early warning information based on each abnormal area when the current operating state of the oil level gauge component is abnormal.

[0017] Furthermore, the exhaust contact seat portion includes a rotating connecting frame and a lower contact, the rotating connecting frame is fixedly connected to the lower contact, and the lower contact is locked and connected to the melt carrier portion.

[0018] Furthermore, the melt-carrying part includes a pull ring casting, an arc shortening rod, a lined metal tube and a melting tube. The melting tube is locked and connected to the lower contact. The lined metal tube is fixedly connected to the melting tube. The arc shortening rod is fixedly connected to the lined metal tube. The pull ring casting is fixedly connected to the upper part of the melting tube.

[0019] Furthermore, the contact part includes a conductive cap and a connecting buckle, the conductive cap is fixedly connected to the upper end of the melting tube, one end of the connecting buckle is fixedly connected to the insulator, and the other end of the connecting buckle is clamped with the conductive cap, and the conductive cap is plum blossom-shaped.

[0020] Furthermore, the method of acquiring multidimensional data of the fiber Bragg grating sensor and synchronously collecting a plurality of multidimensional data based on wavelength division multiplexing includes:

[0021] Connecting a plurality of fiber Bragg grating sensors in series to the same optical fiber, and dividing the composite reflection spectrum into independent channels based on an optical fiber demodulator;

[0022] Each of the independent channels is converted into an electrical signal, the central wavelength of the reflection peak of the fiber Bragg grating sensor is determined by a sliding window Gaussian fitting algorithm, and the wavelength offset of the fiber Bragg grating sensor is obtained.

[0023] Furthermore, when obtaining the temperature data and strain data of the melt carrier based on the multidimensional data, the method includes:

[0024] The temperature sensitivity coefficient and the strain sensitivity coefficient of the fiber Bragg grating sensor are obtained, and a linear equation group is constructed, and the temperature data and the strain data are determined based on the linear equation group.

[0025] Furthermore, the temperature data, the strain data, and the voltage data are preprocessed and aggregated to obtain a data set, including:

[0026] Repairing missing data based on linear interpolation, and interpolating the voltage data to the same time series based on the timestamp of the fiber Bragg grating sensor;

[0027] Using a sliding average filter to suppress the instantaneous temperature rise of the temperature data, and removing high-frequency noise based on a wavelet transform;

[0028] Extracting derived features of the temperature data, the strain data, and the voltage data based on a sliding window, wherein the derived features include time domain features based on the temperature data, the strain data, and the voltage data and frequency domain features extracted based on Fourier transform;

[0029] The temperature data, the strain data, the voltage data, and the derived features are aggregated to form the data set, and the temperature data, the strain data, the voltage data, and the derived features are normalized to obtain the data set.

[0030] Furthermore, when training a condition monitoring model based on historical data, it includes:

[0031] Dividing the historical data into normal operating condition data and fault operating condition data, and obtaining Joule heat and thermal expansion data of the melt carrier material;

[0032] The dataset of the historical data is feature decomposed and model training is performed, the time series of the training data is cross-validated 5-fold, and the model parameters with the highest F1-Score are selected as the parameters of the condition monitoring model.

[0033] Furthermore, when the data set is input into the condition monitoring model for comparison, the following steps are included:

[0034] Based on the state monitoring model, all abnormal state probabilities are obtained and an abnormal state threshold is determined. When the comparison result is greater than the abnormal state threshold, abnormality judgment is entered, and all abnormal state probabilities are traversed, and the abnormal state with the highest probability is selected as the comparison result;

[0035] All abnormal conditions are divided into overload, short circuit and poor contact;

[0036] When the temperature rises slowly, the Joule heat residual is small, and the strain increases uniformly, it is determined to be overloaded;

[0037] When the temperature increases suddenly, the Joule heat residual is large, and the strain changes suddenly, it is determined to be a short circuit;

[0038] When the local temperature is abnormal, the Joule heat residual value is large, and the strain amount fluctuates, it is determined to be a poor contact.

[0039] Furthermore, when abnormal events are determined based on the comparison results and corresponding warning information is sent based on each abnormal event, the method includes:

[0040] When the comparison result is overload, a level 1 warning is generated;

[0041] When the comparison result is poor contact, a secondary warning is generated;

[0042] When the comparison result is a short circuit, a third-level warning is generated.

[0043] The present invention discloses the following technical effects: by embedding fiber grating sensors inside the fuse carrier, multi-dimensional physical parameters such as temperature and strain can be collected in real time. Compared with traditional single parameter monitoring, the synchronous acquisition of multi-dimensional data improves the perception accuracy of the fuse operating status. By synchronously collecting multi-channel fiber grating data through wavelength division multiplexing technology and combining it with the real-time collection of voltage data, the system can fully capture the correlation between electrical and mechanical states, providing a more complete basis for fault diagnosis. The processing unit pre-processes the raw data to eliminate the influence of environmental interference. At the same time, by aggregation, a high signal-to-noise ratio data set is generated, which improves the data availability and lays a reliable foundation for subsequent analysis. The state monitoring model trained based on historical data can autonomously learn the normal behavior pattern and abnormal characteristics of the fuse. By comparing the real-time data set with the model threshold, the system can identify abnormal signals, such as local overheating or mechanical deformation, avoiding the hysteresis of traditional threshold judgment methods. The early warning unit can locate the type of abnormal event based on the model comparison results and send early warning information in a hierarchical manner. Through multi-dimensional perception, intelligent analysis and active early warning, the real-time monitoring capability, fault response speed and long-term operation reliability of the fuse are improved, and the equipment maintenance efficiency is improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] The accompanying drawings, which constitute part of this application, are intended to provide a further understanding of this application. The exemplary embodiments and descriptions of this application are intended to explain this application and do not constitute an improper limitation on this application. In the accompanying drawings:

[0045] Figure 1 An overall schematic diagram of a high-voltage fuse provided by an embodiment of the present invention;

[0046] Figure 2 A cross-sectional view of the fuse-carrying portion of a high-voltage fuse provided in an embodiment of the present invention;

[0047] Figure 3 This is a functional block diagram of a high-voltage fuse provided by an embodiment of the present invention.

[0048] In the figure: 1. Insulator; 2. Fiber Bragg grating sensor; 210. Ceramic sleeve; 220. Silicone sleeve; 3. Contact part; 310. Conductive cap; 320. Connecting buckle; 4. Exhaust contact seat; 410. Rotating connecting frame; 420. Lower contact; 5. Melt carrier; 510. Pull ring casting; 520. Arc shortening rod; 530. Lined metal tube; 540. Melt tube. DETAILED DESCRIPTION

[0049] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0050] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0051] Exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided to enable a more thorough understanding of the present disclosure and to fully convey the scope of the present disclosure to those skilled in the art. It should be noted that, unless there is a conflict, the embodiments of the present disclosure and the features in the embodiments can be combined with each other. The present invention will be described in detail below with reference to the accompanying drawings and in conjunction with the embodiments.

[0052] In some embodiments of the present application, see Figure 1-3 As shown, a high voltage fuse includes:

[0053] The fuse mechanism and the control module are electrically connected to the fuse mechanism, and the control module is used to control the fuse mechanism. The fuse mechanism includes:

[0054] The contact part 3, the exhaust contact seat part 4, the insulator 1, the melt carrier part 5 and the fiber optic Bragg grating sensor 2, the lower part of the insulator 1 and the lower part of the melt carrier part 5 are connected through the exhaust contact seat part 4, the fiber optic Bragg grating sensor 2 is located in the melt carrier part 5, and the fiber optic Bragg grating sensor 2 is fixedly connected to the melt carrier part 5 through the ceramic sleeve 210, and the upper part of the insulator 1 and the upper part of the melt carrier part 5 are connected through the contact part 3.

[0055] The control module includes:

[0056] The acquisition unit is configured to acquire multi-dimensional data of the fiber Bragg grating sensor 2 and synchronously acquire multiple multi-dimensional data based on wavelength division multiplexing. The acquisition unit is also configured to acquire voltage data.

[0057] The processing unit is configured to obtain temperature data and strain data of the melt carrier 5 based on the multidimensional data. The processing unit is also configured to preprocess the temperature data, strain data and voltage data, and aggregate them to obtain a data set.

[0058] The judgment unit is configured to train a condition monitoring model based on historical data, and the judgment unit is further configured to input the data set into the condition monitoring model for comparison.

[0059] The early warning unit is configured to determine abnormal events based on the comparison results and send corresponding early warning information based on each abnormal event.

[0060] Specifically, the fuse mechanism is composed of a contact part 3, an exhaust contact seat part 4, an insulator 1, a melt carrier 5 and a fiber Bragg grating sensor 2. The lower ends of the insulator 1 and the melt carrier 5 are connected through the exhaust contact seat part 4, wherein the insulator 1 is fixedly connected to the exhaust contact seat part 4, and the melt carrier 5 is rotatably connected to the exhaust contact seat part 4. The contact part 3 is used to contact the circuit. Two fiber Bragg grating sensors 2 are provided, which are arranged inside the melt carrier 5. One fiber Bragg grating sensor 2 is arranged in the melt carrier 5 through a ceramic sleeve 210, and the other fiber Bragg grating sensor 2 is also arranged in the melt carrier 5 through the ceramic sleeve 210, but a silicone sleeve 220 is further provided on the outer surface of the ceramic sleeve 210. The fiber Bragg grating sensor 2 without the silicone sleeve 220 is used to sense the temperature and strain of the melt carrier 5, while the fiber Bragg grating sensor 2 with the silicone sleeve 220 is used to sense only the temperature. Since the silicone sleeve 220 isolates the strain but not the heat, it will not affect the acquisition of its thermal data. FBG is an optical device that uses ultraviolet laser to form periodic refractive index modulation in the core of an optical fiber. During the manufacturing process of the melt carrier 5, a pre-buried ceramic sleeve 210 is embedded with a fiber optic Bragg grating sensor 2 therein to ensure that it is in direct contact with the melt carrier 5 to sense temperature and strain, while avoiding damage to the optical fiber during high-temperature melting. At the same time, multiple fiber optic Bragg grating sensors 2 are arranged along the length of the melt to form a distributed monitoring network to capture heat data and strain data. The fiber optic Bragg grating sensor 2 detects the reflection wavelength offset of each sensor through an optical fiber demodulator, and simultaneously realizes the synchronous acquisition of multiple sensor signals on a single optical fiber through wavelength division multiplexing technology. The temperature and strain data are then inversely solved by the processing unit, and the inversely solved data is aggregated to obtain a data set. The judgment unit then inputs the data set into the status monitoring model for comparison. The comparison results can be used to determine whether the current operating status of the fuse is abnormal, and different warning information can be provided based on the abnormal points.

[0061] As can be understood, the differentiated design of the dual fiber Bragg grating sensors 2 (with or without the silicone sleeve 220 for strain isolation) enables measurement of both temperature and strain, avoiding signal crosstalk caused by physical coupling in a single sensor. The purity of temperature data, combined with the dynamic changes in strain, improves the resolution of abnormal conditions, particularly distinguishing between different fault modes such as overload and short circuit. Multiple fiber Bragg grating sensors 2 are arranged along the melt carrier 5, forming a distributed sensing network that captures melt thermal and mechanical deformation data. This overcomes the limitations of traditional single-point monitoring, enabling earlier detection of localized overheating or stress concentration hazards and enhancing the ability to predict potential risks. The use of ceramic sleeve 210 packaging technology ensures direct contact between the fiber sensor and the melt carrier, improving heat transfer efficiency, while also insulating the sensor from physical damage caused by the extreme environment of the melting instant through high-temperature resistant materials. The silicone sleeve 220 further isolates mechanical vibration from interference with strain measurement, ensuring long-term data acquisition stability. Leveraging wavelength division multiplexing technology, a single optical fiber can transmit multiple sensor signals in parallel, simplifying the complexity of traditional multi-cable cabling. This reduces the space utilization of the device's internal structure while also minimizing attenuation and noise interference during signal transmission. A preprocessing module aggregates and standardizes temperature, strain, and voltage data to construct a multimodal dataset, providing high-quality input for subsequent analysis. A condition monitoring model trained with historical data adaptively learns the fuse's characteristic patterns under different operating conditions, gradually improving its ability to generalize and recognize complex anomaly patterns. The early warning unit triggers differentiated response strategies based on the severity of the comparison results, such as distinguishing between overload warnings, poor contact warnings, and short-circuit emergency measures. This hierarchical mechanism optimizes the allocation of maintenance resources and avoids response delays and resource waste caused by "one-size-fits-all" alerts. By real-time monitoring of the temperature and mechanical condition of the fuse carrier 5, abnormal trends, such as gradual temperature rise caused by overload or accumulated deformation caused by mechanical aging, can be identified before the fuse reaches its critical melting point. This proactive warning provides a window for manual or automated intervention, reducing the probability of unplanned melting. Fiber optic sensing technology's inherent resistance to electromagnetic interference enables stable operation in harsh electromagnetic environments such as high voltage, strong electric fields, and lightning surges. The corrosion- and moisture-resistant ceramic and silicone materials ensure the sensor's long-term reliability in extreme outdoor conditions, such as high humidity and salt spray. The non-invasive fiber optic sensing solution enables condition monitoring without damaging the fuse's original structure, reducing the risk of human intervention during installation and maintenance. Furthermore, by locating the type of anomaly, targeted component replacement or local overhaul can be guided, avoiding the high cost of replacing the entire device.

[0062] In some embodiments of the present application, the exhaust contact seat portion 4 includes a rotating connecting frame 410 and a lower contact 420 . The rotating connecting frame 410 is fixedly connected to the lower contact 420 , and the lower contact 420 is locked and connected to the melt carrier 5 .

[0063] In some embodiments of the present application, the melt-carrying part 5 includes a pull ring casting 510, an arc shortening rod 520, a lined metal tube 530 and a melting tube 540. The melting tube 540 is locked and connected to the lower contact 420. The lined metal tube 530 is fixedly connected to the melting tube 540. The arc shortening rod 520 is fixedly connected to the lined metal tube 530. The upper part of the melting tube 540 is fixedly connected to the pull ring casting 510.

[0064] In some embodiments of the present application, the contact part 3 includes a conductive cap 310 and a connecting buckle 320, the conductive cap 310 is fixedly connected to the upper end of the melting tube 540, one end of the connecting buckle 320 is fixedly connected to the insulator 1, and the other end of the connecting buckle 320 is clamped with the conductive cap 310, and the conductive cap 310 is in a plum blossom shape.

[0065] Specifically, the insulator 1 is fixed by casting with an insert, and the melt carrier 5 is connected to the insulator 1 through a rotating connecting frame 410, allowing it to rotate around the axis to realize the falling action. After the melt carrier is installed, it forms an angle of 15°~30° with the plumb line to ensure that it falls naturally by gravity after melting. The insulator 1 provides vertical support, and the rotating shaft is located at the bottom of the melt carrier so that it can rotate freely around this point. The lower contact 420 provides contact pressure through a stainless steel spring to ensure that the melt carrier is in close contact with the lower contact 420. The lower contact 420 is located below the falling direction of the melt carrier. The melt carrier is locked by a spring before melting. When melting, the fuse is broken. The melting causes the fuse-carrying component to lose its supporting force. After the spring is released, the fuse-carrying component rotates around the rotating axis and separates from the lower contact 420. During operation, the fuse-carrying part 5 is fixed in the closed position by the spring pressure of the lower contact 420, and its upper end is engaged with the connecting buckle 320. The current passes through the conductive cap 310 to the melting tube 540, shortens the arc through the arc shortening rod 520, and then is turned on through the lower contact 420. The fuse-carrying part 5 rotates downward around the rotating axis under the action of gravity, and is completely separated from the lower contact 420, forming a visible fracture. The opening at the bottom of the melting tube 540 is exposed as the fuse-carrying part 5 falls, and the fault gas is discharged through a single-end downward exhaust method.

[0066] As can be understood, the locking connection between the rotating connecting frame 410 and the lower contact 420, as well as the drop mechanism of the axial rotation of the fuse carrier 5, ensure the triggering and separation of the melting action. The introduction of the rotating structure enhances the stability of the fuse carrier 5 in the normally closed state. At the same time, the synergistic effect of gravity and spring force achieves rapid release during melting, reducing the risk of sticking or malfunction. The lower contact 420 uses a stainless steel spring to provide constant contact pressure, offsetting the loosening of the contact surface caused by thermal expansion or mechanical vibration. This not only ensures electrical continuity during operation, but also allows the fuse carrier 5 to quickly disengage at the moment of melting through spring release, improving breaking efficiency. The preset angle (15° to 30°) formed by the fuse carrier 5 and the plumb line naturally guides the melting tube 540's drop through the action of gravity, ensuring a clearly visible physical fracture after melting. The combination of the lined metal tube 530 and the arc shortening rod 520 accelerates the contraction and cooling of the arc through physical constraints and electromagnetic field optimization, shortens the arc burning time, reduces the arc energy's ablation of the melting tube 540 and surrounding components, and improves the equipment life and disconnection safety. The geometric structure of the plum blossom-shaped conductive cap 310 increases the contact area, balances the current distribution, reduces the contact resistance and local temperature rise. Combined with the metal lining structure of the melting tube 540, the heat dissipation path is further improved, avoiding material aging or delayed melting caused by hot spot accumulation. The single-ended downward exhaust channel at the bottom of the melting tube 540, combined with the opening exposure mechanism after the melting part 5 falls, realizes the directional and rapid discharge of fault gas, prevents the pressure shock or secondary discharge risk caused by gas accumulation in a closed space, and improves the safety of the operating environment.

[0067] In some embodiments of the present application, obtaining multidimensional data of the fiber Bragg grating sensor 2 and synchronously collecting multiple multidimensional data based on wavelength division multiplexing includes:

[0068] Several fiber Bragg grating sensors 2 are connected in series to the same optical fiber, and the composite reflection spectrum is split into independent channels based on a fiber optic interrogator.

[0069] Each independent channel is converted into an electrical signal, and the central wavelength of the reflection peak of the fiber Bragg grating sensor 2 is determined by a sliding window Gaussian fitting algorithm, and the wavelength offset of the fiber Bragg grating sensor 2 is obtained.

[0070] In some embodiments of the present application, when obtaining the temperature data and strain data of the melt carrier 5 based on the multidimensional data, the process includes:

[0071] The temperature sensitivity coefficient and strain sensitivity coefficient of the fiber Bragg grating sensor 2 are obtained, and a linear equation group is constructed, and the temperature data and strain data are determined based on the linear equation group.

[0072] Specifically, two fiber Bragg grating sensors 2 are connected in series to the same optical fiber. Each sensor is preset with a different Bragg wavelength, and the wavelength interval between adjacent sensors must be greater than the maximum expected offset to avoid signal overlap. The optical signal emitted by the broadband light source enters the sensor network through the optical fiber, and then each sensor reflects the light of its specific wavelength, and the remaining wavelengths are projected. The reflected light then returns to the fiber demodulator through the circulator. The demodulator divides the composite reflection spectrum into independent channels through an arrayed waveguide grating or a tunable filter. The independent channel represents the reflection wavelength range of the i-th fiber Bragg grating sensor 2 corresponding to the i-th channel. At the same time, the optical signal of each channel is converted into an electrical signal and output as a voltage-wavelength sequence. The electrical signal of each channel is intercepted by sliding the window width to capture the local spectral data: ,in, is the electrical signal sequence of the i-th channel, is the wavelength value of the kth sampling point, N is the total number of sampling points, W is the window width, and then the spectral data in each window is fitted: , where A is the reflection peak amplitude, is the fitting center wavelength, is the half-height width of the reflection peak, B is the background noise baseline, represents the reflection spectrum intensity of the fiber Bragg grating sensor 2, and then the wavelength offset is calculated according to the fitting result: ,in, is the initial Bragg wavelength of the i-th fiber Bragg grating sensor 2, is the real-time Bragg wavelength of the i-th fiber Bragg grating sensor 2, is the wavelength offset of the i-th fiber Bragg grating sensor 2. In this embodiment, two fiber Bragg grating sensors 2 are provided, namely FBG1 (without the silicone sleeve 220) and FBG2 (with the silicone sleeve 220). FBG1 senses both temperature and strain, and the wavelength offset is: ,in, is the wavelength offset of FBG1, is the temperature sensitivity coefficient of the FBG sensor, is the strain sensitivity coefficient of the FBG sensor, is the temperature change, is the strain change, which represents the difference between the current mechanical strain and the reference strain. FBG2 only senses temperature, and the wavelength shift is: ,in, is the wavelength offset of FBG2, and then the strain is solved by the simultaneous equations to eliminate the temperature effect and obtain the pure strain change: .

[0073] As can be understood, the differentiated design of the dual fiber Bragg grating sensor 2 (with or without the silicone sleeve 220 for strain isolation) achieves physically separate measurement of temperature and strain. The sensor without the silicone sleeve 220 directly senses the combined effect of temperature and strain, while the sensor with the silicone sleeve 220 only senses temperature, eliminating temperature interference with mechanical strain measurement through simultaneous equations. This fundamentally addresses the measurement errors caused by cross-sensitivity in traditional sensors and improves the accuracy of identifying abnormal conditions (such as overload and mechanical fatigue). By connecting multiple sensors in series on the same optical fiber and independently collecting data using wavelength division multiplexing technology, the temperature and strain distribution at different locations on the fuse can be simultaneously acquired. This allows for the detection of early abnormal signals in local hot spots or areas of stress concentration, avoiding the risk of missed detection due to blind spots in single-point monitoring, and improving the comprehensiveness and reliability of condition perception. A sliding window Gaussian fitting algorithm is used to extract the central wavelength of the reflection peak, and mathematical modeling is used to separate the effective signal from the background noise, suppressing the interference introduced by ambient light fluctuations, circuit noise and optical fiber transmission loss, ensuring high-precision detection of weak wavelength offsets, and providing a technical basis for early warning of subtle anomalies (such as local temperature rise caused by poor contact).

[0074] In some embodiments of the present application, the temperature data, strain data, and voltage data are preprocessed and aggregated to obtain a data set, including:

[0075] The missing data are repaired based on linear interpolation, and the voltage data are interpolated to the same time series based on the timestamp of the fiber Bragg grating sensor 2.

[0076] The sliding average filter is used to suppress the instantaneous temperature rise of the temperature data, and the high-frequency noise is removed based on the wavelet transform.

[0077] Derivative features of temperature data, strain data and voltage data are extracted based on a sliding window. The derived features include time domain features based on temperature data, strain data and voltage data and frequency domain features extracted based on Fourier transform.

[0078] The temperature data, strain variable data, voltage data, and derived features are aggregated to form a data set, and the temperature data, strain variable data, voltage data, and derived features are normalized to obtain a data set.

[0079] Specifically, the acquired temperature data, strain data, and voltage data are timestamp-aligned. Missing values ​​in the temperature and strain data are filled in through linear interpolation of adjacent points. Furthermore, to suppress instantaneous temperature rise noise caused by direct sunlight, the average value within the window width is taken: ,in, Expressed as the mean of the window width, is the time window width, Indicates that the repaired temperature data is missing the timestamp The value at , and then the high-frequency noise is removed by wavelet transform. Since wavelet transform is an existing mature technology, it will not be described here. After that, the time domain and frequency domain features of the temperature data, strain data and voltage data are extracted within the window width. The time domain features specifically include but are not limited to the temperature change rate, strain accumulation, and voltage fluctuation standard deviation. The frequency domain features are Fourier transformed on the temperature data, strain data and voltage data respectively to extract their main frequency components, and then the original data and its derived features are normalized and feature aggregation is performed to obtain the aggregated data set.

[0080] As can be understood, linear interpolation is used to correct missing values ​​in the temperature and strain data, ensuring the integrity and coherence of the time series. This process avoids model misjudgments caused by data interruptions and provides a continuous and reliable data foundation for subsequent analysis. Using the fiber optic sensor timestamps as a benchmark, the voltage data is aligned using cubic spline interpolation to eliminate temporal asynchrony caused by differences in sampling frequencies between sensors, ensuring spatiotemporal consistency among temperature, strain, and voltage data and enhancing the ability to correlate multimodal data. A sliding average filter smooths transient temperature fluctuations caused by external factors such as direct sunlight, preserving the true temperature rise trend. Combined with a wavelet transform to remove high-frequency noise, the signal-to-noise ratio (SNR) is further improved, making the data more consistent with actual physical processes. Frequency domain features are extracted using a Fourier transform to analyze the signal's periodicity from an energy distribution perspective, reducing the interference of random noise on feature representation. The complementary nature of time and frequency domain features enhances the robustness of anomaly detection. A sliding window mechanism extracts features such as the temperature change rate, strain accumulation, and voltage fluctuation standard deviation in the time domain, dynamically reflecting both transient and cumulative changes in device status. The dominant frequency component in the frequency domain reveals the periodic patterns underlying the data, providing a deeper basis for fault pattern identification. Time domain features (such as the temperature change rate) are directly related to changes in the physical state of the equipment, while frequency domain features (such as the proportion of high-frequency energy) map the frequency band characteristics of mechanical vibration or arc discharge. The combination of the two gives the feature set both intuitive physical meaning and the ability to represent hidden patterns. Normalization unifies temperature, strain, voltage, and their derivative features to the same scale, avoiding model weight bias caused by dimensional differences. This step significantly improves the convergence speed and stability of machine learning algorithms. The aggregated dataset contains both raw data and derived features, providing unified input for various analytical tasks such as classification, regression, and clustering, and is adaptable to different scenarios such as overload warning, life prediction, and fault diagnosis.

[0081] In some embodiments of the present application, when training a condition monitoring model based on historical data, the following steps are included:

[0082] The historical data is divided into normal operating condition data and fault operating condition data, and the Joule heat and thermal expansion data of the material of the melt carrier 5 are obtained.

[0083] The historical data set is feature decomposed and the model is trained. The time series of the training data is cross-validated 5-fold, and the model parameters with the highest F1-Score are selected as the parameters of the condition monitoring model.

[0084] Specifically, by obtaining the Joule heat data and thermal expansion data of the material of the melt carrier 5 and the data set in the historical data, the features are decomposed and a feature matrix is ​​constructed. The feature matrix is: , where F is the feature matrix, The heat power of the melting part is 5 joules. is the accumulated Joule heat, is the temperature change rate, is the mechanical strain separation, The energy ratio of the main frequency is used, and the feature matrix is ​​divided into 5 subsets according to time for cross-validation training. Four subsets are used as training sets and one subset is used as validation set. The model is then trained and the F1-Score of the validation set is calculated. , where F1 is the F1-Score result, is the accuracy, is the recall rate. The higher the F1-Score, the better the model achieves in reducing false positives and missing negatives. Therefore, the parameter with the highest F1-Score is selected as the parameter of the condition monitoring model.

[0085] It's easy to understand that by incorporating Joule heating and thermal expansion data from the fuse-carrying section 5 material, the physical nature of device operation is embedded in feature engineering. Joule heating power reflects the cumulative energy effect of current overload, while thermal expansion strain correlates with the coupled relationship between material deformation and temperature change. This deep integration of physical quantities enhances the model's understanding of fuse operating mechanisms, making state judgments more closely aligned with actual physical processes. The clear physical quantity definitions (such as temperature change rate and mechanical strain separation) and frequency-domain energy distribution characteristics (main frequency energy percentage) within the feature matrix provide a clear physical explanation for model decision-making. Time-domain features (such as Joule heating accumulation) capture the gradual evolution of device status, while frequency-domain features (such as main frequency energy percentage) reveal the frequency characteristics of periodic or sudden faults. The multi-dimensional complementarity of these two features enhances the model's adaptability to complex operating conditions, enabling it to identify both slowly developing overload hazards and transient short circuits or arcing. A 5-fold time series cross-validation approach is employed, strictly following the chronological order of the data to divide the training and validation sets, thereby preventing future information leakage and potentially falsely high accuracy. Ensure the model maintains reliable performance on data from unknown time intervals, reducing the risk of misjudgment caused by data distribution shifts. Using F1-Score as the core evaluation metric, it balances precision (reducing false positives) and recall (reducing false negatives), preventing the model from favoring the majority class (normal data) while increasing sensitivity to the minority class (faults).

[0086] In some embodiments of the present application, when a data set is input into a condition monitoring model for comparison, the process includes:

[0087] Based on the condition monitoring model, all abnormal state probabilities are obtained and the abnormal state threshold is determined. When the comparison result is greater than the abnormal state threshold, the abnormal judgment is entered, and all abnormal state probabilities are traversed, and the abnormal state with the highest probability is selected as the comparison result.

[0088] All abnormal conditions are divided into overload, short circuit and poor contact.

[0089] When the temperature rises slowly, the Joule heat residual is small, and the strain increases uniformly, it is determined to be overloaded.

[0090] When the temperature increases suddenly, the Joule heat residual is large, and the strain changes suddenly, it is determined to be a short circuit.

[0091] When the local temperature is abnormal, the Joule heat residual value is large, and the strain amount fluctuates, it is determined to be a poor contact.

[0092] In some embodiments of the present application, determining abnormal events based on the comparison results and sending corresponding warning information based on each abnormal event includes:

[0093] When the comparison result is overload, a level 1 warning is generated.

[0094] When the comparison result is poor contact, a second-level warning is generated.

[0095] When the comparison result is a short circuit, a third-level warning is generated.

[0096] As can be seen, the multi-dimensional analysis of temperature trends, Joule heating residuals, and strain characteristics avoids the limitations of single-parameter misjudgment. For example, an overload condition requires a gradual temperature rise, a small Joule heating residual, and a uniform strain increase. This validates the abnormality pattern from multiple perspectives, such as energy accumulation and material deformation, enhancing the physical plausibility of the judgment. A threshold judgment mechanism based on the probability of an abnormal state distinguishes normal fluctuations from true faults. An alert is triggered only when the probability exceeds a preset threshold, reducing false alarms caused by brief disturbances (such as transient voltage surges) while ensuring sensitive detection of potential risks. Abnormal events are clearly classified into overload, short circuit, and poor contact, with differentiated judgment criteria defined (e.g., a sudden temperature increase corresponds to a short circuit, while a localized temperature anomaly corresponds to poor contact). This enables the model to accurately locate the root cause of the fault and provide a basis for targeted remediation. Alerts are divided into Level 1 (overload), Level 2 (poor contact), and Level 3 (short circuit) based on the severity of the abnormality, enabling dynamic calibration of risk levels. Level 1 warnings indicate potential risks, level 2 warnings require planned maintenance, and level 3 warnings trigger emergency responses, optimizing the prioritization of O&M resources. The system traverses all abnormal state probabilities and selects the highest-probability type as the result, reducing reliance on manual review. The system automatically associates warning levels, shortening the time delay from fault identification to action initiation and improving emergency response efficiency. Early warnings for overload and poor contact provide proactive warnings of equipment degradation, enabling load adjustment or component replacement before a fault occurs, avoiding power outages caused by unplanned downtime and reducing associated risks.

[0097] In summary, the beneficial effects of the present invention are as follows: by embedding the fiber Bragg grating sensor 2 inside the fuse carrier 5, multi-dimensional physical parameters such as temperature and strain can be collected in real time. Compared with traditional single parameter monitoring, the synchronous acquisition of multi-dimensional data improves the perception accuracy of the fuse operating status. By synchronously collecting multi-channel fiber Bragg grating data through wavelength division multiplexing technology and combining it with the real-time collection of voltage data, the system can fully capture the correlation between electrical and mechanical states, providing a more complete basis for fault diagnosis. The processing unit pre-processes the raw data to eliminate the influence of environmental interference. At the same time, by aggregating and generating a high signal-to-noise ratio data set, the data availability is improved, laying a reliable foundation for subsequent analysis. The state monitoring model trained based on historical data can autonomously learn the normal behavior patterns and abnormal characteristics of the fuse. By comparing real-time data sets with model thresholds, the system can identify abnormal signals, such as local overheating or mechanical deformation, avoiding the lag of traditional threshold judgment methods. The early warning unit can locate the type of abnormal event through model comparison results and send early warning information in a graded manner. Through multi-dimensional perception, intelligent analysis and active early warning, the real-time monitoring capability, fault response speed and long-term operation reliability of the fuse are improved, thereby improving equipment maintenance efficiency.

[0098] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware. Furthermore, the present application may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0099] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems) and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0100] These computer program instructions may also be stored in a computer-readable storage device that can direct a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable storage device produce an article of manufacture comprising an instruction device that implements the process. Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0101] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0102] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, ordinary technicians in the field should understand that the specific implementation methods of the present invention can still be modified or replaced by equivalents. Any modification or equivalent replacement that does not depart from the spirit and scope of the present invention should be covered by the scope of protection of the claims of the present invention.

Claims

1. A high voltage fuse, characterized in that: include: A fuse mechanism and a control module, wherein the control module is electrically connected to the fuse mechanism and is used to control the fuse mechanism; the fuse mechanism includes: A contact portion (3), an exhaust contact seat portion (4), an insulator (1), a melt carrier portion (5) and a fiber Bragg grating sensor (2), wherein the lower portion of the insulator (1) and the lower portion of the melt carrier portion (5) are connected via the exhaust contact seat portion (4), the fiber Bragg grating sensor (2) is located in the melt carrier portion (5), and the fiber Bragg grating sensor (2) is fixedly connected to the melt carrier portion (5) via a ceramic sleeve (210), and the upper portion of the insulator (1) and the upper portion of the melt carrier portion (5) are connected via the contact portion (3); The control module includes: An acquisition unit configured to acquire multidimensional data of the fiber Bragg grating sensor (2) and synchronously acquire a plurality of the multidimensional data based on wavelength division multiplexing, and the acquisition unit is further configured to acquire voltage data; A processing unit is configured to obtain temperature data and strain data of the melt carrier (5) based on the multidimensional data, and the processing unit is further configured to pre-process the temperature data, the strain data and the voltage data, and aggregate them to obtain a data set; a judgment unit configured to train a condition monitoring model based on historical data, and further configured to input the data set into the condition monitoring model for comparison; an early warning unit configured to determine abnormal events based on the comparison results and send corresponding early warning information based on each abnormal event; When the temperature data and strain data of the melt carrier (5) are obtained based on the multidimensional data, the method includes: Obtaining the temperature sensitivity coefficient and the strain sensitivity coefficient of the fiber Bragg grating sensor (2), and constructing a linear equation group, and determining the temperature data and the strain data based on the linear equation group; Preprocessing the temperature data, the strain data, and the voltage data and aggregating them to obtain a data set includes: Repairing missing data based on linear interpolation, and interpolating the voltage data to the same time series based on the timestamp of the fiber Bragg grating sensor (2); Sliding average filtering is used to suppress the instantaneous temperature rise of the temperature data, and high-frequency noise is removed based on wavelet transform; Extracting derived features of the temperature data, the strain data, and the voltage data based on a sliding window, wherein the derived features include time domain features based on the temperature data, the strain data, and the voltage data and frequency domain features extracted based on Fourier transform; aggregating the temperature data, the strain data, the voltage data, and the derived features to form the data set, and normalizing the temperature data, the strain data, the voltage data, and the derived features to obtain the data set; When training a condition monitoring model based on historical data, this includes: Dividing the historical data into normal operating condition data and fault operating condition data, and obtaining Joule heat and thermal expansion data of the material of the melt carrier (5); The dataset of the historical data is feature decomposed and model training is performed, the time series of the training data is cross-validated 5-fold, and the model parameters with the highest F1-Score are selected as the parameters of the condition monitoring model.

2. The high-voltage fuse according to claim 1, characterized in that: The exhaust contact seat portion (4) comprises a rotating connection frame (410) and a lower contact (420), the rotating connection frame (410) is fixedly connected to the lower contact (420), and the lower contact (420) is locked and connected to the melt carrier portion (5).

3. The high voltage fuse according to claim 2, characterized in that: The melt-carrying part (5) comprises a pull ring casting (510), an arc shortening rod (520), a lining metal tube (530) and a melting tube (540); the melting tube (540) is locked and connected to the lower contact (420); the lining metal tube (530) is fixedly connected inside the melting tube (540); the arc shortening rod (520) is fixedly connected inside the lining metal tube (530); and the pull ring casting (510) is fixedly connected to the upper part of the melting tube (540).

4. The high-voltage fuse according to claim 3, characterized in that: The contact part (3) comprises a conductive cap (310) and a connecting buckle (320), wherein the conductive cap (310) is fixedly connected to the upper end of the melting tube (540), one end of the connecting buckle (320) is fixedly connected to the insulator (1), and the other end of the connecting buckle (320) is clamped to the conductive cap (310), and the conductive cap (310) is in a plum blossom shape.

5. The high voltage fuse according to claim 4, characterized in that: Acquiring multidimensional data of the fiber grating sensor (2) and synchronously collecting a plurality of the multidimensional data based on wavelength division multiplexing comprises: Connecting a plurality of fiber Bragg grating sensors (2) in series to the same optical fiber, and dividing the composite reflection spectrum into independent channels based on an optical fiber demodulator (7); Each of the independent channels is converted into an electrical signal, the central wavelength of the reflection peak of the fiber grating sensor (2) is determined by a sliding window Gaussian fitting algorithm, and the wavelength offset of the fiber grating sensor (2) is obtained.

6. The high-voltage fuse according to claim 5, characterized in that: When the data set is input into the condition monitoring model for comparison, it includes: Based on the state monitoring model, all abnormal state probabilities are obtained and an abnormal state threshold is determined. When the comparison result is greater than the abnormal state threshold, abnormality judgment is entered, and all abnormal state probabilities are traversed, and the abnormal state with the highest probability is selected as the comparison result; All abnormal conditions are divided into overload, short circuit and poor contact; When the temperature rises slowly, the Joule heat residual is small, and the strain increases uniformly, it is determined to be overloaded; When the temperature increases suddenly, the Joule heat residual is large, and the strain changes suddenly, it is determined to be a short circuit; When the local temperature is abnormal, the Joule heat residual value is large, and the strain amount fluctuates, it is determined to be a poor contact.

7. The high-voltage fuse according to claim 6, characterized in that: When abnormal events are determined based on the comparison results and corresponding warning information is sent based on each abnormal event, it includes: When the comparison result is overload, a level 1 warning is generated; When the comparison result is poor contact, a secondary warning is generated; When the comparison result is a short circuit, a third-level warning is generated.

Images