Transformer vibration and partial discharge on-line monitoring device and method based on fiber optic sensor

Abstract

The application relates to a transformer vibration and partial discharge online monitoring device based on an optical fiber sensor, and the device method comprises the following steps: a laser emits a laser signal; an acousto-optic modulator responds to a parameter control signal to adjust the frequency and wavelength of the laser signal; an optical fiber sensor is disturbed by the vibration or partial discharge of a transformer, so that the reflected light of the laser signal is subjected to Bragg wavelength shift; the optical fiber sensor comprises multiple groups of symmetrical fiber gratings and counterweights fixed at both ends of the fiber gratings; the counterweights adopt a hollow thin-wall structure; a photoelectric detector receives the returned light signal of the optical fiber sensor and converts the light signal into an electric signal. Through the application, the problem of low accuracy of the transformer monitoring device is solved; the counterweight with the hollow thin-wall structure improves the sensing ability of the sensor to slight vibration; the symmetrical structure of the fiber grating enables the sensor to more sensitively respond to vibration in different directions, and improves the adaptability and detection capability of the sensor in a complex vibration environment.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of device monitoring, and in particular to an online monitoring device and method for transformer vibration and partial discharge based on an optical fiber sensor. BACKGROUND

[0002] As a key device in the power system, transformers are widely used in power transmission and distribution. The running state of the transformer directly affects the stability and safety of the power system, so monitoring and fault warning of the transformer are crucial.

[0003] Existing transformer vibration monitoring usually uses acceleration sensors or speed sensors. Acceleration sensors are mainly used to capture low-frequency vibrations, but their sensitivity is poor for high-frequency vibrations and weak electrical discharge signals, and they cannot provide comprehensive fault diagnosis information. In terms of partial discharge monitoring, traditional technologies mainly use ultrasonic sensors, electromagnetic sensors and capacitive sensors. Ultrasonic sensors and electromagnetic sensors depend on the stability of the external environment and have high requirements for spatial arrangement, often failing to cover all possible partial discharge sources inside the transformer. The sensitivity and accuracy of capacitive sensors are low, and they cannot effectively distinguish between normal operation and fault of the transformer, resulting in inaccurate fault diagnosis. SUMMARY

[0004] The embodiments of the present application provide an online monitoring method, system, electronic device and storage medium for transformer vibration and partial discharge based on an optical fiber sensor, to at least solve the problem of low accuracy of the transformer monitoring device in related technologies.

[0005] In a first aspect, the embodiments of the present application provide an online monitoring device for transformer vibration and partial discharge based on an optical fiber sensor, comprising:

[0006] a laser for emitting a laser signal;

[0007] an acousto-optic modulator for adjusting the frequency and wavelength of the laser signal in response to a parameter control signal;

[0008] an optical fiber sensor for being disturbed by the vibration or partial discharge of the transformer, causing the reflected light of the laser signal to have a Bragg wavelength shift, the optical fiber sensor comprising a plurality of symmetric optical fiber gratings, and counterweights fixed at both ends of each optical fiber grating, the counterweights adopting a hollow thin-walled structure;

[0009] a photodetector for receiving the optical signal returned by the optical fiber sensor and converting the optical signal into an electrical signal.

[0010] In some embodiments, the fiber sensor further comprises a sensitive film fixed on the package shell, and the weight is fixed to the sensitive film by epoxy resin.

[0011] In some embodiments, the device further comprises:

[0012] an arbitrary waveform generator configured to generate a parameter control signal and send the parameter control signal to the acousto-optic modulator;

[0013] a first doped fiber amplifier configured to amplify the laser signal modulated by the acousto-optic modulator;

[0014] an amplified spontaneous emission filter configured to remove spontaneous emission noise generated in the amplification process of the laser signal;

[0015] a circulator configured to transmit the processed laser signal to the fiber sensor and receive the optical signal returned by the fiber sensor;

[0016] a second doped fiber amplifier configured to amplify the optical signal returned by the fiber sensor and send the amplified optical signal to the photodetector;

[0017] a data acquisition card configured to convert the electrical signal obtained by the photodetector into a digital signal.

[0018] In a second aspect, the embodiments of the present application provide a transformer vibration and partial discharge online monitoring method based on a fiber sensor, the method comprising:

[0019] The transformer data includes vibration signal data of a surface of a transformer tank and partial discharge signal data of the transformer.

[0020] The transformer data and the optical signal data are analyzed by an abnormality prediction model based on a random forest algorithm to obtain vibration characteristics, partial discharge characteristics, and optical signal characteristics of the transformer.

[0021] An early warning signal is generated according to the vibration characteristics, the partial discharge characteristics, the optical signal characteristics, and a preset grading early warning mechanism.

[0022] In some embodiments, the analysis of the transformer data and the optical signal data comprises:

[0023] The vibration signal data is subjected to feature extraction to obtain vibration characteristics, the vibration characteristics including vibration amplitude and vibration frequency.

[0024] extracting features from the partial discharge signal data to obtain partial discharge features, the partial discharge features including a partial discharge signal frequency, a discharge signal intensity change, and a signal-to-noise ratio;

[0025] obtaining temperature change data by using a temperature-compensated optical fiber sensor, and temperature-compensating the vibration features and the partial discharge features according to the temperature change data to obtain corrected vibration features and partial discharge features;

[0026] extracting features from the optical signal data to obtain optical signal features, the optical signal features including an optical intensity change rate.

[0027] In some embodiments, the generating of the early warning signal according to the vibration features, the partial discharge features, the optical signal features, and a preset hierarchical early warning mechanism comprises:

[0028] in response to any one of the vibration amplitude, the vibration frequency, and the optical intensity change rate being greater than a corresponding critical vibration threshold value, performing a critical vibration early warning; and / or

[0029] in response to any one of the vibration amplitude, the vibration frequency, and the optical intensity change rate being greater than a corresponding warning vibration threshold value, and all of the vibration amplitude, the vibration frequency, and the optical intensity change rate being less than or equal to the corresponding critical vibration threshold value, performing a warning vibration early warning; and / or

[0030] in response to any one of the partial discharge signal frequency, the discharge signal intensity change, and the signal-to-noise ratio being greater than a corresponding first partial discharge threshold value, performing a high-intensity partial discharge early warning; and / or

[0031] in response to any one of the partial discharge signal frequency, the discharge signal intensity change, and the signal-to-noise ratio being greater than a corresponding second partial discharge threshold value, and all of the partial discharge signal frequency, the discharge signal intensity change, and the signal-to-noise ratio being less than the corresponding first partial discharge threshold value, performing a low-intensity partial discharge early warning.

[0032] In some embodiments, the method further comprises:

[0033] generating parameter adjustment instructions for the laser and the acousto-optic modulator based on the optical signal features; and / or

[0034] detecting an optical fiber link of the optical fiber sensor-based transformer vibration and partial discharge online monitoring device of the first aspect, obtaining loss data of the optical fiber link, and generating a link alarm signal in a case where the loss data is greater than a preset loss threshold value.

[0035] In some embodiments, the method further comprises:

[0036] collecting a transformer dataset for training the anomaly prediction model, and performing normal or abnormal labeling on data in the transformer dataset according to historical fault data and an actual state of the transformer to obtain a labeled dataset;

[0037] constructing an anomaly prediction model by using a random forest algorithm, and training and evaluating the anomaly prediction model based on the labeled dataset.

[0038] In some embodiments, the performing of the normal or abnormal labeling on the data in the transformer dataset comprises:

[0039] extracting historical fault records of a substation where the transformer is located, and obtaining transformer monitoring data corresponding to a timestamp of the historical fault records;

[0040] obtaining a transformer operation log, and performing normal or abnormal labeling on the data in the transformer dataset based on a preset labeling rule and according to the transformer monitoring data and the transformer operation log to obtain a labeled dataset.

[0041] In some embodiments, the labeled dataset comprises a training set and a test set, and the training and evaluation of the anomaly prediction model based on the labeled dataset comprises:

[0042] testing the trained anomaly prediction model based on the test set to obtain evaluation indexes of the anomaly prediction model, the evaluation indexes comprising a prediction accuracy, a recall rate, an F1-score, a specificity and an AUC-ROC, and evaluating the anomaly prediction model based on the evaluation indexes; and / or

[0043] testing the anomaly prediction model based on the test set to obtain an accuracy of the anomaly prediction model, determining a drift rate of the accuracy, and triggering an update of the anomaly prediction model in a case where the drift rate is greater than a preset drift threshold.

[0044] In a third aspect, an embodiment of the present application provides a computer device, which comprises a memory, a processor, and a computer program stored in the memory and capable of running on the processor, and the processor implements the transformer vibration and partial discharge online monitoring method based on an optical fiber sensor when implementing the computer program.

[0045] In a fourth aspect, an embodiment of the present application provides a computer readable storage medium, which stores a computer program, and the program is executed by a processor to implement the transformer vibration and partial discharge online monitoring method based on an optical fiber sensor.

[0046] Compared with the related art, the transformer vibration and partial discharge online monitoring device based on the optical fiber sensor provided by the embodiment of the application comprises a laser, an acousto-optic modulator, an optical fiber sensor and a photodetector, wherein the optical fiber sensor comprises a plurality of groups of symmetrical fiber gratings and counterweights fixed at both ends of each fiber grating, and the counterweight adopts a hollow thin-wall structure. The device solves the problem of low accuracy of the transformer monitoring device. The counterweight adopting the hollow thin-wall structure improves the sensing ability of the sensor to slight vibration. The symmetrical structure of the fiber grating enables the sensor to respond more sensitively to vibration in different directions, improving the adaptability and detection capability of the sensor in a complex vibration environment. The symmetrical structure combined with the multi-point connection can enhance the stability of the entire sensor structure. For example, under high-frequency or large-amplitude vibration, the symmetrical multi-point connection structure can disperse stress and avoid stress concentration in a certain part, thereby prolonging the service life of the sensor. BRIEF DESCRIPTION OF DRAWINGS

[0047] The drawings described herein are used to provide further understanding of the application, constitute a part of the application, the illustrative embodiments of the application and the description thereof are used to explain the application, and do not constitute improper limitation on the application. In the drawings:

[0048] Figure 1 is a structural block diagram of the transformer vibration and partial discharge online monitoring device based on the optical fiber sensor according to the embodiment of the application;

[0049] Figure 2 is a schematic diagram of an optical fiber sensor according to the embodiment of the application;

[0050] Figure 3 is a connection schematic diagram of a counterweight and a sensitive film according to the embodiment of the application;

[0051] Figure 4 is a flowchart of the transformer vibration and partial discharge online monitoring method based on the optical fiber sensor according to the embodiment of the application;

[0052] Figure 5 is an internal structure schematic diagram of an electronic device according to the embodiment of the application.

[0053] In each of the above drawings, the meanings of the reference signs are as follows:

[0054] 11, laser; 12, acousto-optic modulator; 13, optical fiber sensor; 131, fiber grating; 132, counterweight; 133, sensitive film; 134, packaging shell; 135, epoxy resin; 14, photodetector; 21, arbitrary waveform generator; 22, first doped fiber amplifier; 23, amplified spontaneous emission filter; 24, circulator; 25, second doped fiber amplifier; 26, data acquisition card; 31, transformer; 32, control system. DETAILED DESCRIPTION

[0055] In order to make the purposes, technical solutions, and advantages of the present application clearer, the present application is described and explained below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application, and should not be used to limit the present application. Based on the embodiments provided in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort belong to the scope of the present application.

[0056] It is obvious that the accompanying drawings in the following description are only some examples or embodiments of the present application, and for those of ordinary skill in the art, the present application can be applied to other similar scenarios without creative effort based on the accompanying drawings. In addition, it can be understood that although the efforts made in the development process can be complex and lengthy, for those of ordinary skill in the art related to the content disclosed in the present application, some design, manufacture or production changes based on the technical content disclosed in the present application are only routine technical means, and should not be understood as insufficient disclosure of the content disclosed in the present application.

[0057] In the present application, the phrase "embodiments" means that the specific features, structures or properties described in conjunction with the embodiments can be included in at least one embodiment of the present application. The appearance of this phrase at various places in the specification does not necessarily mean the same embodiment, nor is it an independent or alternative embodiment to other embodiments. It is explicitly and implicitly understood by those of ordinary skill in the art that the embodiments described in the present application can be combined with other embodiments without conflict.

[0058] Unless otherwise defined, technical terms and scientific terms used in the present application shall have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains. Unless otherwise defined, the terms "one" and "a" or "an" used in this application do not denote a limitation of quantity, but rather denote the presence of at least one of the referenced item. The terms "including" and / or "containing", or "having" and variations thereof herein are meant to encompass the presence of stated features, steps, or components without limitation as to the number of such features, steps, or components that can be present. The terms "connected", "coupled", or "pathway" are not limited to direct connections, but can also include indirect connections unless otherwise noted. The term "plurality" means two or more. The term "and / or" describes associated objects in association relationships, which means that there can be three relationships, for example, "A and / or B" can mean that A exists alone, A and B exist together, and B exists alone. The character " / " generally represents an "or" relationship between the front and rear associated objects. The terms "first", "second", "third", and the like are merely used to distinguish similar objects, and do not represent a specific order of the objects.

[0059] The embodiment provides an on-line monitoring device for transformer vibration and partial discharge based on an optical fiber sensor. As used below, the terms "module", "unit", "sub-unit", and the like can be a combination of software and / or hardware that implements a predetermined function. Although the device described in the following embodiments is preferably implemented in software, hardware, or a combination of software and hardware is also possible and contemplated.

[0060] Figure 1 is a structural block diagram of an on-line monitoring device for transformer vibration and partial discharge based on an optical fiber sensor according to the embodiment of the present application, as shown in Figure 1 The device includes a laser 11, an acousto-optic modulator 12, an optical fiber sensor 13, and a photodetector 14.

[0061] The laser 11 is configured to emit a laser signal.

[0062] The acousto-optic modulator 12 is configured to adjust the frequency and wavelength of the laser signal in response to a parameter control signal.

[0063] The optical fiber sensor 13 is configured to be disturbed by the vibration or partial discharge of the transformer, causing the reflected light of the laser signal to have a Bragg wavelength shift. The optical fiber sensor 13 includes a plurality of symmetric optical fiber gratings 131, and counterweights 132 fixed at both ends of each optical fiber grating. The counterweights 132 adopt a hollow thin-walled structure.

[0064] A photodetector 14 is used to receive the optical signal returned by the fiber sensor and convert it into an electrical signal.

[0065] The laser 11 emits a laser signal, the wavelength, intensity and frequency of which can be precisely adjusted by the acousto-optic modulator 12. The adjusted laser signal passes through the fiber sensor 13, and the optical signal returned by the fiber sensor 13 is transmitted to the photodetector 14. The photodetector 14 is responsible for converting the received optical signal into an electrical signal. The high sensitivity of the detector enables it to capture weak optical signals and convert them into electrical signals, ensuring effective measurement even in low light intensity and weak fluctuations.

[0066] The fiber sensor of this embodiment, i.e. the WFBG sensor, ensures that the device has good temperature and strain sensitivity. The WFBG sensor can detect physical quantities such as temperature, strain and vibration through wavelength changes, thus achieving the same high sensitivity vibration detection as fiber gratings. WFBG has higher integration and anti-interference capability, which can meet the requirements of long-distance monitoring and high precision.

[0067] The fiber grating uses periodic changes in refractive index (Bragg wavelength changes) to sense changes in physical quantities such as temperature, strain and vibration, and can detect weak external disturbances. Even in complex environments such as high temperature and strong electromagnetic interference, it can still maintain high measurement accuracy.

[0068] The use of high spatial resolution fiber grating array layout and precise light source pulse design enables the device to achieve sub-millimeter spatial resolution and accurately locate multiple disturbance events. This technology not only enables the device to monitor multiple events simultaneously, but also enables real-time accurate positioning of the location of each event, making it suitable for multi-point detection tasks in dynamic environments.

[0069] The counterweight 132 in this embodiment can adopt a hollow thin-walled structure, such as a hollow resin sleeve. The hollow thin-walled structure is more conducive to stress concentration at the stress concentration point of the fiber grating during vibration. Stress concentration causes the strain on the fiber grating to increase further, thereby improving the sensor's ability to perceive slight vibrations, and even extremely subtle vibration changes can be detected more accurately.

[0070] The hollow thin-walled structure has a more reasonable mass distribution and larger moment of inertia than the solid structure while ensuring a certain mass. During vibration, due to its mass distribution characteristics, it can more effectively transmit vibration to the packaging structure and the internal fiber grating connected thereto, resulting in less energy loss during vibration transmission and improving the efficiency and accuracy of vibration transmission.

[0071] This embodiment also employs a symmetrical structure and multi-point connection design, which is beneficial for improving the response to vibrations in different directions, improving the uniformity of vibration response, and enhancing structural stability.

[0072] The hollow, thin-walled mass block offers greater design flexibility, allowing for connection to the encapsulation structure through multi-point connections by studying the location and number of connection points. This flexible connection method helps optimize vibration transmission paths, enabling the sensor to respond more sensitively to vibrations in different directions, thus improving the sensor's adaptability and detection capabilities in complex vibration environments.

[0073] The symmetrical structure makes the sensor's response to vibration more uniform in all directions. When external vibrations are transmitted to the sensor, the symmetrical layout ensures that the vibration excitation received by each part is transmitted to the fiber Bragg grating in a similar way, avoiding the situation where the vibration response is strong in some directions and weak in others due to structural asymmetry, thereby comprehensively improving the detection sensitivity of vibrations in different directions.

[0074] A symmetrical structure combined with multi-point connections enhances the stability of the entire sensor structure. When facing complex and variable vibration environments, this structure can better resist external interference and reduce the risk of structural deformation or damage caused by vibration. For example, under high-frequency or high-amplitude vibrations, the symmetrical multi-point connection structure can disperse stress, preventing stress concentration in a single area, thereby extending the sensor's lifespan and ensuring its long-term stable operation.

[0075] It should be noted that, besides using resin tubing as the material for the counterweight, piezoelectric materials, such as piezoelectric ceramics, can also be used to fabricate the mass. Piezoelectric materials generate an electric charge when subjected to vibration, a property that can directly convert the mechanical energy of vibration into an electrical signal. By rationally designing the shape and size of the piezoelectric mass and matching it with a fiber optic grating, the electrical signal generated by the piezoelectric effect alters the electric field environment around the fiber optic grating, thereby affecting its refractive index and achieving highly sensitive vibration detection.

[0076] In some embodiments, the fiber optic sensor 13 further includes a sensitive film 133 and an encapsulation housing 134, wherein the sensitive film is fixed to the encapsulation housing and the counterweight is fixed to the sensitive film by epoxy resin.

[0077] Figure 2 This is a schematic diagram of an optical fiber sensor according to an embodiment of this application, as shown below. Figure 2 As shown, the fiber optic sensor includes a fiber optic grating 131, a counterweight 132, a sensitive film 133, and an encapsulation shell 134.

[0078] Figure 3 This is a schematic diagram illustrating the connection between a counterweight and a sensitive film according to an embodiment of this application, as shown below. Figure 3As shown, the weight block 132 is adhered to the sensitive membrane 133 by epoxy 135. Preferably, the epoxy coating has a radius of 1 mm and a thickness of 0.3 mm.

[0079] Epoxy has high viscosity, high elasticity, chemical stability and electrical insulation, which can well meet the needs of fiber optic sensor.

[0080] The high viscosity of epoxy can form a strong adhesive force between the mass and the packaging structure, ensuring a tight connection between the two. This tight connection can effectively prevent relative displacement between the mass and the packaging structure during vibration, ensuring efficient transmission of vibration energy, thereby helping to improve the sensor's ability to perceive small vibrations.

[0081] Epoxy retains a certain elasticity after curing, which allows the mass to move relatively flexibly during vibration while accurately transmitting the vibration to the packaging structure and the internal fiber Bragg grating. It plays a role in buffering and elastic transmission, optimizing the vibration transmission path and improving the sensor's response sensitivity to vibrations in different directions.

[0082] Epoxy has good chemical stability and can resist the corrosion of various chemicals. In complex environments that the sensor may face, such as humid and chemical gas environments, epoxy will not easily react chemically to cause performance degradation, ensuring the long-term stability of the connection between the mass and the packaging structure, and helping to maintain the stability of the sensor's detection sensitivity.

[0083] Good electrical insulation performance ensures that epoxy does not interfere with the transmission of optical or electrical signals inside the sensor during bonding. The fiber Bragg grating and other components in the vibration sensor have high requirements for signal transmission. The electrical insulation properties of epoxy avoid the impact of problems such as electrical leakage or electromagnetic interference on the performance of the sensor, ensuring that the sensor detects vibration signals stably and accurately.

[0084] It should be noted that in addition to using epoxy to connect the weight block and the sensitive membrane, magnetic coupling or shape memory alloy connection can also be used.

[0085] Magnetic coupling connection: Magnetic materials are provided on the mass and the packaging structure respectively, and the connection is achieved through magnetic coupling. When the mass vibrates, the change in the magnetic field will be transmitted to the packaging structure through magnetic coupling, and then act on the fiber Bragg grating. This connection method has the advantages of non-contact and no wear, which can improve the stability and durability of the connection to some extent, and may have better response to certain specific frequencies of vibration.

[0086] Shape memory alloy connection: using shape memory alloy as connecting material, such as nickel-titanium alloy. Shape memory alloy will restore to the pre-set shape when the temperature changes. In the sensor assembly process, the shape memory alloy is first connected in a low-temperature deformed state, and then heated to restore its shape, thereby tightly connecting the mass and the packaging structure. This connection method can provide a larger pre-tightening force during the connection process, and maintain a relatively stable connection state under different temperature environments, which helps to optimize the vibration transmission path.

[0087] An example of the specific processing steps of an optical fiber sensor:

[0088] Process the sensitive film and the counterweight. Use a laser cutting machine to cut the sensitive film, and process a circular through-hole in the center of the sensitive film for the optical fiber to pass through; use 3D printing technology to make a small sleeve for the optical fiber to pass through as a counterweight.

[0089] For example, a single crystal silicon wafer with a radius of 7mm and a thickness of 120μm is processed into a sensitive film by processing a circular through-hole with a diameter of 0.4mm in the center of the wafer; four counterweights are evenly placed 4mm away from the center, wherein the single crystal silicon wafer has a Poisson's ratio μ=0.28, a Young's modulus E=190GPa, and a density ρ=2.33g / cm3. Use 3D printing technology to print a hollow resin sleeve with a radius of 0.25mm and a length of 3mm as a counterweight.

[0090] Process the fiber Bragg grating. Use femtosecond inscription technology to process the fiber Bragg grating on a single-mode optical fiber.

[0091] For example, use femtosecond inscription technology to process a fiber Bragg grating with a grating length of 10mm, a center wavelength of 1550nm, a reflectivity of >90%, and a 3dB bandwidth of <0.3nm.

[0092] Process the packaging shell. Use 3D printing technology to make a hollow resin cylinder as the packaging shell of the sensor.

[0093] For example, use 3D printing technology to process a hollow resin cylinder with an inner diameter of 7mm, an outer diameter of 10mm, and a height of 25mm as the packaging shell of the sensor.

[0094] Assemble the optical fiber sensing device. Use epoxy resin coating to fix the counterweight and the sensitive film, and use ultraviolet curing glue to fix the counterweight and the optical fiber.

[0095] The installation of the optical fiber sensor specifically includes: cleaning the key detection parts of the transformer using fiber-free cloth and appropriate cleaning agents, ensuring that the surface is free of oil, dust and other contaminants, to ensure that the optical fiber sensor can be firmly attached. According to the structure and vibration, partial discharge distribution characteristics of the transformer, determine the installation position of the optical fiber sensor, preferentially select the parts that are more affected. Evenly apply an appropriate amount of adhesive (such as epoxy resin or special optical fiber adhesive) on the selected optical fiber sensor installation position, to prepare for the sensor attachment. Attach the optical fiber sensor to the surface coated with adhesive, ensuring that the sensor is in full contact with the surface to improve the accuracy of signal acquisition. Use tape or other fixing devices to temporarily fix the sensor to ensure that the sensor does not move during the curing of the adhesive.

[0096] It should be noted that the present embodiment also provides a temperature compensation optical fiber sensor for monitoring changes in ambient temperature. The same type of sensor as the optical fiber sensor used to monitor the transformer is selected as the temperature compensation optical fiber sensor to monitor temperature changes to eliminate the influence of temperature on vibration and partial discharge measurement.

[0097] The temperature compensation optical fiber sensor is installed in a position not affected by vibration and partial discharge, so as to detect the change of ambient temperature alone. Use appropriate methods to ensure the stable installation of the temperature compensation optical fiber sensor, and avoid displacement during monitoring.

[0098] Protection measures such as optical fiber protection tubes are added to the installed optical fiber sensor and its connecting parts to prevent damage caused by external environmental factors. Connect the optical fiber of the optical fiber sensor to the optical fiber coupler to ensure firm connection and smooth optical path, and check all connection points to confirm that there is no looseness or optical fiber breakage.

[0099] When collecting data, start the optical fiber sensor-based monitoring device to start real-time recording of vibration and partial discharge data. Ensure that the device is running normally and continuously monitor the data to discover and solve possible problems in a timely manner.

[0100] The optical fiber sensor provided by the embodiment significantly improves the detection sensitivity. In terms of vibration detection, the counterweight adopts a hollow thin-walled structure, which is carefully designed to generate a stress concentration point that is more conducive to being transmitted to the fiber grating during vibration, further increasing the strain on the fiber grating, thereby greatly improving the perception of slight vibrations. At the same time, the sensor adopts a symmetrical structure, and the vibration transmission path is optimized through a multi-point connection, so that the sensor has higher sensitivity to vibrations in different directions. In terms of partial discharge detection, the special process and material are used to realize efficient capture and accurate detection of the partial discharge signal, and the high-viscosity and high-elasticity epoxy resin is selected to ensure that the mass and the packaging structure are tightly and elastically connected. Reliable data support is provided for the operation state evaluation of power equipment.

[0101] In some embodiments, the apparatus further comprises:

[0102] An arbitrary waveform generator 21 is configured to generate a parameter control signal and send the parameter control signal to the acousto-optic modulator.

[0103] The acousto-optic modulator receives the control signal from the arbitrary waveform generator and adjusts the frequency and intensity of the laser according to the signal, thereby changing the waveform mode of the laser, so that the laser signal meets the experimental or measurement requirements.

[0104] A first doped fiber amplifier 22 is configured to amplify the laser signal modulated by the acousto-optic modulator;

[0105] An amplified spontaneous emission filter 23 is configured to remove the spontaneous emission noise generated in the amplification process of the laser signal;

[0106] A circulator 24 is configured to transmit the processed laser signal to the optical fiber sensor and receive the optical signal returned by the optical fiber sensor;

[0107] A second doped fiber amplifier 25 is configured to amplify the optical signal returned by the optical fiber sensor and send the amplified optical signal to a photodetector;

[0108] A data acquisition card 26 is configured to convert the electrical signal obtained by the photodetector into a digital signal.

[0109] The laser 11 emits a laser signal, which is modulated into periodic pulsed light by the acousto-optic modulator 12 driven by the arbitrary waveform generator 21. Subsequently, the pulsed light is amplified by the first doped fiber amplifier 22, and the amplified spontaneous emission filter 23 removes the spontaneous emission noise generated during amplification. The processed optical signal is transmitted to the optical fiber sensor 13 through the circulator 24. When the transformer 31 generates vibration or partial discharge, the Bragg wavelength of the optical fiber sensor 13 changes due to external disturbance. The changed optical signal is returned through the circulator 24, amplified again by the second doped fiber amplifier 25, detected by the photodetector 14, and finally transmitted to the control system 32 for analysis and processing through the data acquisition card 26.

[0110] The data processed by the data acquisition card will be sent back to the control system for real-time feedback and adjustment. The control system automatically adjusts the parameters of the laser and acousto-optic modulator according to the measured optical signal characteristics, ensuring that the system can always maintain the best working state. The data acquisition card not only realizes accurate signal conversion, but also plays a role in monitoring and controlling the real-time performance of the system, ensuring that the system can quickly respond and make adjustments under changing working conditions.

[0111] The frequency, wavelength and intensity of the laser signal are adjusted by the acousto-optic modulator, allowing the laser signal to flexibly adapt to different experimental or industrial application requirements. The combination of the photodetector and the data acquisition card makes signal conversion and data processing more efficient, allowing accurate measurement results to be obtained in real time, and enabling immediate adjustment of each module through the monitoring system.

[0112] The above device ensures high precision, high sensitivity measurement results in various complex environments through high-precision optical signal control, sensitive photodetection, fast data acquisition and processing, and real-time system feedback and adjustment, and improves the dynamic response capability and stability of the system.

[0113] It should be noted that the above-mentioned modules can be functional modules or program modules, which can be implemented by software or hardware. For modules implemented by hardware, the above-mentioned modules can be located in the same processor; or the above-mentioned modules can also be located in different processors in any combination.

[0114] The embodiment also provides a transformer vibration and partial discharge online monitoring method based on an optical fiber sensor. Figure 4 is a flowchart of the transformer vibration and partial discharge online monitoring method based on an optical fiber sensor according to the embodiment of the present application, as Figure 4 shown, the flow includes the following steps:

[0115] Step S401, based on the above-mentioned transformer vibration and partial discharge online monitoring device based on optical fiber sensor, collect transformer data, and get the optical signal data returned by the device, the transformer data includes vibration signal data on the surface of the transformer tank and partial discharge signal data of the transformer.

[0116] The transformer data collected in this embodiment includes vibration signal, partial discharge signal and auxiliary parameters (environmental temperature and transformer oil temperature). The collection accuracy of each signal is as follows:

[0117] Vibration signal: the optical fiber grating sensor collects the vibration of the transformer tank surface, covering the amplitude range of 0.1~50mm / s (resolution 0.01mm / s) and the frequency range of 0.1~100Hz (focus on monitoring 1~10Hz low frequency band).

[0118] Partial discharge signal: the partial discharge optical fiber sensor captures the change of partial discharge light intensity (0.1%~10%, resolution 0.1%) and the frequency range of 10kHz~500kHz.

[0119] Auxiliary parameters: synchronous collection of environmental temperature (-40℃~85℃, accuracy ±0.5℃), transformer oil temperature (0℃~100℃) and load rate (0%~120%).

[0120] Optionally, sampling rate: vibration signal 2.5MS / s, partial discharge signal 1MS / s (satisfying Nyquist theorem).

[0121] Storage mode: raw data is written into local SSD (storage capacity ≥1TB) in binary format in real time, and is named by timestamp (such as 20231001_120000.bin).

[0122] In this embodiment, the implementation steps of vibration and partial discharge measurement are as follows:

[0123] Step 1: record the initial reflection wavelength and temperature, the initial vibration fiber grating reflection wavelength λ 0，vibration And the reflection wavelength of the partial discharge optical fiber grating λ 0,partial .

[0124] Step 2: record the current reflection wavelength, the reflection wavelength of the vibration fiber grating λ vibration And the reflection wavelength of the partial discharge optical fiber grating λ partial .

[0125] Step 3: calculate the wavelength drift of the vibration fiber grating:

[0126] Step 4: calculate the wavelength drift of the partial discharge optical fiber grating:

[0127] Step 5: Calculate temperature change, calculate temperature change by referring to wavelength shift signal of fiber Bragg grating (used for temperature measurement): ΔT = Δλ temp / K T , where K T is the temperature sensitivity coefficient, which is calibrated according to actual experimental conditions, and Δλ temp represents the shift value of fiber Bragg grating reflection wavelength caused by temperature change.

[0128] Step 6: Calculate vibration and partial discharge after temperature compensation. Using the obtained wavelength shift and temperature change, temperature compensation is performed to obtain the vibration change and partial discharge change after temperature compensation.

[0129] Step S402, analyze the transformer data and optical signal data through the abnormality prediction model constructed based on the random forest algorithm to obtain the vibration characteristics and partial discharge characteristics of the transformer, and the optical signal characteristics.

[0130] The collected data is preprocessed, including data cleaning and feature extraction. Data cleaning: remove missing values and outliers to ensure data accuracy. Feature extraction: extract key features from raw data, such as vibration amplitude, frequency, partial discharge amplitude, etc.

[0131] Missing value processing: divided into short-term missing and long-term missing. Short-term missing: if the missing value time is less than the preset threshold (for example, 5 seconds), it is supplemented by linear interpolation or forward filling method. Long-term missing: if the missing value time is greater than or equal to the preset threshold, it is marked as invalid data segment.

[0132] Outlier detection includes vibration amplitude anomaly detection and partial discharge light intensity change anomaly detection. Vibration amplitude anomaly: detected by IQR method (such as Q3+1.5IQR as the upper limit, Q1-1.5IQR as the lower limit, and not within the upper and lower limit interval is considered abnormal). Partial discharge light intensity change: detected by 3σ criterion (outside the μ±3σ range is considered abnormal).

[0133] In some embodiments, step S402 specifically includes:

[0134] Step S4021, feature extraction is performed on the vibration signal data to obtain vibration characteristics, including vibration amplitude and vibration frequency.

[0135] According to the real-time vibration amplitude and vibration amplitude root mean square, the vibration amplitude is obtained.

[0136] Vibration amplitude root mean square (RMS) formula:

[0137]

[0138] The power spectral density (PSD) of the 1Hz-100Hz frequency band is extracted by FFT transformation to identify the dominant frequency component.

[0139] In step S4022, the partial discharge signal data is subjected to feature extraction to obtain partial discharge features, including partial discharge signal frequency, discharge signal intensity variation and signal-to-noise ratio.

[0140] Frequency separation: the signal is decomposed into 10kHz-100kHz (low-intensity partial discharge) and 100kHz-500kHz (high-intensity partial discharge) frequency bands using wavelet transformation.

[0141] The signal-to-noise ratio (SNR) calculation formula is as follows:

[0142]

[0143] Where P signal represents the partial discharge signal power, P noise represents the noise power.

[0144] The above data is subjected to standardization processing, and is normalized to the [0, 1] interval by the normalization formula as follows:

[0145]

[0146] Where x is the original value, x min is the minimum value in the original value, x max is the maximum value in the original value, and x norn is the normalized value of x.

[0147] In step S4023, temperature change data is obtained by a temperature-compensated optical fiber sensor, and the vibration features and partial discharge features are temperature-compensated according to the temperature change data to obtain corrected vibration features and partial discharge features.

[0148] The temperature compensation correction formula for the vibration signal is:

[0149]

[0150] The temperature compensation correction formula for the partial discharge signal is:

[0151]

[0152] Where A c is the corrected vibration feature value, B c is the corrected partial discharge feature value, ΔT is the temperature change, K T is the temperature sensitivity coefficient, K vibration and K partial are the sensitivity coefficients for the vibration signal and the partial discharge signal, respectively.

[0153] For example, the amplitude drift is corrected according to the vibration signal correction formula, and the amplitude drift correction formula is as follows:

[0154]

[0155] wherein A comp is the vibration amplitude after temperature compensation, and the unit is mm / s; Δλ vib is the wavelength change value measured by the sensor, and the unit is pm (picometer), which contains the results of vibration and temperature joint effects; ΔT is the temperature change, and the unit is ℃; K T is the temperature sensitivity coefficient, which represents the wavelength change when the temperature changes by 1 ℃, and the unit is pm / ℃; K vib is the vibration sensitivity coefficient, which represents the wavelength change when the vibration speed changes by 1 mm / s, and the unit is pm / (mm / s).

[0156] In step S4024, the light signal data is subjected to feature extraction to obtain light signal features, and the light signal features include light intensity change rate.

[0157] Light intensity change rate calculation: the light intensity change rate is calculated by real-time monitoring of the reflected light intensity change, and the calculation formula is as follows:

[0158]

[0159] wherein P current is the reflected light intensity measured at the current time, which represents the light signal intensity detected at the current time; P0 is the initial reference light intensity, which represents the light signal intensity under a certain initial state or reference state; ΔP% is the light intensity change rate, which represents the percentage change of the light intensity relative to the initial light intensity, and is used to reflect the degree of light intensity change.

[0160] In step S403, the pre-warning signal is generated according to the vibration features, partial discharge features, light signal features and preset grading pre-warning mechanism.

[0161] If the model output is abnormal, the pre-warning mechanism is triggered to prompt the maintenance personnel to check or handle it. The pre-warning event is recorded as the basis for subsequent analysis and improvement of the model.

[0162] According to the actual fault condition and the monitoring data, feedback information is collected, and the random forest model is optimized and adjusted in time to improve its prediction accuracy and robustness.

[0163] Through the above steps, the transformer data is collected by the transformer vibration and partial discharge online monitoring device based on fiber optic sensor. The monitoring device uses high spatial resolution fiber optic grating array layout and precise light source pulse design, which can achieve sub-meter spatial resolution and accurately locate multiple disturbance events. The system can not only monitor multiple events simultaneously, but also accurately locate the position of each event in real time, which is suitable for multi-point detection tasks in dynamic environment.

[0164] The random forest algorithm is used for signal classification and analysis, which can efficiently process echo signals from fiber optic gratings. As a powerful classifier based on decision tree ensemble learning, random forest algorithm has excellent performance in identifying multiple disturbance events. Random forest algorithm can automatically extract features from high-dimensional signals and classify them, greatly improving the ability to extract effective signals from complex signals such as vibration, noise, etc., thereby improving the accuracy and robustness of the system.

[0165] The system can transmit the measurement data collected by the fiber optic sensor to the remote platform (control system) in real time through the remote data transmission module. Combined with real-time analysis of data using random forest algorithm, remote monitoring and alarm processing of data can be effectively realized. Through the Internet of Things (IoT) technology, the system can upload the analysis results to the cloud platform and perform long-term storage, trend analysis and alarm notification, which is suitable for large-scale monitoring systems and security warning applications.

[0166] In some embodiments, step S403 specifically includes:

[0167] Step S4031, in response to any one of the vibration amplitude, vibration frequency and light intensity change rate being greater than the corresponding critical vibration threshold, a critical vibration warning is performed.

[0168] Step S4032, in response to any one of the vibration amplitude, vibration frequency and light intensity change rate being greater than the corresponding warning vibration threshold, and the vibration amplitude, vibration frequency and light intensity change rate being less than or equal to the corresponding critical vibration threshold, a warning vibration warning is performed.

[0169] Step S4033, in response to any one of the partial discharge signal frequency, discharge signal intensity change and signal-to-noise ratio being greater than the corresponding first partial discharge threshold, a high-intensity partial discharge warning is performed.

[0170] Step S4034, in response to any one of the partial discharge signal frequency, discharge signal intensity change and signal-to-noise ratio being greater than the corresponding second partial discharge threshold, and the partial discharge signal frequency, discharge signal intensity change and signal-to-noise ratio being less than the corresponding first partial discharge threshold, a low-intensity partial discharge warning is performed.

[0171] The abnormality detection result is divided into normal situation, serious situation and critical situation.

[0172] Vibration early warning is divided into two categories: critical vibration early warning and warning vibration early warning. When the critical vibration occurs, the early warning result is emergency shutdown or maintenance inspection. When the warning vibration occurs, the early warning result is routine inspection and monitoring of the vibration condition.

[0173] Partial discharge early warning is divided into two categories: high-intensity partial discharge early warning and low-intensity partial discharge early warning. When the high-intensity partial discharge occurs, the early warning result is immediate shutdown and electrical insulation test to confirm the partial discharge source. When the low-intensity partial discharge occurs, the early warning result is further monitoring and insulation performance inspection of the equipment.

[0174] For example, the specific early warning threshold settings are as follows:

[0175] Warning vibration threshold: vibration amplitude is 1-2 mm / s; optical signal intensity change (change of fiber Bragg grating reflection signal) is about 1-5%; vibration frequency is low frequency vibration 1 Hz to 10 Hz.

[0176] Critical vibration threshold: vibration amplitude exceeds 5 mm / s; optical signal intensity change (change of fiber reflection signal) is greater than 5%; vibration frequency exceeds 10 Hz and may be accompanied by a large mechanical impact.

[0177] Low-intensity partial discharge early warning: discharge signal intensity change (change of fiber reflection signal) is 1%-5%; frequency range is 10 kHz-100 kHz; signal-to-noise ratio > 10 dB.

[0178] High-intensity partial discharge early warning: discharge signal intensity change (change of fiber reflection signal) is greater than 5%, frequency range is 100 kHz-500 kHz, signal-to-noise ratio > 15 dB.

[0179] In some embodiments, the method further comprises:

[0180] Step S404, generating parameter adjustment instructions of the laser and the acousto-optic modulator based on the optical signal characteristics.

[0181] Automatically modulating the parameters of the monitoring device according to the measured optical signal characteristics, including the parameters of the laser and the acousto-optic modulator.

[0182] Step S405, detecting the optical fiber link of the transformer vibration and partial discharge online monitoring device based on the optical fiber sensor, obtaining the loss data of the optical fiber link, and generating a link alarm signal when the loss data is greater than a preset loss threshold.

[0183] The present embodiment further ensures the accuracy of the collected data through sensor calibration and link monitoring.

[0184] Sensor calibration: calibration is performed using a standard vibration table (accuracy ±0.5%) and a partial discharge simulator (error ≤3%) at regular intervals (e.g., every month).

[0185] Link monitoring: real-time detection of fiber link loss, triggering an alarm when the loss data is greater than the preset loss threshold (e.g., 3dB).

[0186] In some embodiments, the method further comprises:

[0187] Step S501, collecting a transformer dataset for training an abnormality prediction model, and marking the data in the transformer dataset as normal or abnormal according to historical fault data and actual transformer state to obtain a marked dataset.

[0188] In some embodiments, the marking of the data in the transformer dataset as normal or abnormal in step S501 comprises:

[0189] Step S5011, extracting historical fault records of the transformer substation where the transformer is located, and obtaining transformer monitoring data corresponding to the time stamp of the historical fault records.

[0190] Step S5012, obtaining a transformer operation log, and marking the data in the transformer dataset as normal or abnormal according to the transformer monitoring data and the transformer operation log based on a preset marking rule to obtain a marked dataset.

[0191] Extracting historical fault records of the substation (e.g., winding deformation, core loosening, partial discharge), and associating the monitoring data corresponding to the time stamp. Optionally, marking according to the transformer operation log (e.g., oil temperature mutation, relay protection action).

[0192] The marking rule can be set as follows:

[0193] Vibration amplitude <1mm / s and partial discharge light intensity change <1% is normal state.

[0194] If any of the following conditions is met, it is an abnormal state:

[0195] Vibration amplitude ≥1mm / s or partial discharge light intensity change ≥1%.

[0196] Oil temperature >85℃ or load rate >120%.

[0197] The Kappa coefficient calculation formula for consistency test is as follows:

[0198]

[0199] Where, P o is the observed consistency, i.e., the proportion of agreement between evaluators in actual observation; P eFor chance agreement, i.e., the expected agreement that raters would achieve by chance (or by accident) in a random (or accidental) situation.

[0200] Optionally, K ≥ 0.8 is considered to pass the consistency test.

[0201] The model construction module constructs an anomaly prediction model through a random forest algorithm, and trains and evaluates the anomaly prediction model based on the labeled data set.

[0202] The labeled data set is divided into a training set and a test set. Optionally, the training set (70%) and the test set (30%) are divided in a ratio of 7:3 to ensure that the proportions of the two types of samples are consistent. Example: total samples 10,000 (normal 7,000, abnormal 3,000) → training set 7,000 (normal 4,900, abnormal 2,100), test set 3,000 (normal 2,100, abnormal 900).

[0203] 10% of the training set is further divided as a validation set (e.g., 700), which is used for model parameter tuning.

[0204] Using the training set data, an anomaly prediction model is constructed through a random forest algorithm. Appropriate parameters (such as the number of trees, maximum depth, etc.) are set to optimize the performance of the model. Table 1 is a parameter setting table of an anomaly prediction model according to an embodiment of the present application.

[0205] Table 1

[0206]

[0207] Multi-threading (4 cores) is used to accelerate training, and the training time of a single tree is less than 100 ms.

[0208] The training set is used to train the random forest model so that it can learn the features of normal and abnormal states in the data.

[0209] Grid search (GridSearchCV) is used to iterate and optimize the parameter combinations, and the F1-score of the validation set is used as the evaluation index.

[0210] Example optimal parameters: n_estimators=100, max_depth=10, max_features='sqrt'.

[0211] Training process: input the training set feature vector (vibration amplitude, frequency, temperature compensation value + partial discharge light intensity change rate, frequency, oil temperature, load rate, total 7 dimensions). Output binary classification probability value (normal / abnormal), and the threshold is initially set to 0.5.

[0212] In step S502, an anomaly prediction model is constructed through a random forest algorithm, and the anomaly prediction model is trained and evaluated based on the labeled data set.

[0213] In some embodiments, the labeled dataset includes a training set and a test set; training and evaluating the anomaly prediction model based on the labeled dataset in step S502 includes:

[0214] In step S5021, the trained anomaly prediction model is tested based on the test set to obtain evaluation indicators of the anomaly prediction model, including prediction accuracy, recall, F1-score, specificity, and AUC-ROC, and the anomaly prediction model is evaluated based on the evaluation indicators.

[0215] Recall reflects the detection ability of faults, and power scenarios need to prioritize high recall to avoid missed reports leading to equipment damage. F1-score balances precision and recall to balance fault detection efficiency and maintenance cost. Specificity measures the recognition ability of normal state to reduce false alarms of normal equipment. AUC-ROC is a comprehensive indicator, and the closer the value is to 1, the stronger the model's ability to distinguish between normal and abnormal.

[0216] Optionally, the model is qualified if it meets the following parameters: model accuracy ≥ 95%, recall ≥ 90%, F1-score ≥ 90%, specificity ≥ 95%, and AUC-ROC ≥ 0.98.

[0217] In step S5022, the anomaly prediction model is tested based on the test set to obtain the accuracy of the anomaly prediction model, and the drift rate of the accuracy is determined. If the drift rate is greater than a preset drift threshold, the anomaly prediction model is updated.

[0218] The model performance is detected for degradation, including drift monitoring and distribution detection.

[0219] Drift monitoring: calculate the accuracy drift rate, formula:

[0220]

[0221] When the drift exceeds 5%, the model is updated.

[0222] Distribution detection: use KL divergence (D KL (P||Q)) to monitor the difference between the new and old data distributions. If D KL >0.1, data augmentation is started.

[0223] According to the actual fault conditions and monitoring data, feedback information is collected to optimize and adjust the random forest model in a timely manner, improving its prediction accuracy and robustness.

[0224] Collect vibration / partial discharge data (such as winding deformation, partial discharge) when the transformer actually fails, and ensure that the proportion of fault samples in the new data is ≥20% (formula: fault data ratio = fault sample number / total sample number × 100%).

[0225] Incrementally train the random forest model using new data, with the number of new decision trees being 10%-30% of the original model, to avoid retraining with full data.

[0226] Optionally, the features are screened and expanded. Based on Gini importance, core features are screened, and features with a contribution of less than 5% (such as redundant vibration frequency components) are removed. Among them,

[0227]

[0228] Fusion of oil temperature, load rate and other auxiliary parameters to construct a multi-dimensional feature vector x:

[0229] x=[A comp , ΔP%, oil temperature, load rate]

[0230] It should be noted that the threshold in this application can be adaptively adjusted. The dynamic threshold adaptive adjustment method is as follows:

[0231] False / missed alarm control: use binary search to find the optimal classification threshold θ * , minimize the weighted false / missed alarm rate θ * =argmin(λ˙FRP+(1-λ)FNP), where λ=0.3, FRP and FNP represent the false / missed alarm rate, respectively, and focus on reducing the false / missed alarm rate.

[0232] Threshold range: the vibration amplitude threshold is dynamically adjusted in the range of 1-5 mm / s, and the partial discharge light intensity change threshold is 1%-5%.

[0233] By using the random forest algorithm to classify and analyze the signals, it can efficiently process the echo signals from the fiber grating. The random forest algorithm can automatically extract features from high-dimensional signals and classify them, greatly improving the ability to extract effective signals from complex signals (such as vibration, noise, etc.), thereby improving the accuracy and robustness of the system.

[0234] The measurement data collected by the fiber grating sensor is transmitted to the remote platform in real time. Combined with the random forest algorithm, the data is analyzed in real time, which can effectively realize the remote monitoring and alarm processing of the data. Through the Internet of Things (IoT) technology, the system can upload the analysis results to the cloud platform and perform long-term storage, trend analysis and alarm notification, which is suitable for large-scale monitoring systems and security warning applications.

[0235] It should be noted that the steps shown in the above flow or the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from here.

[0236] The embodiment also provides an electronic device including a memory and a processor, the memory storing a computer program, and the processor being configured to execute the computer program to perform the steps in any of the method embodiments described above.

[0237] Optionally, the electronic device described above can further include a transmission device connected with the processor and an input and output device connected with the processor.

[0238] Optionally, in the embodiment, the processor can be configured to execute the following steps through the computer program:

[0239] S1, based on the transformer vibration and partial discharge online monitoring device based on the optical fiber sensor of the first aspect, collecting transformer data and obtaining the optical signal data returned by the device, the transformer data including the vibration signal data of the transformer tank surface and the partial discharge signal data of the transformer.

[0240] S2, analyzing the transformer data and the optical signal data through the abnormality prediction model constructed based on the random forest algorithm to obtain the vibration characteristics and the partial discharge characteristics of the transformer, and the optical signal characteristics.

[0241] S3, generating a warning signal according to the vibration characteristics, the partial discharge characteristics, the optical signal characteristics and the preset grading warning mechanism.

[0242] It should be noted that the specific examples in the embodiment can refer to the examples described in the above embodiments and optional implementation manners, which will not be described here again.

[0243] In one embodiment, Figure 5 is a schematic diagram of the internal structure of an electronic device according to an embodiment of the present application, as Figure 5 shown, an electronic device is provided, which can be a server, and the internal structure diagram thereof can be as Figure 5As shown. The electronic device includes a processor, a memory, a network interface and a database connected through a system bus. Among them, the processor of the electronic device is used to provide computing and control capabilities. The memory of the electronic device includes a non-volatile storage medium, an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operating system and the computer program in the non-volatile storage medium to run. The database of the electronic device is used to store data. The network interface of the electronic device is used to communicate with the external terminal through the network connection. The computer program is executed by the processor to implement a transformer vibration and partial discharge online monitoring method based on an optical fiber sensor.

[0244] Those skilled in the art can understand that, Figure 5 The structure shown in the figure is only a block diagram of part of the structure related to the scheme of the present application, and does not constitute a limitation on the electronic device to which the scheme of the present application is applied. The specific electronic device can include more or fewer components than those shown in the figure, or combine certain components, or have a different component arrangement.

[0245] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing related hardware through a computer program. The computer program can be stored in a non-volatile computer readable storage medium. When the computer program is executed, it can include the processes of the above-mentioned embodiments of each method. Any reference to memory, storage, database or other medium used in each embodiment provided by the present application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. As an illustration but not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDR SDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM) and the like.

[0246] Those skilled in the art should understand that each technical feature of the above-described embodiments can be combined arbitrarily, and for the sake of brevity, each technical feature of the above-described embodiments is not described in all possible combinations, however, as long as the combinations of the technical features do not exist, it should be considered that it is within the scope of the description.

[0247] The above-described embodiments only express several implementation manners of the present application, the description is more specific and detailed, but it should not be understood as a limitation on the patent scope of the application. It should be pointed out that for ordinary skilled in the art, without departing from the concept of the present application, a number of modifications and improvements can be made, which are within the protection scope of the present application. Therefore, the protection scope of the patent of the present application should be subject to the appended claims.

Claims

1. A fiber optic sensor based transformer vibration and partial discharge on-line monitoring device, characterized in that, The device comprises: a laser for emitting a laser signal; an acousto-optic modulator for adjusting the frequency and wavelength of the laser signal in response to a parameter control signal; an optical fiber sensor for reflecting the laser signal and causing Bragg wavelength shift of the reflected laser signal in response to vibration or partial discharge of a transformer, the optical fiber sensor comprising a plurality of symmetric fiber gratings and counterweights fixed at both ends of each of the fiber gratings, the plurality of fiber gratings being symmetrically arranged to enable the sensor to respond sensitively to vibration in different directions, improve the adaptability and detection capability of the sensor in a complex vibration environment, and enhance the stability of the entire sensor structure, and the counterweights being hollow thin-walled structures that are more conducive to stress concentration on the fiber gratings during vibration, improving the sensor's ability to perceive slight vibrations while reducing energy loss during vibration transmission; the optical fiber sensor further comprising a sensitive film and an encapsulating shell, the sensitive film being fixed to the encapsulating shell, and the counterweights being fixed to the sensitive film by epoxy resin; a photodetector for receiving the optical signal returned by the optical fiber sensor and converting the optical signal into an electrical signal.

2. The apparatus of claim 1, wherein, The device further comprises: an arbitrary waveform generator for generating a parameter control signal and sending the parameter control signal to the acousto-optic modulator; a first doped fiber amplifier for amplifying the laser signal modulated by the acousto-optic modulator; an amplified spontaneous emission filter for removing spontaneous emission noise generated during amplification of the laser signal; a circulator for transmitting the processed laser signal to the optical fiber sensor and receiving the optical signal returned by the optical fiber sensor; a second doped fiber amplifier for amplifying the optical signal returned by the optical fiber sensor and sending the amplified optical signal to the photodetector; a data acquisition card for converting the electrical signal obtained by the photodetector into a digital signal.

3. A method for on-line monitoring of transformer vibrations and partial discharges based on fiber-optic sensors, characterized in that, The method comprises: collecting transformer data and obtaining optical signal data returned by the device based on the optical fiber sensor-based transformer vibration and partial discharge online monitoring device according to any one of claims 1-2, wherein the transformer data comprises vibration signal data of the surface of the transformer tank and partial discharge signal data of the transformer; analyzing the transformer data and the optical signal data by an abnormality prediction model constructed based on a random forest algorithm to obtain vibration characteristics, partial discharge characteristics, and optical signal characteristics of the transformer; generating an early warning signal according to the vibration characteristics, the partial discharge characteristics, the optical signal characteristics, and a preset grading early warning mechanism.

4. The method of claim 3, wherein, The analysis of the transformer data and the optical signal data comprises: extracting features from the vibration signal data to obtain vibration characteristics, wherein the vibration characteristics comprise vibration amplitude and vibration frequency; extracting features from the partial discharge signal data to obtain partial discharge characteristics, wherein the partial discharge characteristics comprise partial discharge signal frequency, discharge signal intensity variation, and signal-to-noise ratio; obtaining temperature change data by a temperature-compensated optical fiber sensor, and temperature-compensating the vibration characteristics and the partial discharge characteristics according to the temperature change data to obtain corrected vibration characteristics and partial discharge characteristics; extracting features of the optical signal data to obtain optical signal features, the optical signal features including a light intensity change rate.

5. The method of claim 4, wherein, The generating a warning signal according to the vibration features, the partial discharge features, the optical signal features, and a preset hierarchical warning mechanism comprises: performing a critical vibration warning in response to any one of the vibration amplitude, the vibration frequency, and the light intensity change rate being greater than a corresponding critical vibration threshold value; and / or performing a warning vibration warning in response to any one of the vibration amplitude, the vibration frequency, and the light intensity change rate being greater than a corresponding warning vibration threshold value, and all of the vibration amplitude, the vibration frequency, and the light intensity change rate being less than or equal to the corresponding critical vibration threshold value; and / or performing a high-intensity partial discharge warning in response to any one of the partial discharge signal frequency, the discharge signal intensity change, and the signal-to-noise ratio being greater than a corresponding first partial discharge threshold value; and / or performing a low-intensity partial discharge warning in response to any one of the partial discharge signal frequency, the discharge signal intensity change, and the signal-to-noise ratio being greater than a corresponding second partial discharge threshold value, and all of the partial discharge signal frequency, the discharge signal intensity change, and the signal-to-noise ratio being less than the corresponding first partial discharge threshold value.

6. The method of claim 3, wherein, The method further comprises: generating parameter adjustment instructions for the laser and the acousto-optic modulator based on the optical signal features; and / or detecting an optical fiber link of an optical fiber sensor-based transformer vibration and partial discharge online monitoring device to obtain loss data of the optical fiber link, and generating a link alarm signal in a case where the loss data is greater than a preset loss threshold value.

7. The method of claim 3, wherein, The method further comprises: collecting a transformer data set for training the abnormality prediction model, and performing normal or abnormal labeling on data in the transformer data set to obtain a labeled data set based on historical fault data and an actual state of the transformer; constructing an abnormality prediction model through a random forest algorithm, and training and evaluating the abnormality prediction model based on the labeled data set.

8. The method of claim 7, wherein, The performing normal or abnormal labeling on data in the transformer data set comprises: extracting historical fault records of a substation where the transformer is located to obtain transformer monitoring data corresponding to time stamps of the historical fault records; obtaining a transformer operation log, and performing normal or abnormal labeling on data in the transformer data set based on a preset labeling rule and based on the transformer monitoring data and the transformer operation log to obtain a labeled data set.

9. The method of claim 7, wherein, The labeled data set includes a training set and a test set; and the training and evaluating the abnormality prediction model based on the labeled data set comprises: testing the trained abnormality prediction model based on the test set to obtain evaluation indexes of the abnormality prediction model, the evaluation indexes including a prediction accuracy, a recall rate, an F1-score, a specificity, and an AUC-ROC, and evaluating the abnormality prediction model based on the evaluation indexes; and / or Test the anomaly prediction model based on the test set to obtain an accuracy rate of the anomaly prediction model, determine a drift rate of the accuracy rate, and trigger updating the anomaly prediction model in a case where the drift rate is greater than a preset drift threshold.

Images