Method and system for predicting health state of transformer based on fiber grating sensor

Abstract

This invention discloses a method and system for predicting the health status of a transformer based on fiber Bragg grating sensors. The method includes: deploying fiber Bragg grating sensors at key locations inside the transformer to collect temperature and strain data, obtaining temperature and strain data ranges for each key location measuring point under different operating conditions; calculating a weighted transformer health status index system; constructing a transformer internal index prediction model based on a recurrent neural network, using the weighted transformer health status index system and operating environment parameters as training data to train the prediction model; setting fiber Bragg grating sensors at key measuring points of the transformer to be predicted, obtaining predicted temperature and predicted strain data for the transformer in a future time period through the trained transformer internal index prediction model; and combining the temperature and strain data ranges to obtain the predicted health status of the transformer. This invention can accurately predict the internal state of a transformer.

Description

Technical Field

[0001] This invention belongs to the technical field of online monitoring of power equipment, and more specifically, relates to a method and system for predicting the health status of transformers based on fiber optic grating sensors. Background Technology

[0002] Fiber Bragg grating (FBG) sensors offer advantages such as strong resistance to electromagnetic interference, flexible deployment, and high measurement accuracy. They can simultaneously measure temperature and stress data. When a transformer experiences faults such as partial discharge, winding deformation, or insulation damage, FBG sensors can detect temperature changes, vibrations, and stress changes. This information effectively reflects the internal state of the transformer and is of great significance for ensuring its normal operation.

[0003] Deep learning technology is a method by which computers utilize large amounts of existing data to analyze and calculate mathematical models, thereby understanding the inherent patterns in the data. The concept of deep learning originated from artificial neural networks, and today various neural network models have been developed for different scenarios. These models possess strong feature extraction capabilities and are suitable for processing large amounts of complex data. Transformers have complex internal structures, require numerous measurement points, and have many fault parameters. Therefore, assessing their condition and diagnosing faults requires strong data processing and feature extraction capabilities, which traditional data processing methods struggle to meet.

[0004] Traditional methods for assessing the health status of transformers include: 1) Electrical signal sensor measurement, such as thermocouples and resistance temperature detectors (RTDs), which are susceptible to electromagnetic interference, have limited lifespan, and produce less than ideal measurement results; 2) Infrared thermometry, which is a non-contact measurement method that is easy to operate manually, but is susceptible to background noise and electromagnetic interference; and 3) Point-based temperature measurement using fluorescent optical fibers, which has limited measurement points and makes it difficult to accurately determine the internal health status of the transformer. Therefore, new methods for measuring the internal condition of transformers are needed.

[0005] Prior art document 1 discloses a power transformer fault diagnosis method based on acoustic features and neural networks. This method involves establishing and training a GRU neural network model; acquiring the sound signal of the power transformer to be diagnosed; preprocessing the signal; and inputting it into the trained GRU neural network model. The fault diagnosis of the power transformer is then completed based on the output of the GRU neural network model. Prior art document 2 discloses an online vibration monitoring system for oil-immersed transformers with built-in fiber optic grating sensors. This system arranges fiber optic grating accelerometers inside the transformer and uses wavelength division / space division hybrid multiplexing technology to form a quasi-distributed sensor network. This enables direct measurement of the three-dimensional vibration acceleration of the transformer body during operation, providing a new approach for real-time online monitoring of transformer vibration and an effective technical means for accurate assessment of transformer operating status.

[0006] The existing technology has the following technical defects:

[0007] (1) The location of the sensor cannot accurately reflect the internal environment of the transformer, resulting in low measurement accuracy and susceptibility to interference;

[0008] (2) Existing diagnostic methods are not suitable for handling the complex internal state of transformers, and the diagnostic accuracy cannot meet the requirements.

[0009] (3) The measured characteristic quantities are relatively simple and cannot fully reflect the health status of the transformer under various operating environments. Summary of the Invention

[0010] To address the shortcomings of existing technologies, this invention provides a transformer health status prediction method based on fiber optic grating sensors. This method can effectively assess the internal health status of the transformer and identify faults and potential hazards after acquiring temperature and stress data at various measuring points inside the transformer.

[0011] The present invention adopts the following technical solution.

[0012] A transformer health status prediction design and module based on fiber Bragg grating sensors includes the following steps:

[0013] Step 1: Place fiber Bragg grating sensors at key locations inside the transformer. The fiber Bragg grating sensors collect temperature and strain data at the key measurement points according to a preset acquisition frequency.

[0014] Step 2: Simulate the operating environment of the transformer and obtain the temperature and strain data of key measuring points of the transformer under different operating conditions through fiber optic grating sensors to obtain the temperature range and strain data range of each key measuring point of the transformer under different operating conditions.

[0015] Step 3: Normalize and weight the operating environment simulation parameters used in Step 2, as well as the temperature and strain data collected by the fiber optic grating sensor at each measuring point, to obtain a weighted transformer health status index system.

[0016] Step 4: Construct a transformer internal index prediction model based on a recurrent neural network. Use the weighted transformer health status index system and operating environment parameters obtained in Step 3 as training data to train the prediction model and obtain the trained transformer internal index prediction model.

[0017] Step 5: Set up fiber optic grating sensors at key measurement points of the transformer to be predicted, collect temperature and strain data of each measurement point over a period of time, as well as the operating environment parameters of the location of the transformer to be predicted, and obtain the predicted temperature and predicted strain data of the transformer to be predicted in the future time period through the trained transformer internal index prediction model.

[0018] Step 6: Based on the predicted temperature and predicted strain data, and combined with the temperature range and strain data range of each key measuring point of the transformer under different operating conditions obtained in Step 2, the health status prediction result of the transformer to be predicted is obtained.

[0019] Preferably, in step 1, the fiber optic grating sensor is placed at a key measuring point inside the transformer, and the acquisition frequency is set to acquire temperature and strain data at the key measuring point according to the acquisition frequency.

[0020] Key locations of a transformer include the upper, lower, and middle ends of the transformer windings, the top of the transformer tank, and the ends of the core.

[0021] Preferably, in step 2, the relevant parameters of the simulated transformer operating environment include: ambient temperature, load current, and voltage; the simulated transformer operating state includes normal transformer state and fault state, and the fault state also includes: loose transformer windings, loose iron core, and overheating operation; under different transformer operating environments, the temperature and strain data of key measuring points are collected at a preset time period using fiber optic grating sensors according to a preset acquisition frequency for transformers in different operating states, and the temperature range and strain data range of each key measuring point under different operating states are obtained based on the maximum and minimum values ​​of the temperature and strain data of each measuring point.

[0022] Preferably, in step 2, when simulating the transformer operating environment parameters, the simulated ambient temperature includes the average temperature of summer, winter and spring / autumn in the location of the transformer under test, the simulated load current includes 40%, 60%, 80% and 100% of the rated load current of the transformer, and the simulated voltage includes the voltage of each tap of the transformer.

[0023] Preferably, step 3 further includes:

[0024] Step 3-1: Normalize the temperature and strain data of key transformer measurement points obtained by fiber optic grating sensors under different operating environments in Step 2, as well as the corresponding external factor data of ambient temperature, load current, and voltage.

[0025] Step 3-2: By randomly configuring the network and normalizing the data, the temperature and strain data of each measuring point of the transformer are weighted and calculated to obtain a weighted transformer health status index system corresponding to different operating states and different operating environments.

[0026] Preferably, step 3-2 includes: for transformers in different operating states, the normalized data obtained in step 2 and the transformer operating state are input into a random configuration network. The normalized data includes normalized data of ambient temperature, load current, and voltage related to the operating environment, as well as normalized data of temperature and strain data at each measuring point of the transformer. The correlation between temperature and strain data at each measuring point and external influencing factors and various typical faults is extracted through the random configuration network, and the specific weights of temperature and strain data at each measuring point are obtained.

[0027] Preferably, step 5 further includes: collecting temperature and strain data of each measuring point of the transformer to be predicted during time period T1, normalizing the data, and combining it with the weight values ​​calculated in step 3 to form a weighted transformer health status index system; collecting and normalizing the operating environment parameters of the location of the transformer to be predicted, including ambient temperature, transformer load current and voltage; and using the weighted transformer health status index system and operating environment parameters obtained after processing as input to the trained transformer internal index prediction model to obtain the predicted temperature and predicted strain data of the transformer in the next time period T2.

[0028] Preferably, step 6 further includes: comparing the range of temperature and strain data at key points in normal and fault states with the predicted temperature and strain data to determine whether the internal operating state of the transformer to be predicted is normal, and if a fault exists, specifically predicting the type of fault state, including transformer winding loosening, core loosening, and overheating operation.

[0029] The present invention also provides a transformer health status prediction system based on fiber Bragg grating sensors, comprising:

[0030] The data acquisition module includes a fiber optic grating sensor installed inside the transformer to collect temperature and strain data at the measuring point, as well as sensors for collecting ambient temperature, current, and voltage.

[0031] The data processing module performs row normalization on the collected temperature and strain data;

[0032] Run the simulation module to simulate the operating environment and state of the transformer, and obtain the temperature range and strain data range of each key measuring point of the transformer under different operating conditions;

[0033] The weight calculation module calculates the weights of temperature and strain data at various measuring points of the transformer by randomly configuring the network and normalizing the data, and obtains a weighted transformer health status index system corresponding to different operating states and different operating environments.

[0034] The prediction module constructs a transformer internal index prediction model based on a recurrent neural network. It is trained using data processed by the data processing module and a weighted transformer health status index system obtained by the weight calculation module. The trained transformer internal index prediction model can predict the temperature and strain data of each measuring point of the transformer. Based on the predicted temperature and strain data of each measuring point of the transformer, and the temperature range and strain data range of each key measuring point obtained by the simulation module, the predicted operating status of the transformer is obtained.

[0035] The beneficial effects of this invention are that, compared with the prior art, this invention can effectively assess the internal health status of a transformer after acquiring temperature and stress data at various measuring points inside the transformer, identify faults and potential hazards, realize automated detection of power equipment, improve the reliability and stability of transformer operation, promptly detect potential faults and hazards, reduce the probability of faults, reduce maintenance time and downtime due to faults, and improve the economy and operating efficiency of the power system. Attached Figure Description

[0036] Figure 1 This is a flowchart of the transformer internal health state prediction method in this invention;

[0037] Figure 2 This is a schematic diagram of the network structure of the randomly configured network in this invention;

[0038] Figure 3 This is a schematic diagram of the neural unit structure of the gated recurrent neural network in this invention;

[0039] Figure 4 This is a structural diagram of the transformer internal health status prediction system in this invention. Detailed Implementation

[0040] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of this invention. The embodiments described in this application are merely some embodiments of this invention, and not all embodiments. Based on the spirit of this invention, other embodiments obtained by those skilled in the art without creative effort are all within the protection scope of this invention.

[0041] like Figure 1 As shown, this invention provides a method for predicting the health status of a transformer based on a fiber Bragg grating sensor. The method specifically includes the following steps:

[0042] Step 1: Place fiber Bragg grating sensors at key locations inside the transformer. The fiber Bragg grating sensors collect temperature and strain data at the key measurement points according to a preset acquisition frequency.

[0043] In step 1, fiber Bragg grating sensors need to be installed inside the transformer. Fiber Bragg grating sensors can be directly installed inside the transformer and are easy to deploy a large number of measurement points. The fiber Bragg grating sensors are set at key measurement points inside the transformer, and the acquisition frequency is set. Temperature and strain data at the key measurement points are collected according to the acquisition frequency.

[0044] Key locations include the upper end, lower end, and middle point of the transformer windings, the top of the transformer tank, and the ends of the core. Specifically:

[0045] According to finite element simulation, the temperature is highest near the top of the winding, making it a key area for temperature monitoring; the temperature is lowest at the bottom of the winding, and a large temperature difference between the top and bottom will seriously affect the service life of the transformer, so the temperature at the bottom of the winding must be monitored simultaneously. According to Ampere force calculation, the radial Ampere force in the middle of the winding is the largest, making it a key area for deformation monitoring. Excessive transformer oil temperature will accelerate oil deterioration and affect heat dissipation of the winding and core, with the highest oil temperature at the top of the transformer, so the oil temperature at the top of the transformer needs to be monitored. The transformer core is tightened by end clamps, and if the clamps loosen, it will aggravate core vibration, so the deformation at the ends of the core needs to be monitored.

[0046] By deploying multiple sets of fiber optic grating sensors to measure temperature and strain data at several key locations, the obtained internal condition information of the transformer is more comprehensive and can better reflect the internal health status of the transformer.

[0047] Step 2: Simulate the operating environment of the transformer and obtain the temperature and strain data of key measuring points of the transformer under different operating conditions through fiber optic grating sensors to obtain the temperature range and strain data range of each key measuring point of the transformer under different operating conditions.

[0048] Specifically, the relevant parameters of the simulated transformer operating environment include: ambient temperature, load current, and voltage; the simulated transformer operating states include normal and fault states, with fault states including: loose transformer windings, loose iron core, and overheating operation; under different transformer operating environments, temperature and strain data at key measuring points are collected at a preset acquisition frequency and within a preset time period for transformers in different operating states using fiber optic grating sensors, and based on the maximum and minimum values ​​of temperature and strain data at each measuring point, the temperature and strain data ranges of each key measuring point of the transformer under different operating states are obtained.

[0049] Furthermore, when simulating the operating environment parameters of the transformer, the ambient temperature includes the average temperature of summer, winter and spring / autumn in the location of the transformer under test, the load current includes 40%, 60%, 80% and 100% of the rated load current of the transformer, and the voltage includes the voltage of each tap.

[0050] Based on the maximum and minimum values ​​of the temperature and strain data collected from each measuring point, the temperature range and strain data range of each key measuring point under different operating conditions are obtained. The upper and lower limits of the range are the maximum and minimum values ​​of the data collected at that measuring point, respectively.

[0051] Step 3: Normalize and weight the operating environment simulation parameters used in Step 2, as well as the temperature and strain data collected by the fiber optic grating sensor at each measuring point, to obtain a weighted transformer health status index system.

[0052] Step 3 specifically includes:

[0053] Step 3-1: Normalize the temperature and strain data of key transformer measurement points obtained by fiber optic grating sensors under different operating environments in Step 2, as well as the corresponding external factor data of ambient temperature, load current, and voltage.

[0054] Specifically, to facilitate subsequent processing, the various data collected in step 2 are normalized to make the data dimensionless. The normalization formula is:

[0055]

[0056] In the formula, ω represents the raw data, including the data collected by the fiber optic grating sensor in step 2 or external factor data. For the normalized data, ω max and ω min These are the maximum and minimum numbers in the original data, respectively.

[0057] Step 3-2: By randomly configuring the network and normalizing the data, the temperature and strain data of each measuring point of the transformer are weighted and calculated to obtain a weighted transformer health status index system corresponding to different operating states and different operating environments.

[0058] like Figure 2 As shown, in this invention, a randomized network (SCN) is used to calculate the weight values ​​of temperature and strain data at each measuring point of the transformer under different operating environments and states. Specifically, for transformers in different operating states, the normalized data obtained in step 2 and the transformer operating state are input into the randomized network. The normalized data includes normalized data of ambient temperature, load current, and voltage related to the operating environment, as well as normalized data of temperature and strain data at each measuring point of the transformer. The correlation between temperature and strain data at each measuring point and external influencing factors and various typical faults is extracted through the randomized network, and the specific weights of temperature and strain data at each measuring point are obtained.

[0059] By randomly configuring the network, the weight values ​​corresponding to the temperature and strain data of each measuring point of the transformer under different operating conditions are obtained. The weighted transformer health status index system includes the temperature and strain data of each measuring point of the transformer and their corresponding weights.

[0060] The randomized network employs a three-layer feedforward neural network model and uses an incremental method to build the network step by step, starting with a small-scale network and gradually adding hidden nodes until an acceptable error is obtained.

[0061] Given an objective function f: R n →R m Suppose a single-layer feedforward network (SLFN) with P-1 hidden nodes has been constructed, and the training output value f obtained through the network is... p-1 (x) is:

[0062]

[0063] Where P = 1, 2, ..., n-1, f0 = 0, B j =[B j-1 ...B j.m ] T .

[0064] The current remaining error is:

[0065] e L-l =ff L-1 =[e L-1.1 , ..., e L-1.m ]

[0066] Where f is the actual value. And at this time ||e L-1 || The acceptable tolerance level has not been reached. At this point, add the Pth node and configure ω. p and b p Let f p =f p-1 +B p g p This ensures that the error reaches the expected deviation ε.

[0067] SCN uses a supervisory mechanism (inequality constraints) to randomly assign new input weights and biases.

[0068] Step 4: Construct a transformer internal index prediction model based on a recurrent neural network. Use the weighted transformer health status index system and operating environment parameters obtained in Step 3 as training data to train the prediction model and obtain the trained transformer internal index prediction model.

[0069] In step 4, this invention uses a GRU network to construct a transformer internal index prediction model, and uses the weighted transformer health status index system processed in step 3 as the input to the prediction model to obtain the predicted value. Specifically, this includes the following steps:

[0070] like Figure 3 As shown, the GRU (Gate Recurrent Unit) network is an improved RNN neural network based on LSTM, which solves the long-term dependency problem of ordinary RNNs. GRU uses an "update gate" z t And "Reset Gate" t This is used to control the transmission of information. During calculation, the reset gate is first used to reset the previous state h. t-1 Reset to get h t-1 '=h t-1 ⊙r, then h t-1 'With input x t The data is concatenated and then scaled to the range [-1, 1] using an activation function tanh to obtain h. t The calculation formula is:

[0071] h t '=tanh(W h x t +U h (h t-1 ⊙r t ))

[0072] h t 'Mainly contains the current input x' t Data is selectively added to the hidden state, which is equivalent to memorizing the current input. Finally, the memory is updated. The mathematical expression for this is:

[0073] h t =(1-z) t )⊙h t-1 +z t ⊙h′

[0074] Among them, the gate signal z t The range is [0, 1], and the gate signal z t The closer z is to 1, the more likely it is to retain the state from the previous time step during memory updates. t The closer it is to 0, the more likely it is to update to the current state.

[0075] The training of the transformer internal index prediction model also includes training the transformer internal index prediction model based on weighted transformer health status index system data under normal operation, transformer winding loosening, core loosening and overheating operation, as well as operating environment parameters.

[0076] Specifically, during the training process, the weighted transformer health status index system data and operating environment parameters of the transformer in the previous time period T1 are used as the input of the model to predict the temperature and strain data of each key measuring point of the transformer in the next time period T2. The model parameters are then adjusted according to the prediction results until the prediction accuracy is met.

[0077] The weighted transformer health status index system data includes temperature and strain data at key locations of the transformer, as well as their corresponding weights. The operating environment parameters include the transformer's ambient temperature, load current, and voltage. The time periods T1 and T2 and the prediction accuracy requirements can be set by technicians according to the actual situation.

[0078] Step 5: Set up fiber optic grating sensors at key measurement points of the transformer to be predicted, collect temperature and strain data of each measurement point over a period of time, as well as the operating environment parameters of the location of the transformer to be predicted, and obtain the predicted temperature and predicted strain data of the transformer to be predicted in the future time period through the trained transformer internal index prediction model.

[0079] Specifically, temperature and strain data of each measuring point of the transformer to be predicted are collected during time period T1. The data are normalized and combined with the weight values ​​calculated in step 3 to form a weighted transformer health status index system. The operating environment parameters of the location of the transformer to be predicted are collected and normalized, including ambient temperature, transformer load current and voltage. The weighted transformer health status index system and operating environment parameters are used as inputs to the trained transformer internal index prediction model to obtain the predicted temperature and predicted strain data of the transformer in the next time period T2.

[0080] Step 6: Based on the predicted temperature and predicted strain data, and combined with the temperature range and strain data range of each key measuring point of the transformer under different operating conditions obtained in Step 2, the health status prediction result of the transformer to be predicted is obtained.

[0081] Specifically, by comparing the range of temperature and strain data at key points under normal and fault conditions, as well as the predicted temperature and strain data, it is determined whether the internal operating state of the transformer to be predicted is normal, and if a fault exists, it is predicted to be of what type, including transformer winding loosening, core loosening, and overheating. Furthermore, based on the predicted fault type, staff can deploy countermeasures in advance to ensure the safe and stable operation of the transformer.

[0082] like Figure 4 As shown, this invention proposes a transformer internal health status identification system using fiber Bragg grating sensors. The aforementioned transformer internal health status identification method can be implemented based on this system, which includes:

[0083] The data acquisition module includes a fiber optic grating sensor installed inside the transformer to collect temperature and strain data at the measuring point, as well as sensors for collecting ambient temperature, current, and voltage.

[0084] The data processing module performs row normalization on the collected temperature and strain data;

[0085] Run the simulation module to simulate the operating environment and state of the transformer, and obtain the temperature range and strain data range of each key measuring point of the transformer under different operating conditions;

[0086] The weight calculation module calculates the weights of temperature and strain data at various measuring points of the transformer by randomly configuring the network and normalizing the data, and obtains a weighted transformer health status index system corresponding to different operating states and different operating environments.

[0087] The prediction module constructs a transformer internal index prediction model based on a recurrent neural network. It is trained using data processed by the data processing module and a weighted transformer health status index system obtained by the weight calculation module. The trained transformer internal index prediction model can predict the temperature and strain data of each measuring point of the transformer. Based on the predicted temperature and strain data of each measuring point of the transformer, and the temperature range and strain data range of each key measuring point obtained by the simulation module, the predicted operating status of the transformer is obtained.

[0088] This invention can effectively assess the internal health status of a transformer after acquiring temperature and stress data at various measuring points inside the transformer, identify faults and potential hazards, realize automated detection of power equipment, improve the reliability and stability of transformer operation, promptly detect potential faults and hazards, reduce the probability of faults, reduce maintenance time and downtime due to faults, and improve the economy and operating efficiency of the power system.

[0089] This disclosure can be a system, method, and / or computer program product. A computer program product may include a computer-readable storage medium having computer-readable program instructions loaded thereon for causing a processor to implement various aspects of this disclosure.

[0090] Computer-readable storage media can be tangible devices capable of holding and storing instructions for use by an instruction execution device. Computer-readable storage media can be, for example—but not limited to—electrical storage devices, magnetic storage devices, optical storage devices, electromagnetic storage devices, semiconductor storage devices, or any suitable combination of the foregoing. More specific examples (a non-exhaustive list) of computer-readable storage media include: portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), static random access memory (SRAM), portable compact disc read-only memory (CD-ROM), digital multifunction disc (DVD), memory sticks, floppy disks, mechanical encoding devices, such as punch cards or recessed protrusions storing instructions thereon, and any suitable combination of the foregoing. The computer-readable storage media used herein are not to be construed as transient signals themselves, such as radio waves or other freely propagating electromagnetic waves, electromagnetic waves propagating through waveguides or other transmission media (e.g., light pulses through fiber optic cables), or electrical signals transmitted through wires.

[0091] The computer-readable program instructions described herein can be downloaded from computer-readable storage media to various computing / processing devices, or downloaded via a network, such as the Internet, local area network, wide area network, and / or wireless network, to an external computer or external storage device. The network may include copper transmission cables, fiber optic transmission, wireless transmission, routers, firewalls, switches, gateway computers, and / or edge servers. A network adapter card or network interface in each computing / processing device receives the computer-readable program instructions from the network and forwards them to the computer-readable storage media in the respective computing / processing device.

[0092] Computer program instructions used to perform the operations of this disclosure may be assembly instructions, instruction set architecture (ISA) instructions, machine instructions, machine-dependent instructions, microcode, firmware instructions, status setting data, or source code or object code written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Smalltalk, C++, etc., and conventional procedural programming languages ​​such as the "C" language or similar programming languages. The computer-readable program instructions may execute entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving a remote computer, the remote computer may be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or may be connected to an external computer (e.g., via the Internet using an Internet service provider). In some embodiments, electronic circuitry, such as programmable logic circuitry, field-programmable gate arrays (FPGAs), or programmable logic arrays (PLAs), is personalized by utilizing the status information of the computer-readable program instructions to implement various aspects of this disclosure.

[0093] 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, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the protection scope of the claims of the present invention.

Claims

1. A method for transformer health state prediction based on fiber grating sensors, characterized in that, Includes the following steps: Step 1: Place fiber Bragg grating sensors at key locations inside the transformer. The fiber Bragg grating sensors collect temperature and strain data at the key measurement points according to a preset acquisition frequency. Key locations of a transformer include the upper end, lower end, and midpoint of the transformer windings, the top of the transformer tank, and the ends of the core. Step 2: Simulate the operating environment of the transformer and obtain the temperature and strain data of key measuring points of the transformer under different operating conditions through fiber optic grating sensors to obtain the temperature range and strain data range of each key measuring point of the transformer under different operating conditions. Step 3: Normalize and weight the operating environment simulation parameters used in Step 2, as well as the temperature and strain data collected by the fiber optic grating sensor at each measuring point, to obtain a weighted transformer health status index system. Step 3 includes: Step 3-2: By randomly configuring the network and normalizing the data, the temperature and strain data of each measuring point of the transformer are weighted and calculated to obtain a weighted transformer health status index system corresponding to different operating states and different operating environments. Step 4: Construct a transformer internal index prediction model based on a recurrent neural network. Use the weighted transformer health status index system and operating environment parameters obtained in Step 3 as training data to train the prediction model and obtain the trained transformer internal index prediction model. Step 5: Set up fiber optic grating sensors at key measurement points of the transformer to be predicted, collect temperature and strain data of each measurement point over a period of time, as well as the operating environment parameters of the location of the transformer to be predicted, and obtain the predicted temperature and predicted strain data of the transformer to be predicted in the future time period through the trained transformer internal index prediction model. Step 6: Based on the predicted temperature and predicted strain data, and combined with the temperature range and strain data range of each key measuring point of the transformer under different operating conditions obtained in Step 2, the health status prediction result of the transformer to be predicted is obtained.

2. The transformer health status prediction method based on fiber optic grating sensors according to claim 1, characterized in that: In step 1, fiber optic grating sensors are placed at key measuring points inside the transformer, and a sampling frequency is set to collect temperature and strain data at the key measuring points according to the sampling frequency.

3. The transformer health status prediction method based on fiber Bragg grating sensors according to claim 1, characterized in that: In step 2, the relevant parameters of the simulated transformer operating environment include: ambient temperature, load current, and voltage; the simulated transformer operating state includes normal transformer state and fault state, and the fault state also includes: loose transformer windings, loose iron core, and overheating operation; under different transformer operating environments, the temperature and strain data of key measuring points are collected at a preset time period using fiber optic grating sensors at a preset acquisition frequency for transformers in different operating states, and the temperature and strain data ranges of each key measuring point under different operating states are obtained based on the maximum and minimum values ​​of the temperature and strain data of each measuring point.

4. The transformer health status prediction method based on fiber Bragg grating sensors according to claim 3, characterized in that: In step 2, when simulating the transformer operating environment parameters, the simulated ambient temperature includes the average summer, winter, and spring / autumn temperatures of the location of the transformer under test; the simulated load current includes 40%, 60%, 80%, and 100% of the transformer's rated load current; and the simulated voltage includes the voltage of each tap of the transformer.

5. The transformer health status prediction method based on fiber Bragg grating sensors according to claim 1, characterized in that: Step 3 also includes: Step 3-1: Normalize the temperature and strain data of key transformer measurement points obtained by fiber optic grating sensors under different operating environments in Step 2, as well as the corresponding external factor data of ambient temperature, load current, and voltage. Step 3-2: By randomly configuring the network and normalizing the data, the temperature and strain data of each measuring point of the transformer are weighted and calculated to obtain a weighted transformer health status index system corresponding to different operating states and different operating environments.

6. The transformer health status prediction method based on fiber Bragg grating sensors according to claim 5, characterized in that: Step 3-2 includes: for transformers in different operating states, the normalized data obtained in step 2 and the transformer operating state are input into a random configuration network. The normalized data includes normalized data of ambient temperature, load current and voltage related to the operating environment, as well as normalized data of temperature and strain data at each measuring point of the transformer. The correlation between temperature and strain data at each measuring point and external influencing factors and various typical faults is extracted through the random configuration network, and the specific weights of temperature and strain data at each measuring point are obtained.

7. The transformer health status prediction method based on fiber Bragg grating sensors according to claim 1, characterized in that: Step 5 further includes: collecting temperature and strain data of each measuring point of the transformer to be predicted during time period T1, normalizing the data, and combining it with the weight values ​​calculated in step 3 to form a weighted transformer health status index system; collecting and normalizing the operating environment parameters of the location of the transformer to be predicted, including ambient temperature, transformer load current and voltage; and using the weighted transformer health status index system and operating environment parameters obtained after processing as input to the trained transformer internal index prediction model to obtain the predicted temperature and predicted strain data of the transformer in the next time period T2.

8. The transformer health status prediction method based on fiber Bragg grating sensors according to claim 1, characterized in that: Step 6 further includes: comparing the range of temperature and strain data at key points in normal and fault states with the predicted temperature and strain data to determine whether the internal operating state of the transformer to be predicted is normal, and if a fault exists, predicting which type of fault state it is, including transformer winding loosening, core loosening and overheating operation.

9. A transformer health status prediction system based on a fiber Bragg grating sensor, utilizing the transformer health status prediction method according to any one of claims 1-8, characterized in that, include: The data acquisition module includes a fiber optic grating sensor installed inside the transformer to collect temperature and strain data at the measuring point, as well as sensors for collecting ambient temperature, current, and voltage. The data processing module performs row normalization on the collected temperature and strain data; Run the simulation module to simulate the operating environment and state of the transformer, and obtain the temperature range and strain data range of each key measuring point of the transformer under different operating conditions; The weight calculation module calculates the weights of temperature and strain data at various measuring points of the transformer by randomly configuring the network and normalizing the data, and obtains a weighted transformer health status index system corresponding to different operating states and different operating environments. The prediction module constructs a transformer internal index prediction model based on a recurrent neural network. It is trained using data processed by the data processing module and a weighted transformer health status index system obtained by the weight calculation module. The trained transformer internal index prediction model can predict the temperature and strain data of each measuring point of the transformer. Based on the predicted temperature and strain data of each measuring point of the transformer, and the temperature range and strain data range of each key measuring point obtained by the simulation module, the prediction result of the transformer's operating status is obtained.

10. A terminal, comprising a processor and a storage medium; characterized in that: The storage medium is used to store instructions; The processor is configured to operate according to the instructions to perform the steps of the method according to any one of claims 1-8.

11. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the program implements the steps of the method according to any one of claims 1-8.

Images