Transformer winding deformation monitoring method and system based on Internet of Things

Abstract

The invention discloses a transformer winding deformation monitoring method and system based on the Internet of Things, and relates to the technical field of winding monitoring, and the method comprises the following steps: installing each sensor, and connecting an edge computing device through a bus; the edge device calculates and corrects the measurement drift of the vibration sensor; analyzing the compensated vibration spectrum, and extracting power frequency electromagnetic vibration energy; calculating the phase difference of the winding vibration signal relative to the fuel tank reference signal; the cloud terminal dynamically adjusts a phase difference alarm threshold value according to the current salt mist concentration and the oil temperature; when the phase difference exceeds the standard, all sensor signals are used for calculating the sound wave transmission relation; and the edge node uploads alarm information, and the cloud performs comprehensive judgment by combining the operation state of the fan, triggers graded early warning and limits the power of the fan. The sensor base line is reconstructed in real time by establishing the compensation model, and measurement deviation introduced by environmental corrosion can be stripped from a signal source.

Description

Technical Field

[0001] This invention belongs to the field of winding monitoring technology, and in particular relates to a method and system for monitoring transformer winding deformation based on the Internet of Things. Background Technology

[0002] Traditional methods for monitoring transformer winding deformation and vibration face serious failure risks in highly corrosive environments such as offshore wind power. High concentrations of salt spray and humid heat can cause irreversible drift of the sensor's measurement reference, leading to systematic deviations in the vibration signal acquisition process. Existing vibration amplitude detection mechanisms cannot distinguish between actual winding deformation and signal distortion caused by environmental corrosion, resulting in continuous misjudgments by the monitoring system under harsh conditions. This inherent defect stems from the coupling effect of salt spray deposition and humid heat on the sensor's physical characteristics, while conventional temperature compensation methods lack effective decoupling capabilities for such multi-physical field interference. Therefore, the following solutions are proposed to address these issues. Summary of the Invention

[0003] The purpose of this invention is to provide a transformer winding deformation monitoring method and system based on the Internet of Things. By establishing a compensation model to reconstruct the sensor baseline in real time, the measurement deviation introduced by environmental corrosion can be removed from the signal source, thus solving the problem of the surge in false alarm rate caused by sensor drift in existing monitoring methods.

[0004] To solve the above-mentioned technical problems, the present invention is achieved through the following technical solution:

[0005] This invention relates to a method for monitoring transformer winding deformation based on the Internet of Things (IoT). The monitoring method specifically includes the following steps:

[0006] Step S1, Sensor Deployment: Install vibration, temperature and humidity, and salt spray sensors at specific locations in the transformer tank and connect them to edge computing devices via a bus;

[0007] Step S2, Salt spray drift compensation: The edge device calculates and corrects the measurement drift of the vibration sensor based on the real-time salt spray concentration and humidity;

[0008] Step S3, Electromagnetic vibration identification: Analyze the compensated vibration spectrum and extract the power frequency electromagnetic vibration energy as the triggering reference for winding state analysis;

[0009] Step S4, Phase offset detection: Calculate the phase difference between the winding vibration signal and the oil tank reference signal, which serves as the main criterion for winding deformation;

[0010] Step S5, Dynamic Threshold Adjustment: The cloud dynamically adjusts the phase difference alarm threshold based on the current salt spray concentration and oil temperature to adapt to environmental changes;

[0011] Step S6, Deformation Location Positioning: When the phase difference exceeds the standard, calculate the acoustic wave transmission relationship using all sensor signals to accurately locate the deformation point inside the winding;

[0012] Step S7, Deformation Location Positioning: Intelligent Early Warning Linkage: Edge nodes upload alarm information, and the cloud makes a comprehensive judgment based on the wind turbine's operating status, triggering graded early warnings and limiting wind turbine power.

[0013] Further, step S1, sensor deployment specifically includes the following steps:

[0014] Step S11: At 12 predetermined three-dimensional spatial coordinate points on the outer wall of the transformer tank, a triaxial MEMS vibration sensor, a temperature and humidity composite sensor, and an electrochemical impedance spectroscopy salt spray deposition rate sensor are simultaneously deployed at each point.

[0015] Step S12: All sensors are connected to the RS485 industrial bus network via shielded twisted-pair cables, and finally connected to the edge computing node based on the ARM Cortex-A53 architecture. The system is uniformly configured with a vibration signal sampling rate of 10,000 data points per second, forming the physical layer infrastructure of the distributed multimodal sensor array.

[0016] This step involves deploying a network of 12 triaxial vibration, temperature and humidity, and salt spray deposition sensors on the transformer tank wall, which are then connected to edge nodes via an RS485 bus to establish a foundation for the synchronous acquisition of multiple physical quantities under salt spray conditions.

[0017] Furthermore, in step S2, the edge nodes in the salt spray drift compensation process perform the following steps every 5 minutes:

[0018] Step S21: Read the salt spray deposition amount C s and ambient humidity H;

[0019] Step S22: Using these two real-time measurements, the node calculates the sensor baseline drift compensation coefficient, which accounts for the impact of the current environment on the vibration sensor's measurement accuracy, based on a preset compensation model.

[0020] K comp =α(C s ) 2 ×e βH ;

[0021] In the formula, K comp C is the sensor baseline drift compensation coefficient. s denoted as salt spray deposition, H as ambient relative humidity, e as a natural constant, and α and β as empirical parameters fitted from measured data of offshore wind power platforms.

[0022] Step S23: Apply the calculated compensation coefficient to the original vibration signal V rawReal-time adjustments and corrections are made to offset the sensor baseline drift error caused by the combined effects of salt spray deposition and humidity, thereby obtaining a more accurate and reliable vibration signal.

[0023]

[0024] In the formula, V corr (t) represents the compensated vibration signal, V raw (t) represents the original vibration signal at time t. Let ξ be the second time derivative of the vibration signal, and τ be intermediate variables for integration.

[0025] This step, based on electrochemical salt spray deposition data and ambient humidity, dynamically corrects the baseline drift of the vibration signal through a nonlinear compensation formula, thereby eliminating sensor measurement distortion caused by salt spray corrosion.

[0026] Furthermore, step S3, electromagnetic vibration identification, specifically includes the following steps:

[0027] Step S31: Separate the 100Hz power frequency electromagnetic vibration component by FFT, and perform a 4096-point FFT on the compensated signal by applying a Hanning window;

[0028] Step S32: Extract energy in the 99-101Hz frequency band

[0029] In the formula, X(k) is the complex amplitude of the k-th frequency point after FFT transformation, and k is the frequency point index;

[0030] Step S33: Compare the calculated power frequency electromagnetic vibration energy value with a pre-set threshold value on the order of the square of the voltage. When E em >0.8V 2 When the transformer is in a loaded operating state, the winding analysis is triggered. This mechanism effectively eliminates unnecessary triggering conditions such as no-load or light-load conditions.

[0031] This step uses FFT spectrum analysis to separate the 100Hz power frequency electromagnetic vibration component, sets an energy threshold to automatically trigger winding analysis, eliminates invalid operating conditions such as transformer no-load, and ensures that monitoring is only started under effective electromagnetic excitation.

[0032] Further, step S4, phase offset detection, specifically includes the following steps:

[0033] Step S41: Analyze the winding vibration signal V winding (t) and fuel tank reference signal V tank (t) Perform cross-correlation to reveal the relative positional relationship between the two signals on the time axis:

[0034] R(τ)=∫V winding (t)×Vtank (t+γ)dt;

[0035] In the formula, R(τ) is the cross-correlation function between the winding vibration signal and the oil tank reference signal, and V winding (t) represents the winding vibration signal, V tank (t) represents the reference vibration signal of the oil tank wall, and γ represents the time delay.

[0036] Step S42: Based on the time correlation characteristics, calculate the time lead or lag of the winding vibration relative to the tank reference vibration on the waveform, and convert this time difference into a phase difference value representing the angular offset between the two:

[0037] Δθ = arctan2(Q,I);

[0038] I = R(0);

[0039]

[0040] In the formula, Δθ is the phase offset angle between the winding and the oil tank vibration, arctan 2 is the arctangent function in the four quadrants, I is the real part of the cross-correlation function at zero delay, Q is the imaginary part of the derivative of the cross-correlation function at zero delay, and f0 is the power frequency.

[0041] This step uses cross-correlation function and instantaneous frequency differentiation to calculate the phase offset angle Δθ between the winding vibration and the tank reference signal, which can improve the sensitivity to early deformation.

[0042] Further, step S5, the dynamic threshold adjustment, specifically involves the following steps:

[0043] Based on historical operational data accumulated over a long period, the cloud platform pre-constructs a mapping table reflecting the correlation between salt spray concentration, transformer top oil temperature, and winding vibration phase shift. When the salt spray concentration value monitored by the system in real time exceeds the set critical level, the cloud platform dynamically adjusts the phase shift alarm threshold based on the currently measured transformer top oil temperature value. The specific rules are as follows:

[0044] Δθ threshold =ε+∈×(T-κ);

[0045] In the formula, Δθ threshold ε is the dynamic phase offset alarm threshold, T is the transformer top oil temperature, ε is the basic threshold, ∈ is the compensation coefficient, and κ is the reference critical point for the temperature change of the transformer winding.

[0046] If the real-time salt spray concentration does not exceed the critical value, the standard phase offset threshold (e.g., 1.2 degrees) is used directly for judgment.

[0047] This step adjusts the phase offset alarm threshold in real time based on the salt spray concentration-temperature historical mapping table. When the salt spray concentration exceeds the set critical level, a temperature adaptive threshold model is used to solve the problem of false triggering of fixed thresholds in harsh environments.

[0048] Furthermore, step S6, the deformation position positioning, specifically includes the following steps:

[0049] Step S61: When the system detects that the phase offset angle of a certain sensor exceeds the dynamically adjusted alarm threshold, all 12 sensors deployed on the transformer tank wall are immediately activated, and the collaborative analysis network is started.

[0050] Step S62: Calculate the coordinates of the anomaly source using the transfer function matrix:

[0051] [H] 12×12 ×[P]=[Δθ];

[0052]

[0053] In the formula, [H] is the transfer function matrix among the 12 sensors, [P] is the location vector of the anomaly source inside the winding, and [Δθ] is the vector of phase offset measurements of each sensor. ij Let be the transfer function from the i-th sensor to the j-th sensor, ω be the angular frequency, and r be the angular frequency. ij Let c be the straight-line distance between sensors i and j, c be the speed of sound in the oil, and j be the imaginary unit.

[0054] Step S63: Solve the ill-conditioned equations using Tikhonov regularization to calculate the three-dimensional coordinates of the points where deformation or loosening occurs inside the transformer windings;

[0055] This step involves constructing the acoustic wave transmission matrix equation based on a 12-sensor array, and using a regularization algorithm to solve for the three-dimensional coordinates of the deformation points inside the winding, thereby enabling spatial positioning.

[0056] Furthermore, step S7, the deformation position positioning, specifically includes the following steps:

[0057] Step S71: The edge computing node uploads an alarm data packet containing the three-dimensional coordinate information of the winding deformation point, the calculated specific value of the phase offset angle, and the real-time salt spray concentration measurement value to the cloud monitoring platform via narrowband Internet of Things (NB-IoT) communication.

[0058] Step S72: After receiving the alarm data packet, the cloud platform automatically associates and retrieves the real-time operation monitoring data (SCADA system data) of the wind turbine where the transformer is located, and performs a comprehensive status judgment. Only when the following three conditions are met simultaneously: the winding hot spot temperature exceeds 75 degrees Celsius, the current load rate of the transformer is higher than 85%, and the detected phase offset angle continues to exceed the dynamic threshold for more than 10 minutes, will the cloud platform push the highest level three alarm (identification code F3) to the terminal equipment of the operation and maintenance personnel, and at the same time issue an instruction to the wind turbine control system to prohibit any operation that increases the wind turbine's power generation, so as to prevent the fault from escalating.

[0059] This step involves uploading alarm data via NB-IoT, performing multi-condition collaborative judgment (temperature / load / duration) with the wind turbine SCADA system, automatically triggering hierarchical alarms and locking power commands, thus forming a closed-loop control link.

[0060] A transformer winding deformation monitoring system based on the Internet of Things, the monitoring system comprising:

[0061] The multimodal sensing and acquisition module is used to deploy a sensor array on the surface of the transformer tank to collect vibration signals, ambient temperature and humidity and salt spray deposition in real time, and transmit the raw data through the industrial bus.

[0062] The edge computing processing module is used to compensate for salt spray drift of vibration signals, separate power frequency electromagnetic vibration components, and extract the vibration phase difference characteristics between the winding and the oil tank through cross-correlation calculation.

[0063] The dynamic threshold management module is used to dynamically adjust the phase offset threshold by combining salt spray concentration and oil temperature parameters, establish a mapping relationship between environmental parameters and deformation criteria, and adapt to changes in harsh working conditions.

[0064] The topology positioning and calculation module is used to activate the sensor network when an anomaly is detected, and calculate the three-dimensional spatial coordinates of the deformation point inside the winding through the acoustic wave transmission matrix to achieve positioning.

[0065] The IoT communication module is used to upload encrypted alarm data to the cloud via narrowband IoT to ensure reliable transmission in harsh marine environments.

[0066] The cloud-based decision-making and linkage module is used to integrate wind turbine operation data, generate hierarchical alarm strategies, trigger wind turbine power lockout commands, and link with the operation and maintenance system.

[0067] The present invention has the following beneficial effects:

[0068] 1. This invention effectively eliminates the sensor baseline drift problem caused by the highly corrosive marine environment through a salt spray humidity compensation algorithm. By analyzing the coupling interference of salt spray deposition and humidity on vibration signals in real time, the true vibration waveform is reconstructed using a second-order differential compensation mechanism. This method removes environmental noise from the signal source, enabling the monitoring system to maintain stable data acquisition capabilities under highly corrosive and high-humidity conditions, reducing the risk of misjudgment due to sensor failure, and providing a high-quality data foundation for subsequent analysis.

[0069] 2. This invention uses vibration phase shift decoupling technology to replace traditional amplitude monitoring, capturing the microscopic features of changes in the winding mechanical structure; by extracting the instantaneous phase difference between the winding vibration and the oil tank reference signal, and combining the differential operation of the cross-correlation function, a phase shift quantization model is constructed; this algorithm has unique response characteristics to early latent deformations such as winding inter-turn displacement and insulation compression, and can achieve early warning before mechanical deformation causes significant amplitude changes, improving the time margin of defect detection.

[0070] 3. This invention solves the problem of fixed alarm threshold failure under harsh environments by establishing a multi-dimensional dynamic threshold model of salt spray concentration, temperature, and phase shift. The system automatically adjusts the phase shift criterion according to the real-time salt spray deposition and oil temperature parameters, so that the diagnostic criteria always evolve in sync with the environmental conditions. This design can effectively overcome the problem of frequent false alarms when the environment fluctuates drastically, and improve the alarm accuracy and system robustness.

[0071] 4. This invention constructs a set of acoustic propagation path equations for a multi-sensor array and combines them with regularization methods to solve ill-conditioned inversion problems, thereby resolving the three-dimensional spatial coordinates of deformation points inside the transformer. This design can improve fault location accuracy and provide direct spatial information support for subsequent targeted maintenance.

[0072] Of course, any product implementing this invention does not necessarily need to achieve all of the advantages described above at the same time. Attached Figure Description

[0073] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0074] Figure 1 This is a schematic flowchart of a transformer winding deformation monitoring method based on the Internet of Things according to the present invention.

[0075] Figure 2 This is a schematic diagram of the structure of a transformer winding deformation monitoring system based on the Internet of Things according to the present invention. Detailed Implementation

[0076] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0077] Please see Figure 1 As shown, this invention is a method for monitoring transformer winding deformation based on the Internet of Things, comprising the following steps:

[0078] Step S1, Sensor Deployment: Install vibration, temperature and humidity, and salt spray sensors at specific locations in the transformer tank and connect them to edge computing devices via a bus;

[0079] Step S1, sensor deployment specifically includes the following steps:

[0080] Step S11: At 12 predetermined three-dimensional spatial coordinate points on the outer wall of the transformer tank, a triaxial MEMS vibration sensor, a temperature and humidity composite sensor, and an electrochemical impedance spectroscopy salt spray deposition rate sensor are simultaneously deployed at each point.

[0081] Step S12: All sensors are connected to the RS485 industrial bus network via shielded twisted-pair cables, and finally connected to the edge computing node based on the ARM Cortex-A53 architecture. The system is uniformly configured with a vibration signal sampling rate of 10,000 data points per second, forming the physical layer infrastructure of the distributed multimodal sensor array.

[0082] Step S2, Salt spray drift compensation: The edge device calculates and corrects the measurement drift of the vibration sensor based on the real-time salt spray concentration and humidity;

[0083] Step S2, in salt spray drift compensation, the edge nodes perform the following steps every 5 minutes:

[0084] Step S21: Read the salt spray deposition amount C s and ambient humidity H;

[0085] Step S22: Using these two real-time measurements, the node calculates the sensor baseline drift compensation coefficient, which accounts for the impact of the current environment on the vibration sensor's measurement accuracy, based on a preset compensation model.

[0086] K comp =α(C s ) 2 ×e βH ;

[0087] In the formula, K compC is the sensor baseline drift compensation coefficient. s denoted as salt spray deposition, H as ambient relative humidity, e as a natural constant, and α and β as empirical parameters fitted from measured data of offshore wind power platforms.

[0088] Step S23: Apply the calculated compensation coefficient to the original vibration signal V raw Real-time adjustments and corrections are made to offset the sensor baseline drift error caused by the combined effects of salt spray deposition and humidity, thereby obtaining a more accurate and reliable vibration signal.

[0089]

[0090] In the formula, V corr (t) represents the compensated vibration signal, V raw (t) represents the original vibration signal at time t. Let ξ be the second-order time derivative of the vibration signal, and τ be intermediate variables for integration.

[0091] Step S3, Electromagnetic vibration identification: Analyze the compensated vibration spectrum and extract the power frequency electromagnetic vibration energy as the triggering reference for winding state analysis;

[0092] Step S3, electromagnetic vibration identification specifically includes the following steps:

[0093] Step S31: Separate the 100Hz power frequency electromagnetic vibration component by FFT, and perform a 4096-point FFT on the compensated signal by applying a Hanning window;

[0094] Step S32: Extract energy in the 99-101Hz frequency band

[0095] In the formula, X(k) is the complex amplitude of the k-th frequency point after FFT transformation, and k is the frequency point index;

[0096] Step S33: Compare the calculated power frequency electromagnetic vibration energy value with a pre-set threshold value on the order of the square of the voltage. When E em >0.8V 2 When the transformer is in a loaded operating state, the winding analysis is triggered. This mechanism effectively eliminates unnecessary triggering conditions such as no-load or light-load conditions.

[0097] Step S4, Phase offset detection: Calculate the phase difference between the winding vibration signal and the oil tank reference signal, which serves as the main criterion for winding deformation;

[0098] Step S4, phase offset detection specifically includes the following steps:

[0099] Step S41: Analyze the winding vibration signal V winding (t) and fuel tank reference signal V tank(t) Perform cross-correlation to reveal the relative positional relationship between the two signals on the time axis:

[0100] R(τ)=∫V winding (t)×V tank (t+γ)dt;

[0101] In the formula, R(τ) is the cross-correlation function between the winding vibration signal and the oil tank reference signal, and V winding (t) represents the winding vibration signal, V tank (t) represents the reference vibration signal of the oil tank wall, and γ represents the time delay.

[0102] Step S42: Based on the time correlation characteristics, calculate the time lead or lag of the winding vibration relative to the tank reference vibration on the waveform, and convert this time difference into a phase difference value representing the angular offset between the two:

[0103] Δθ = arctan2(Q,I);

[0104] I = R(0);

[0105]

[0106] In the formula, Δθ is the phase offset angle between the winding and the oil tank vibration, arctan 2 is the arctangent function in the four quadrants, I is the real part of the cross-correlation function at zero delay, Q is the imaginary part of the derivative of the cross-correlation function at zero delay, and f0 is the power frequency.

[0107] Step S5, Dynamic Threshold Adjustment: The cloud dynamically adjusts the phase difference alarm threshold based on the current salt spray concentration and oil temperature to adapt to environmental changes;

[0108] Step S5, the specific steps for dynamic threshold adjustment are as follows:

[0109] Based on historical operational data accumulated over a long period, the cloud platform pre-constructs a mapping table reflecting the correlation between salt spray concentration, transformer top oil temperature, and winding vibration phase shift. When the salt spray concentration value monitored by the system in real time exceeds the set critical level, the cloud platform dynamically adjusts the phase shift alarm threshold based on the currently measured transformer top oil temperature value. The specific rules are as follows:

[0110] Δθ threshold =ε+∈×(T-κ);

[0111] In the formula, Δθ threshold ε is the dynamic phase offset alarm threshold, T is the transformer top oil temperature, ε is the basic threshold, ∈ is the compensation coefficient, and κ is the reference critical point for the temperature change of the transformer winding.

[0112] If the real-time salt spray concentration does not exceed the critical value, the standard phase offset threshold (e.g., 1.2 degrees) is used directly for judgment.

[0113] Step S6, Deformation Location Positioning: When the phase difference exceeds the standard, calculate the acoustic wave transmission relationship using all sensor signals to accurately locate the deformation point inside the winding;

[0114] Step S6, the deformation location positioning specifically includes the following steps:

[0115] Step S61: When the system detects that the phase offset angle of a certain sensor exceeds the dynamically adjusted alarm threshold, all 12 sensors deployed on the transformer tank wall are immediately activated, and the collaborative analysis network is started.

[0116] Step S62: Calculate the coordinates of the anomaly source using the transfer function matrix:

[0117] [H] 12×12 ×[P]=[Δθ];

[0118]

[0119] In the formula, [H] is the transfer function matrix among the 12 sensors, [P] is the location vector of the anomaly source inside the winding, and [Δθ] is the vector of phase offset measurements of each sensor. ij Let be the transfer function from the i-th sensor to the j-th sensor, ω be the angular frequency, and r be the angular frequency. ij Let c be the straight-line distance between sensors i and j, c be the speed of sound in the oil, and j be the imaginary unit.

[0120] Step S63: Solve the ill-conditioned equations using Tikhonov regularization to calculate the three-dimensional coordinates of the points where deformation or loosening occurs inside the transformer windings.

[0121] Step S7, Deformation Location Positioning: Intelligent Early Warning Linkage: Edge nodes upload alarm information, and the cloud makes a comprehensive judgment based on the wind turbine's operating status, triggering graded early warnings and limiting wind turbine power.

[0122] Step S7, the deformation location positioning specifically includes the following steps:

[0123] Step S71: The edge computing node uploads an alarm data packet containing the three-dimensional coordinate information of the winding deformation point, the calculated specific value of the phase offset angle, and the real-time salt spray concentration measurement value to the cloud monitoring platform via narrowband Internet of Things (NB-IoT) communication.

[0124] Step S72: After receiving the alarm data packet, the cloud platform automatically associates and retrieves the real-time operation monitoring data (SCADA system data) of the wind turbine where the transformer is located, and performs a comprehensive status judgment. Only when the following three conditions are met simultaneously: the winding hot spot temperature exceeds 75 degrees Celsius, the current load rate of the transformer is higher than 85%, and the detected phase offset angle continues to exceed the dynamic threshold for more than 10 minutes, will the cloud platform push the highest level three alarm (identification code F3) to the terminal equipment of the operation and maintenance personnel, and at the same time issue an instruction to the wind turbine control system to prohibit any operation that increases the wind turbine's power generation, so as to prevent the fault from escalating.

[0125] Please see Figure 2 As shown, this invention is an Internet of Things (IoT) based transformer winding deformation monitoring system. The monitoring system includes:

[0126] The multimodal sensing and acquisition module is used to deploy a sensor array on the surface of the transformer tank to collect vibration signals, ambient temperature and humidity and salt spray deposition in real time, and transmit the raw data through the industrial bus.

[0127] The edge computing processing module is used to compensate for salt spray drift of vibration signals, separate power frequency electromagnetic vibration components, and extract the vibration phase difference characteristics between the winding and the oil tank through cross-correlation calculation.

[0128] The dynamic threshold management module is used to dynamically adjust the phase offset threshold by combining salt spray concentration and oil temperature parameters, establish a mapping relationship between environmental parameters and deformation criteria, and adapt to changes in harsh working conditions.

[0129] The topology positioning and calculation module is used to activate the sensor network when an anomaly is detected, and calculate the three-dimensional spatial coordinates of the deformation point inside the winding through the acoustic wave transmission matrix to achieve positioning.

[0130] The IoT communication module is used to upload encrypted alarm data to the cloud via narrowband IoT to ensure reliable transmission in harsh marine environments.

[0131] The cloud-based decision-making and linkage module is used to integrate wind turbine operation data, generate hierarchical alarm strategies, trigger wind turbine power lockout commands, and link with the operation and maintenance system.

[0132] One specific application of this embodiment is:

[0133] Background: On July 15, 2025, the 66kV step-up transformer of Unit #5 at an offshore wind farm was exposed to a high salt spray environment following a typhoon (salt spray deposition rate 2.1 mg / cm³). 2 (Humidity 90%), load rate 92%, oil temperature 68℃.

[0134] Implementation steps:

[0135] S1. Sensor array data acquisition:

[0136] Data was collected simultaneously from 12 monitoring points (sampling rate 10kHz):

[0137] Vibration signal at point P3 corresponding to the winding: V raw (t)=0.35sin(2π·100t)+0.08e -0.3t (V);

[0138] Salt spray deposition C s =2.1 mg / cm³ 2 ;

[0139] Ambient humidity H = 90%;

[0140] S2. Salt spray drift compensation calculation:

[0141] Calculate the compensation coefficient: K comp =1.23×10 -3 ×(2.1) 2 ×e 0.07×90 =0.181;

[0142] Perform vibration signal correction:

[0143] Calculate the second derivative:

[0144] Double integral:

[0145] Output compensation signal: V corr (t)=[0.35sin(200πt)+0.08e -0.3t ]-0.181×(0.35sin(200πt)-0.08e -0.3t );

[0146] S3. Electromagnetic vibration feature extraction:

[0147] For V corr (t) Perform a 4096-point FFT (Hunting window);

[0148] Calculate the energy in the 100Hz frequency band: E em =|X(99.98)| 2 +|X(100)| 2 +|X(100.02)| 2 =0.97,V 2 ;

[0149] Judgment: 0.97 > 0.8 → Start winding analysis;

[0150] S4, Phase Shift Detection:

[0151] Fuel tank reference point P1 signal: V tank (t)=0.32sin(200πt);

[0152] Calculate the peak offset of the cross-correlation function:

[0153] R(τ) max It occurs at τ = -1.85 ms;

[0154] Phase offset angle:

[0155] S5, Dynamic Threshold Adjustment:

[0156] Salt spray concentration 28 μg / m³ 3 >15, μg / m 3 ;

[0157] Calculate the dynamic threshold: Δθ threshold =1.5 + 0.03 × (68 - 40) = 2.34°;

[0158] Judgment: |Δθ|=3.33°>2.34°→Exceeds the standard;

[0159] S6. Topological positioning of deformation points:

[0160] Phase offset vectors of 12 sensors: [Δθ] = [-3.33°, -2.41°, ..., 1.05°] T ;

[0161] Construct the transfer function matrix (c = 1420, m / s, oil temperature 68℃):

[0162]

[0163] Solve for the coordinates of the anomaly source:

[0164] Solving for (x,y,z) gives (0.71,1.53,-0.22) → the 7th turn of phase B winding;

[0165] S7, IoT early warning linkage:

[0166] Edge node upload alarm: CODE:F3-B-071153022 (F3 level alarm, phase B, coordinates 071 / 153 / 022);

[0167] Cloud platform detection parameters: winding temperature: 79℃; load rate: 92%; duration of phase exceedance: 18min;

[0168] Perform the following operation:

[0169] ① Push the following message to the operation and maintenance terminal: "Emergency alarm: B phase winding deformation (coordinates 071, 153, 022)";

[0170] ② Send the interlock command to the SCADA system: "MAX_POWER_LIMIT = 85%".

[0171] In the description of this specification, references to terms such as "an embodiment," "example," "specific example," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0172] The preferred embodiments of the present invention disclosed above are merely illustrative of the invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the invention to the specific implementations described. Clearly, many modifications and variations can be made based on the content of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize the invention. The invention is limited only by the claims and their full scope and equivalents.

Claims

1. A method for monitoring transformer winding deformation based on the Internet of Things, characterized in that, The monitoring method specifically includes the following steps: Step S1, Sensor Deployment: Install vibration, temperature and humidity, and salt spray sensors in the transformer tank and connect them to edge computing devices via a bus; Step S2, Salt spray drift compensation: The edge device calculates and corrects the measurement drift of the vibration sensor based on the real-time salt spray concentration and humidity; Step S3, Electromagnetic vibration identification: Analyze the compensated vibration spectrum and extract the power frequency electromagnetic vibration energy as the triggering reference for winding state analysis; Step S4, Phase offset detection: Calculate the phase difference between the winding vibration signal and the oil tank reference signal, which serves as the main criterion for winding deformation; Step S5, Dynamic Threshold Adjustment: The cloud dynamically adjusts the phase difference alarm threshold based on the current salt spray concentration and oil temperature to adapt to environmental changes; Step S6, Deformation Location Positioning: When the phase difference exceeds the standard, calculate the acoustic wave transmission relationship using all sensor signals to accurately locate the deformation point inside the winding; Step S7, Intelligent Early Warning Linkage: Edge nodes upload alarm information, and the cloud makes a comprehensive judgment based on the wind turbine operating status, triggering graded early warnings and limiting wind turbine power.

2. The method for monitoring transformer winding deformation based on the Internet of Things according to claim 1, characterized in that, Step S1, sensor deployment, specifically includes the following steps: Step S11: At multiple three-dimensional spatial coordinate points predetermined on the outer wall of the transformer tank, a vibration sensor, a temperature and humidity composite sensor, and an electrochemical impedance spectroscopy salt spray deposition rate sensor are simultaneously deployed at each point. Step S12: All sensors are connected to the industrial bus network via shielded twisted-pair cables and then connected to the edge computing node.

3. The method for monitoring transformer winding deformation based on the Internet of Things according to claim 1, characterized in that, In step S2, the edge nodes in the salt spray drift compensation process perform the following steps at regular intervals: Step S21: Read the salt spray deposition amount and ambient humidity at the current monitoring point from the connected sensors; Step S22: The node calculates the sensor baseline drift compensation coefficient based on the preset compensation model to determine the impact of the current environment on the measurement accuracy of the vibration sensor. Step S23: Apply the calculated compensation coefficients to adjust and correct the original vibration signal in real time.

4. The method for monitoring transformer winding deformation based on the Internet of Things according to claim 1, characterized in that, Step S3, electromagnetic vibration identification, specifically includes the following steps: Step S31: Apply a window function to the vibration signal after salt spray compensation and perform high-precision spectrum conversion; Step S32: From the converted spectrum, identify and accumulate all energy values ​​within a narrow frequency band centered on the power grid frequency and extended 1 Hz above and below it; Step S33: Compare the calculated power frequency electromagnetic vibration energy value with a pre-set threshold of the square of the voltage. When the energy value exceeds the threshold, it is determined that the transformer is in a load operation state, thereby triggering the subsequent winding deformation analysis process.

5. The method for monitoring transformer winding deformation based on the Internet of Things according to claim 1, characterized in that, Step S4, phase shift detection, specifically includes the following steps: Step S41: Perform cross-correlation calculation on the vibration signal measured at a specific location of the winding and the reference vibration signal obtained from the transformer tank wall to obtain the time correlation characteristics of the two signals; Step S42: Based on the time correlation characteristics, calculate the time lead or lag of the winding vibration relative to the oil tank reference vibration on the waveform, and convert this time difference into a phase difference value representing the angular offset between the two.

6. The method for monitoring transformer winding deformation based on the Internet of Things according to claim 1, characterized in that, Step S5, the specific steps for dynamic threshold adjustment, are as follows: Based on historical operating data accumulated over a long period of time, the cloud platform pre-constructs a mapping table reflecting the relationship between salt spray concentration, transformer top oil temperature, and winding vibration phase shift. When the salt spray concentration value monitored by the system in real time exceeds the set critical level, the cloud will dynamically adjust the alarm threshold for phase shift based on the currently measured transformer top oil temperature value.

7. The method for monitoring transformer winding deformation based on the Internet of Things according to claim 1, characterized in that, Step S6, the deformation position positioning, specifically includes the following steps: Step S61: When the system detects that the phase offset angle of a certain sensor exceeds the dynamically adjusted alarm threshold, all sensors deployed on the transformer tank wall are immediately activated, and the collaborative analysis network is started. Step S62: Calculate the coordinates of the anomaly source using the transfer function matrix; Step S63: Solve the ill-conditioned equations using Tikhonov regularization to calculate the three-dimensional coordinates of the points where deformation or loosening occurs inside the transformer windings.

8. The method for monitoring transformer winding deformation based on the Internet of Things according to claim 1, characterized in that, Step S7, the deformation position positioning, specifically includes the following steps: Step S71: Edge nodes upload alarm data packets via narrowband IoT; Step S72: The cloud platform automatically associates with the wind turbine SCADA system. When the winding hot spot temperature exceeds the threshold, the current load rate of the transformer is higher than the threshold, and the detected phase offset angle continues to exceed the dynamic threshold for a period of time, a level 3 alarm is pushed to the operation and maintenance terminal, and the wind turbine power increase command is blocked.

9. A transformer winding deformation monitoring system based on the Internet of Things, characterized in that, The monitoring system includes: The multimodal sensing and acquisition module is used to deploy a sensor array on the surface of the transformer tank to collect vibration signals, ambient temperature and humidity and salt spray deposition in real time, and transmit the raw data through the industrial bus. The edge computing processing module is used to compensate for salt spray drift of vibration signals, separate power frequency electromagnetic vibration components, and extract the vibration phase difference characteristics between the winding and the oil tank through cross-correlation calculation. The dynamic threshold management module is used to dynamically adjust the phase offset threshold by combining salt spray concentration and oil temperature parameters, establish a mapping relationship between environmental parameters and deformation criteria, and adapt to changes in harsh working conditions. The topology positioning and calculation module is used to activate the sensor network when an anomaly is detected, and calculate the three-dimensional spatial coordinates of the deformation point inside the winding through the acoustic wave transmission matrix to achieve positioning. The IoT communication module is used to upload encrypted alarm data to the cloud via narrowband IoT to ensure reliable transmission in harsh marine environments. The cloud-based decision-making and linkage module is used to integrate wind turbine operation data, generate hierarchical alarm strategies, trigger wind turbine power lockout commands, and link with the operation and maintenance system.

Images