Transformer with high-temperature early warning device

Abstract

The invention relates to the technical field of transformers, in particular to a transformer with a high-temperature early warning device, which comprises a base and a transformer body fixedly mounted on the base, and further comprises a temperature control system, and the temperature control system is used for monitoring the temperature of the transformer body and dynamically analyzing the transformer body. A plurality of sets of cooling fins are fixedly connected to the outer portion of the transformer body, and a plurality of sets of mounting frames are fixedly connected to the base. In the invention, aiming at the defect of incomplete monitoring of a traditional temperature sensor, the temperature control system can realize comprehensive temperature monitoring and dynamic operation analysis of the transformer body, accurately capture internal complex temperature field distribution, and avoid the problem of insulation aging acceleration caused by local high-temperature leak detection; compared with an existing single threshold value control mode, intelligent early warning is achieved through a dynamic analysis technology, a control strategy can be adjusted according to real-time temperature changes, and the limitation that a traditional method lacks state perception is overcome.

Description

Technical Field

[0001] This invention relates to the field of transformer technology, and in particular to a transformer with a high-temperature early warning device. Background Technology

[0002] In power systems, transformers, as core equipment for energy conversion and transmission, directly affect the reliability of the entire power system through their safe and stable operation. During operation, transformers inevitably generate heat due to core losses, winding copper losses, and other additional losses, causing the internal temperature of the transformer to rise. When the temperature exceeds a certain limit, it accelerates the aging of insulation materials, reduces the transformer's service life, and in severe cases, can even lead to insulation breakdown, short circuits, and other faults, causing large-scale power outages and resulting in huge losses to social production and people's lives.

[0003] Currently, existing transformer temperature monitoring and control methods mainly rely on traditional temperature sensors (such as platinum resistance thermometers and thermocouples) to collect temperature data at single or multiple points, and then use simple threshold comparisons to achieve high-temperature early warning and control. However, this method has many limitations. On the one hand, traditional temperature sensors cannot fully reflect the complex temperature field distribution inside the transformer; on the other hand, existing control strategies often use a single threshold triggering method, lacking dynamic analysis of the transformer's operating status. Summary of the Invention

[0004] To address the technical problem that existing transformer temperature monitoring systems lack dynamic analysis of transformer operating status, this invention provides a transformer with a high-temperature early warning device.

[0005] The technology used in this embodiment of the invention is: a transformer with a high temperature early warning device, including a base and a transformer body fixedly installed on the base, and a temperature control system. The temperature control system is used to monitor the temperature of the transformer body and perform dynamic analysis on the transformer body. Multiple sets of heat sinks are fixedly connected to the outside of the transformer body. Multiple sets of mounting brackets are fixedly connected to the base. Cooling fans are installed on the mounting brackets. A water-cooled heat dissipation component is also provided on the outside of the transformer body.

[0006] In one embodiment, the water-cooled heat dissipation assembly includes a water tank fixedly connected to a base and a water pump fixedly installed inside the water tank. The water tank stores coolant. Multiple sets of heat-conducting pipes are provided through the heat sink. A first connecting pipe and a second connecting pipe are fixedly connected to both ends of the heat-conducting pipes, respectively. Both the first connecting pipe and the second connecting pipe are located inside the water tank. An output pipe is fixedly connected to the outside of the first connecting pipe, and an input pipe is fixedly connected to the outside of the second connecting pipe. The output end of the water pump is fixedly connected to the input pipe. A cooling fan is also fixedly installed on the base, and the cooling fan corresponds to the water tank. Multiple sets of heat dissipation slots are provided on the outside of the water tank.

[0007] The temperature control system includes a three-dimensional temperature field model construction module, a distributed optical fiber sensor network deployment module, a temperature acquisition and preprocessing module, a module for inverting the internal heat source distribution of the transformer based on temperature data, a transformer temperature prediction model establishment module, a dynamic early warning threshold module, a high temperature early warning judgment and level classification module, a graded control module, a temperature feedback adjustment and strategy optimization module, and a system self-diagnosis and maintenance reminder module.

[0008] In one embodiment, the three-dimensional temperature field model construction module describes the heat conduction process inside the transformer body using heat conduction equations based on the principles of heat transfer, as follows:

[0009] ;

[0010] in, For temperature, For time, For material density, For the specific heat capacity of the material, The thermal conductivity of the material, For the intensity of the internal heat source, For Hamiltonian operators, Represents the heat conduction term. Represents the change in internal energy;

[0011] Internal heat source intensity It is caused by winding losses and core losses, winding losses Calculated using the following formula:

[0012] ;

[0013] in, For winding current, For winding resistance, and winding resistance The relationship with temperature is as follows:

[0014] ;

[0015] In the formula, Reference temperature The resistance at that time, α is the temperature coefficient of resistance, and the core loss. Including hysteresis loss and eddy current loss, calculated using the following formula:

[0016] ;

[0017] in, For hysteresis loss, For eddy current losses, , These are the hysteresis loss coefficient and the eddy current loss coefficient, respectively. For power supply frequency, For the maximum magnetic flux density of the iron core, The Steinmetz coefficient;

[0018] After calculating the losses of the windings and the core, and then uniformly distributing them to the corresponding areas, the intensity of the internal heat source can be obtained. Distribution;

[0019] Then, by combining the geometric parameters of the transformer, the thermal properties of the material, and the boundary conditions, the heat conduction equation was solved using finite element analysis software, and a three-dimensional temperature field model of the transformer was constructed.

[0020] In one embodiment, the distributed optical fiber sensor network deployment module determines the fiber optic laying path and sensor point distribution based on the structural characteristics of the transformer body and the analysis results of the three-dimensional temperature field model. The fiber optic laying covers the winding hot spot area, core surface, and tank wall of the transformer body. The temperature measurement principle of the distributed optical fiber sensor is based on the Raman scattering effect in the optical fiber, and its temperature measurement formula is:

[0021] ;

[0022] in, The change in temperature For reference temperature, For temperature The ratio of the intensity of Stokes light to the intensity of anti-Stokes light at the time of the event. Reference temperature Strength ratio at time It is a proportionality constant.

[0023] In one embodiment, the temperature acquisition and preprocessing module uses a distributed optical fiber sensor network to acquire the temperature inside the transformer body in real time.

[0024] The collected temperature data is filtered to eliminate noise interference. A moving average filtering algorithm is used, and its calculation formula is as follows:

[0025] ;

[0026] in, For the first Filtered temperature values ​​for each data point. To adjust the sliding window size, For the first Original temperature data; by selecting the window size It can filter out high-frequency noise while preserving the temperature change trend;

[0027] Secondly, it also includes outlier detection and processing of the filtered temperature data, using an outlier detection method based on standard deviation. When a certain temperature data... A value is considered an outlier if it meets the following condition:

[0028] ;

[0029] in, The mean of the temperature data series. To find the standard deviation, linear interpolation is used to correct outliers, i.e.:

[0030] ;

[0031] In the formula, The temperature value after repair. and These are the normal temperature data before and after the outlier, respectively.

[0032] In one embodiment, the temperature data-based transformer internal heat source distribution inversion module inverts the heat source distribution inside the transformer body based on the preprocessed temperature data and a three-dimensional temperature field model.

[0033] By adjusting the intensity of the internal heat source The distribution of minimizes the error between the temperature field calculated by the heat conduction equation and the actual measured temperature field. The objective function is defined as:

[0034] ;

[0035] in, The objective function value, The number of measurement points, Based on the distribution of heat sources Calculated Temperature value at that location The actual measured temperature value;

[0036] The conjugate gradient method is used to optimize the objective function and find the optimal heat source distribution. .

[0037] The iterative formula for the conjugate gradient method is:

[0038] ;

[0039] in, For the first Heat source distribution in the next iteration Step size factor For the first The gradient of the next iteration, gradient Pass

[0040] By differentiating the objective function, we obtain:

[0041] ;

[0042] Through continuous iteration, until the objective function is achieved. The heat source distribution converges to the minimum value. This is the result of the inversion.

[0043] In one embodiment, the transformer temperature prediction model building module constructs a temperature prediction model based on the heat source distribution and historical temperature data obtained by inversion, using time series analysis methods combined with neural network algorithms.

[0044] First, the historical temperature data undergoes stationarity testing and preprocessing to transform non-stationary temperature series into stationary ones. This preprocessing is performed using the finite difference method. Its first-order difference sequence is:

[0045] ;

[0046] If the first-order difference sequence is still not stationary, then perform the second-order difference until a stationary sequence is obtained;

[0047] Then, a temperature prediction model is constructed using a backpropagation (BP) neural network. The BP neural network consists of an input layer, a hidden layer, and an output layer, and its prediction formula is as follows:

[0048] ;

[0049] in, For prediction Temperature value at time The input values ​​are the input values ​​of the input layer neurons. These are the connection weights from the input layer to the hidden layer. These are the connection weights from the hidden layer to the output layer. This represents the bias value of the hidden layer neurons. This represents the bias value of the output layer neurons. quantity, The number of neurons in the input layer. The sigmoid function is used as the activation function. ;

[0050] A genetic algorithm is used to optimize the initial weights and biases of the BP neural network. The objective function is defined as the mean square error between the predicted and actual values.

[0051] ;

[0052] in, For the sample size, For the first The predicted temperature value for each sample. For the first The actual temperature value of each sample.

[0053] In one embodiment, the dynamic early warning threshold module sets a dynamic early warning threshold based on the transformer's operating parameters, environmental conditions, aging characteristics of the insulation material, and the equipment's service life.

[0054] First, consider the impact of the transformer's rated capacity and load factor on the early warning threshold. The transformer's load factor... Defined as:

[0055] ;

[0056] in, This refers to the actual operating capacity of the transformer itself. Rated capacity;

[0057] The aging degree of the insulation material is also considered. The relationship between the aging life of the insulation material and temperature follows the Arrhenius equation:

[0058] ;

[0059] in, For temperature Insulation life at that time Reference temperature Insulation life at that time For activation energy, The value is Boltzmann's constant. Based on the remaining life of the insulating material, when the remaining life is lower than a certain value, the warning threshold should be lowered and a warning should be issued in advance.

[0060] At the same time, a correction factor for the service life of the transformer body is introduced. The longer the equipment has been in operation, the more serious the aging of internal components, and the warning threshold for the decline in heat resistance needs to be further lowered.

[0061] Taking all the above factors into account, the dynamic early warning threshold The calculation formula is optimized as follows:

[0062] ;

[0063] in, As the baseline warning threshold, , , , These are the influence coefficients of load rate, ambient temperature, insulation aging, and service life, respectively. For reference, ambient temperature For the remaining life of the insulation material, The actual number of years of operation. For the designed service life.

[0064] In one embodiment, the high temperature warning judgment and level classification module judges the high temperature warning based on the obtained temperature prediction value and the set dynamic warning threshold, and classifies the warning level.

[0065] First, predict the temperature With dynamic early warning threshold Compare;

[0066] like This indicates that the transformer body temperature is within a safe range, and no warning will be issued.

[0067] like A high-temperature warning will be issued.

[0068] The warning levels are divided into three levels:

[0069] Level 1 warning: When And the rate of temperature rise At that time, a Level 1 warning was triggered;

[0070] in, This is the temperature deviation value for a Level 1 warning. The threshold for the rate of temperature rise in the first-level warning;

[0071] Level 2 warning: When or rate of temperature rise At that time, a level-two warning was triggered; among them, The temperature deviation value for Level II warning (unit: K). The threshold for the rate of temperature rise during the Level II warning period (unit: K / s) is given, and

[0072] Level 3 Warning: When or rate of temperature rise At that time, a Level 3 warning was triggered.

[0073] In one embodiment, the hierarchical control module formulates corresponding hierarchical control strategies for different warning levels. For a level one warning, although the transformer body temperature slightly exceeds the warning threshold, the control strategy focuses on enhancing heat dissipation.

[0074] Specific measures include improving the operating efficiency of the cooling system, increasing the number of operating cooling fans, or increasing the speed of the water pump. (Number of operating cooling fans) Adjustments can be made based on the difference between the current temperature and the warning threshold. The calculation formula is as follows:

[0075] ;

[0076] in, The current number of cooling fans in operation. The temperature reduction achieved by adding one fan. It is a rounding function;

[0077] When a Level 2 warning is triggered, the transformer body temperature exceeds the warning threshold by a significant amount or rises at a rapid rate. In addition to enhancing heat dissipation, the load on the transformer body is adjusted, and the calculation formula for the load reduction is optimized as follows:

[0078] ;

[0079] in, For the load adjustment coefficient of the secondary early warning ( ), Actual cooling medium flow rate: To design cooling flow rate, Traffic influencing factors;

[0080] When the actual flow rate is lower than the design value, the insufficient heat dissipation is compensated by increasing the load reduction ratio.

[0081] If a Level 3 warning is triggered, the load on the transformer itself will be significantly reduced, or even the power supply to some critical loads will be suspended.

[0082] Load reduction The calculation formula is:

[0083] ;

[0084] in, The load adjustment coefficient for Level 3 early warning ( If the temperature continues to rise after load reduction measures are taken, the emergency shutdown procedure for the transformer body will be initiated to disconnect the transformer body from the power grid.

[0085] After implementing the hierarchical control strategy, the temperature feedback regulation and strategy optimization module needs to collect the temperature data of the transformer body in real time through a distributed optical fiber sensor network to form a temperature feedback closed loop and evaluate the control effect.

[0086] The control strategy is dynamically adjusted and optimized based on the deviation between the feedback temperature data and the predicted temperature data.

[0087] Define temperature deviation Actual temperature With predicted temperature The difference:

[0088] ;

[0089] when When the actual temperature is higher than the predicted temperature, the control measures have not been effective as expected, and the control efforts need to be strengthened. If the actual temperature is lower than the predicted temperature, the control measures are too strict and can be appropriately weakened based on the temperature deviation. The proportional-integral-derivative control algorithm is used to adjust the control parameters, and a dynamic weight of the temperature rise rate is introduced.

[0090] The control formula for adjusting the number of operating cooling fans is optimized as follows:

[0091] ;

[0092] in, This is an adjustment for the number of fans in operation. This is the proportionality coefficient. The integral coefficient is... Differential coefficients For rate weighting coefficients, This represents the rate of temperature increase.

[0093] The system self-diagnosis and maintenance reminder module performs regular self-diagnosis of the entire system and issues maintenance reminders based on the diagnosis results and the operating status of the transformer body.

[0094] The system self-diagnosis includes functional checks on each component, such as the distributed fiber optic sensor network, data acquisition and processing module, and control actuator.

[0095] For fiber optic sensor networks, faults such as fiber breakage and excessive attenuation are determined by detecting parameters such as optical signal intensity and signal-to-noise ratio. The fault determination formula is as follows: or in, The current optical signal-to-noise ratio, To achieve the minimum permissible signal-to-noise ratio, Current optical signal power The minimum permissible optical signal power is defined as follows: When the above condition is met, the fiber optic sensor network is considered to be faulty.

[0096] When the cooling fan operates current Exceeding the rated current 1.2 times or action response time Exceeding the set threshold At that time, it was determined that the fan was faulty.

[0097] When the remaining life of the insulation material Below the set maintenance threshold When necessary, a reminder to replace the insulation material will be issued.

[0098] The beneficial effects of this invention are:

[0099] Compared with existing technologies, this invention addresses the shortcomings of traditional temperature sensors in terms of comprehensive monitoring. The temperature control system enables comprehensive temperature monitoring and dynamic operation analysis of the transformer body, accurately capturing the complex internal temperature field distribution and avoiding accelerated insulation aging caused by missed detection of localized high temperatures. Compared to existing single-threshold control methods, intelligent early warning is achieved through dynamic analysis technology, adjusting the control strategy based on real-time temperature changes and overcoming the limitations of traditional methods that lack state awareness. Simultaneously, multiple heat sinks, cooling fans, and water-cooled components form a multi-layered heat dissipation system, rapidly enhancing heat dissipation capacity as temperatures rise.

[0100] II. Compared with existing technologies, this invention addresses the shortcomings of traditional temperature monitoring, which suffers from incompleteness and insufficient accuracy, by employing a technical approach combining a three-dimensional temperature field model with a distributed optical fiber sensor network. The three-dimensional temperature field model is constructed based on heat transfer principles and uses finite element analysis software to accurately calculate the internal temperature distribution of the transformer, overcoming the limitations of single-point temperature measurement. The distributed optical fiber sensor network utilizes the Raman scattering effect to achieve continuous distributed temperature measurement, exhibiting strong anti-electromagnetic interference capabilities and covering key areas such as winding hotspots and the core surface, enabling real-time capture of temperature changes across the entire domain. Regarding the accuracy and timeliness of early warning, this embodiment uses a temperature prediction model optimized by a BP neural network and genetic algorithm, combined with inverted heat source distribution data, to predict future temperature trends in advance. The dynamic early warning threshold comprehensively considers load rate, ambient temperature, insulation aging degree, and equipment operating years, employing a multi-factor correction formula for precise setting, avoiding false alarms and missed alarms caused by a single threshold. Addressing the issues of single control measures and insufficient economy, this embodiment designs a hierarchical control and dynamic optimization system. The temperature feedback closed loop uses an improved PID algorithm, introducing a temperature rise rate weighting coefficient, solving the problem of lag in traditional control response and achieving precise temperature regulation. This tiered strategy ensures equipment safety while minimizing the impact on power system operation, thus balancing safety and economy. Attached Figure Description

[0101] Figure 1 This is a schematic diagram of the structure of the present invention;

[0102] Figure 2 This is a schematic diagram of the transformer body in this invention. Figure 1 ;

[0103] Figure 3 This is a schematic diagram of the transformer body in this invention. Figure 2 ;

[0104] Figure 4 This is a schematic diagram of the heat pipe structure in the transformer body of the present invention.

[0105] The following are marked in the diagram: 1. Transformer body; 2. Heat sink; 3. Base; 4. Fixing frame; 5. Cooling fan; 6. Water tank; 7. Cooling fan; 8. Heat dissipation groove; 9. Heat conduction pipe; 10. First connecting pipe; 11. Second connecting pipe; 12. Output pipe; 13. Input pipe. Detailed Implementation

[0106] In the description of this invention, it should be noted that the terms "front", "up", "down", "left", "right", "vertical", "horizontal", etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this invention.

[0107] refer to Figures 1-4 In order to solve the problems existing in the background technology, this application proposes the following technology in this embodiment: a transformer with a high temperature early warning device, including a base 3 and a transformer body 1 fixedly installed on the base 3, and also including a temperature control system, which is used to monitor the temperature of the transformer body 1 and perform dynamic analysis on the transformer body 1. Multiple sets of heat sinks 2 are fixedly connected to the outside of the transformer body 1, and multiple sets of mounting brackets are fixedly connected to the base 3. Cooling fans 5 are installed on the mounting brackets, and water-cooled heat dissipation components are also provided on the outside of the transformer body 1.

[0108] In one embodiment, the water-cooled heat dissipation assembly includes a water tank 6 fixedly connected to a base 3 and a water pump fixedly installed inside the water tank 6. The water tank 6 stores coolant. Multiple sets of heat-conducting pipes 9 are provided through the heat sink 2. A first connecting pipe 10 and a second connecting pipe 11 are fixedly connected to the two ends of the heat-conducting pipe 9, respectively. Both the first connecting pipe 10 and the second connecting pipe 11 are located inside the water tank 6. An output pipe 12 is fixedly connected to the outside of the first connecting pipe 10, and an input pipe 13 is fixedly connected to the outside of the second connecting pipe 11. The output end of the water pump is fixedly connected to the input pipe 13. A cooling fan 7 is also fixedly installed on the base 3. The cooling fan 7 corresponds to the water tank 6. Multiple sets of heat dissipation slots 8 are provided on the outside of the water tank 6.

[0109] The above technical solution is explained as follows: The heat generated by the transformer body 1 during operation is initially dissipated through multiple sets of externally fixed heat sinks 2. The temperature control system monitors and dynamically analyzes the temperature of the transformer body 1 in real time, accurately monitoring changes in the internal temperature field and providing a basis for heat dissipation regulation.

[0110] When the temperature rises, the cooling fan 5 on the mounting bracket on the base 3 starts, accelerating the airflow around the heat sink 2 and enhancing the convection cooling effect. At the same time, the water cooling components operate synchronously: the water pump in the water tank 6 pumps the coolant through the inlet pipe 13 into the second connecting pipe 11, and then into the multiple sets of heat conduction pipes 9 that run through the heat sink 2. After absorbing the heat transferred by the heat sink 2 in the heat conduction pipes 9, the coolant flows back to the water tank 6 through the first connecting pipe 10 and the outlet pipe 12.

[0111] The cooling fan 7 on the base 3 blows air towards the water tank 6, which, together with the multiple heat dissipation slots 8 on the outside of the water tank 6, quickly dissipates the heat carried by the coolant, ensuring the efficiency of coolant circulation and heat dissipation, forming a multi-layer heat dissipation closed loop of "transformer body heat dissipation - heat sink conduction - air cooling + water cooling enhanced heat dissipation", achieving efficient temperature control of the transformer.

[0112] The temperature control system includes a three-dimensional temperature field model construction module, a distributed fiber optic sensor network deployment module, a temperature acquisition and preprocessing module, a module for inverting the internal heat source distribution of the transformer based on temperature data, a transformer temperature prediction model establishment module, a dynamic early warning threshold module, a high-temperature early warning judgment and level classification module, a graded control module, a temperature feedback adjustment and strategy optimization module, and a system self-diagnosis and maintenance reminder module.

[0113] To comprehensively and accurately reflect the internal temperature distribution of a transformer, the three-dimensional temperature field model construction module first needs to construct a three-dimensional temperature field model of the transformer. The temperature field distribution of a transformer is affected by various factors, including winding losses and core losses.

[0114] First, based on the principles of heat transfer, the heat conduction process inside the transformer body 1 is described using the heat conduction equation, as follows:

[0115] ;

[0116] in, Temperature (unit: K) Time (unit: seconds) Material density (unit: kg / m³) , Specific heat capacity of the material (unit: J / (kg·K)) Thermal conductivity of the material (unit: W / (m·K)) Internal heat source intensity (unit: W / m) , For Hamiltonian operators, Represents the heat conduction term. Represents the change in internal energy;

[0117] Internal heat source intensity It is caused by winding losses and core losses, winding losses Calculated using the following formula:

[0118] ;

[0119] in, Winding current (unit: A) Winding resistance (unit: ), while winding resistance The relationship with temperature is as follows:

[0120] ;

[0121] In the formula, Reference temperature The resistance (unit: Ω) at (usually taken as 20°C), α is the temperature coefficient of resistance (unit: ... Core loss Including hysteresis loss and eddy current loss, calculated using the following formula:

[0122] ;

[0123] in, Hysteresis loss (unit: W) Eddy current loss (unit: W) , These are the hysteresis loss coefficient and the eddy current loss coefficient, respectively. Power supply frequency (unit: Hz) The maximum magnetic flux density of the iron core (unit: T). This is the Steinmetz coefficient, which is usually taken as 1.5-2.5.

[0124] After calculating the losses of the winding and core using the above formula, and then uniformly distributing them to the corresponding areas, the intensity of the internal heat source can be obtained. The distribution of heat conduction is determined. Then, combining the transformer's geometric parameters, material thermal properties, and boundary conditions (such as the convective heat transfer coefficient between the coolant and the transformer body surface), the heat conduction equation is solved using finite element analysis software (such as ANSYS, COMSOL, etc.), constructing a three-dimensional temperature field model of the transformer. Solving the heat conduction equation using finite element analysis software (such as ANSYS, COMSOL, etc.) to construct the three-dimensional temperature field model of the transformer is a current technology. A brief method is as follows: First, import the three-dimensional geometric model of the transformer and clarify the material properties of each component, such as density, specific heat capacity, and thermal conductivity, including the core, windings, and tank. Next, mesh the model, discretizing it into a large number of elements. Then, define boundary conditions, including the convective heat transfer coefficient between the coolant and the surface. Then, calculate the internal heat source intensity based on the winding and core loss formulas and apply it to the model. Finally, solve the heat conduction equation; the software iteratively calculates the temperature of each element to obtain the three-dimensional temperature field distribution of the entire transformer, which can intuitively display results such as temperature cloud maps.

[0125] A three-dimensional temperature field model of a transformer can intuitively display the temperature values ​​at various points inside the transformer, providing a basis for subsequent temperature monitoring and early warning.

[0126] The above-mentioned technology is explained in this embodiment as follows: Constructing a three-dimensional temperature field model of the transformer is the fundamental step of the entire control method and has significant technical advantages. Traditional temperature monitoring often relies on single-point sensors, which are difficult to reflect the complex temperature distribution inside the transformer. However, based on the principles of heat transfer, the three-dimensional model constructed by combining the heat conduction equation with winding loss and core loss calculations can intuitively display the temperature values ​​at various points inside the transformer. The winding loss formula considers the change of resistance with temperature, and the core loss formula distinguishes between hysteresis and eddy current losses, making the calculation of internal heat source intensity more accurate. With the help of finite element software, the model can accurately present the temperature conditions of key areas such as winding hot spots and core surfaces, providing a theoretical reference for subsequent temperature monitoring. This model overcomes the limitations of traditional temperature measurement, can identify potential high-temperature hazards in advance, lays the foundation for temperature field verification, heat source inversion, and early warning strategy formulation, improves the depth of understanding of the transformer heating mechanism, and ensures that subsequent monitoring and control are more targeted.

[0127] To verify and monitor the three-dimensional temperature field model of transformer body 1 in real time, the distributed optical fiber sensor network deployment module needs to deploy a distributed optical fiber sensor network. Distributed optical fiber sensing technology has advantages such as wide measurement range, high spatial resolution, and strong anti-electromagnetic interference capability, enabling continuous distributed monitoring of the internal temperature of transformer body 1. When deploying the optical fiber sensor network, the fiber laying path and sensor point distribution need to be determined based on the structural characteristics of transformer body 1 and the analysis results of the three-dimensional temperature field model. The fiber laying covers the winding hotspot area, core surface, and tank wall of transformer body 1. The temperature measurement principle of the distributed optical fiber sensor is based on the Raman scattering effect in the optical fiber, and its temperature measurement formula is:

[0128] ;

[0129] in, Temperature change (unit: K) Reference temperature (unit: K), For temperature The ratio of the intensity of Stokes light to the intensity of anti-Stokes light at the time of the event. Reference temperature Strength ratio at time It is a proportionality constant, which is related to the characteristics of the optical fiber material and the light source.

[0130] This formula allows calculation of the temperature value at a corresponding location based on the ratio of light signal intensity collected by the fiber optic sensor. Simultaneously, to improve measurement accuracy, the fiber optic sensing network needs to be calibrated to eliminate the influence of factors such as fiber loss and connector loss on the measurement results. During calibration, a standard heat source of known temperature can be used to calibrate the sensing points, establishing an accurate correspondence between the light signal intensity ratio and temperature.

[0131] The above-mentioned technologies are explained in this embodiment as follows: Real-time acquisition and preprocessing of temperature data are crucial to ensuring the accuracy of subsequent analysis. Raw temperature data is susceptible to noise and outliers. A moving average filtering algorithm, using an appropriate window size, filters out high-frequency noise while preserving the temperature change trend, making the data smoother and more stable. An outlier detection method based on standard deviation can accurately identify data deviating from the normal range, and then repair it through linear interpolation, avoiding interference from outliers in subsequent analysis. The quality of the preprocessed temperature data is significantly improved, providing high-quality input for heat source inversion and temperature prediction. Traditional methods often lead to analytical biases due to data noise or anomalies, but through systematic processing, data accuracy and reliability are ensured, reducing false alarms caused by data problems and providing data support for the accuracy of the entire control method.

[0132] The temperature acquisition and preprocessing module utilizes a distributed fiber optic sensor network to acquire the temperature inside the transformer body 1 in real time. The acquired temperature data contains a large amount of raw information, but may also contain noise, outliers, and other interference factors, requiring preprocessing. First, the acquired temperature data is filtered to eliminate noise interference. A moving average filtering algorithm is used, and its calculation formula is as follows:

[0133] ;

[0134] in, For the first Filtered temperature values ​​(unit: K) for each data point To adjust the sliding window size, For the first The original temperature data (unit: K) is used to select an appropriate window size. This method can effectively filter out high-frequency noise while preserving the temperature change trend. Secondly, it performs outlier detection and processing on the filtered temperature data, employing an outlier detection method based on standard deviation. When a certain temperature data... A value is considered an outlier if it meets the following condition:

[0135] ;

[0136] in, This represents the mean of the temperature data series (unit: K). The standard deviation is K. Outliers detected are corrected using linear interpolation, i.e.:

[0137] ;

[0138] In the formula, The temperature value after repair (unit: K). and These are the normal temperature data before and after the outliers (unit: K). The preprocessed temperature data is more accurate and reliable, providing high-quality data support for subsequent temperature field inversion and state assessment.

[0139] The above-mentioned technology is explained in this embodiment as follows: Real-time acquisition and preprocessing of temperature data are crucial to ensuring the accuracy of subsequent analysis. Raw temperature data is susceptible to noise and outliers. A moving average filtering algorithm, using an appropriate window size, filters out high-frequency noise while preserving the temperature change trend, making the data smoother and more stable. An outlier detection method based on standard deviation can accurately identify data deviating from the normal range, and then linear interpolation is used to repair it, avoiding interference from outliers in subsequent analysis. The quality of the preprocessed temperature data is significantly improved, providing high-quality input for heat source inversion and temperature prediction. Traditional methods often lead to analytical biases due to data noise or anomalies, but through systematic processing, data accuracy and reliability are ensured, reducing false alarms caused by data problems and providing data support for the accuracy of the entire control method.

[0140] The module for inverting the internal heat source distribution of the transformer based on temperature data inverts the heat source distribution inside the transformer body 1 by combining the preprocessed temperature data with a three-dimensional temperature field model.

[0141] The inversion of heat source distribution helps to gain a deeper understanding of the loss distribution and heating mechanism of transformer body 1, providing a more accurate basis for high-temperature early warning and control. The heat source inversion problem can be reduced to an optimization problem, namely, adjusting the intensity of internal heat sources. The distribution of minimizes the error between the temperature field calculated by the heat conduction equation and the actual measured temperature field. The objective function is defined as:

[0142] ;

[0143] in, The objective function value, The number of measurement points, Based on the distribution of heat sources Calculated Temperature value at (unit: K), The temperature value is the actual measured value (unit: K). The conjugate gradient method is used to optimize the objective function and find the optimal heat source distribution. .

[0144] The iterative formula for the conjugate gradient method is:

[0145] ;

[0146] in, For the first Heat source distribution in the next iteration Step size factor For the first The gradient of the next iteration, gradient Accessible

[0147] By differentiating the objective function, we obtain:

[0148] ;

[0149] Through continuous iteration, until the objective function is achieved. The heat source distribution converges to the minimum value. This is the inversion result. The heat source distribution obtained from the inversion can accurately reflect the loss distribution inside the transformer body 1, providing important parameters for subsequent temperature prediction and condition assessment.

[0150] The above-mentioned technology is explained in this embodiment as follows: Inverting the internal heat source distribution of a transformer based on temperature data deepens the understanding of the transformer's heating characteristics. Traditional methods cannot directly obtain the internal heat source distribution and can only indirectly infer it through external temperature measurement, resulting in limited accuracy. The heat source inversion is transformed into an optimization problem. By minimizing the error between the calculated and measured temperatures using the conjugate gradient method, the internal heat source intensity distribution is obtained. The objective function accurately quantifies the temperature deviation, and the iterative process continuously optimizes the heat source distribution. The final result accurately reflects the loss distribution of the windings and core. This process reveals the correlation between internal transformer losses and heating, providing key parameters for temperature prediction models and offering in-depth evidence for assessing the transformer's operating status. Compared to traditional methods, the inversion results allow high-temperature early warning to move beyond surface temperature monitoring to analyze the root causes of heating, improving the scientific rigor and relevance of the warning and laying the foundation for developing effective control strategies.

[0151] To predict the future temperature trend of transformer body 1 and issue early high-temperature warnings, a temperature prediction model for transformer body 1 needs to be established. This model establishment module uses time series analysis combined with a neural network algorithm to construct a temperature prediction model based on the inverted heat source distribution and historical temperature data. First, the historical temperature data undergoes stationarity testing and preprocessing to transform non-stationary temperature sequences into stationary ones. Then, a difference method is used for preprocessing the temperature sequence... Its first-order difference sequence is:

[0152] ;

[0153] If the first-order difference sequence is still not stationary, a second-order difference is performed until a stationary sequence is obtained. Then, a backpropagation (BP) neural network is used to construct a temperature prediction model. The BP neural network consists of an input layer, hidden layers, and an output layer, and its prediction formula is as follows:

[0154] ;

[0155] in, For prediction Temperature value at any given time (unit: K), The input values ​​for the neurons in the input layer (such as historical temperature data, heat source distribution parameters, etc.) These are the connection weights from the input layer to the hidden layer. These are the connection weights from the hidden layer to the output layer. This represents the bias value of the hidden layer neurons. This represents the bias value of the output layer neurons. quantity, The number of neurons in the input layer. For activation functions, the Sigmoid function is typically used.

[0156] To improve the prediction accuracy of the model, a genetic algorithm is used to optimize the initial weights and biases of the BP neural network. The genetic algorithm searches for the optimal parameter combination through selection, crossover, and mutation operations to minimize the prediction error of the neural network; the objective function is defined as the mean squared error between the predicted and actual values.

[0157] ;

[0158] in, For the sample size, For the first Predicted temperature values ​​(in K) for each sample. For the first The actual temperature value of each sample (unit: K); by training and optimizing the BP neural network model, it is possible to accurately predict the temperature change trend of each point inside the transformer body 1 in the future period based on the current and historical temperature data and the distribution of heat sources, providing advance warning for high temperature.

[0159] The above-mentioned technology is explained in this embodiment as follows: The establishment of a transformer temperature prediction model enables early prediction of high-temperature risks. Traditional early warning systems rely on real-time temperature threshold comparisons, lacking predictive capabilities and making it difficult to address rising temperature trends in advance. This embodiment employs a BP neural network combined with a genetic algorithm for optimization. First, the non-stationary temperature sequence is processed using a difference method, and then the optimized neural network is used to predict future temperatures. The genetic algorithm optimizes the initial weights and biases, reducing the risk of the neural network getting trapped in local optima. The objective function is guided by minimizing the mean square error, improving prediction accuracy. The model can accurately predict future temperature change trends based on historical temperatures and heat source distribution, providing advance warning for high-temperature warnings. Compared to traditional methods, this model achieves a shift from "passive response" to "active early warning," issuing warnings before temperatures exceed limits, buying time for control measures, and significantly improving the timeliness and effectiveness of early warnings.

[0160] Traditional high-temperature warning for transformer body 1 uses a fixed threshold, which is difficult to adapt to the temperature change requirements of transformer body 1 under different operating conditions. The dynamic warning threshold module sets a dynamic warning threshold based on the operating parameters of transformer body 1, environmental conditions, aging characteristics of insulation materials and equipment operating years.

[0161] First, consider the impact of the rated capacity and load rate of transformer body 1 on the early warning threshold. The load rate of transformer body 1... Defined as:

[0162] ;

[0163] in, The actual operating capacity of transformer body 1 (unit: kVA) The rated capacity is given in kVA. The higher the load rate, the greater the losses in transformer body 1 and the faster the temperature rises. Therefore, the warning threshold should be appropriately reduced as the load rate increases. Secondly, ambient temperature should be considered. Impact on early warning thresholds.

[0164] The higher the ambient temperature, the worse the heat dissipation of the transformer body 1, and the higher the temperature under the same load. Therefore, the warning threshold should decrease as the ambient temperature increases. In addition, the degree of aging of the insulation material also needs to be considered.

[0165] The relationship between the aging life of insulation materials and temperature follows the Arrhenius equation:

[0166] ;

[0167] in, For temperature Insulation life (unit: h) Reference temperature Insulation life (unit: h) Activation energy (unit: J / mol) Boltzmann constant ( ).

[0168] Based on the remaining life of the insulation material, when the remaining life falls below a certain value, the warning threshold should be lowered, and an early warning should be issued. Simultaneously, a correction factor for the transformer body's service life is introduced.

[0169] The longer the equipment has been in operation, the more severe the aging of its internal components, leading to a decline in heat resistance. Therefore, the warning threshold for this necessitates further lowering. Considering all these factors, the dynamic warning threshold... The calculation formula is optimized as follows:

[0170] ;

[0171] in, The baseline warning threshold (unit: K) , , , These are the influence coefficients of load rate, ambient temperature, insulation aging, and service life, respectively. For reference ambient temperature (unit: K), The remaining life of the insulation material (unit: hours). Actual operating years (unit: years) The design service life (unit: years) is given. Compared with existing technologies, this formula adds a factor on the impact of operating years, which more comprehensively reflects the equipment's degradation characteristics and makes the warning threshold setting more accurate.

[0172] The above technology is explained in this embodiment as follows: Setting a dynamic early warning threshold solves the problem of poor adaptability of traditional fixed thresholds. Traditional thresholds do not consider dynamic factors such as load rate and ambient temperature, which can easily lead to late warnings or false alarms. A dynamic threshold formula is constructed by comprehensively considering load rate, ambient temperature, insulation aging, and service life. When the load rate increases or the ambient temperature rises, the threshold decreases accordingly; when insulation aging or service life increases, the threshold is further lowered. The Arrhenius equation is introduced to quantify the relationship between insulation life and temperature, combined with a service life correction factor, so that the threshold is more in line with the actual state of the equipment. The dynamic threshold is adjusted in real time according to the transformer's operating conditions, and reasonable early warning standards can be set under different operating conditions. Compared with fixed thresholds, it can reduce unnecessary warnings while ensuring safety, and provide early warnings when the equipment deteriorates, improving the accuracy and pertinence of the warnings and avoiding the rigidity problem of traditional thresholds.

[0173] The high temperature warning judgment and level classification module judges the high temperature warning based on the obtained temperature prediction value and the set dynamic warning threshold, and classifies the warning level.

[0174] First, predict the temperature With dynamic early warning threshold Compare them.

[0175] like This indicates that the temperature of transformer body 1 is within a safe range, and no warning will be issued; if A high-temperature warning will be issued.

[0176] To more effectively guide the operation and maintenance of transformer body 1, the warning level is divided into three levels based on the degree to which the predicted temperature exceeds the warning threshold and the rate of temperature rise:

[0177] Level 1 warning: When And the rate of temperature rise At that time, a Level 1 warning was triggered.

[0178] in, The temperature deviation value for Level 1 warning (unit: K). The threshold for the rate of temperature rise during the first-level warning (unit: K / s).

[0179] Level 2 warning: When or rate of temperature rise At that time, a level-two warning was triggered. Among them, The temperature deviation value for Level II warning (unit: K). The threshold for the rate of temperature rise during the Level II warning period (unit: K / s) is given, and Level 3 Warning: When or rate of temperature rise At that time, a Level 3 warning was triggered.

[0180] Different warning levels correspond to different emergency response measures, so that operators can take appropriate control strategies according to the warning level.

[0181] The above-mentioned technology is explained in this embodiment as follows: The high-temperature warning judgment and classification make the warnings more instructive and operational. Traditional warnings are mostly single-level, unable to distinguish the severity of high-temperature risks, resulting in insufficient targeted control measures. Based on the predicted temperature exceeding the threshold and the rate of increase, the warnings are divided into three levels, each clearly defined and corresponding to different levels of risk. Level 1 warnings are for mild exceedances and slow temperature rises; Level 2 warnings cover moderate exceedances or relatively rapid temperature rises; and Level 3 warnings correspond to severe exceedances or rapid temperature rises. This classification method allows operators to quickly determine the risk level and take corresponding measures. Traditional methods, due to their ambiguous classifications, may lead to measures that are either too lenient or too severe. By clearly defining the classifications, it is ensured that the warning information can effectively guide decision-making, reducing losses caused by inappropriate responses and improving the scientific nature and efficiency of high-temperature emergency response.

[0182] The hierarchical control module formulates corresponding hierarchical control strategies for different warning levels to achieve effective control of the temperature of the transformer body 1, while taking into account the economic efficiency and reliability of the power system operation.

[0183] For a Level 1 warning, although the temperature of transformer body 1 slightly exceeds the warning threshold, the rate of increase is slow, and the situation is relatively stable. The control strategy focuses on enhancing heat dissipation to minimize impact on the normal operation of transformer body 1. Specific measures include improving the operating efficiency of the cooling system, such as increasing the number of operating cooling fans 5 or increasing the speed of the water pump. The number of operating cooling fans 5... Adjustments can be made based on the difference between the current temperature and the warning threshold. The calculation formula is as follows:

[0184] ;

[0185] in, This represents the current number of 5 cooling fans in operation. The temperature reduction (in Kelvin) that can be achieved by adding one cooling fan. This is the floor function.

[0186] Meanwhile, the operating parameters of transformer body 1 are monitored in real time, and the temperature change trend is closely watched.

[0187] When a Level II warning is triggered, if the temperature of transformer body 1 exceeds the warning threshold by a large margin or rises at a rapid rate, more proactive control measures are required.

[0188] Based on enhanced heat dissipation, the load on transformer body 1 should be adjusted appropriately.

[0189] Considering the impact of the cooling medium (including coolant and the airflow from cooling fan 5) on heat dissipation, the load reduction calculation formula is optimized as follows:

[0190] ;

[0191] in, For the load adjustment coefficient of the secondary early warning ( ), Actual cooling medium flow rate (unit: m³) 3 / h: The unit for design cooling flow rate is m. 3 / h, Traffic Influence Factor ( When the actual flow rate is lower than the design value, the insufficient heat dissipation is compensated by increasing the load reduction ratio, which is more in line with the actual heat dissipation conditions than existing technologies.

[0192] By coordinating with the power dispatching system, some non-critical loads are transferred or reduced to lower the load rate of transformer body 1, thereby reducing losses and heat generation.

[0193] If a Level 3 warning is triggered, it indicates that transformer body 1 is facing a serious risk of high temperature, and emergency control measures must be taken to prevent a fault from occurring.

[0194] At this point, the load on transformer body 1 should be drastically reduced immediately, or even the power supply to some important loads should be suspended.

[0195] Load reduction The calculation formula is:

[0196] ;

[0197] in, The load adjustment coefficient for Level 3 early warning ( If the temperature continues to rise after load reduction measures are implemented, the emergency shutdown procedure for transformer body 1 will be initiated to disconnect transformer body 1 from the power grid, preventing serious faults such as insulation breakdown. Simultaneously, an emergency alarm will be issued to the power dispatch center and operation and maintenance personnel, notifying them to promptly implement repair and emergency power supply measures.

[0198] The above-mentioned technology is explained in this embodiment as follows: A tiered control strategy achieves a balance between safety and economy. Traditional control measures are often simplistic, primarily employing load reduction or shutdown, without considering differences in risk levels, which can easily lead to unnecessary power outages. Differentiated measures are developed for the three levels of early warning: Level 1 warnings enhance heat dissipation to minimize operational impact; Level 2 warnings adjust the load based on heat dissipation, combined with cooling medium compensation; Level 3 warnings involve emergency load reduction or shutdown. The Level 2 load reduction formula incorporates a cooling flow factor, increasing the reduction ratio when actual flow is insufficient to ensure control effectiveness. The tiered strategy prioritizes power supply during periods of mild risk and decisively stops losses during periods of severe risk, avoiding the drawbacks of a "one-size-fits-all" approach. Compared to traditional methods, this approach ensures transformer safety while minimizing load reduction and cooling energy consumption, balancing the reliability and economy of the power system, and improving the rationality and effectiveness of control measures.

[0199] After implementing the hierarchical control strategy, the temperature feedback adjustment and strategy optimization module needs to collect the temperature data of the transformer body 1 in real time through a distributed fiber optic sensor network to form a temperature feedback closed loop and evaluate the control effect. Based on the deviation between the feedback temperature data and the predicted temperature data, the control strategy is dynamically adjusted and optimized. The temperature deviation is defined. Actual temperature With predicted temperature The difference:

[0200] ;

[0201] when When the actual temperature is higher than the predicted temperature, the control measures have not been effective as expected, and stronger control measures are needed, such as further increasing the number of cooling fans or increasing the load reduction. If the actual temperature is lower than the predicted temperature, it means that the control measures are too strict and the control can be appropriately weakened, such as reducing the number of cooling fans running.

[0202] Based on temperature deviation An improved proportional-integral-derivative (PID) control algorithm is used to adjust the control parameters, introducing a dynamic weight for the temperature rise rate. Taking the adjustment of the number of operating cooling fans 5 as an example, the optimized control formula is as follows:

[0203] ;

[0204] in, This is an adjustment for the number of fans in operation. This is the proportionality coefficient. The integral coefficient is... Differential coefficients For rate weighting coefficients ( ), The temperature rise rate is expressed in K / s. When the temperature rises rapidly, the proportional term weight is increased to accelerate the adjustment response, addressing the issue of lag in the response of existing PID controllers to sudden temperature increases. Through PID regulation, the temperature deviation is quickly converged to zero, ensuring that the transformer body temperature remains stable within a safe range. Simultaneously, based on long-term operating data and control performance, machine learning algorithms are used to optimize the parameters in the hierarchical control strategy (such as load adjustment coefficients and temperature deviation values). Control performance evaluation indicators are established, such as temperature control accuracy, load reduction, and cooling energy consumption. With the optimization of these evaluation indicators as the goal, the parameters are iteratively updated to continuously improve the adaptability and effectiveness of the control strategy.

[0205] The above technology is explained in this embodiment as follows: Temperature feedback regulation and strategy optimization enable dynamic and precise adjustment of control measures. Traditional control measures lack closed-loop feedback after implementation, making it difficult to evaluate and optimize their effects. By collecting temperature data in real time, the deviation between the actual and predicted temperatures is calculated, and control parameters are adjusted using an improved PID algorithm. PID regulation introduces a weight for the rate of temperature rise, enhancing the regulation force when the temperature rises too quickly, thus solving the problem of lag in traditional PID response. Simultaneously, machine learning is used to optimize control parameters based on long-term data, continuously iterating with indicators such as control accuracy and energy consumption. The feedback loop ensures that control measures are dynamically adjusted according to actual effects, avoiding insufficient or excessive measures. Compared to traditional methods, this transforms control from "static execution" to "dynamic optimization," continuously improving the adaptability and efficiency of the control strategy, ensuring that the transformer temperature remains stable within a safe range, and reducing energy waste and unnecessary load reduction.

[0206] To ensure the long-term reliable operation of the high-temperature early warning device and control method, the system self-diagnosis and maintenance reminder module performs regular self-diagnosis of the entire system and issues maintenance reminders based on the diagnosis results and the operating status of the transformer body 1.

[0207] System self-diagnosis includes functional checks on various components such as the distributed fiber optic sensor network, data acquisition and processing module, and control actuators. For the fiber optic sensor network, by detecting parameters such as optical signal intensity and signal-to-noise ratio, it determines whether there are faults such as fiber breaks or excessive attenuation. The fault diagnosis formula is: or in, The current optical signal-to-noise ratio, To achieve the minimum permissible signal-to-noise ratio, Current optical signal power The minimum permissible optical signal power is defined as follows: When the above condition is met, the fiber optic sensor network is considered to be faulty.

[0208] Taking cooling fan 5 as an example, when the operating current of cooling fan 5... Exceeding the rated current 1.2 times or action response time Exceeding the set threshold At that time, it was determined that the fan was faulty.

[0209] Based on the self-diagnostic results and historical temperature data of transformer body 1, as well as information such as the remaining lifespan of the insulation materials, a maintenance plan is formulated. When the remaining lifespan of the insulation materials... Below the set maintenance threshold When necessary, a reminder to replace the insulation material will be issued.

[0210] The above-mentioned technology is explained in this embodiment as follows: System self-diagnosis and maintenance reminders ensure the long-term reliable operation of the device. Traditional systems lack automatic diagnostic capabilities, making it difficult to detect faults in a timely manner, and maintenance relies on periodic inspections, which are prone to oversights. Regular self-diagnosis of the fiber optic network and actuators is performed, detecting fiber optic faults through signal-to-noise ratio and optical power, and judging the status of actuators by monitoring current and response time. Based on the diagnostic results and historical data on insulation life and temperature, maintenance reminders are issued, such as insulation replacement and equipment overhaul. Self-diagnosis can promptly detect potential faults, avoiding early warning or control failures due to device malfunctions; maintenance reminders are based on the actual condition of the equipment, achieving precise maintenance and avoiding over-maintenance or under-maintenance. Compared with traditional methods, this improves system reliability, extends transformer lifespan, reduces fault risk and maintenance costs, and provides data support and decision-making basis for the full life cycle management of transformers.

[0211] In summary, this embodiment effectively solves the core problems existing in the current transformer high temperature early warning and control methods, and significantly improves the safety, economy and reliability of transformer operation.

[0212] To address the shortcomings of traditional temperature monitoring methods, such as incompleteness and insufficient accuracy, this embodiment employs a technical approach combining a three-dimensional temperature field model with a distributed fiber optic sensor network. The three-dimensional temperature field model, constructed based on heat transfer principles, accurately calculates the internal temperature distribution of the transformer using finite element analysis software, overcoming the limitations of single-point temperature measurement. The distributed fiber optic sensor network utilizes the Raman scattering effect to achieve continuous distributed temperature measurement, exhibiting strong resistance to electromagnetic interference and covering key areas such as winding hotspots and the core surface. It can capture real-time temperature changes across the entire temperature range, overcoming the drawbacks of traditional sensors that are fixed in place and susceptible to environmental interference, thus providing comprehensive and accurate temperature data support for high-temperature early warning.

[0213] In terms of accuracy and timeliness of early warning, this embodiment breaks through the limitations of traditional fixed thresholds. Through a temperature prediction model optimized by a BP neural network and genetic algorithm, combined with inverted heat source distribution data, future temperature change trends can be predicted in advance. The dynamic early warning threshold comprehensively considers load rate, ambient temperature, insulation aging degree, and equipment operating years, and is precisely set using a multi-factor correction formula, avoiding false alarms and missed alarms caused by a single threshold. The three-level early warning classification, combined with temperature deviation and rate of rise, achieves refined risk identification, significantly improving the targeting and proactiveness of early warnings compared to existing technologies, and buying valuable time for fault prevention.

[0214] To address the issues of simplistic and inefficient control measures, this embodiment designs a hierarchical control and dynamic optimization system. Level 1 early warning enhances heat dissipation without affecting power supply by finely adjusting the number of operating cooling fans 5. Level 2 early warning introduces a cooling flow compensation mechanism, dynamically calculating load reduction based on actual heat dissipation conditions to minimize unnecessary load losses. Level 3 early warning implements emergency temperature control and shutdown protection to prevent major faults. The temperature feedback closed loop employs an improved PID algorithm, introducing a temperature rise rate weighting coefficient to solve the problem of lag in traditional control response and achieve precise temperature control. This hierarchical strategy ensures equipment safety while minimizing the impact on power system operation, balancing safety and economy. Furthermore, the system's self-diagnosis and maintenance reminder functions effectively address the problem of delayed equipment maintenance. By monitoring the operating parameters of key components such as the fiber optic sensor network and cooling fans 5 in real time, faults are detected and warnings are issued promptly. Combined with insulation remaining life calculation and accumulated operating data, intelligent maintenance plans are pushed, avoiding the blindness of traditional periodic maintenance. Full lifecycle data management provides a scientific basis for equipment condition assessment, extends transformer lifespan, and reduces maintenance costs.

[0215] By designing comprehensive temperature monitoring, dynamic early warning strategies, refined control measures, and intelligent operation and maintenance management, the problems of incomplete temperature monitoring, insufficient early warning accuracy, and single control measures in the background technology have been completely solved. This provides all-round protection for the safe and stable operation of transformers, significantly reduces the risk of failure and economic losses, and has important practical significance for improving the reliability of power systems.

[0216] Although embodiments of the invention have been shown and described, the scope of the invention will be defined by the appended claims and their equivalents by those skilled in the art.

Claims

1. A transformer with high temperature early warning device, comprising a base (3) and a transformer body (1) fixedly installed on the base (3), characterized in that, Also include temperature control system, the temperature control system is used for temperature monitoring to transformer body (1) and dynamic analysis to transformer body (1), the outside of the transformer body (1) is fixedly connected with multiple groups of radiating fins (2), the base (3) is fixedly connected with multiple groups of mounting racks, the mounting rack is installed with cooling fan (5), the outside of the transformer body (1) is also provided with water-cooling heat dissipation assembly.

2. The transformer with high temperature early warning device according to claim 1, characterized in that, The water-cooling heat dissipation assembly includes a water tank (6) fixedly connected to the base (3) and a water pump fixedly installed in the water tank (6), the water tank (6) stores a cooling liquid, a plurality of heat conducting pipes (9) are arranged through the radiating fins (2), the two ends of the heat conducting pipe (9) are fixedly connected with a first connecting pipe (10) and a second connecting pipe (11) respectively, the first connecting pipe (10) and the second connecting pipe (11) are located in the water tank (6), the outside of the first connecting pipe (10) is fixedly connected with an output pipe (12), the outside of the second connecting pipe (11) is fixedly connected with an input pipe (13), the output end of the water pump is fixedly connected with the input pipe (13), the base (3) is further fixedly installed with a cooling fan (7), the cooling fan (7) corresponds to the water tank (6), the outside of the water tank (6) is provided with a plurality of heat dissipation grooves (8). The temperature control system includes a three-dimensional temperature field model construction module, a distributed optical fiber sensing network arrangement module, a temperature acquisition preprocessing module, a transformer internal heat source distribution inversion module based on temperature data, a transformer temperature prediction model establishment module, a dynamic early warning threshold module, a high temperature early warning judgment and grade division module, a hierarchical control module, a temperature feedback adjustment and strategy optimization module, and a system self-diagnosis and maintenance reminding module.

3. The transformer with high temperature early warning device according to claim 2, characterized in that, The three-dimensional temperature field model construction module describes the heat conduction process inside the transformer body (1) with a heat conduction equation according to the heat transfer principle, as follows: ； wherein, T is temperature, t is time, p is material density, c is material specific heat capacity, k is material thermal conductivity, q is internal heat source intensity, H is Hamiltonian operator, represents heat conduction term, represents internal energy change term; Inner heat source intensity generated by winding losses and core losses in the transformer body, winding losses are calculated by the following formula: ； wherein is the winding current in the transformer body, is the winding resistance, while the winding resistance is a function of temperature as ； In the formula, Reference temperature The resistance at time α is the temperature coefficient of resistance, and the core loss in the transformer body is... Including hysteresis loss and eddy current loss, calculated using the following formula: ； wherein, is the hysteresis loss, is the eddy current loss, , are the hysteresis loss coefficient and the eddy current loss coefficient, respectively, is the power supply frequency, is the maximum magnetic flux density of the core, is the Steinmetz coefficient; After the losses of the winding and the core in the transformer body are calculated, the losses are evenly distributed to the corresponding regions, and the distribution of the inner heat source intensity is obtained ​ Then, the heat conduction equation is solved by using a finite element analysis software in combination with the geometric structure parameters of the transformer, the material thermal physical property parameters and the boundary conditions, and a three-dimensional temperature field model of the transformer is constructed.

4. The transformer with high temperature early warning device according to claim 3, characterized in that, The distributed optical fiber sensing network arrangement module determines the laying path of the optical fiber and the distribution of the sensing points according to the structural features of the transformer body (1) and the analysis results of the three-dimensional temperature field model, the laying of the optical fiber covers the winding hot spot area, the core surface and the oil tank wall of the transformer body (1), and the temperature measurement principle of the distributed optical fiber sensor is based on the Raman scattering effect in the optical fiber, and the temperature measurement formula is as follows: ； wherein, is the amount of temperature change, is the reference temperature, is the temperature is the intensity ratio of the Stokes light and the anti-Stokes light at the time of the temperature is the reference temperature is the intensity ratio at the time of the reference temperature, is a proportional constant.

5. The transformer having high temperature early warning device as claimed in claim 4 wherein, The temperature acquisition preprocessing module acquires the temperature inside the transformer body (1) in real time by using the arranged distributed optical fiber sensing network; The collected temperature data are filtered to eliminate noise interference, and a sliding average filtering algorithm is adopted, and the calculation formula is as follows: ； wherein, is the filtered temperature value for the th data point, is the size of the sliding window, is the original temperature data for the th data point; by choosing the window size , high-frequency noise is filtered out while the temperature change trend is preserved; Secondly, it also includes the abnormal value detection and processing of the filtered temperature data, and the abnormal value detection method based on standard deviation is adopted. When a certain temperature data When the following formula is met, it is determined as an abnormal value: ； wherein, is the mean of the temperature data sequence, is the standard deviation, the outliers are repaired by linear interpolation, i.e.: ； In the formula, is the temperature value after repair, and are normal temperature data before and after the abnormal value, respectively.

6. The transformer having high temperature early warning device as claimed in claim 5 wherein, The transformer internal heat source distribution inversion module inverses the heat source distribution inside the transformer body (1) according to the preprocessed temperature data in combination with the three-dimensional temperature field model; By adjusting the distribution of the inner heat source intensity such that the error between the temperature field calculated by the heat conduction equation and the actually measured temperature field is minimized, the objective function is defined as: ； wherein, is the target function value, is the number of measurement points, is the temperature value at the point according to the heat source distribution calculated from the heat source distribution at the point, is the actually measured temperature value; The conjugate gradient method is used to optimize and solve the objective function to find the optimal heat source distribution ; The iteration formula of the conjugate gradient method is as follows: ； wherein, is the first iteration of the heat source distribution, is the first iteration of the gradient, is a step factor, is the first iteration of the heat source distribution, is the first iteration of the gradient, pass The derivative of the objective function is as follows: ； By iterating until the objective function converges to a minimum, the resulting heat source distribution is the inversion result.

7. The transformer with high temperature early warning device according to claim 6, characterized in that, The transformer temperature prediction model establishment module adopts a time series analysis method combined with a neural network algorithm to construct a temperature prediction model based on the obtained heat source distribution and historical temperature data by inversion; First, the historical temperature data undergoes stationarity testing and preprocessing to transform non-stationary temperature series into stationary ones. This preprocessing is performed using the finite difference method. Its first-order difference sequence is: ； If the first-order difference sequence is still not stationary, second-order difference is performed until a stationary sequence is obtained; Then, a BP neural network is used to construct a temperature prediction model, the BP neural network is composed of an input layer, a hidden layer and an output layer, and the prediction formula is: ； wherein, is the predicted temperature value at the time instant, is the input value of the input layer neuron, is the connection weight from the input layer to the hidden layer, is the connection weight from the hidden layer to the output layer, is the bias value of the hidden layer neuron, is the bias value of the output layer neuron, quantity, is the number of input layer neurons, is the activation function, a sigmoid function is used: ; The initial weight and bias values of the BP neural network are optimized by using a genetic algorithm, and the objective function is defined as the mean square error between the predicted value and the actual value: ； wherein, is the number of samples, is the predicted temperature value for the is the actual temperature value for the is the actual temperature value for the is the actual temperature value for the 8. The transformer with high temperature early warning device according to claim 7, characterized in that, The dynamic early warning threshold module sets a dynamic early warning threshold according to the operating parameters of the transformer body (1), environmental conditions, aging characteristics of insulating materials and equipment operating life; First, considering the influence of the rated capacity and the load rate of the transformer body (1) on the early warning threshold, the load rate of the transformer body (1) is defined as: ； wherein, is the actual operating capacity of the transformer body (1), is the rated capacity; The aging degree of the insulating material is also considered, and the relationship between the service life of the insulating material and the temperature follows the Arrhenius equation: ； Wherein, is the insulation life at temperature is the insulation life at reference temperature is the insulation life at reference temperature is the insulation life at reference temperature is the activation energy, is the Boltzmann constant, according to the remaining life of the insulation material, when the remaining life is lower than a certain value, the warning threshold should be reduced, and the warning should be given in advance; At the same time, a transformer body (1) operating life correction factor is introduced, the longer the equipment operating life, the more serious the internal component aging, and the heat resistance ability decreases, so the early warning threshold needs to be further adjusted downward; Taking all the above factors into account, the calculation formula of the dynamic early warning threshold is optimized as: ； Wherein, is the reference early warning threshold, , , , are the load rate, ambient temperature, insulation aging and operating age influence coefficients, respectively, is the reference ambient temperature, is the remaining life of the insulation material, is the actual operating age, is the design service life.

9. The transformer having high temperature early warning device as claimed in claim 7 wherein, The high-temperature early warning judgment and grade division module performs high-temperature early warning judgment and divides the early warning grade according to the obtained temperature prediction value and the set dynamic early warning threshold; First, the predicted temperature is compared to a dynamic warning threshold ​ If , it indicates that the temperature of the transformer body (1) is within the safe range, and no warning is issued. If , a high temperature warning is issued; The early warning grade is divided into three levels: Level 1 warning: when and the temperature rise rate is triggered; wherein, is a first level warning temperature deviation value, is a first level warning temperature rise rate threshold value; Secondary warning: when or the temperature rise rate a secondary warning is triggered; wherein, is a secondary warning temperature deviation value, is a secondary warning temperature rise rate threshold, and Level 3 warning: When or temperature rise rate is triggered, level 3 warning is triggered.

10. The transformer with a high-temperature early warning device according to claim 9, characterized in that, The hierarchical control module formulates corresponding hierarchical control strategies for different early warning grades, for a first-level early warning, although the temperature of the transformer body (1) slightly exceeds the early warning threshold, the control strategy is mainly to enhance heat dissipation; Specific measures include increasing the operating efficiency of the cooling system, increasing the number of operating cooling fans (5), or increasing the rotational speed of the water pump; the number of operating cooling fans (5) Adjustment is made according to the difference between the current temperature and the early warning threshold, and the calculation formula is: ； wherein, is the current number of operating cooling fans (5), is the temperature value to be reduced for each additional fan, is a ceiling function; When a second-level early warning is triggered, the temperature of the transformer body (1) exceeds the early warning threshold by a larger margin or has a faster rising rate, on the basis of enhancing heat dissipation, the load of the transformer body (1) is adjusted, and the load reduction calculation formula is optimized as: ； wherein, is a secondary warning load adjustment coefficient , is the actual cooling medium flow rate: is the design cooling flow rate, is the flow rate influence factor; When the actual flow is lower than the design value, the insufficient heat dissipation is compensated by increasing the load reduction proportion; If a third-level early warning is triggered, the load of the transformer body (1) is greatly reduced, and even the power supply of part of important loads is suspended; Load reduction The calculation formula is: ； wherein, is a tertiary early warning load adjustment coefficient ( ) ; if the temperature continues to rise after the load reduction measures are taken, an emergency shutdown procedure of the transformer body (1) is started, and the transformer body (1) is cut off from the power grid; The temperature feedback adjustment and strategy optimization module needs to collect the temperature data of the transformer body (1) in real time through the distributed optical fiber sensing network after implementing the hierarchical control strategy, forms a temperature feedback closed loop, and evaluates the control effect; According to the deviation between the feedback temperature data and the predicted temperature data, the control strategy is dynamically adjusted and optimized; define a temperature deviation from an actual temperature from a predicted temperature the difference ； When , it indicates that the actual temperature is higher than the predicted temperature, the effect of the control measures is not as expected, and the control strength needs to be strengthened. When , it indicates that the actual temperature is lower than the predicted temperature, the control measures are too strict, and the control needs to be weakened. Based on the temperature deviation , a proportional-integral-derivative control algorithm is used to adjust the control parameters, and a temperature rise rate dynamic weight is introduced. The control formula of the number of operating cooling fans (5) is optimized as: ； wherein, is an adjustment amount for the number of fan operations, is a proportional coefficient, is an integral coefficient, is a differential coefficient is a rate weight coefficient, is a temperature rise rate; The system self-diagnosis and maintenance reminding module performs periodic self-diagnosis on the entire system, and sends a maintenance reminder according to the diagnosis result and the operating state of the transformer body (1); System self-diagnosis includes functional inspection of each component part such as the distributed optical fiber sensing network, the data acquisition and processing module and the control execution mechanism; For the optical fiber sensing network, by detecting the intensity, signal-to-noise ratio and other parameters of the optical signal, it is judged whether the optical fiber has faults such as breakage and excessive attenuation, and the fault judgment formula is: or wherein, is the current optical signal signal-to-noise ratio, is the minimum allowed signal-to-noise ratio, is the current optical signal power is the minimum allowed optical signal power; when the above conditions are met, it is determined that the optical fiber sensing network has a fault; When the cooling fan (5) runs current over 1.2 times the rated current or the response time of the action exceeds a set threshold , it is determined that the fan is malfunctioning; When the remaining life of the insulation material is below a set maintenance threshold a replacement of the insulation material is prompted.

Images