Transformer dehumidifying device parameter self-adaptive adjusting method, medium and transformer dehumidifying device parameter self-adaptive adjusting system

Abstract

The invention provides a transformer dehumidification device parameter self-adaptive adjusting method, medium and system, belongs to the technical field of transformer dehumidification, is used for adjusting parameters of an electroosmosis dehumidification device arranged in a transformer, and comprises the following steps: obtaining the micro-water content in transformer insulating oil collected by each micro-water monitoring unit, establishing a micro-water content matrix of insulating oil in the transformer; determining a dehumidification period for the micro-water content matrix by using an empirical formula; calculating the optimal voltage of the electroosmosis dehumidification device and the maximum loading duration of the optimal voltage by utilizing the determined dehumidification period and the micro-water content matrix and combining with a pre-established electric field intensity-dehumidification efficiency model; and the optimal voltage is loaded to the electroosmosis dehumidification device, and after the optimal voltage is continuously loaded for the maximum duration, the operation is cyclically executed. According to the method, the technical problem that the existing electroosmosis dehumidification device is difficult to adapt to the dynamic change of the running state and the environmental condition of the transformer because the dehumidification parameters are often set by adopting established empirical parameters in the practical application is solved.

Description

Technical Field

[0001] The present invention belongs to the technical field of transformer dehumidification, and in particular, relates to a method, medium and system for adaptively adjusting parameters of a transformer dehumidification device. Background Art

[0002] As an important part of the power system, the insulation performance of the transformer is directly related to the safe operation of the entire power grid. The water content in the insulating oil inside the transformer is one of the key factors affecting the insulation performance. Excessive water content will accelerate the aging of the insulating material, increase the risk of partial discharge, and ultimately lead to a decrease in the withstand voltage performance of the transformer insulation system. Therefore, effective dehumidification of the insulating oil inside the transformer has always been a key technical issue of concern to the power industry. At present, the main methods of transformer dehumidification include adsorption dehumidification, thermal cycle dehumidification and electro-osmosis dehumidification. Among them, electro-osmosis dehumidification is increasingly widely used due to its low energy consumption and high dehumidification efficiency. The principle of electro-osmosis dehumidification is to use the electric field force to drive the directional movement of charged ions in porous media, thereby realizing the directional migration and removal of water in the insulating oil. This method can not only effectively reduce the water content in the insulating oil, but also prevent water from re-penetrating.

[0003] However, in actual applications, the dehumidification parameters of existing electro-osmosis dehumidification devices are often set using established empirical parameters, which has the technical problem of being difficult to adapt to the dynamic changes in transformer operating conditions and environmental conditions. Summary of the invention

[0004] In view of this, the present invention provides a method, medium and system for adaptively adjusting parameters of a transformer dehumidification device, which can solve the technical problem that the setting of dehumidification parameters of existing electro-osmosis dehumidification devices in actual applications often adopts established empirical parameters, which is difficult to adapt to the dynamic changes of transformer operating conditions and environmental conditions.

[0005] The present invention is achieved in that:

[0006] A first aspect of the present invention provides a method for adaptively adjusting parameters of a transformer dehumidification device, wherein the method is used to adaptively adjust parameters of an electro-osmosis dehumidification device disposed in a transformer, and the specific steps include:

[0007] S10, obtaining the micro-water content in the transformer insulating oil collected by each micro-water monitoring unit, and establishing a micro-water content matrix of the insulating oil in the transformer;

[0008] S20, determining a dehumidification cycle for the micro-water content matrix using an empirical formula;

[0009] S30, using the determined dehumidification cycle and the micro-water content matrix, combined with a pre-established electric field strength-dehumidification efficiency model, to calculate the optimal voltage and the maximum duration of optimal voltage loading of the electro-osmosis dehumidification device;

[0010] S40, loading the optimal voltage on the electro-osmosis dehumidification device, and after the optimal voltage is continuously loaded for a maximum time, looping through steps S10-S40.

[0011] Specifically, the step S10 specifically includes: collecting the micro-water content data of each monitoring position in the insulating oil through multiple micro-water monitoring units arranged inside the transformer, and establishing a micro-water content matrix. Each micro-water monitoring unit includes a capacitive sensor, a temperature sensor and a pressure sensor, which are used to measure the dielectric constant, temperature and pressure of the insulating oil at the position in real time. Then, using the pre-established micro-water content calculation equation, the micro-water content in the insulating oil at the position is calculated according to the measured dielectric constant, temperature and pressure data. After measuring the micro-water content at all monitoring positions, a matrix of the micro-water distribution of the insulating oil inside the transformer can be constructed. The purpose of this step is to fully grasp the spatiotemporal distribution of the micro-water content in the insulating oil inside the transformer, and provide a basis for the subsequent dehumidification control strategy.

[0012] Wherein, the step S20 specifically includes: using an empirical formula to determine the dehumidification cycle. The process of constructing the empirical formula is as follows: First, collect a large amount of historical operation data, including transformer load, ambient temperature and humidity, and changes in micro-water content. Then, analyze these historical data to find the correlation between changes in micro-water content and other factors. Based on the correlation analysis results, establish an empirical formula for predicting a suitable dehumidification cycle based on the current transformer operating status and environmental conditions. The empirical formula can be constructed using methods such as multivariate linear regression or nonlinear fitting. After determining the dehumidification cycle, electro-osmosis dehumidification can be carried out in a targeted manner to improve dehumidification efficiency.

[0013] Wherein, the step S30 specifically includes: using the determined dehumidification cycle and micro-water content matrix, combined with the pre-established electric field strength-dehumidification efficiency model, to calculate the optimal voltage and the maximum duration of optimal voltage loading of the electro-osmotic dehumidification device. The electric field strength-dehumidification efficiency model adopts an embedded multi-branch neural network structure, including a screening module, an embedded equation group, a global micro-water content branch network, a local micro-water content branch network, and a fusion network. Using this model, the optimal dehumidification parameters can be dynamically calculated according to the current micro-water distribution of the transformer to improve the overall dehumidification effect.

[0014] Wherein, the step S40 specifically includes: loading the optimal voltage calculated in step S30 onto the electro-osmosis dehumidification device, and continuously loading the optimal voltage for the maximum time. During this process, the device will continue to work, continuously separating and discharging the micro-water in the insulating oil. After the maximum loading time is over, step S10 is entered again to monitor and update the micro-water distribution inside the transformer in real time, so as to optimize the parameters of the next dehumidification cycle. This cycle process can ensure that the dehumidification device can dynamically adjust the dehumidification strategy as the operating state of the transformer changes, and continuously optimize the dehumidification effect.

[0015] Among them, the electro-osmosis dehumidification device specifically includes a drainage system, a porous ceramic tube, an electrically connected built-in electrode, an external electrode, a power control unit, and multiple micro-water monitoring units; the principle is to use the electric field force to drive the directional movement of charged ions in the porous medium, thereby realizing the directional migration and removal of micro-water in the insulating oil.

[0016] The parameter adaptive adjustment of the electro-osmosis dehumidification device mainly includes the following aspects:

[0017] Electric field strength regulation: According to the water content in the insulating oil, the voltage between the inner and outer electrodes is dynamically adjusted to optimize the dehumidification efficiency.

[0018] Dehumidification cycle adjustment: According to the distribution of micro-water, the length of the dehumidification cycle is adaptively adjusted to balance the dehumidification effect and energy consumption.

[0019] Adjustment of micro-water monitoring frequency: According to the transformer operating status and environmental factors, the micro-water monitoring frequency is dynamically adjusted to provide timely feedback on the dehumidification effect.

[0020] Furthermore, the optimal voltage includes 0 voltage, that is, the electro-osmosis dehumidification device does not work.

[0021] Wherein, the micro-water monitoring unit includes a capacitive sensor, a temperature sensor, a pressure sensor and a micro-water monitoring chip.

[0022] Furthermore, a micro-water monitoring module is provided in the micro-water monitoring chip, and the micro-water monitoring module is used to perform the following steps:

[0023] Step 1, collecting the dielectric constant of the insulating oil measured by the capacitive sensor, and the temperature and pressure of the insulating oil measured by the temperature sensor and the pressure sensor;

[0024] Step 2: Calculate the water content in the insulating oil according to the measured dielectric constant, temperature and pressure using a preset water content calculation equation;

[0025] Step 3, calculating the change rate between the current trace water content and the trace water content measured last time;

[0026] Step 4: Dynamically adjust the micro-water monitoring frequency according to the change rate: if the change rate is greater than the preset high threshold, increase the monitoring frequency; if the change rate is less than the preset low threshold, reduce the monitoring frequency; if the change rate is between the high threshold and the low threshold, keep the current monitoring frequency unchanged;

[0027] Step 5: Apply the adjusted monitoring frequency to the next sampling cycle, and transmit the trace water content data, current monitoring frequency and other information to the power control unit.

[0028] Wherein, the step 1 specifically includes: each micro-water monitoring unit measures the dielectric constant, temperature and pressure of the insulating oil respectively through a capacitive sensor, a temperature sensor and a pressure sensor. The capacitive sensor uses an LCR tester to measure the capacitance value of the insulating oil at a frequency of 1kHz, and the temperature and pressure sensors directly measure the corresponding physical quantities. These data collection tasks are completed by the micro-water monitoring module built into the micro-water monitoring chip, and the data is transmitted to the power control unit.

[0029] Wherein, the step 2 specifically includes: the micro-water monitoring module in the micro-water monitoring chip calculates the micro-water content in the insulating oil at the current position according to the dielectric constant, temperature and pressure data collected in step 1, using the pre-established micro-water content calculation equation. The micro-water content calculation equation comprehensively considers multiple factors such as capacitance change rate, dielectric properties, pressure and temperature, and can more accurately reflect the change of micro-water content in the insulating oil.

[0030] The micro-water content calculation equation is specifically expressed as follows:

[0031]

[0032] Where W is the water content in the insulating oil (ppm); C is the capacitance (pF); t is the time (s); A is the real part of the dielectric constant; B is the imaginary part of the dielectric constant; i is the imaginary unit; P k is the pressure of the kth measuring point (Pa); T k is the temperature of the kth measuring point (K); α, β, γ, δ are unknown coefficients; n is the power; m is the number of measuring points.

[0033] The parameter acquisition method is:

[0034] 1. Obtaining C:

[0035] The specific steps for measuring with capacitive sensors are as follows:

[0036] Step 1: Immerse the capacitive sensor in insulating oil;

[0037] Step 2: Use an LCR meter to measure the capacitance at 1kHz.

[0038] Step 3: Record the capacitance values ​​at multiple time points for calculation

[0039] 2. Calculation:

[0040] The central difference method is used for calculation, and the formula is as follows:

[0041]

[0042] Among them, C t+Δt and C t-Δt are the capacitance values ​​at time points t+Δt and t-Δt respectively, and Δt is the time interval.

[0043] 3. Acquisition of A and B:

[0044] It is obtained by dielectric spectrum measurement. The specific steps are as follows:

[0045] Step 1: Use a dielectric spectrum analyzer to scan the insulating oil sample in the frequency range of 0.1 Hz to 1 MHz;

[0046] Step 2: Extract the complex dielectric constant at 1kHz based on the measurement results;

[0047] Step 3: The real part of the complex dielectric constant is A, and the imaginary part is B.

[0048] 4.P k and T k Obtaining:

[0049] Use pressure sensor and temperature sensor to measure. The specific steps are as follows:

[0050] Step 1: Install m pairs of pressure and temperature sensors at different locations in the transformer tank;

[0051] Step 2: Each sensor pair measures the pressure and temperature at that point simultaneously;

[0052] Step 3: Record the pressure and temperature data of all measuring points.

[0053] 5. Determination of α, β, γ, δ:

[0054] Determined by multivariate nonlinear regression analysis, the specific steps are as follows:

[0055] Step 1: Collect a large number of insulating oil sample data with known trace water content, including A, B, P k ,T k And other parameters;

[0056] Step 2: Use the least squares method to fit and optimize the objective function:

[0057]

[0058] Where N is the number of samples, W is j is the actual trace water content of the jth sample.

[0059] 6. Determination of n:

[0060] During the optimization process in step 5, the best n value is also determined. Usually, different integer values ​​(such as 1, 2, 3, etc.) can be tried to select the one with the best fitting effect.

[0061] It is important to note that the following special functions or algorithms are used in this formula:

[0062] - Partial derivatives It represents the partial derivative with respect to time and is used to describe the rate of change of capacitance with time.

[0063] - Complex number operations (A+Bi) n : represents the exponential operation of the complex dielectric constant, which is used to capture the nonlinear effect of dielectric properties on the micro-water content.

[0064] -Summation symbol ∑: It is used to accumulate the pressure-temperature ratio of multiple measuring points, reflecting the comprehensive influence of temperature and pressure on the trace water content.

[0065] This micro-water content calculation equation comprehensively considers multiple factors such as capacitance change rate, dielectric properties, pressure and temperature, and can more comprehensively reflect the changes in micro-water content in insulating oil. By obtaining parameters through experiments and mathematical modeling, an accurate micro-water content estimation model suitable for a specific transformer system can be obtained.

[0066] Wherein, the step 3 specifically includes: the micro-water monitoring module in the micro-water monitoring chip calculates the change rate between the current micro-water content and the micro-water content measured last time. This change rate reflects the change of the micro-water content between two measurements, and can be used to indicate whether the micro-water content of the insulating oil has changed significantly.

[0067] Wherein, the step 4 specifically includes: according to the change rate of the micro-water content calculated in step 3, the micro-water monitoring module will dynamically adjust the micro-water monitoring frequency. The specific adjustment strategy includes: when the change rate is greater than the preset high threshold, the monitoring frequency is increased; when the change rate is less than the preset low threshold, the monitoring frequency is reduced; when the change rate is between the high and low thresholds, the current frequency is kept unchanged. Such an adaptive adjustment strategy can ensure that the monitoring frequency is increased when the micro-water content changes dramatically, and the frequency is reduced when the micro-water content is relatively stable, thereby reducing unnecessary energy consumption.

[0068] Furthermore, it also includes step 6, further adjusting the monitoring frequency according to the operating status and environmental factors of the transformer: when the current load of the transformer is higher than the preset first load threshold, increasing the monitoring frequency according to the difference ratio between the current load and the first load threshold; when the current load of the transformer is lower than the preset second load threshold, reducing the monitoring frequency according to the difference ratio between the current load and the first load threshold.

[0069] The steps of determining the high threshold and the low threshold specifically include:

[0070] Step 11: Collect historical trace water content data and corresponding dehumidification effect data.

[0071] Step 12: Calculate the statistical distribution of the rate of change of the trace water content.

[0072] Step 13: Classify the change rate into three categories according to the effect: "need to increase the monitoring frequency", "can maintain the current frequency" and "can reduce the monitoring frequency".

[0073] Step 14: Use a clustering algorithm (such as K-means) to cluster the three types of data.

[0074] Step 15: Take the minimum value of the "need to increase monitoring frequency" category as the high threshold.

[0075] Step 16: Take the maximum value of the "monitoring frequency can be reduced" category as the low threshold.

[0076] Step 17: Regularly update these thresholds using new data to adapt to system changes.

[0077] The step of determining the first load threshold and the second load threshold specifically includes:

[0078] Step 21: Collect historical transformer load data and corresponding trace water content change data.

[0079] Step 22: Analyze the correlation between the load and the change of the trace water content.

[0080] Step 23: Use regression analysis to determine the relationship function between the load and the rate of change of the trace water content.

[0081] Step 24: According to the relationship function, select the load point where the rate of change of the trace water content increases significantly as the first load threshold.

[0082] Step 25: Select a load point where the rate of change of the trace water content is significantly reduced as the second load threshold.

[0083] Step 26: Verify the effectiveness of these thresholds by simulating different scenarios.

[0084] Step 27: Regularly update these thresholds using new data to adapt to system changes.

[0085] Furthermore, the electric field strength-dehumidification efficiency model adopts an embedded multi-branch neural network model, including a screening module, an embedded equation group, a global micro-water content branch network, a local micro-water content branch network and a fusion network.

[0086] Furthermore, the screening module is used to screen multiple local micro-water content matrices by clustering the input micro-water content matrix: specifically, the micro-water content matrix is ​​divided into multiple clusters using a K-means clustering algorithm, each cluster represents a local micro-water content matrix;

[0087] The embedded equation group includes an initial range equation, a dehumidification efficiency equation, a deviation equation and an adjustment equation, and each equation of the embedded equation group has its own error term;

[0088] The initial range equation is used to determine the initial value range of the electric field strength according to the dehumidification cycle and the micro-water content matrix, the input is the dehumidification cycle and the micro-water content matrix, and the output is the upper and lower limits of the initial value of the electric field strength;

[0089] The dehumidification efficiency equation is used to calculate the theoretical dehumidification efficiency under a given electric field strength, the input is the electric field strength, the micro-water content, the temperature, and the pressure, and the output is the theoretical dehumidification efficiency;

[0090] The deviation equation is used to estimate the deviation between the actual dehumidification efficiency and the theoretical efficiency, the input is the theoretical dehumidification efficiency and the historical actual efficiency data, and the output is the efficiency deviation estimate;

[0091] The adjustment equation is used to adjust the electric field strength according to the efficiency deviation, the input is the current electric field strength and the efficiency deviation, and the output is the adjusted electric field strength;

[0092] The following are the formulas for each equation in the embedded system:

[0093] 1. Initial range equation:

[0094]

[0095] In the formula, E min and E max are the lower and upper limits of the electric field strength (V / m); T d is the dehumidification cycle (h); W i is the water content (ppm) at the ith monitoring point; n is the total number of monitoring points; k1, k2, k3, k4, k5, k6, α, β are unknown coefficients; ε1 and ε2 are error terms.

[0096] Parameter acquisition method:

[0097] T d Determined by the empirical formula in step S20. i It is directly measured by the micro-water monitoring unit. The undetermined coefficient is obtained through historical data regression analysis. The specific steps are as follows:

[0098] Step 1: Collect historical data, including dehumidification cycle, micro-water content matrix and corresponding effective electric field strength range;

[0099] Step 2: Use nonlinear least squares method to fit the data and optimize the unknown coefficients;

[0100] Step 3: Use cross-validation to evaluate the fitting effect and adjust the equation form or introduce new variables if necessary.

[0101] 2. Dehumidification efficiency equation:

[0102]

[0103] Where, η is the theoretical dehumidification efficiency (%); E is the electric field strength (V / m); W is the water content (ppm); T is the temperature (℃); P is the pressure (Pa); k7, k8, k9, k 10 ,ω,φ are unknown coefficients; ε3 is the error term.

[0104] Parameter acquisition method:

[0105] E is the current applied electric field strength, W, T, and P are directly measured by the micro-water monitoring unit. The coefficients to be determined are obtained by fitting the experimental data. The specific steps are as follows:

[0106] Step 1: Design an orthogonal experiment to measure the dehumidification efficiency under different electric field strengths, trace water contents, temperatures, and pressures;

[0107] Step 2: Use nonlinear least squares method to fit the experimental data and optimize the unknown coefficients;

[0108] Step 3: Use leave-one-out cross-validation to evaluate the fitting effect and adjust the equation form if necessary.

[0109] 3. Deviation equation:

[0110]

[0111] Where Δη is the estimated value of efficiency deviation (%); η theo is the theoretical dehumidification efficiency (%); η real is the actual dehumidification efficiency (%); is the average value of historical actual efficiency (%); t is time (h); k 11 ,k 12 ,k 13is the unknown coefficient; ε4 is the error term.

[0112] Parameter acquisition method:

[0113] η theo Calculated by the dehumidification efficiency equation. η rea l Calculate by real-time monitoring of the rate of change of micro-water content:

[0114]

[0115] Where W0 is the initial water content, W t is the water content at time t.

[0116] Obtained by calculating the average of historical actual efficiency data. Compute by numerical differentiation:

[0117]

[0118] The integral term is calculated by numerical integration (such as trapezoidal method). The unknown coefficients are obtained by historical data regression analysis, and the steps are the same as the parameter acquisition method of the initial range equation.

[0119] 4. Adjust the equation:

[0120]

[0121] In the formula, E new is the adjusted electric field strength (V / m); E old is the current electric field strength (V / m); Δη is the estimated value of efficiency deviation (%); t is time (h); k 14 ,k 15 ,k 16 is the unknown coefficient; ε5 is the error term.

[0122] Parameter acquisition method:

[0123] E old is the electric field strength currently applied. Δη is calculated from the deviation equation. Compute by numerical differentiation:

[0124]

[0125] The integral term is calculated by numerical integration (such as trapezoidal method). The unknown coefficients are obtained by historical data regression analysis, and the steps are the same as the parameter acquisition method of the initial range equation.

[0126] The global micro-water content branch network is used to process the global micro-water distribution characteristics of the entire transformer, the input is the global micro-water content matrix, temperature distribution, pressure distribution, and the output is the global dehumidification efficiency prediction;

[0127] The local micro-water content branch network is used to process local high-concentration micro-water areas, the input is the local micro-water content matrix, the local temperature, and the local pressure, and the output is the local dehumidification efficiency prediction;

[0128] The fusion network is used to integrate global and local prediction results, the input is the global dehumidification efficiency prediction, multiple local dehumidification efficiency predictions, and the output is the final dehumidification efficiency prediction and the optimal electric field strength recommendation.

[0129] The steps of constructing the training data set of the electric field strength-dehumidification efficiency model specifically include:

[0130] Step 31, collecting historical operation data, including micro-water content, temperature, pressure, electric field strength and actual dehumidification efficiency;

[0131] Step 32: Perform data cleaning to remove outliers and missing values;

[0132] Step 33: standardize the data;

[0133] Step 34: Generate time series features using a sliding window method;

[0134] Step 35: Perform feature selection to remove redundant and irrelevant features;

[0135] Step 36: Perform stratified sampling on the data set and divide it into a training set, a validation set, and a test set;

[0136] Step 37: Perform data enhancement on the training set, such as adding noise, generating synthetic samples, etc.

[0137] Step 38: Construct negative samples, i.e., invalid or inefficient dehumidification cases.

[0138] The training process of the electric field intensity-dehumidification efficiency model adopts a staged training strategy. First, each equation in the embedded equation group is pre-trained, and the equation parameters are optimized using historical data. Then, the parameters of the embedded equation group are fixed, and the global and local micro-water content branch networks are trained. Finally, all parameters are unfrozen and the entire model is fine-tuned end-to-end. The Adam optimizer is used in the training process, an adaptive learning rate strategy is adopted, and an early stopping method is applied to prevent overfitting. The model performance is evaluated by cross-validation, and an ensemble learning method (such as bagging) is used to improve the generalization ability and stability of the model.

[0139] A second aspect of the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium stores program instructions, and when the program instructions are run in a computer, they are used to execute the above-mentioned method for adaptively adjusting parameters of a transformer dehumidification device.

[0140] The third aspect of the present invention provides a transformer dehumidification device parameter adaptive adjustment system, which includes a control chip and the above-mentioned electro-osmosis dehumidification device; the control chip is provided with a memory and a CPU, the memory stores program instructions, and when the CPU runs the program instructions, it is used to execute the above-mentioned transformer dehumidification device parameter adaptive adjustment method; in the electro-osmosis dehumidification device, the built-in electrode is installed at the inner center position of the porous ceramic tube, the external electrode surrounds the outside of the porous ceramic tube and fits tightly with the outer wall of the ceramic tube, the built-in electrode and the external electrode are respectively connected to the power control unit through insulated wires, and multiple micro-water monitoring units are evenly distributed at different positions of the transformer for monitoring the micro-water content in the insulating oil. The drainage system of the electro-osmosis dehumidification device includes a water collection tank and a drainage pipe. The water collection tank is an annular water collection tank arranged at the bottom of the porous ceramic tube for collecting water discharged from the ceramic tube, and the drainage pipe is a drainage pipe connected from the bottom of the water collection tank for draining the collected water out of the transformer oil tank.

[0141] Compared with the prior art, the method, medium and system for adaptively adjusting parameters of a transformer dehumidification device provided by the present invention have the following beneficial effects:

[0142] 1. Dynamic optimization of dehumidification parameters is achieved. This method combines the micro-water distribution characteristics of the insulating oil inside the transformer and uses the pre-established electric field strength-dehumidification efficiency model to calculate the optimal voltage and optimal loading time of the electro-osmosis dehumidification device based on real-time monitoring data, ensuring the continuous improvement of the dehumidification effect.

[0143] 2. Enhanced the accuracy of micro-water monitoring. By setting up multiple micro-water monitoring units inside the transformer, the temporal and spatial distribution information of the micro-water content in the insulating oil can be fully obtained, providing a more accurate basis for the optimization of dehumidification parameters. At the same time, the monitoring frequency can be dynamically adjusted according to the changes in micro-water to improve the timeliness of micro-water monitoring.

[0144] 3. An adaptive control mechanism is implemented. Based on the micro-water monitoring and dehumidification parameter calculation, this method also designs a corresponding adjustment strategy. When a large change in the micro-water content is detected, the working state of the dehumidification device can be automatically adjusted to ensure that the best dehumidification effect is always maintained.

[0145] In summary, the present invention solves the technical problem that the dehumidification parameters of the existing electro-osmosis dehumidification device are often set using established empirical parameters in actual applications, which makes it difficult to adapt to the dynamic changes of the transformer operating state and environmental conditions. BRIEF DESCRIPTION OF THE DRAWINGS

[0146] Figure 1 A flow chart of a method for adaptively adjusting parameters of a transformer dehumidification device provided by the present invention;

[0147] Figure 2 A schematic diagram of the composition of a transformer dehumidification device parameter adaptive adjustment system provided by the present invention;

[0148] Figure 3 Schematic diagram of the composition of the micro-water monitoring unit. DETAILED DESCRIPTION

[0149] In order to make the purpose, technical solution and advantages of the embodiments of the present invention more clear, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention.

[0150] like Figure 1 As shown, it is a flow chart of a method for adaptively adjusting parameters of a transformer dehumidification device provided by the present invention. The method is used to adjust the parameters of an electro-osmosis dehumidification device arranged in a transformer. The specific steps include:

[0151] S10, obtaining the micro-water content in the transformer insulating oil collected by each micro-water monitoring unit, and establishing a micro-water content matrix of the insulating oil in the transformer;

[0152] S20, for the micro-water content matrix, determine the dehumidification cycle using an empirical formula;

[0153] S30, using the determined dehumidification cycle and the micro-water content matrix, combined with a pre-established electric field strength-dehumidification efficiency model, to calculate the optimal voltage and the maximum duration of optimal voltage loading of the electro-osmotic dehumidification device;

[0154] S40, loading the optimal voltage on the electro-osmosis dehumidification device, and after the optimal voltage is continuously loaded for a maximum time, steps S10-S40 are executed in a loop.

[0155] The specific implementation methods of the above steps are described in detail below:

[0156] like Figure 2 As shown, the electro-osmosis dehumidification device specifically includes a porous ceramic tube, an internal electrode, an external electrode, a power control unit, a plurality of micro-water monitoring units and a drainage system; Figure 3 As shown, the micro-water monitoring unit includes a capacitive sensor, a temperature sensor, a pressure sensor and a micro-water monitoring chip.

[0157] Furthermore, a micro-water monitoring module is provided in the micro-water monitoring chip, and the micro-water monitoring module is used to perform the following steps:

[0158] Step 1, collecting the dielectric constant of the insulating oil measured by the capacitive sensor, and the temperature and pressure of the insulating oil measured by the temperature sensor and the pressure sensor;

[0159] Step 2: Calculate the water content in the insulating oil according to the measured dielectric constant, temperature and pressure using a preset water content calculation equation;

[0160] Step 3, calculating the change rate between the current trace water content and the trace water content measured last time;

[0161] Step 4: Dynamically adjust the micro-water monitoring frequency according to the change rate: if the change rate is greater than the preset high threshold, increase the monitoring frequency; if the change rate is less than the preset low threshold, reduce the monitoring frequency; if the change rate is between the high threshold and the low threshold, keep the current monitoring frequency unchanged;

[0162] Step 5: Apply the adjusted monitoring frequency to the next sampling cycle, and transmit the trace water content data, current monitoring frequency and other information to the power control unit.

[0163] Furthermore, it also includes step 6, further adjusting the monitoring frequency according to the operating status and environmental factors of the transformer: when the current load of the transformer is higher than the preset first load threshold, increasing the monitoring frequency according to the difference ratio between the current load and the first load threshold; when the current load of the transformer is lower than the preset second load threshold, reducing the monitoring frequency according to the difference ratio between the current load and the first load threshold.

[0164] The following is a detailed description of the above steps:

[0165] The specific implementation method of step S10 is to collect the micro-water content data of each monitoring position in the insulating oil through multiple micro-water monitoring units set inside the transformer, and establish a micro-water content matrix. Each micro-water monitoring unit includes a capacitive sensor, a temperature sensor and a pressure sensor, which are used to measure the dielectric constant, temperature and pressure of the insulating oil at the position in real time. Then, the micro-water content in the insulating oil at the position is calculated based on the measured dielectric constant, temperature and pressure data using the pre-established micro-water content calculation equation. After measuring the micro-water content at all monitoring positions, a matrix of the micro-water distribution of the insulating oil inside the transformer can be constructed. The purpose of this step is to fully grasp the spatiotemporal distribution of the micro-water content in the insulating oil inside the transformer, and provide a basis for the subsequent dehumidification control strategy.

[0166] The specific implementation method of step S20 is to use an empirical formula to determine the dehumidification cycle. The construction process of the empirical formula is as follows: First, collect a large amount of historical operation data, including transformer load, ambient temperature and humidity, and changes in micro-water content. Then, analyze these historical data to find the correlation between changes in micro-water content and other factors. Based on the correlation analysis results, establish an empirical formula to predict the appropriate dehumidification cycle based on the current transformer operating status and environmental conditions. The empirical formula can be constructed using methods such as multivariate linear regression or nonlinear fitting. After determining the dehumidification cycle, electro-osmosis dehumidification can be carried out in a targeted manner to improve dehumidification efficiency.

[0167] The specific implementation method of step S30 is to use the determined dehumidification cycle and micro-water content matrix, combined with the pre-established electric field strength-dehumidification efficiency model, to calculate the optimal voltage and the maximum duration of the optimal voltage loading of the electro-osmotic dehumidification device. The electric field strength-dehumidification efficiency model adopts an embedded multi-branch neural network structure, including a screening module, an embedded equation group, a global micro-water content branch network, a local micro-water content branch network and a fusion network. Among them, the embedded equation group includes 4 sub-equations: an initial range equation, a dehumidification efficiency equation, a deviation equation and an adjustment equation. These sub-equations determine their respective parameters through historical data regression analysis to form a differentiable end-to-end model. The global and local branch networks are used to process the global micro-water distribution and the local high-concentration micro-water area respectively, and the fusion network integrates the prediction results of the two branches to give the final dehumidification efficiency prediction and the optimal electric field strength recommendation. Using this model, the optimal dehumidification parameters can be dynamically calculated according to the current micro-water distribution of the transformer to improve the overall dehumidification effect.

[0168] The specific implementation of step S40 is to load the optimal voltage calculated in step S30 onto the electro-osmosis dehumidification device, and continue to load the optimal voltage for the maximum time. During this process, the device will continue to work, continuously separating and discharging the micro-water in the insulating oil. After the maximum loading time is over, step S10 is entered again to monitor and update the micro-water distribution inside the transformer in real time, so as to optimize the parameters of the next dehumidification cycle. This cycle process can ensure that the dehumidification device can dynamically adjust the dehumidification strategy as the operating state of the transformer changes, and continuously optimize the dehumidification effect.

[0169] The specific implementation method of step 1 is that each micro-water monitoring unit measures the dielectric constant, temperature and pressure of the insulating oil through a capacitive sensor, a temperature sensor and a pressure sensor. The capacitive sensor uses an LCR tester to measure the capacitance value of the insulating oil at a frequency of 1kHz, and the temperature and pressure sensors directly measure the corresponding physical quantities. These data collection tasks are completed by the micro-water monitoring module built into the micro-water monitoring chip, and the data is transmitted to the power control unit.

[0170] The specific implementation method of step 2 is that the micro-water monitoring module in the micro-water monitoring chip calculates the micro-water content in the insulating oil at the current position based on the dielectric constant, temperature and pressure data collected in step 1, using the pre-established micro-water content calculation equation. The micro-water content calculation equation comprehensively considers multiple factors such as the capacitance change rate, dielectric properties, pressure and temperature, and can more accurately reflect the change in the micro-water content in the insulating oil. Specifically, the calculation formula is as follows:

[0171]

[0172] Where W is the water content in the insulating oil (unit: ppm), C is the capacitance (unit: pF), t is the time (unit: seconds), A is the real part of the dielectric constant, B is the imaginary part of the dielectric constant, i is the imaginary unit, P k is the pressure at the kth measuring point (unit: Pa), T k is the temperature of the kth measuring point (unit: K), α, β, γ, δ are unknown coefficients, n is the power, and m is the number of measuring points.

[0173] This micro-water content calculation equation comprehensively considers multiple factors such as capacitance change rate, dielectric properties, pressure and temperature, and can more comprehensively reflect the change of micro-water content in insulating oil. Through experimental parameter acquisition and mathematical modeling, an accurate micro-water content estimation model suitable for a specific transformer system can be obtained.

[0174] The specific implementation method of step 3 is that the micro-water monitoring module in the micro-water monitoring chip calculates the current micro-water content W and the micro-water content W measured last time. last The rate of change between This rate of change reflects the change in the moisture content between two measurements and can be used to indicate whether the moisture content of the insulating oil has changed significantly.

[0175] The specific implementation method of step 4 is that the micro-water monitoring module will dynamically adjust the micro-water monitoring frequency according to the micro-water content change rate calculated in step 3. The specific adjustment strategy is as follows:

[0176] - If the rate of change is greater than the preset high threshold θ h , then increase the monitoring frequency to track the rapid changes in trace water content more timely.

[0177] - If the rate of change is less than the preset low threshold θ l , then reduce the monitoring frequency and save energy.

[0178] - If the rate of change is above the high threshold θ h and low threshold θ l The current monitoring frequency remains unchanged.

[0179] Such an adaptive adjustment strategy can ensure that the monitoring frequency is increased when the water content changes dramatically to capture more effective information, and the frequency is reduced when the water content is relatively stable to reduce unnecessary energy consumption. h and θ l The appropriate reference value can be determined by analyzing historical monitoring data and using clustering algorithms.

[0180] The specific implementation method of step 5 is that the micro-water monitoring chip transmits the adjusted monitoring frequency and the currently measured micro-water content data to the power control unit through the data line. The power control unit applies this information to the subsequent dehumidification parameter optimization process.

[0181] In summary, the specific implementation of steps S10-S40 and steps 1-5 involves multiple sub-steps, including the construction of the micro-water content matrix, the determination of the dehumidification cycle, the establishment and application of the electric field strength-dehumidification efficiency model, the adaptive adjustment of the micro-water content monitoring frequency, etc. The purpose of these steps is to make full use of the micro-water distribution information inside the transformer and dynamically optimize the operating parameters of the electro-osmotic dehumidification device to improve the overall dehumidification effect. Among them, the key technologies involved include the micro-water content calculation equation, the embedded neural network model, the adaptive monitoring frequency adjustment strategy, etc.

[0182] A second aspect of the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium stores program instructions, and when the program instructions are run in a computer, they are used to execute the above-mentioned method for adaptively adjusting parameters of a transformer dehumidification device.

[0183] A third aspect of the present invention provides a transformer dehumidification device parameter adaptive adjustment system, such as Figure 2 As shown, it includes a control chip and the above-mentioned electro-osmosis dehumidification device; the control chip is provided with a memory and a CPU, the memory stores program instructions, and when the CPU runs the program instructions, it is used to execute the above-mentioned method for adaptively adjusting parameters of a transformer dehumidification device; in the electro-osmosis dehumidification device, the built-in electrode is installed at the inner center position of the porous ceramic tube, the external electrode surrounds the outside of the porous ceramic tube and fits tightly with the outer wall of the ceramic tube, the built-in electrode and the external electrode are respectively connected to the power control unit through insulated wires, and multiple micro-water monitoring units are evenly distributed at different positions of the transformer to monitor the micro-water content in the insulating oil. The drainage system of the electro-osmosis dehumidification device includes a water collection tank and a drainage pipe. The water collection tank is an annular water collection tank arranged at the bottom of the porous ceramic tube for collecting water discharged from the ceramic tube, and the drainage pipe is a drainage pipe connected from the bottom of the water collection tank to guide the collected water out of the transformer oil tank.

[0184] In order to better understand and implement the present invention, a specific embodiment 1 of the method of the present invention is provided below: The steps included in this embodiment 1 are explained in detail as follows:

[0185] Step S10: First, by setting up multiple micro-water monitoring units inside the transformer, the micro-water content data of each monitoring position in the insulating oil can be obtained, and a micro-water content matrix W can be established:

[0186] Among them, w ij It represents the water content of the ith monitoring point at the jth measurement time (unit: ppm), m is the total number of monitoring points, and n is the length of the measurement time series.

[0187] Each micro-water monitoring unit contains the following sensors and modules:

[0188] 1. Capacitive sensor: used to measure the dielectric constant C (unit: pF) of insulating oil. It can be measured at a frequency of 1kHz using an LCR tester, and the capacitance values ​​at multiple time points can be recorded for calculation.

[0189] 2. Temperature sensor: used to measure the temperature T (unit: ℃) of the insulating oil.

[0190] 3. Pressure sensor: used to measure the pressure P (unit: Pa) of insulating oil.

[0191] 4. Micro-water monitoring chip: Built-in micro-water monitoring module, used to process the data from the above sensors and calculate the micro-water content w of the insulating oil at the current position.

[0192] The specific calculation formula of the micro-water monitoring module is as follows:

[0193]

[0194] Among them, α, β, γ, δ are unknown coefficients, A is the real part of the dielectric constant, B is the imaginary part of the dielectric constant, i is the imaginary unit, and n is the power.

[0195] This formula takes into account the capacitance change rate Complex dielectric constant (A+Bi) n And the pressure-temperature ratio The factors such as α, β, γ and δ can accurately reflect the variation of the water content in insulating oil. The unknown coefficients α, β, γ and δ can be determined by regression analysis of a large amount of measured data.

[0196] By converting the trace water content data of each monitoring point into ijOrganized into a matrix W, the temporal and spatial characteristics of the distribution of micro-water in the insulating oil of the transformer can be fully reflected, which provides an important basis for the subsequent optimization of the dehumidification control strategy.

[0197] Next is the specific implementation of step S20. d (Unit: h), the following empirical formula can be used:

[0198]

[0199] in, is the average temperature of the insulating oil inside the transformer (unit: °C), k1, k2, k3, k4 are unknown coefficients.

[0200] The formula takes into account the statistical characteristics of the distribution of micro-water content, the average temperature of the transformer and some empirical factors, and can accurately predict the appropriate dehumidification cycle. The coefficients to be determined can be determined by regression analysis of a large amount of historical operating data.

[0201] Determine the dehumidification cycle T d After that, we can proceed to step S30 and use the electric field strength-dehumidification efficiency model to calculate the optimal voltage and optimal voltage loading time of the electro-osmotic dehumidification device. This model adopts an embedded multi-branch neural network structure, including the following key components:

[0202] 1. Screening module:

[0203] - Input: Micro-water content matrix W

[0204] - Output: Multiple local micro-water content matrices {W (k)};

[0205] -The K-means clustering algorithm is used to divide the global micro-water content matrix W into K local matrices to reflect the local micro-water distribution characteristics.

[0206] 2. Embed the system of equations:

[0207] (1) Initial range equation:

[0208]

[0209] Among them, E min and E max are the lower and upper limits of the electric field strength (unit: V / m), k1, k2, k3, k4, k5, k6, α, β are unknown coefficients, and ε1 and ε2 are error terms. The equation is based on the dehumidification cycle T d and the statistical characteristics of the water content distribution to determine the reasonable initial range of the electric field intensity.

[0210] (2) Dehumidification efficiency equation:

[0211]

[0212] Where η is the theoretical dehumidification efficiency (%),

[0213] E is the electric field strength (V / m), w is the water content (ppm), T is the temperature (℃), P is the pressure (Pa), k7, k8, k9, k 10 ,ω,φ are unknown coefficients, and ε3 is an error term. This equation describes the influence of electric field intensity, micro-water content, temperature and pressure on dehumidification efficiency.

[0214] (3) Deviation equation:

[0215]

[0216] Where Δη is the estimated value of efficiency deviation (%), η theo is the theoretical dehumidification efficiency (%), η real is the actual dehumidification efficiency (%), is the average value of historical actual efficiency (%), t is time (h), k 11 ,k 12 ,k 13 is the unknown coefficient, and ε4 is the error term. This equation is used to estimate the deviation between the actual dehumidification efficiency and the theoretical efficiency.

[0217] (4) Adjustment equation:

[0218]

[0219] Among them, E new is the adjusted electric field strength (V / m), E old is the current electric field strength (V / m), Δη is the estimated value of efficiency deviation (%), t is the time (h), k 14 ,k 15 ,k 16 is the unknown coefficient, and ε5 is the error term. The equation adaptively adjusts the electric field strength according to the efficiency deviation to optimize the dehumidification effect.

[0220] 3. Global micro-water content branch network:

[0221] -Input: global micro-water content matrix W, temperature distribution T, pressure distribution P;

[0222] -Output: Global dehumidification efficiency prediction η global

[0223] 4. Local micro-water content branch network:

[0224] - Input: Local micro-water content matrix W (k), local temperature T (k) , local pressure P (k) ;

[0225] - Output: Local dehumidification efficiency prediction

[0226] 5. Converged Network:

[0227] - Input: Global dehumidification efficiency prediction η global ,Prediction of multiple local dehumidification efficiency

[0228] -Output: Final dehumidification efficiency prediction η final , optimal electric field strength E opt ;

[0229] The entire electric field strength-dehumidification efficiency model can make full use of the global and local micro-water distribution information through the fusion of the embedded equation group and the multi-branch neural network, dynamically predict the optimal dehumidification parameters, and provide a basis for the execution of step S40.

[0230] A specific embodiment 2 of the micro-water monitoring module of the present invention is provided below: The steps included in this embodiment 2 are explained in detail as follows:

[0231] In step 1, each micro-water monitoring unit collects the dielectric constant C, temperature T and pressure P data of the insulating oil through sensors and transmits them to the micro-water monitoring chip.

[0232] In step 2, the micro-water monitoring module built into the micro-water monitoring chip uses the aforementioned micro-water content calculation formula based on the data collected in step 1 to obtain the micro-water content w of the insulating oil at the current position:

[0233]

[0234] in, Calculated by the central difference method:

[0235]

[0236] A and B are the real and imaginary parts of the complex dielectric constant obtained by dielectric spectroscopy measurement.

[0237] In step 3, the micro-water monitoring module calculates the current micro-water content w and the micro-water content w measured last time. last The rate of change between

[0238] In step 4, according to the change rate obtained in step 3, the micro-water monitoring module will dynamically adjust the micro-water monitoring frequency. The specific strategy is as follows:

[0239] - If the rate of change is greater than the preset high threshold θh , then increase the monitoring frequency;

[0240] - If the rate of change is less than the preset low threshold θ l , then reduce the monitoring frequency;

[0241] - If the rate of change is θ h and θ l The current frequency remains unchanged.

[0242] These thresholds θ h and θ l Appropriate reference values ​​can be determined by clustering analysis of historical monitoring data.

[0243] In step 5, the micro-water monitoring chip transmits the adjusted monitoring frequency and the currently measured micro-water content data to the power control unit through the data line, providing input for subsequent dehumidification parameter optimization.

[0244] In summary, in the specific implementation of steps S10-S40 and steps 1-5, a large number of mathematical formulas and algorithm models are used, the micro-water distribution characteristics of the insulating oil inside the transformer are fully utilized, and the dynamic adaptive adjustment of the dehumidification device parameters is realized. These innovative methods can effectively improve the overall effect of electro-osmosis dehumidification and ensure the long-term stable operation of the transformer insulation system.

[0245] The following is a specific embodiment 3 of the electro-osmosis dehumidification device of the present invention: The main components of the electro-osmosis dehumidification device of this embodiment 3 include a porous ceramic tube, an internal electrode, an external electrode, a power control unit, a plurality of micro-water monitoring units, and a drainage system. The connection relationship between these components is as follows:

[0246] 1. Connection between porous ceramic tube and electrode:

[0247] The built-in electrode is installed at the inner center of the porous ceramic tube and keeps a certain distance from the inner wall of the ceramic tube.

[0248] The external electrode surrounds the outside of the porous ceramic tube and fits tightly to the outer wall of the ceramic tube.

[0249] An electric field is formed between the built-in electrode and the external electrode to drive the directional movement of charged ions in the porous medium.

[0250] 2. Connection between electrode and power control unit:

[0251] The internal electrode and the external electrode are respectively connected to the power control unit through insulated wires.

[0252] The power control unit provides an adjustable voltage to the electrodes through these connections, thereby controlling the electric field strength.

[0253] 3. Installation of micro-water monitoring unit:

[0254] Multiple trace water monitoring units are evenly distributed at different locations of the transformer to monitor the trace water content in the insulating oil.

[0255] Each micro-water monitoring unit contains a capacitive sensor, a temperature sensor and a pressure sensor, which are in direct contact with the insulating oil.

[0256] 4. Connection between micro-water monitoring unit and power control unit:

[0257] Each micro-water monitoring unit is connected to the power control unit via a data line.

[0258] These connections are used to transmit data such as moisture content, temperature, pressure, etc., as well as to receive monitoring frequency adjustment instructions.

[0259] 5. Integration of power control unit:

[0260] The power control unit integrates a micro-water monitoring chip, a data processing module and a voltage regulation module.

[0261] The micro-water monitoring chip is used to process the data from each micro-water monitoring unit.

[0262] The data processing module performs calculations of the electric field strength-dehumidification efficiency model.

[0263] The voltage regulating module adjusts the voltage applied to the electrodes according to the calculation results.

[0264] 6. Layout of the overall system:

[0265] The porous ceramic tube and its electrode system are installed at appropriate positions in the transformer tank to ensure full contact with the insulating oil.

[0266] The power control unit is installed in the control box outside the transformer for easy operation and maintenance.

[0267] Micro-water monitoring units are distributed in key locations inside the transformer tank to obtain comprehensive micro-water distribution information.

[0268] 7. Safety measures:

[0269] Install overcurrent protection devices and insulation isolation devices between the electrodes and the power control unit to prevent short circuits and electrical faults.

[0270] Install temperature sensors at key points of the system to monitor the system operating temperature and prevent overheating.

[0271] 8. Communication interface:

[0272] The power control unit is equipped with a communication interface, which can exchange data with the transformer monitoring system.

[0273] Through this interface, the operating status of the dehumidification device can be remotely monitored and necessary parameter adjustments can be made.

[0274] 9. Design and connection of drainage system:

[0275] Water collecting trough: An annular water collecting trough is set at the bottom of the porous ceramic tube to collect the water discharged from the ceramic tube.

[0276] Drain pipe: A drain pipe is connected from the bottom of the sump to conduct the collected water out of the transformer tank.

[0277] Check valve: Install a check valve on the drain pipe to prevent external air and moisture from flowing back into the tank.

[0278] Drain pump: If necessary, a small drain pump can be installed in the drain line to assist in the drainage of water, especially when the transformer is located at a low position.

[0279] Water Storage Container: The drain pipe is ultimately connected to a sealed water storage container outside the transformer to collect the drained water.

[0280] 10. Connection between drainage system and other components:

[0281] Water collection trough and porous ceramic tube: The water collection trough fits tightly to the bottom of the porous ceramic tube to ensure effective collection of seeping water.

[0282] Drain pipe and oil tank: The drain pipe needs to pass through the oil tank wall, so a good sealing design is required at the passage point to prevent oil leakage.

[0283] Drainage system and power control unit: If a drain pump is used, it needs to be connected to the power control unit so that the drainage can be controlled as needed.

[0284] Water level sensor: Install water level sensors in the water collection tank and water storage container, connected to the power control unit, to monitor the water level.

[0285] 11. Monitoring and control of drainage system:

[0286] The water level sensor transmits data to the power control unit.

[0287] The power control unit decides whether to start the drain pump (if installed) based on the water level.

[0288] The system can set alarm thresholds to notify operators when the water storage container is close to full.

[0289] Specifically, the principle of the present invention is to make full use of the temporal and spatial characteristics of the micro-water distribution in the insulating oil of the transformer, establish a mathematical model of electric field strength-dehumidification efficiency, and realize intelligent optimization of dehumidification parameters. The specific implementation principle is as follows:

[0290] First, by setting up multiple micro-water monitoring units inside the transformer, the micro-water content data of each monitoring position in the insulating oil can be fully obtained and organized into a micro-water content matrix. These monitoring units contain capacitive sensors, temperature sensors, and pressure sensors, which can measure the dielectric constant, temperature, and pressure of the insulating oil in real time. Then, using the pre-established micro-water content calculation formula, these measurement data are converted into the micro-water content value of the current position.

[0291] Based on the obtained micro-water content matrix and the determined dehumidification cycle, the present invention establishes an embedded neural network model of electric field strength-dehumidification efficiency. The model includes components such as a screening module, an embedded equation group, a global and local micro-water branch network, and a fusion network. Among them, the embedded equation group contains four sub-equations: an initial range equation, a dehumidification efficiency equation, a deviation equation, and an adjustment equation. These equations determine the corresponding parameters by performing regression analysis on historical data to form a differentiable end-to-end model.

[0292] The global and local micro-water branch networks are used to process the overall micro-water distribution characteristics of the transformer and the local high-concentration area, respectively. The fusion network combines the prediction results of the two branches to give the final dehumidification efficiency prediction and optimal electric field strength recommendation. Such a multi-branch neural network structure can make full use of global and local micro-water information and improve the prediction accuracy of the dehumidification effect.

[0293] In the actual operation process, the electro-osmosis dehumidification device first dehumidifies the inside of the transformer according to the optimal voltage and optimal loading time calculated in step S30. During this period, the micro-water monitoring unit will continuously monitor the changes in the micro-water content in the insulating oil and transmit the monitoring data to the power control unit through steps 1-5.

[0294] The micro-water monitoring module built into the power control unit will dynamically adjust the monitoring frequency according to the real-time monitoring of the micro-water content change rate. When the micro-water content changes significantly, the monitoring frequency is increased to capture the change information in time; when the micro-water content is relatively stable, the monitoring frequency is reduced to save energy. At the same time, the control unit will also input these monitoring data into the aforementioned electric field strength-dehumidification efficiency model, calculate the latest dehumidification parameters in real time, and automatically adjust the working state of the electro-osmosis dehumidification device.

[0295] Such a parameter adaptive adjustment mechanism can ensure that the electro-osmosis dehumidification device always maintains the optimal working state and continuously improves the overall dehumidification performance.

[0296] In order to further better understand and implement the present invention, the following provides Example 4 of a specific application scenario of the present invention: The transformer monitoring center of a certain power company recently received a report on the abnormal insulation performance of a main transformer. Through remote monitoring data analysis, the micro-water content in the insulating oil inside the transformer showed a continuous upward trend and has exceeded the warning value. In order to take measures as soon as possible to optimize the insulation state of the transformer, the power company decided to test run the transformer internal dehumidification device parameter adaptive adjustment method proposed by the present invention on the transformer.

[0297] 1. Installation of micro-water monitoring unit

[0298] First, the staff installed 10 micro-water monitoring units at key locations inside the transformer tank. Each monitoring unit includes the following sensors and modules:

[0299] 1) Capacitive sensor: used to measure the dielectric constant C of insulating oil.

[0300] 2) Temperature sensor: used to measure the temperature T of the insulating oil.

[0301] 3) Pressure sensor: used to measure the pressure P of insulating oil.

[0302] 4) Micro-water monitoring chip: Built-in micro-water monitoring module, used to process the data from the above sensors and calculate the micro-water content w of the insulating oil at the current position.

[0303] The 10 micro-water monitoring units are evenly distributed at the top, middle and bottom of the transformer tank to fully reflect the micro-water distribution in different areas. The monitoring unit is connected to the transformer control system through a data line to transmit the monitoring data to the power control unit.

[0304] 2. Construction of Micro-water Content Matrix

[0305] After the monitoring unit was installed, the staff started the dehumidification device in the transformer and began to monitor the micro-water content in the insulating oil in real time. Through 10 micro-water monitoring units, data was collected every 30 minutes for 24 hours, and a total of 48 sets of data were obtained. These data were organized into a micro-water content matrix W as follows:

[0306]

[0307] Each row corresponds to a monitoring point, each column corresponds to a time point, and the matrix element w ijIt represents the micro-water content of the i-th monitoring point at the j-th time point (unit: ppm). It can be seen from the matrix that as time goes by, the micro-water content in each area of ​​the transformer shows an upward trend, and the micro-water content in the lower area is significantly higher than that in the upper area. This provides an important basis for the subsequent optimization of dehumidification parameters.

[0308] 3. Determination of dehumidification cycle

[0309] According to the above empirical formula, combined with the above micro-water content matrix W, the staff calculated the dehumidification cycle T suitable for the current transformer operation state. d :

[0310]

[0311] in, is the average temperature of the insulating oil inside the transformer.

[0312] According to the calculation results, the optimal dehumidification cycle T of the transformer d That is to say, the electro-osmosis dehumidification device needs to be operated once every 24 hours to keep the water content in the insulating oil within a reasonable range.

[0313] 4. Establishment of electric field intensity-dehumidification efficiency model

[0314] Based on the determined dehumidification cycle T d With the aforementioned micro-water content matrix W, the staff began to construct the electric field strength-dehumidification efficiency model. The specific process is as follows:

[0315] (1) Screening module:

[0316] Using the K-means algorithm, the micro-water content matrix W of the 10 monitoring points is divided into three local matrices {W (1) ,W (2) ,W (3)}, representing the micro-water distribution characteristics in the top, middle and bottom areas of the transformer, respectively.

[0317] (2) Embedded equations:

[0318] Based on historical data, the staff determined the parameters of each embedded equation through nonlinear regression analysis:

[0319] Initial range equation:

[0320]

[0321] Dehumidification efficiency equation:

[0322]

[0323] Deviation equation:

[0324]

[0325] Adjustment equation:

[0326]

[0327] (3) Global and local micro-water branch networks:

[0328] After training, the structure of the global micro-water branch network is as follows:

[0329] Input layer: W, T, P;

[0330] Hidden layer 1: 64 neurons, using ReLU activation function;

[0331] Hidden layer 2: 32 neurons, using ReLU activation function;

[0332] Output layer: 1 neuron, output η global ;

[0333] The structure of the local micro-water branch network is as follows:

[0334] Input layer: W (k) ,T (k) ,P (k) ;

[0335] Hidden layer 1: 32 neurons, using ReLU activation function;

[0336] Hidden layer 2: 16 neurons, using ReLU activation function;

[0337] Output layer: 1 neuron, output

[0338] (4) Converged Network:

[0339] The structure of the fusion network is as follows:

[0340] Input layer: η global , Hidden layer 1: 24 neurons, using Sigmoid activation function

[0341] Output layer: 2 neurons, output η final ,E opt

[0342] Through end-to-end training and optimization, the prediction performance of the entire electric field strength-dehumidification efficiency model has been greatly improved.

[0343] 5. Dynamic optimization of dehumidification parameters

[0344] With the electric field strength-dehumidification efficiency model established above, the staff can start to dynamically optimize the parameters of the electro-osmosis dehumidification device. The specific process is as follows:

[0345] (1) Initial dehumidification: According to the initial range of electric field strength given by the model, E min =5.8kV / m,E max =9.2kV / m, the staff set the voltage of the dehumidification device to E=7.5kV / m, and the dehumidification time lasted for 24 hours.

[0346] (2) Micro-water monitoring and frequency adjustment: During the dehumidification process, 10 micro-water monitoring units collect data every 30 minutes to form a new micro-water content matrix W. The monitoring module built into the micro-water monitoring chip calculates the change rate based on these data:

[0347]

[0348] Since the rate of change is above the preset high threshold θ h = 0.1 and a low threshold θ l =0.05, the monitoring frequency remains unchanged, that is, collecting data once every 30 minutes.

[0349] (3) Dehumidification effect evaluation: After the 24-hour dehumidification, the staff conducted a comprehensive inspection of the transformer interior. Based on the monitoring data, the actual dehumidification efficiency η was calculated. real =78%, which is consistent with the model prediction of η final =82% has a certain deviation.

[0350] (4) Parameter adjustment: The difference between the actual efficiency and the predicted efficiency Δη = η final -η real =4%, substitute the deviation equation and adjustment equation to calculate the new electric field strength E new =8.1kV / m.

[0351] (5) Next dehumidification: At the beginning of the next 24-hour dehumidification cycle, the staff set the voltage of the electro-osmotic dehumidification device to E = 8.1 kV / m and continued to operate for 24 hours. During the whole process, the micro-water monitoring unit continued to collect data at a frequency of 30 minutes, and the power control unit tracked the trend of micro-water changes in real time and adjusted the dehumidification parameters in a timely manner.

[0352] Through multiple rounds of cycle optimization, the water content in the insulating oil inside the transformer has been continuously reduced, and finally stabilized in a reasonable range below 50ppm. During the entire dehumidification process, the voltage of the dehumidification device is dynamically adjusted between 7.5kV / m and 8.5kV / m, and the frequency of water monitoring is always maintained at once every 30 minutes, which fully meets the changing needs of the transformer's operating status.

[0353] The above description is only a specific implementation mode of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A method for adaptively adjusting parameters of a transformer dehumidification device, characterized in that: It is used to adaptively adjust the parameters of the electro-osmosis dehumidification device arranged in the transformer, and the specific steps include: S10, obtaining the micro-water content in the transformer insulating oil collected by each micro-water monitoring unit, and establishing a micro-water content matrix of the insulating oil in the transformer; S20, determining a dehumidification cycle for the micro-water content matrix using an empirical formula; S30, using the determined dehumidification cycle and the micro-water content matrix, combined with a pre-established electric field strength-dehumidification efficiency model, to calculate the optimal voltage and the maximum duration of optimal voltage loading of the electro-osmosis dehumidification device; S40, loading the optimal voltage on the electro-osmosis dehumidification device, and after the optimal voltage is continuously loaded for a maximum time, looping through steps S10-S40.

2. A method for adaptively adjusting parameters of a transformer dehumidification device according to claim 1, characterized in that: The electro-osmosis dehumidification device specifically comprises: a drainage system, a porous ceramic tube, an internal electrode, an external electrode, an electrically connected power supply control unit, and a plurality of micro-water monitoring units.

3. A method for adaptively adjusting parameters of a transformer dehumidification device according to claim 2, characterized in that: The optimal voltage includes 0 voltage, that is, the electro-osmosis dehumidification device does not work.

4. A method for adaptively adjusting parameters of a transformer dehumidification device according to claim 1, characterized in that: The micro-water monitoring unit comprises: an electrically connected capacitive sensor, a temperature sensor, a pressure sensor and a micro-water monitoring chip.

5. A method for adaptively adjusting parameters of a transformer dehumidification device according to claim 4, characterized in that: The micro-water monitoring chip is provided with a micro-water monitoring module, and the micro-water monitoring module is used to perform the following steps: Step 1: collecting the dielectric constant of the insulating oil measured by the capacitive sensor, and the temperature and pressure of the insulating oil measured by the temperature sensor and the pressure sensor; Step 2: Calculate the water content in the insulating oil according to the measured dielectric constant, temperature and pressure using a preset water content calculation equation; Step 3, calculating the change rate between the current trace water content and the trace water content measured last time; Step 4: Dynamically adjust the micro-water monitoring frequency according to the change rate: if the change rate is greater than the preset high threshold, increase the monitoring frequency; if the change rate is less than the preset low threshold, reduce the monitoring frequency; If the rate of change is between the high threshold and the low threshold, the current monitoring frequency remains unchanged; Step 5: Apply the adjusted monitoring frequency to the next sampling period, and transmit the trace water content data and current monitoring frequency information to the power control unit.

6. A method for adaptively adjusting parameters of a transformer dehumidification device according to claim 5, characterized in that: It also includes step 6, further adjusting the monitoring frequency according to the operating status and environmental factors of the transformer: when the current load of the transformer is higher than the preset first load threshold, the monitoring frequency is increased according to the difference ratio between the current load and the first load threshold; when the current load of the transformer is lower than the preset second load threshold, the monitoring frequency is reduced according to the difference ratio between the current load and the first load threshold.

7. A method for adaptively adjusting parameters of a transformer dehumidification device according to claim 5, characterized in that: The electric field intensity-dehumidification efficiency model adopts an embedded multi-branch neural network model, including a screening module, an embedded equation group, a global micro-water content branch network, a local micro-water content branch network and a fusion network.

8. A method for adaptively adjusting parameters of a transformer dehumidification device according to claim 7, characterized in that: The screening module is used to screen multiple local micro-water content matrices using the following clustering method for the input micro-water content matrix: Specifically, the K-means clustering algorithm is used to divide the micro-water content matrix into multiple clusters, each cluster represents a local micro-water content matrix; The embedded equation group includes an initial range equation, a dehumidification efficiency equation, a deviation equation and an adjustment equation, and each equation of the embedded equation group has its own error term; The initial range equation is used to determine the initial value range of the electric field strength according to the dehumidification cycle and the micro-water content matrix, the input is the dehumidification cycle and the micro-water content matrix, and the output is the upper and lower limits of the initial value of the electric field strength; The dehumidification efficiency equation is used to calculate the theoretical dehumidification efficiency under a given electric field strength, the input is the electric field strength, the micro-water content, the temperature, and the pressure, and the output is the theoretical dehumidification efficiency; The deviation equation is used to estimate the deviation between the actual dehumidification efficiency and the theoretical efficiency, the input is the theoretical dehumidification efficiency and the historical actual efficiency data, and the output is the efficiency deviation estimate; The adjustment equation is used to adjust the electric field strength according to the efficiency deviation, the input is the current electric field strength and the efficiency deviation, and the output is the adjusted electric field strength; The global micro-water content branch network is used to process the global micro-water distribution characteristics of the entire transformer, the input is the global micro-water content matrix, temperature distribution, pressure distribution, and the output is the global dehumidification efficiency prediction; The local micro-water content branch network is used to process the local high-concentration micro-water area, the input is the local micro-water content matrix, the local temperature, the local pressure, and the output is the local dehumidification efficiency prediction; The fusion network is used to integrate global and local prediction results, the input is the global dehumidification efficiency prediction, multiple local dehumidification efficiency predictions, and the output is the final dehumidification efficiency prediction and the optimal electric field strength recommendation.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores program instructions, and when the program instructions are executed in a computer, they are used to execute the method for adaptively adjusting parameters of a transformer dehumidification device according to any one of claims 1 to 8.

10. A transformer dehumidification device parameter adaptive adjustment system, characterized in that: It comprises a control chip and the electro-osmosis dehumidification device as claimed in claim 2; the control chip is provided with a memory and a CPU, the memory stores program instructions, and when the CPU runs the program instructions, it is used to execute a transformer dehumidification device parameter adaptive adjustment method as claimed in any one of claims 3-8; in the electro-osmosis dehumidification device, the built-in electrode is installed at the inner center position of the porous ceramic tube, the external electrode surrounds the outside of the porous ceramic tube and fits tightly with the outer wall of the ceramic tube, the built-in electrode and the external electrode are respectively connected to the power control unit through insulated wires, and a plurality of micro-water monitoring units are evenly distributed at different positions in the transformer for monitoring the micro-water content in the insulating oil. The drainage system of the electro-osmosis dehumidification device comprises a water collection tank and a drainage pipe, the water collection tank is a water collection tank arranged at the bottom of the porous ceramic tube for collecting water discharged from the ceramic tube, and the drainage pipe is a drainage pipe connected from the bottom of the water collection tank for draining the collected water out of the transformer oil tank.

Images