A transformer operation intelligent control method and system

Abstract

This application provides an intelligent control method and system for transformer operation, applied in the field of transformer technology. It acquires real-time multimodal physical characteristic data of transformer regulating components through multimodal sensors, performs fusion processing, and calculates a comprehensive degradation index and cumulative degradation trend. This trend is used as feedback to dynamically correct traditional regulation logic. When cumulative degradation reaches the warning condition, priority is given to scheduling flexible resources such as user-side energy storage and controllable loads for voltage regulation, reducing the number of mechanical adjustments. Flexible resources have higher priority than mechanical adjustments; even if flexibility is insufficient, only the minimum number of mechanical adjustments is triggered within the gap. This scheme enables transformer operation control to intelligently perceive the deep health status of equipment and incorporate it into decision-making, effectively reducing mechanical wear, extending equipment life, improving grid reliability and stability, and avoiding sudden failures caused by hidden degradation.

Description

Technical Field

[0001] This application relates to the field of transformer technology, and in particular to an intelligent control method and system for transformer operation. Background Technology

[0002] In modern distribution networks with high penetration of distributed energy resources (such as photovoltaics and energy storage) on the user side, bidirectional power flow patterns cause frequent operation of on-load tap changers (OLTCs) on transformers. In response to economic incentives such as time-of-use pricing, users actively adjust their electricity consumption or generation behavior, resulting in a persistent deviation between actual power output and the predicted curve.

[0003] To maintain voltage stability, the power flow bidirectional control system continuously issues voltage regulation commands. Frequent operations not only accelerate the mechanical wear of the OLTC (Optical Voltage Transformer) but also interact with the response characteristics of some inverters, generating additional harmonic currents. These harmonic currents cause slow saturation of the voltage and current transformer cores, leading to a gradual drift in measurement accuracy. Consequently, the controller makes further misjudgments based on distorted data, exacerbating unnecessary OLTC switching. The arc generated by each switching accelerates the decomposition and carbonization of the insulating oil in the OLTC's independent oil chamber, resulting in a continuous decrease in dielectric strength.

[0004] However, existing monitoring systems only focus on abnormal electrical parameters or macroscopic oil quality indicators during a single operation, leaving blind spots for localized, cumulative insulation degradation. Such hidden declines cannot be detected. When the power grid undergoes routine capacity expansion or load testing, the OLTC performs conventional voltage regulation operations. Due to the severely insufficient dielectric strength of the insulating oil, it cannot effectively extinguish the arc, ultimately causing an internal short circuit, resulting in transformer tripping and regional power outages.

[0005] The root cause of the failure lies in the chain reaction of changes in user behavior patterns, interaction between control logic and equipment characteristics, accumulation of measurement errors, and monitoring blind spots, which exposes the inadequacy of traditional power flow control methods in their lack of intelligent perception of the deep health status of key components.

[0006] Therefore, existing technologies urgently need to be improved to address the aforementioned problems. Summary of the Invention

[0007] In view of the shortcomings of the prior art, this application provides a method and system for intelligent control of transformer operation. By intelligently sensing the deep health status of transformer regulating components and incorporating it into the power flow control logic, it has the advantages of effectively reducing the mechanical regulation frequency of regulating components, extending equipment life, improving the reliability and stability of power grid operation, and avoiding sudden failures caused by hidden degradation.

[0008] Firstly, a method for intelligent control of transformer operation, the method comprising the following steps: S1: Obtain multimodal physical characteristic data of the insulating medium inside the transformer regulating component. The multimodal physical characteristic data includes at least acoustic emission signals, dielectric constant variation data, and distribution density data of solid micro / nano particles. S2: Extract the energy evolution characteristics of the acoustic emission signal, the polarity change trend of the dielectric constant, and the concentration growth rate of the solid micro / nano particles, fuse the three to determine the comprehensive degradation index, and take the change trend of the comprehensive degradation index over time as the cumulative degradation trend. S3: The cumulative degradation trend is input as feedback information into the power flow control program of the transformer to correct the mechanical regulation logic triggered only by voltage deviation. S4: When the cumulative degradation trend reaches the preset warning condition, the flexible adjustment resources are prioritized for voltage regulation to reduce the number of mechanical adjustments of the adjustment components, and the scheduling priority of the flexible adjustment resources is higher than that of the mechanical adjustments of the adjustment components.

[0009] Furthermore, step S1 includes: S11: Acoustic emission signals are acquired by a piezoelectric acoustic sensor deployed in the independent oil chamber of the on-load tap changer, the acoustic emission signals including ultrasonic frequency bands reflecting partial discharge or mechanical friction; S12: The dielectric constant change data is measured in real time by a microfluidic dielectric constant sensor based on the microstrip line resonance principle. The dielectric constant change data is used to reflect the change in the concentration of polar molecules in the insulating oil. S13: The distribution density data of solid micro-nano particles is monitored online by a micro-nano particle counting sensor based on the principle of laser scattering. The distribution density data includes particle concentration and particle size distribution.

[0010] Furthermore, step S2 includes: S21: Perform a fast Fourier transform on the acoustic emission signal to extract the energy evolution characteristics of the ultrasonic frequency band; S22: Perform time series analysis on the dielectric constant change data and extract the rate of change of polar molecule concentration as the polarity change trend; S23: Statistically analyze the distribution density data of the solid micro-nano particles, and extract the particle concentration growth rate and the average particle size change trend; S24: The energy evolution characteristics, the polarity change trend, the particle concentration growth rate, and the average particle size change trend are weighted and fused to obtain the comprehensive deterioration index; S25: The cumulative increment of the comprehensive degradation index within the sliding time window is taken as the cumulative degradation trend.

[0011] Furthermore, in step S24, the weighted fusion calculation formula is: Id=wa·Ta+wc·Tc+wp·Tp+wd·Td, where Ta is the normalized value of the energy evolution characteristic, Tc is the normalized value of the polarity change trend, Tp is the normalized value of the particle concentration growth rate, Td is the normalized value of the average particle size change trend, and wa, wc, wp, and wd are preset weighting coefficients; the preset weighting coefficients are preset based on the sensitivity of each characteristic parameter to changes during the degradation process of the insulating medium and its correlation with the decrease in dielectric strength.

[0012] Furthermore, step S3 includes: S31: Real-time monitoring of voltage deviation at the transformer output terminal; S32: The cumulative degradation trend and the voltage deviation are used together as input variables for action triggering judgment; S33: When the cumulative degradation trend does not reach the preset warning condition, the first dead zone value is used as the dead zone threshold of the voltage regulation. When the voltage deviation exceeds the first dead zone value, the mechanical adjustment of the regulation component is triggered. S34: When the cumulative degradation trend reaches the preset warning condition, the dead zone threshold of the voltage regulation is expanded from the first dead zone value to the second dead zone value, and the mechanical adjustment of the regulation component is triggered only when the voltage deviation exceeds the expanded second dead zone value.

[0013] Furthermore, step S4 includes: S41: Obtain the state of charge of the user-side energy storage system and the adjustment margin of the controllable load; S42: Determine the risk level of the regulating component based on the cumulative degradation trend; S43: When the risk level reaches the preset level, determine whether the combined regulation capability of the state of charge and the regulation margin meets the current voltage regulation requirements; S44: If satisfied, only the user-side energy storage system and controllable load are scheduled for voltage regulation, without triggering the mechanical regulation of the regulation component; S45: If not satisfied, calculate the adjustment capacity gap and trigger the adjustment component to perform mechanical adjustment within the gap range; The user-side energy storage system and controllable load constitute flexible regulation resources, and the scheduling priority of the flexible regulation resources is higher than that of the mechanical regulation of the regulation components.

[0014] Furthermore, step S44 includes: S441: Real-time monitoring of the deviation between the actual voltage value and the target value at the transformer output terminal; S442: When the deviation exceeds the allowable range and the risk level reaches the preset level, determine whether the state of charge of the user-side energy storage system is in the charge-discharge range. S443: If the voltage is in the chargeable / dischargeable range, when the actual voltage value is higher than the target value, a charging command is sent to the energy storage system to absorb excess power and reduce the voltage; when the actual voltage value is lower than the target value, a discharging command is sent to the energy storage system to supplement the power deficit and increase the voltage. S444: When the regulation capability of the energy storage system is insufficient, if the actual voltage value is higher than the target value, a power reduction command is issued to the controllable load to reduce the electrical load, thereby reducing the voltage; S445: Adjust the charging and discharging power of the energy storage system and the adjustment amount of the controllable load in real time according to the change of the voltage deviation, until the voltage is restored to the allowable range.

[0015] Furthermore, step S45 includes: S451: Obtain the deviation amount monitored in real time; S452: Calculate the voltage adjustment amount corresponding to the combined regulation capability of the state of charge and the regulation margin; S453: If the absolute value of the voltage deviation is greater than the absolute value of the voltage adjustment, then the gap is calculated as the difference between the absolute value of the voltage deviation and the absolute value of the voltage adjustment. S454: Read the pre-stored single-level voltage adjustment step size of the adjustment component, and calculate the required number of mechanical adjustment actions based on the absolute value of the ratio of the difference to the single-level voltage adjustment step size; the single-level voltage adjustment step size is a fixed parameter set when the transformer leaves the factory. S455: Control the regulating component to perform mechanical adjustment not exceeding the number of calculated actions, and maintain the flexible regulating resources to output at maximum capacity, together restoring the voltage to the allowable range.

[0016] Furthermore, step S455 includes: S4551: Initialize the action counter to 0, and set the number of calculated actions to the maximum allowed number of actions N; S4552: While maintaining the flexible adjustment resources to output at maximum capacity, perform a mechanical adjustment, the mechanical adjustment including: driving a switching switch to move the transformer tap one position in the direction of reducing voltage deviation, thereby changing the voltage ratio; S4553: After each mechanical adjustment is completed, the count value of the action counter is increased by 1, and the voltage deviation at the transformer output terminal is re-monitored; S4554: If the count value is less than the maximum allowable number of operations N and the current voltage deviation is still not within the allowable range, then repeat the next mechanical adjustment; S4555: If the count value is equal to the maximum allowable number of actions N or the current voltage deviation has entered the allowable range, then mechanical adjustment is stopped, and subsequent voltage deviations are absorbed by the continuous output of flexible adjustment resources.

[0017] Secondly, a transformer operation intelligent control system, the system being used to implement the steps of any of the methods described above, the system comprising: Acquisition module: Acquires multimodal physical characteristic data of the insulating medium inside the transformer regulating component, wherein the multimodal physical characteristic data includes at least acoustic emission signals, dielectric constant variation data, and distribution density data of solid micro-nano particles; Fusion module: Extracts the energy evolution characteristics of the acoustic emission signal, the polarity change trend of the dielectric constant, and the concentration growth rate of the solid micro / nano particles, fuses the three to determine the comprehensive degradation index, and uses the change trend of the comprehensive degradation index over time as the cumulative degradation trend; Correction module: The cumulative degradation trend is input as feedback information into the power flow control program of the transformer to correct the mechanical adjustment logic triggered only by voltage deviation; Regulation module: When the cumulative degradation trend reaches the preset warning condition, the flexible regulation resource is prioritized for voltage regulation to reduce the number of mechanical regulation operations of the regulation component, and the scheduling priority of the flexible regulation resource is higher than that of the mechanical regulation of the regulation component.

[0018] Beneficial Effects: This application proposes an intelligent control method and system for transformer operation. It acquires real-time acoustic, dielectric constant, and particle distribution data of the insulating medium in transformer regulating components using multimodal sensors. This data is then fused and processed to extract energy evolution, polarity changes, and concentration growth rates, calculating a comprehensive degradation index and cumulative degradation trend. This trend is used as feedback to dynamically correct mechanical regulation logic based solely on voltage deviation, such as by expanding the dead zone threshold. When cumulative degradation reaches early warning conditions, priority is given to scheduling flexible resources such as user-side energy storage and controllable loads for voltage regulation, reducing the number of mechanical adjustments. Flexible resources have higher priority than mechanical adjustments; even if flexibility is insufficient, only the minimum number of mechanical adjustments is triggered within the gap. This scheme enables transformer operation control to intelligently perceive the deep health status of equipment and incorporate it into decision-making, effectively reducing mechanical wear, extending equipment life, improving grid reliability and stability, and avoiding sudden failures caused by hidden degradation. Attached Figure Description

[0019] Figure 1This is a flowchart of a transformer operation intelligent control method proposed in this application.

[0020] Figure 2 This is a structural diagram of an intelligent control system for transformer operation proposed in this application.

[0021] Figure 3 This is a schematic diagram of an intelligent control system for transformer operation proposed in this application.

[0022] Labeling explanation: 201, Acquisition module; 202, Fusion module; 203, Correction module; 204, Adjustment module. Detailed Implementation

[0023] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. The components of the embodiments of this application described and marked in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely represents selected embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0024] It should be noted that similar reference numerals and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. Furthermore, in the description of this application, terms such as "first," "second," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0025] Please refer to Figure 1 This application proposes an intelligent control method for transformer operation, the method comprising the following steps: S1: Obtain multimodal physical characteristic data of the insulating medium inside the transformer regulating component. The multimodal physical characteristic data includes at least acoustic emission signals, dielectric constant variation data, and distribution density data of solid micro-nano particles. S2: Extract the energy evolution characteristics of the acoustic emission signal, the polarity change trend of the dielectric constant, and the concentration growth rate of solid micro-nano particles. Combine these three factors to determine the comprehensive degradation index, and use the change trend of the comprehensive degradation index over time as the cumulative degradation trend. S3: Input the cumulative degradation trend as feedback information into the power flow control program of the transformer to correct the mechanical regulation logic triggered only by voltage deviation. S4: When the cumulative deterioration trend reaches the preset warning condition, the flexible regulation resources are prioritized for voltage regulation to reduce the number of mechanical regulation operations of the regulating components, and the scheduling priority of the flexible regulation resources is higher than that of the mechanical regulation of the regulating components.

[0026] In this embodiment, the transformer regulating component is specifically an on-load tap changer (OLTC), whose internal insulating medium is insulating oil. The working principle of this method is as follows: By deploying a multi-modal sensor array in the independent oil chamber of the OLTC, the acoustic emission signal, dielectric constant change data, and distribution density data of solid micro / nano particles of the insulating medium are collected in real time. The multi-modal data are fused to extract the energy evolution characteristics, polarity change trends, and concentration growth rates reflecting the deterioration state of the insulating medium. Based on this, a comprehensive deterioration index and its cumulative deterioration trend over time are calculated. This cumulative deterioration trend is used as a feedback variable input to the transformer's power flow control program for dynamic correction of the control logic. Specifically, when the cumulative deterioration trend does not reach the warning condition, the conventional voltage deviation trigger dead zone is maintained; when the warning condition is reached, the dead zone threshold is expanded or flexible regulating resources (such as user-side energy storage systems and controllable loads) are prioritized, thereby reducing the frequency of mechanical regulation of the OLTC. Even when mechanical regulation is necessary, it is only performed the minimum necessary number of times within the gap of insufficient flexible resource capacity. In this way, the method incorporates the deep health status of the regulating components into the power flow control decision, realizing the transformation from passively responding to voltage deviations to actively managing equipment degradation.

[0027] Specifically, acquiring multimodal physical characteristic data aims to comprehensively capture early signs of insulating oil degradation from different physical dimensions. Multimodal data is used because a single physical characteristic often cannot fully reflect the complex degradation process. For example, acoustic emission signals can sensitively capture weak sound waves generated by partial discharge or mechanical friction, which is direct evidence that the insulating oil has begun to decompose or that minor mechanical anomalies have appeared inside. Changes in the dielectric constant are directly related to the increase in polar molecules in the insulating oil, such as moisture and acidic substances, which are chemical products of insulating oil aging. The distribution density of solid micro / nano particles, especially conductive carbon particles, is a direct consequence of arc erosion, directly affecting the dielectric strength of the insulating oil. By simultaneously monitoring these three different physical characteristics, a three-dimensional, mutually corroborating portrait of the insulating oil's health status can be constructed, greatly improving the accuracy and reliability of condition perception.

[0028] The collected multimodal data is processed and fused to form a quantitative indicator that reflects long-term trends. Raw sensor data is discrete and noisy, making direct use difficult for decision-making. Therefore, it is necessary to extract its core features. For example, for acoustic emission signals, the focus is on how their energy evolves over time in a specific frequency band; a continuous increase in energy may indicate intensified partial discharge activity. For dielectric constants, the focus is on the rate of change, i.e., the trend of polarity change; a continuously decreasing trend indicates accelerated aging. For solid micro / nano particles, the focus is on the rate of concentration increase, reflecting the rate of pollutant formation. Fusing these extracted dynamic features, for example through weighted summation, yields a comprehensive degradation index. This index compresses information from multiple dimensions into a single value, intuitively reflecting the current degree of degradation. However, a single degradation index value has limited significance; more important is its trend over time. Therefore, by observing the cumulative change of this comprehensive degradation index over a period of time, a cumulative degradation trend can be formed. This trend effectively filters out interference from short-term fluctuations and measurement errors, accurately revealing the slow but continuous performance degradation process of insulating oil.

[0029] Traditional control logic is very simple, with only one variable—voltage deviation—as its decision input. When the voltage deviation exceeds a fixed dead-zone threshold, mechanical regulation is triggered. This logic doesn't consider the health of the regulating components themselves, potentially leading to frequent operations on components already in a sub-optimal state, thus accelerating their eventual failure. This method introduces a cumulative degradation trend as a second decision input variable, giving the power flow control program a self-protective mechanism. The control logic is no longer rigid but dynamically adjusts its behavior based on the health status of the regulating components.

[0030] When the degradation of insulating oil reaches a level requiring attention, it means that mechanical impact on on-load tap changers should be minimized as much as possible. At this point, flexible regulation resources in the power grid, such as energy storage and controllable loads on the user side, become ideal alternative regulation tools. These resources can influence local voltage through rapid power throughput or reduction, with fast response times and no mechanical wear. By prioritizing the scheduling of flexible regulation resources over mechanical regulation, voltage can be stabilized using flexible resources in most voltage fluctuation scenarios, allowing on-load tap changers to rest and significantly reducing the frequency of mechanical regulation, effectively slowing down their degradation process.

[0031] Through the steps described above, this method constructs a closed-loop intelligent control process from micro-state perception to macro-control decision-making. It can not only detect and warn of hidden degradations that are undetectable by traditional methods, but also proactively manage equipment lifespan through intelligent control strategy adjustments. This significantly improves the reliability and security of transformers and even the entire distribution network when facing the challenges of complex grid environments brought about by distributed energy resources.

[0032] Further, step S1 includes: S11: Acoustic emission signals are acquired by a piezoelectric acoustic sensor deployed in the independent oil chamber of the on-load tap changer. The acoustic emission signals include ultrasonic frequency bands that reflect partial discharge or mechanical friction. S12: The dielectric constant change data is measured in real time by a microfluidic dielectric constant sensor based on the microstrip line resonance principle. The dielectric constant change data is used to reflect the change in the concentration of polar molecules in the insulating oil. S13: Online monitoring of the distribution density data of solid micro-nano particles using a micro-nano particle counting sensor based on the laser scattering principle. The distribution density data includes particle concentration and particle size distribution.

[0033] To achieve accurate acquisition of acoustic emission signals, a piezoelectric acoustic sensor made of lead zirconate titanate ceramic material can be selected. This sensor features high sensitivity and wide bandwidth response, enabling it to capture even very weak sound wave signals. During installation, the sensor chip, approximately a few millimeters square, can be encapsulated in epoxy resin and non-invasively attached to the outer wall of the independent oil chamber of the on-load tap changer, allowing signal pickup through solid-state conduction of sound waves.

[0034] Alternatively, to achieve a higher signal-to-noise ratio, it can be designed as a sealed probe, directly immersed in the insulating oil through a pre-drilled temperature measurement port or sampling valve. The sensor continuously monitors signals primarily in the ultrasonic frequency band, typically between 20kHz and 200kHz. This is because the acoustic characteristics of arcs or coronas generated by partial discharges in the insulating oil are mainly concentrated in this frequency band, while noise from normal mechanical switching operations is primarily concentrated in the low-frequency band. By selecting the appropriate frequency band, abnormal signals can be initially distinguished from normal operating noise.

[0035] To achieve real-time measurement of dielectric constant changes, a microfluidic dielectric constant sensor based on the microstrip line resonance principle can be used. The core of this sensor is a microstrip line resonator fabricated on a substrate using microfabrication techniques. When insulating oil flows through the microfluidic channel in close contact with the resonator, the oil, acting as a medium, alters the resonator's equivalent capacitance, causing a shift in its resonant frequency. By accurately measuring the change in resonant frequency, the real-time value of the insulating oil's dielectric constant can be deduced. During the aging process of insulating oil, polar molecules such as water, alcohol, and acids are produced. The dielectric constants of these molecules are much higher than that of the insulating oil itself; therefore, their presence significantly affects the overall dielectric constant of the mixed medium. By continuously monitoring changes in the dielectric constant, the accumulation process of polar contaminants in the insulating oil can be effectively tracked.

[0036] To achieve online monitoring of solid micro / nano particles, a micro / nano particle counting sensor based on the principle of laser scattering can be used. Its working principle is to pass a collimated laser beam through a flowing insulating oil sample. When micro / nano particles in the oil, especially carbon particles generated by electric arcs, pass through the laser beam's irradiation area, the laser light is scattered. A highly sensitive photodetector is placed at a certain angle to the laser beam to capture the scattered light signal. The intensity and pulse frequency of the scattered light are directly related to the particle size and concentration. By analyzing the electrical signal output by the photodetector, the number of particles of different size ranges in a unit volume of insulating oil can be counted in real time, thus obtaining the particle concentration and particle size distribution. This online monitoring method avoids the lag and operational complexity of traditional offline sampling and testing, and can reflect the contamination status of the insulating oil in real time.

[0037] Furthermore, step S2 includes: S21: Perform a fast Fourier transform on the acoustic emission signal to extract the energy evolution characteristics of the ultrasonic frequency band; S22: Perform time series analysis on dielectric constant variation data to extract the rate of change in polar molecule concentration as the polarity trend; S23: Statistically analyze the distribution density data of solid micro and nano particles, and extract the particle concentration growth rate and the trend of average particle size change. S24: The comprehensive deterioration index is obtained by weighted and fused calculation of energy evolution characteristics, polarity change trend, particle concentration growth rate and average particle size change trend. S25: The cumulative increment of the overall degradation index within the sliding time window is taken as the cumulative degradation trend.

[0038] First, the time-domain waveform signal acquired by the piezoelectric acoustic sensor can be converted to the frequency domain by performing a fast Fourier transform, thus obtaining the energy distribution of the signal at different frequencies, i.e., the spectrum.

[0039] The analysis focuses on a pre-defined ultrasonic frequency band, such as 100kHz to 150kHz. By calculating the signal energy integral within this frequency band, an instantaneous energy value characterizing the intensity of partial discharge activity at that moment can be obtained. Connecting the continuously acquired instantaneous energy values ​​forms a curve showing how energy changes over time.

[0040] Trend analysis of this curve, such as calculating its average growth rate over the past week or month, is known as the energy evolution characteristic. A consistently positive energy evolution characteristic clearly indicates a worsening trend in partial discharge activity. The energy evolution characteristic is extracted by performing a short-time Fourier transform on the acquired raw acoustic signal. In the frequency domain, a characteristic frequency band from 20kHz to 200kHz is selected, and the sum of squares of the amplitudes of all frequency components within this band is calculated. To eliminate interference from background mechanical noise, this sum of squares is averaged in the time domain to obtain an energy statistic reflecting the partial discharge intensity per unit time. By analyzing the change in the derivative of this energy statistic over continuous operating cycles, the rate of energy enhancement can be identified. This rate, the energy evolution characteristic, directly characterizes the severity of the internal discharge phenomenon within the insulating medium, providing a microscopic physical basis for subsequent fusion assessment.

[0041] Secondly, for the time-series data of the dielectric constant output by the microfluidic dielectric constant sensor, time-series analysis methods, such as the moving average method, are used to extract its long-term trend. This step focuses not on the absolute value of the dielectric constant, but on its rate of change. This rate directly reflects the rate at which the concentration of polar molecules in the insulating oil increases, and is therefore defined as the polarity trend. A continuously negative polarity trend with an ever-increasing absolute value indicates that the chemical aging process of the insulating oil is accelerating.

[0042] Furthermore, statistical analysis was performed on the particle concentration and size distribution data obtained from the micro / nano particle counting sensor. On one hand, the increase in particle concentration per unit time was calculated to obtain the particle concentration growth rate. On the other hand, the trend of the average particle size over time was calculated. The rapid increase in particle concentration and the continuous increase in average particle size both indicate intensified arc erosion and a worsening of the insulating oil contamination.

[0043] Furthermore, in step S24, the weighted fusion calculation formula is: Id=wa·Ta+wc·Tc+wp·Tp+wd·Td, where Ta is the normalized value of energy evolution characteristics, Tc is the normalized value of polarity change trend, Tp is the normalized value of particle concentration growth rate, Td is the normalized value of average particle size change trend, and wa, wc, wp, and wd are preset weighting coefficients; the preset weighting coefficients are preset based on the sensitivity of each characteristic parameter to changes during the degradation process of the insulating medium and its correlation with the decrease in dielectric strength.

[0044] In a preferred embodiment, the weighted fusion calculation formula can be specified as: Id = wa·Ta + wc·Tc + wp·Tp + wd·Td. These weighting coefficients are not set arbitrarily, but are based on extensive experimental data and physical mechanism analysis. Specifically, it is necessary to evaluate the sensitivity of each characteristic parameter to changes during the insulation medium degradation process, as well as its correlation with the key indicator for ultimately measuring insulation performance, namely, the decrease in dielectric strength. For example, if research finds that the increase rate of particle concentration exhibits the strongest linear correlation with the decrease in dielectric strength, then the corresponding weighting coefficient wp should be assigned a larger value. In this way, it is ensured that the calculated comprehensive degradation index can most accurately reflect the true health condition of the insulating oil. The preset weighting coefficients are pre-set based on the sensitivity of each characteristic parameter to changes during the insulation medium degradation process and its correlation with the decrease in dielectric strength. The process of determining the weighting coefficients involves establishing an insulation aging experimental model.

[0045] As a specific implementation method, multiple accelerated aging test benches were constructed in an insulating oil performance evaluation laboratory. Each test bench was equipped with a heating device, a high-voltage discharge electrode, and the piezoelectric acoustic sensor, microfluidic dielectric constant sensor, and micro / nano particle counting sensor described in this application. During the test, one group of insulating oil samples underwent thermal aging at 120 degrees Celsius, while another group of samples underwent discharge aging at a partial discharge intensity of 20 pC (pC represents a unit of picocoulomb).

[0046] Every 24 hours, the output data of each sensor is recorded synchronously, and dielectric breakdown voltage tests are performed on the aged insulating oil samples. The data analysis unit receives this data and calculates the rate of change of acoustic signal energy, dielectric constant, and particle concentration in each aging cycle. A correlation coefficient matrix is ​​obtained by performing Pearson correlation analysis on these rates of change and the percentage decrease in breakdown voltage during the same period. For example, the correlation coefficient between the particle concentration growth rate and the breakdown voltage decrease rate is found to be -0.95, the correlation coefficient between the dielectric constant decrease rate and the acoustic energy growth rate is -0.88, and the correlation coefficient between the acoustic energy growth rate and the breakdown voltage decrease rate is -0.75. Subsequently, an analytic hierarchy process (AHP) is used to construct a judgment matrix based on pairwise comparisons and scores of the importance of these three characteristics by three insulation material experts. The initial weights of each characteristic are obtained by calculating the largest eigenvalue and the corresponding eigenvector of the judgment matrix. The initial weights are then adjusted based on the absolute values ​​of the correlation coefficients, giving greater weight to characteristics highly correlated with the breakdown voltage decrease. Ultimately, the weighting coefficient for particle concentration growth rate was set to 0.45, the weighting coefficient for dielectric constant variation trend was set to 0.35, and the weighting coefficient for energy evolution characteristics was set to 0.20. These weighting coefficients were embedded into the algorithm of the fusion module for real-time calculation of the comprehensive degradation index.

[0047] Finally, to obtain a cumulative degradation trend that reflects the long-term cumulative effect, the calculated composite degradation index needs further processing. A sliding time window method can be used, for example, setting a 30-day time window. The difference between the current composite degradation index and the composite degradation index 30 days ago is calculated; this difference, i.e., the cumulative increment, is defined as the current cumulative degradation trend.

[0048] Furthermore, step S3 includes: S31: Real-time monitoring of voltage deviation at the transformer output terminal; S32: The cumulative degradation trend and voltage deviation are used together as input variables for action triggering judgment; S33: When the cumulative degradation trend does not reach the preset warning condition, the first dead zone value is used as the dead zone threshold of the voltage regulation. When the voltage deviation exceeds the first dead zone value, the mechanical adjustment of the regulation component is triggered. S34: When the cumulative degradation trend reaches the preset warning condition, the dead zone threshold of the voltage regulation is expanded from the first dead zone value to the second dead zone value, and the mechanical adjustment of the regulating component is triggered only when the voltage deviation exceeds the expanded second dead zone value.

[0049] This process involves continuously acquiring the actual voltage at the transformer output terminal using measuring devices such as voltage transformers, comparing it with the target voltage value, and obtaining the real-time voltage deviation.

[0050] Furthermore, the input to the decision-making process has been expanded from a single voltage deviation to two variables: voltage deviation and cumulative degradation trend. This means that the decision-making logic of the control program has changed from one-dimensional to two-dimensional, enabling more comprehensive and intelligent judgments.

[0051] Based on these two input variables, the control logic employs different adjustment strategies depending on the range of the cumulative degradation trend. One or more warning thresholds can be preset. When the calculated cumulative degradation trend is below the lowest warning threshold, it indicates that the on-load tap changer is in good health. In this case, the control program uses a conventional adjustment strategy, namely, using a small, preset first dead zone value. For example, the first dead zone value can be set to ±0.5% of the rated voltage. As soon as the monitored voltage deviation exceeds this range, the control program will immediately trigger the on-load tap changer to perform mechanical adjustment to ensure excellent power quality.

[0052] However, when the cumulative degradation trend continues to grow and exceeds the preset warning conditions, such as reaching the first-level warning threshold, it indicates that the health of the regulating component has shown early signs of degradation and requires protection. At this time, the control program will automatically expand the dead zone threshold of the voltage regulation from the first dead zone value to a larger second dead zone value. For example, the second dead zone value can be set to ±1% of the rated voltage. This means that the control program's tolerance for voltage fluctuations is increased. Mechanical regulation will only be triggered when the voltage deviation exceeds this wider range of ±1%.

[0053] For smaller voltage deviations between ±0.5% and ±1%, the control program will temporarily ignore them and refrain from triggering mechanical regulation. This dynamically expanded dead zone effectively filters out most unnecessary voltage regulation actions, significantly reducing the operating frequency of on-load tap changers and thus slowing their degradation process. This is a protective control strategy that intelligently balances power quality and equipment lifespan. When the cumulative degradation trend reaches a preset warning condition, the voltage regulation dead zone threshold is expanded from the first dead zone value to the second dead zone value.

[0054] Furthermore, step S4 includes: S41: Obtain the state of charge of the user-side energy storage system and the adjustment margin of the controllable load; S42: Determine the risk level of the regulating component based on the cumulative deterioration trend; S43: When the risk level reaches the preset level, determine whether the combined regulation capability of the state of charge and the regulation margin meets the current voltage regulation requirements. S44: If satisfied, only the user-side energy storage system and controllable loads will be dispatched for voltage regulation, without triggering the mechanical regulation of the regulating components; S45: If not satisfied, calculate the adjustment capacity gap and trigger the adjustment component to perform mechanical adjustment within the gap range; Among them, user-side energy storage systems and controllable loads constitute flexible regulation resources, and the scheduling priority of flexible regulation resources is higher than that of mechanical regulation components.

[0055] The control program needs to obtain key status information in real time from the user-side energy storage management unit and load control unit via a communication network. For energy storage, the most critical information is its current state of charge (SOC), which determines how much electricity can be added or released. For controllable loads, such as smart air conditioners and electric vehicle charging station clusters, the key information is its adjustable power margin, that is, how much power consumption can be reduced or increased without affecting the user's basic needs.

[0056] Simultaneously, the control program maps the cumulative degradation trend value to a predefined risk level system. For example, the cumulative degradation trend value can be divided into multiple levels such as normal, attention, alarm, and severe. This risk level provides a clear basis for subsequent adjustment strategy selection.

[0057] When a voltage deviation is detected and the risk level of the regulating component reaches a preset trigger level, such as above the concern level, the control program initiates the flexible resource priority scheduling logic. First, based on the acquired energy storage state of charge and controllable load regulation margin, it calculates the maximum voltage regulation capacity that all flexible resources can provide at the current moment.

[0058] Next, it is determined whether this combined regulation capability is sufficient to completely eliminate the current voltage deviation. If the determination is yes, it means that the problem can be solved using only flexible resources. At this time, the control program will immediately issue regulation commands to the relevant energy storage and controllable loads to regulate the voltage, without triggering the on-load tap changer. This is the ideal situation, achieving complete protection for the mechanical regulating components.

[0059] However, in certain situations, such as when the voltage deviation is too large, or when the flexible resources are in a state of insufficient regulation capacity (e.g., energy storage is near full charge or depleted), the combined regulation capacity of the flexible resources may not be able to fully meet the voltage regulation requirements. In this case, the control program calculates the regulation capacity gap, which is the total voltage deviation minus the portion that the flexible resources can regulate. Then, focusing only on this uncompensated gap, it calculates the minimum regulation amount that the on-load tap changer needs to perform and triggers it to execute the corresponding mechanical regulation. During this process, the flexible regulation resources continuously output at their maximum capacity, working in conjunction with the mechanical regulation to restore the voltage to the normal range. This strategy ensures that mechanical regulation is only activated as a last resort, and its action is limited to the minimum necessary range, thereby maximizing the protection of the regulating components.

[0060] When the cumulative degradation trend reaches the preset warning condition, flexible regulation resources are prioritized for voltage regulation to reduce the number of mechanical adjustments of the regulating components, and the scheduling priority of flexible regulation resources is higher than that of mechanical adjustments of the regulating components.

[0061] Furthermore, step S44 includes: S441: Real-time monitoring of the deviation between the actual voltage value and the target value at the transformer output terminal; S442: When the deviation exceeds the allowable range and the risk level reaches the preset level, determine whether the state of charge of the user-side energy storage system is in the charge-discharge range. S443: If the voltage is in the chargeable / dischargeable range, when the actual voltage value is higher than the target value, a charging command is sent to the energy storage system to absorb excess power and reduce the voltage. When the actual voltage value is lower than the target value, a discharging command is sent to the energy storage system to make up for the power deficit and increase the voltage. S444: When the energy storage system's regulation capability is insufficient, if the actual voltage value is higher than the target value, a power reduction command is issued to the controllable load to reduce the electrical load, thereby lowering the voltage; S445: Adjust the charging and discharging power of the energy storage system and the regulation of the controllable load in real time according to the change in voltage deviation until the voltage recovers to the allowable range.

[0062] This process is a closed-loop, hierarchical adjustment process. First, continuous voltage deviation monitoring is the foundation of closed-loop control, providing real-time targets and feedback for adjustment.

[0063] When the triggering conditions—voltage deviation and high risk level of the regulating component—are met, the control program will first evaluate the highest priority flexible resource, namely energy storage. It will check whether the state of charge of the energy storage is within a reasonable charge / discharge range, such as between 20% and 90%, to avoid damage to the battery from overcharging or over-discharging.

[0064] If energy storage is available, the control program will issue corresponding commands based on the direction of the voltage deviation. If the actual voltage is higher than the target value, it indicates that there is excess reactive or active power on the line. In this case, a charging command is issued to the energy storage to absorb this excess power, acting as a controllable load, thereby effectively reducing the line voltage. Conversely, if the actual voltage is lower than the target value, it indicates that there is a power deficit on the line. In this case, a discharging command is issued to the energy storage to inject power into the grid, acting as a controllable power source, thereby raising the line voltage.

[0065] In some situations, the energy storage's regulation capacity may be exhausted, for example, reaching maximum charging power, but the voltage may still be too high. In this case, the control program will activate the second level of flexible resources, namely controllable loads. Power reduction commands will be issued to pre-contracted controllable loads, requiring them to reduce their operating power within a short period. The reduction in electrical load will directly lead to a decrease in power demand on the line, thereby helping to lower the voltage.

[0066] As a specific implementation, assuming a technology park scenario, when fluctuations in the park's photovoltaic output cause a voltage deviation at the transformer's output terminal, an edge computing unit deployed within the substation continuously receives real-time voltage data from the voltage transformer. The voltage deviation monitoring module operating within this unit compares the real-time voltage value with a preset target voltage value to calculate the current voltage deviation. For example, when the actual voltage value is higher than the target value, the deviation is positive; when the actual voltage value is lower than the target value, the deviation is negative.

[0067] The PID controller is integrated into the control program of the edge computing unit. This controller receives real-time voltage deviation as input and calculates the required charging / discharging power command for the energy storage system and the required adjustment command for the controllable load based on its internal proportional, integral, and derivative gain parameters. For example, when the voltage remains consistently high, the proportional term immediately issues a charging command based on the magnitude of the deviation, the integral term accumulates the deviation to eliminate steady-state error, and the derivative term responds to the rate of change of the deviation to suppress oscillations.

[0068] These commands are sent via communication interfaces to the battery management system and intelligent load control unit of the energy storage power station in the park. The energy storage power station adjusts its charging and discharging power according to the commands, and the controllable load adjusts its power consumption accordingly. The control program repeats the above monitoring, calculation, and command transmission process at millisecond intervals, forming a fast-response closed-loop control loop. Through this iterative adjustment, the voltage deviation gradually decreases until the transformer output voltage recovers to the allowable range of ±0.5% of the rated voltage. The entire process proceeds smoothly, and no drastic voltage fluctuations or oscillations are observed.

[0069] When the capacity of flexible regulation resources is insufficient to fully cope with voltage deviations, mechanical regulation is required as a supplement. At this point, accurately calculating the regulation gap and executing mechanical regulation at minimal cost becomes crucial. Therefore, further, step S45 includes: S451: Obtain the deviation value monitored in real time; S452: The voltage adjustment amount corresponding to the combined regulation capability of the state of charge and regulation margin; S453: If the absolute value of the voltage deviation is greater than the absolute value of the voltage adjustment, then the gap is calculated as the difference between the absolute value of the voltage deviation and the absolute value of the voltage adjustment. S454: Reads the pre-stored single-gear voltage adjustment step size of the regulating component, and calculates the required number of mechanical adjustment actions based on the absolute value of the ratio of the difference to the single-gear voltage adjustment step size; the single-gear voltage adjustment step size is a fixed parameter set at the factory when the transformer leaves the factory. S455: The control and regulation components perform mechanical regulation not exceeding the calculated number of actions, and maintain the flexible regulation resources at maximum capacity output, together restoring the voltage to the allowable range.

[0070] This process first requires quantifying the power regulation capability of flexible resources into their voltage regulation capability. This can be achieved by analyzing the power flow Jacobian matrix of the power grid. First, establish the node admittance matrix of the distribution network. Then, calculate the power flow Jacobian matrix under the current power grid operating state using the Newton-Raphson method or a fast decoupled power flow algorithm.

[0071] The elements ∂V / ∂P and ∂V / ∂Q of this matrix represent the sensitivity of the node voltage to active and reactive power, respectively. For an energy storage system, its currently available charge and discharge power range is determined by its state of charge and battery management system. For example, if the energy storage system can currently provide a maximum discharge power of 500kW, and the node voltage sensitivity to active power is 0.2V per 100kW, then the voltage boost that the energy storage system can achieve is 500kW ÷ 100kW × 0.2V, or 1V.

[0072] For controllable loads, such as an air conditioning load cluster in an industrial park, the adjustable margin is 300kW. If the voltage sensitivity of this node to active power is the same, then the voltage reduction that the controllable load can achieve is 300kW ÷ 100kW × 0.2V, which is 0.6V. By algebraically summing the voltage adjustment amounts of all flexible resources, the total voltage adjustment amount corresponding to the joint regulation capability can be obtained.

[0073] In this way, the maximum available charging and discharging power of energy storage and the maximum regulation margin of controllable load can be uniformly converted into a total voltage regulation amount in volts. This method transforms the flexible resources in the power dimension into the regulation variables in the voltage dimension, achieving a same-dimensional comparison with the voltage regulation step size of mechanical tap changers.

[0074] Then, this calculated maximum voltage adjustment is compared with the real-time monitored voltage deviation. If the absolute value of the voltage deviation is still greater than the absolute value of the maximum voltage adjustment that flexible resources can provide, it indicates that there is a regulation gap that cannot be filled by flexible resources. The size of this gap is the difference between the absolute values ​​of the two.

[0075] Next, the control program needs to read the inherent parameters of the on-load tap changer pre-stored in the database, the most critical of which is the single-stage tap change step size. This parameter is a fixed value set at the transformer's factory, indicating how much voltage can be changed with each tap change. For example, a typical single-stage tap change step size is 1.25% of the rated voltage. By dividing the voltage regulation gap calculated in the previous step by this single-stage tap change step size and rounding up, the minimum number of mechanical adjustment actions required to fill this gap can be obtained.

[0076] Finally, the control program sends a command to the on-load tap changer, instructing it to perform mechanical regulation no more than the calculated number of actions. Simultaneously, it commands all flexible regulating resources to output at their maximum capacity; the two work together to pull the voltage back within the permissible range.

[0077] Furthermore, to ensure that the number of mechanical adjustment actions is strictly controlled and the process is safe and reliable, the specific execution process can be designed as an iterative loop. Step S455 includes: S4551: Initialize the action counter to 0 and set the number of actions to be calculated to the maximum allowed number of actions N; S4552: While maintaining the flexible regulation resources to output at maximum capacity, perform a mechanical regulation, which includes: driving the switching switch to move the transformer tap one position in the direction of reducing voltage deviation, thereby changing the voltage ratio; S4553: After each mechanical adjustment is completed, the count value of the action counter is increased by 1, and the voltage deviation at the transformer output terminal is re-monitored; S4554: If the count value is less than the maximum allowable number of operations N and the current voltage deviation is still not within the allowable range, then repeat the next mechanical adjustment. S4555: If the count value equals the maximum allowable number of operations N or the current voltage deviation has entered the allowable range, then mechanical regulation will stop, and subsequent voltage deviations will be absorbed by the continuous output of flexible regulation resources.

[0078] Before coordinating adjustments begins, the program initializes an action counter with an initial value of zero. Simultaneously, the required number of mechanical adjustment actions calculated in the previous step is set as the maximum permissible number of actions for this adjustment process, denoted as N.

[0079] At the start of the cycle, while maintaining maximum capacity output of the flexible resources, the control program sends a command to the on-load tap changer to perform a mechanical adjustment. This command drives its internal switching switch to move the transformer's tap position one step in the direction that reduces voltage deviation, thereby changing the transformer's voltage ratio.

[0080] After each mechanical adjustment is completed, the program increments the action counter by 1. Then, it waits for a short period of stabilization to allow the power grid to reach a new equilibrium before rereading the current voltage deviation from the voltage transformer.

[0081] Next, a loop is executed. The program checks two conditions: whether the current count value is less than the maximum allowed number of actions N, and whether the current voltage deviation is still outside the allowable range. Only when both conditions are met simultaneously—that is, when there is still room for action and the voltage problem has not been resolved—will the program repeat the next mechanical adjustment.

[0082] There are two termination conditions for the loop. The first is that the count value equals the maximum allowable number of operations N. This means that even if the voltage may not have fully recovered to the ideal range, mechanical regulation will no longer be performed to protect the regulating components. The second is that the current voltage deviation has entered the allowable range. This means that the regulation target has been achieved. Once either of these two conditions is met, the program will immediately stop controlling the on-load tap changer. At this point, if there is still a small residual voltage deviation, it will be completely absorbed and processed by the continuous output of the flexible regulating resources, without causing any disturbance to the mechanical components. This iterative control process ensures the accuracy, necessity, and minimization of mechanical regulation.

[0083] Please refer to Figure 2 , Figure 3 This application also proposes an intelligent control system for transformer operation, which implements the steps of any of the above methods. The system includes: Acquisition Module 201: Acquires multimodal physical characteristic data of the insulating medium inside the transformer regulating component. The multimodal physical characteristic data includes at least acoustic emission signals, dielectric constant variation data, and distribution density data of solid micro-nano particles. Fusion module 202: Extracts the energy evolution characteristics of the acoustic emission signal, the polarity change trend of the dielectric constant, and the concentration growth rate of solid micro-nano particles, and performs fusion processing on the three to determine the comprehensive degradation index, and takes the change trend of the comprehensive degradation index over time as the cumulative degradation trend. Correction module 203: Inputs the cumulative degradation trend as feedback information into the power flow control program of the transformer to correct the mechanical regulation logic triggered only by voltage deviation; Regulation module 204: When the cumulative deterioration trend reaches the preset warning condition, it prioritizes scheduling flexible regulation resources for voltage regulation to reduce the number of mechanical regulation operations of the regulation components, and the scheduling priority of flexible regulation resources is higher than that of mechanical regulation of the regulation components.

[0084] Specifically, the acquisition module 201 can be configured to communicate with external sensor devices via various sensor interfaces to receive multimodal physical characteristic data of the insulating medium inside the transformer regulating components. For example, the acquisition module 201 may include an analog-to-digital converter (ADC) for converting analog signals into digital signals, and a communication interface (such as RS485, Ethernet, or a wireless communication module) for data interaction with acoustic sensors, dielectric constant sensors, and micro / nano particle counting sensors. These sensors can be deployed independently and transmit data to the acquisition module via standard protocols.

[0085] The fusion module 202 can be implemented as a data processing unit, such as a microcontroller, digital signal processor (DSP), or embedded system, which internally runs data processing algorithms. This module 202 receives raw multimodal physical feature data from the acquisition module and performs feature extraction and fusion calculations. The fusion module 202 may include a storage unit for storing historical data and calculation results.

[0086] The correction module 203 can be implemented as a control logic unit, such as a programmable logic controller (PLC) or an industrial computer, which interfaces with the power flow control program of the transformer. This module receives cumulative degradation trend information from the fusion module and corrects the mechanical adjustment logic in the power flow control program according to preset logic rules. For example, the correction module 203 can dynamically adjust the voltage deviation threshold used to trigger mechanical adjustment in the power flow control program, or suppress the issuance of mechanical adjustment commands under specific conditions.

[0087] The regulating module 204 can be implemented as a scheduling and management unit, such as a central controller or distributed control unit, which communicates with flexible regulating resources (such as user-side energy storage systems and controllable loads) and the mechanical regulating actuators of the transformer regulating components. Based on cumulative degradation trend information from the fusion module and combined with the real-time operating status of the power grid, this module determines the scheduling strategy for flexible regulating resources and the triggering timing of mechanical regulation. For example, the regulating module 204 can send charging and discharging commands to the energy storage system or power adjustment commands to the controllable loads to achieve voltage regulation. When flexible regulating capacity is insufficient, the regulating module can send mechanical regulation commands to the actuators of the transformer regulating components.

[0088] The above description is merely an embodiment of this application and is not intended to limit the scope of protection of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. A method for intelligent control of transformer operation, characterized in that, The method includes the following steps: S1: Obtain multimodal physical characteristic data of the insulating medium inside the transformer regulating component. The multimodal physical characteristic data includes at least acoustic emission signals, dielectric constant variation data, and distribution density data of solid micro / nano particles. S2: Extract the energy evolution characteristics of the acoustic emission signal, the polarity change trend of the dielectric constant, and the concentration growth rate of the solid micro / nano particles, fuse the three to determine the comprehensive degradation index, and take the change trend of the comprehensive degradation index over time as the cumulative degradation trend. S3: The cumulative degradation trend is input as feedback information into the power flow control program of the transformer to correct the mechanical regulation logic triggered only by voltage deviation. S4: When the cumulative degradation trend reaches the preset warning condition, the flexible adjustment resources are prioritized for voltage regulation to reduce the number of mechanical adjustments of the adjustment components, and the scheduling priority of the flexible adjustment resources is higher than that of the mechanical adjustments of the adjustment components.

2. The intelligent control method for transformer operation according to claim 1, characterized in that, Step S1 includes: S11: Acoustic emission signals are acquired by a piezoelectric acoustic sensor deployed in the independent oil chamber of the on-load tap changer, the acoustic emission signals including ultrasonic frequency bands reflecting partial discharge or mechanical friction; S12: The dielectric constant change data is measured in real time by a microfluidic dielectric constant sensor based on the microstrip line resonance principle. The dielectric constant change data is used to reflect the change in the concentration of polar molecules in the insulating oil. S13: The distribution density data of solid micro-nano particles is monitored online by a micro-nano particle counting sensor based on the principle of laser scattering. The distribution density data includes particle concentration and particle size distribution.

3. The intelligent control method for transformer operation according to claim 2, characterized in that, Step S2 includes: S21: Perform a fast Fourier transform on the acoustic emission signal to extract the energy evolution characteristics of the ultrasonic frequency band; S22: Perform time series analysis on the dielectric constant change data and extract the rate of change of polar molecule concentration as the polarity change trend; S23: Statistically analyze the distribution density data of the solid micro-nano particles, and extract the particle concentration growth rate and the average particle size change trend; S24: The energy evolution characteristics, the polarity change trend, the particle concentration growth rate, and the average particle size change trend are weighted and fused to obtain the comprehensive deterioration index; S25: The cumulative increment of the comprehensive degradation index within the sliding time window is taken as the cumulative degradation trend.

4. The intelligent control method for transformer operation according to claim 3, characterized in that, In step S24, the weighted fusion calculation formula is: Id=wa·Ta+wc·Tc+wp·Tp+wd·Td, where Ta is the normalized value of the energy evolution characteristic, Tc is the normalized value of the polarity change trend, Tp is the normalized value of the particle concentration growth rate, Td is the normalized value of the average particle size change trend, and wa, wc, wp, and wd are preset weighting coefficients; the preset weighting coefficients are preset based on the sensitivity of each characteristic parameter to changes during the degradation process of the insulating medium and its correlation with the decrease in dielectric strength.

5. The intelligent control method for transformer operation according to claim 3, characterized in that, Step S3 includes: S31: Real-time monitoring of voltage deviation at the transformer output terminal; S32: The cumulative degradation trend and the voltage deviation are used together as input variables for action triggering judgment; S33: When the cumulative degradation trend does not reach the preset warning condition, the first dead zone value is used as the dead zone threshold of the voltage regulation. When the voltage deviation exceeds the first dead zone value, the mechanical adjustment of the regulation component is triggered. S34: When the cumulative degradation trend reaches the preset warning condition, the dead zone threshold of the voltage regulation is expanded from the first dead zone value to the second dead zone value, and the mechanical adjustment of the regulation component is triggered only when the voltage deviation exceeds the expanded second dead zone value.

6. The intelligent control method for transformer operation according to claim 1, characterized in that, Step S4 includes: S41: Obtain the state of charge of the user-side energy storage system and the adjustment margin of the controllable load; S42: Determine the risk level of the regulating component based on the cumulative degradation trend; S43: When the risk level reaches the preset level, determine whether the combined regulation capability of the state of charge and the regulation margin meets the current voltage regulation requirements; S44: If satisfied, only the user-side energy storage system and controllable load are scheduled for voltage regulation, without triggering the mechanical regulation of the regulation component; S45: If not satisfied, calculate the adjustment capacity gap and trigger the adjustment component to perform mechanical adjustment within the gap range; The user-side energy storage system and controllable load constitute flexible regulation resources, and the scheduling priority of the flexible regulation resources is higher than that of the mechanical regulation of the regulation components.

7. The intelligent control method for transformer operation according to claim 6, characterized in that, Step S44 includes: S441: Real-time monitoring of the deviation between the actual voltage value and the target value at the transformer output terminal; S442: When the deviation exceeds the allowable range and the risk level reaches the preset level, determine whether the state of charge of the user-side energy storage system is in the charge-discharge range. S443: If the voltage is in the chargeable / dischargeable range, when the actual voltage value is higher than the target value, a charging command is sent to the energy storage system to absorb excess power and reduce the voltage; when the actual voltage value is lower than the target value, a discharging command is sent to the energy storage system to supplement the power deficit and increase the voltage. S444: When the regulation capability of the energy storage system is insufficient, if the actual voltage value is higher than the target value, a power reduction command is issued to the controllable load to reduce the electrical load, thereby reducing the voltage; S445: Adjust the charging and discharging power of the energy storage system and the adjustment amount of the controllable load in real time according to the change of the voltage deviation, until the voltage is restored to the allowable range.

8. The intelligent control method for transformer operation according to claim 7, characterized in that, Step S45 includes: S451: Obtain the deviation amount monitored in real time; S452: Calculate the voltage adjustment amount corresponding to the combined regulation capability of the state of charge and the regulation margin; S453: If the absolute value of the voltage deviation is greater than the absolute value of the voltage adjustment, then the gap is calculated as the difference between the absolute value of the voltage deviation and the absolute value of the voltage adjustment. S454: Read the pre-stored single-level voltage adjustment step size of the adjustment component, and calculate the required number of mechanical adjustment actions based on the absolute value of the ratio of the difference to the single-level voltage adjustment step size; the single-level voltage adjustment step size is a fixed parameter set when the transformer leaves the factory. S455: Control the regulating component to perform mechanical adjustment not exceeding the number of calculated actions, and maintain the flexible regulating resources to output at maximum capacity, together restoring the voltage to the allowable range.

9. The intelligent control method for transformer operation according to claim 8, characterized in that, Step S455 includes: S4551: Initialize the action counter to 0, and set the number of calculated actions to the maximum allowed number of actions N; S4552: While maintaining the flexible adjustment resources to output at maximum capacity, perform a mechanical adjustment, the mechanical adjustment including: driving a switching switch to move the transformer tap one position in the direction of reducing voltage deviation, thereby changing the voltage ratio; S4553: After each mechanical adjustment is completed, the count value of the action counter is increased by 1, and the voltage deviation at the transformer output terminal is re-monitored; S4554: If the count value is less than the maximum allowable number of operations N and the current voltage deviation is still not within the allowable range, then repeat the next mechanical adjustment; S4555: If the count value is equal to the maximum allowable number of actions N or the current voltage deviation has entered the allowable range, then mechanical adjustment is stopped, and subsequent voltage deviations are absorbed by the continuous output of flexible adjustment resources.

10. An intelligent control system for transformer operation, characterized in that, The system is used to implement the steps of the method according to any one of claims 1-9, and the system includes: Acquisition module: Acquires multimodal physical characteristic data of the insulating medium inside the transformer regulating component, wherein the multimodal physical characteristic data includes at least acoustic emission signals, dielectric constant variation data, and distribution density data of solid micro-nano particles; Fusion module: Extracts the energy evolution characteristics of the acoustic emission signal, the polarity change trend of the dielectric constant, and the concentration growth rate of the solid micro / nano particles, fuses the three to determine the comprehensive degradation index, and uses the change trend of the comprehensive degradation index over time as the cumulative degradation trend; Correction module: The cumulative degradation trend is input as feedback information into the power flow control program of the transformer to correct the mechanical adjustment logic triggered only by voltage deviation; Regulation module: When the cumulative degradation trend reaches the preset warning condition, the flexible regulation resource is prioritized for voltage regulation to reduce the number of mechanical regulation operations of the regulation component, and the scheduling priority of the flexible regulation resource is higher than that of the mechanical regulation of the regulation component.

Images