Method and system for evaluating voltage regulation performance of dry-type transformer for energy storage

Abstract

The application discloses a voltage regulation performance evaluation method and system for dry-type transformers for energy storage, relates to the technical field of power system equipment state perception and operation evaluation, generates a voltage regulation weak point map through a two-dimensional near-field magnetic flux sensor array, realizes event determination in combination with an acoustic implicit saturation precursor index and a voltage threshold value, extracts voltage drop, recovery time and current rise slope to calculate event severity, introduces Arrhenius temperature correction and rain flow counting to form a voltage regulation degradation index, and further generates a voltage regulation health score. The system integrates magnetic flux, acoustic and electric parameter collection and control interfaces, can issue a limited slope to a power conversion device and issue a wind increasing instruction to a ventilation device, realizes an evaluation-early warning-control closed loop, and improves positioning accuracy and operation safety.

Description

Technical Field

[0001] This invention relates to the field of power system equipment status perception and operation assessment technology, specifically to a method and system for evaluating the voltage regulation performance of dry-type transformers for energy storage. Background Technology

[0002] With the increasing demands for grid-friendliness and peak-shaving / frequency regulation in large-scale energy storage power stations, energy storage converters frequently experience rapid power step jumps and power command switching during grid-connected operation. This leads to a greater slope in the secondary current rise, increased harmonic and interharmonic content, and a higher probability of three-phase imbalance. Simultaneously, the winding hotspot temperature exhibits periodic fluctuations due to charge-discharge cycles. Dry-type transformers are widely used in energy storage scenarios due to their flame-retardant and environmentally friendly properties; however, their voltage regulation performance exhibits significant dynamic characteristics and temperature dependence under these operating conditions. Traditional static voltage regulation indicators based on no-load versus load points are insufficient to reflect the true capabilities under such transient and complex disturbances.

[0003] Existing technologies typically rely on type tests and routine tests to obtain parameters such as voltage regulation, impedance voltage, and temperature rise, or on online monitoring of voltage, current, and temperature for operational evaluation. Other approaches include using magnetic shielding design and increasing capacity margin to reduce the risk of local saturation, and using offline acoustic or partial discharge detection to identify insulation defects. While these solutions provide a basis for the design and acceptance phases, they still have shortcomings under the conditions of rapid power changes and high-frequency disturbances in energy storage: it is difficult to spatially locate weak areas near the winding ends and clamps without shutting down the system; there is a lack of quantifiable precursors that can reflect implicit saturation risks in advance; a cumulative correlation between voltage drop events and lifetime depletion is not established, making it difficult to form a long-term degradation view; the evaluation results lack closed-loop linkage with the energy storage converter and ventilation system, making it impossible to promptly translate risk quantification results into mitigation measures such as slope limiting and ventilation enhancement; and there is a lack of unified dimensions and comparable standards between different sites and different equipment.

[0004] Therefore, there is an urgent need for a voltage regulation performance evaluation method and system for energy storage operation conditions. This system should be able to complete the fusion of near-field magnetic flux and acoustic multi-sources without shutting down the system, construct a voltage regulation weakness map and implicit saturation precursor index, establish a voltage regulation degradation index based on event severity and temperature correction, generate a voltage regulation health score, and link with the energy storage converter and ventilation actuator through a control interface to achieve an engineering closed loop from assessment and early warning to control. Summary of the Invention

[0005] The purpose of this invention is to overcome the shortcomings of the prior art and to propose a method and system for evaluating the voltage regulation performance of dry-type transformers for energy storage, so as to solve the above-mentioned problems.

[0006] The objective of this invention is achieved through the following technical solution: a method for evaluating the voltage regulation performance of a dry-type transformer for energy storage, comprising the following steps: S1. Leakage flux is collected by a two-dimensional near-field magnetic flux sensor array deployed on the outer surface of the transformer. Based on the amplitude of the leakage flux and its spatial gradient, it is weighted according to the first weight and the second weight, and then normalized to obtain the voltage regulation weakness map. The weakness map is used as the spatial weight for subsequent processing. S2. Deploy acoustic sensors at the weak points marked on the weakness map, perform spectral analysis on the acoustic signals, and calculate the implicit saturation precursor index for each measurement point based on the changes in the spectral center, bandwidth, and power of the second frequency component relative to the baseline. Then, weight and fuse the precursor indices of each measurement point according to their weights in the weakness map to obtain the weighted precursor index. S3. Synchronously acquire secondary side voltage and current; when the relative voltage drop exceeds the voltage threshold or the weighted precursor index exceeds the first threshold, determine that a voltage regulation event has occurred; and obtain the voltage drop amplitude, recovery time and current rise slope of the event; S4. Combine the voltage drop amplitude, recovery time, and current rise slope according to preset weights to obtain the event severity, and amplify the event severity according to an amplification factor related to the weighted precursor index; S5. Calculate the temperature correction factor based on the winding hot spot temperature, use the Arrhenius model to characterize the effect of temperature on degradation, and take the product of the event severity and the temperature correction factor as the loss of the event; S6. Perform rainflow counting on the voltage regulation event sequence, classify and accumulate the losses of each event to obtain the voltage regulation degradation index; S7. Generate a voltage regulation health score based on the weighted precursor index and voltage regulation degradation index, and output the location information and score of the weak area in the weakness map for operation, maintenance and control strategy formulation.

[0007] In step S1, the first and second weights used for weighting are set to 0.3 to 0.7; the reference for the amplitude of the rated leakage flux is the average value of the leakage flux under rated operating conditions, and the reference for the spatial gradient is the root mean square value of the spatial gradient under no-load operating conditions; the weakness map is generated by bilinear interpolation to produce a two-dimensional grid heat map with a resolution of not less than 5 mm.

[0008] In step S2, the spectral analysis uses an analysis frequency band of 100 Hz to 3 kHz, which is obtained through an FIR bandpass filter with a filter order of not less than 128. The first threshold is set to 0.6 to 0.8. The number of measurement points is 3 to 8. The three weights of the precursor index are equal or adjusted using verification data.

[0009] In step S4, the recovery time is 95% of the recovery time, with a reference value of 50 ms to 200 ms. The reference value for the current rise slope is 10 A / ms to 1 kA / ms. The weights of the three linear combination terms satisfy the following: the weight of the voltage drop term is not less than 0.3, the weights of the recovery time term and the current slope term are not less than 0.2 respectively, and the sum of the three weights is 1. The coupling coefficient used for hazard amplification is 0.3 to 1.0.

[0010] In step S5, the activation energy of the Arrhenius model is set to 0.3–0.7 eV, and the reference temperature is 393 K–413 K. The temperature sensor uses a fiber optic grating or platinum resistance array, and the sampling period is no more than 1 s.

[0011] In step S6, the four-point rainflow identification method is used. The severity of the event is used as the amplitude index. The event sequence is divided into no less than 8 level intervals according to the amplitude. The loss of a single event is accumulated in each interval and then summed to obtain the voltage regulation degradation index.

[0012] In step S7, the health score adopts a combination rule of first applying an exponential weight to the voltage regulation degradation index and then deducting it with a linear coefficient according to the weighted precursor index; when the health score is lower than the set minimum permissible value, an alarm is triggered and the location with the highest score in the weakness map is recorded.

[0013] A voltage regulation performance evaluation system for dry-type transformers used in energy storage includes: Near-field magnetic flux acquisition module: includes a two-dimensional near-field magnetic flux sensor array installed on the outer surface of the transformer and a matching data acquisition channel. The sampling and resolution of the channel can support the generation of a weakness map. Acoustic acquisition module: includes an array of microphones or a laser ultrasonic sensor deployed in the weak area and matched with a data acquisition channel. The sampling parameters of the channel can support the extraction of the spectral center, bandwidth and second frequency components. Electrical parameter acquisition module: used to synchronously acquire secondary side voltage and current; Processing and storage module: Includes processor and storage medium, which stores program instructions. When the processor runs the program instructions, it executes steps S1 to S7 and outputs a weakness map, weighted precursor index, voltage regulation degradation index and voltage regulation health score.

[0014] The near-field magnetic flux acquisition module is equipped with a magnetic shielding backplate and a fixed arc-shaped guide rail to fit the shape of the winding; the system has a built-in calibration process, including no-load baseline acquisition, step load testing and frequency band noise injection, to obtain the baseline value of the acoustic spectrum center, the baseline value of the bandwidth and the baseline power value of the second frequency component.

[0015] The processing and storage module includes a control interface for issuing slope limiting commands to the energy storage converter to reduce the current rise slope, and issuing ventilation gain commands to the fan or duct actuator; when the health score is lower than the set minimum permissible value or the weighted precursor index exceeds the set first threshold, the combined strategy of slope limiting and ventilation increase is automatically executed.

[0016] The beneficial effects of this invention are: This invention addresses the operational characteristics of dry-type transformers used in energy storage. It constructs an integrated evaluation and control closed loop comprising a two-dimensional near-field magnetic flux sensor array, a voltage regulation weakness map, an implicit saturation precursor index, a weighted precursor index, event severity, a temperature correction factor, single-event loss, a voltage regulation degradation index, and a voltage regulation health score. Through processing and storage modules, it issues slope limit commands to the energy storage converter and ventilation gain commands to the ventilation actuator, achieving on-site implementation of "assessment-early warning-control." This invention offers the following beneficial effects: The voltage regulation weakness map integrates the amplitude information of leakage magnetic flux with spatial gradient information, which can form an intuitive distribution of weak areas in structural parts such as winding ends and clamps. This can be used to guide the layout of acoustic sensors and the determination of maintenance points, avoiding missed detections caused by averaging detection.

[0017] Implicit saturation precursor indexes characterize early signs of a magnetic circuit entering implicit saturation by variations in the spectral center, bandwidth, and power of the secondary frequency component. By using weighted precursor indexes, signals in weak regions are spatially amplified, enabling the system to issue early warnings before significant voltage drops occur, thus reducing the risk of sudden events.

[0018] The combination of spatial weighting of the weakness map and baseline processing of acoustic features effectively suppresses the impact of environmental noise, installation differences and operational fluctuations on the judgment results, ensuring the consistency of judgments at different sites and at different times.

[0019] The relative voltage drop, recovery time, and current rise slope are defined as event characteristics, and event severity is formed. This allows events under different equipment and operating conditions to be compared on the same scale, providing a basis for operational review and asset management.

[0020] The temperature correction factor reflects the accelerating effect of high temperature on degradation, the single event loss characterizes the life loss caused by a single impact, and the voltage regulation degradation index reflects the long-term deterioration trend through the hierarchical accumulation of event sequences, forming a measurement framework that connects events and life, which is convenient for traceability and auditing.

[0021] The voltage regulation health score unifies long-term degradation and immediate precursors into a single visual indicator, and triggers alarms in conjunction with the weighted precursor index, facilitating rapid decision-making and scheduling by maintenance personnel.

[0022] The processing and storage module maps the voltage regulation health score and weighted precursor index into slope limit command and ventilation gain command. The control results are reflected in real time on the weakness map, weighted precursor index and voltage regulation degradation index, forming an adaptive closed loop of "measure-feedback-re-optimization".

[0023] The slope limit command suppresses the risk of voltage drop caused by excessive current rise slope, and the ventilation gain command reduces the winding hot spot temperature and implicit saturation risk, which helps to reduce the occurrence of voltage regulation events and slow down the growth of the voltage regulation degradation index.

[0024] A unified time-based data acquisition and processing architecture ensures that the weakness map, weighted precursor index, voltage regulation degradation index, and voltage regulation health score can be continuously updated during operation, adapting to the online operation requirements of energy storage sites.

[0025] The near-field magnetic flux sensor array, combined with the magnetic shielding backplate and the arc-shaped guide rail installation method, as well as the built-in calibration process, enables dry-type transformers of different capacities and winding structures to be quickly deployed and obtain comparable results with minimal modifications.

[0026] The spatial location information provided by the weakness map supports fixed-point inspection and on-demand maintenance, avoiding large-scale disassembly and inspection; sensor resources can be dynamically allocated based on the normalized score of the weak area, improving the efficiency of long-term monitoring.

[0027] By reducing the rapid fluctuations in bus voltage and anomalies caused by local saturation, the coordinated stability of the energy storage converter and the distribution network is improved, reducing malfunctions and protection setting pressure.

[0028] The system timestamps voltage regulation events, weakness maps, weighted precursor indices, voltage regulation degradation indices, voltage regulation health scores, and control commands, facilitating the creation of auditable operation and maintenance records and policy evaluation bases.

[0029] The system architecture allows for the introduction of more sensor channels or algorithm modules without changing the existing processes, maintaining consistency in terminology and interfaces, and facilitating subsequent functional expansion and standard integration. Attached Figure Description

[0030] Figure 1 The process of this invention Figure 1 ; Figure 2 The process of this invention Figure 2 ; Figure 3 This is a system architecture diagram of the present invention. Detailed Implementation

[0031] The technical solution of the present invention will be clearly and completely described below with reference to the embodiments. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0032] It should be noted that the directional concepts of "left", "right", "up", "down", "front", "back", "inner", and "outer" in the following scheme are all relative directions, and will not be listed one by one here.

[0033] Example 1 like Figure 1 As shown, this embodiment addresses the voltage regulation behavior of dry-type transformers used for energy storage under operating conditions including step loads, harmonics, imbalances, and DC bias. It presents an implementation method encompassing spatial localization, acoustic precursor quantization, event determination, and feature extraction. This method uses a weakness map formed by a two-dimensional near-field flux sensor array as spatial weights, employs an implicit saturation precursor index as the precursor quantization quantity, and combines it with a voltage threshold to trigger voltage regulation event determination. This yields three event features: voltage drop amplitude, recovery time, and current rise slope, providing a data foundation for subsequent degradation assessment and control strategies.

[0034] In terms of equipment configuration, a two-dimensional near-field magnetic flux sensor array is installed along the axial and radial directions on the outer surface of the dry-type transformer. The sensors employ anisotropic magnetoresistive devices or Hall effect devices, and the array arrangement covers the sensitive areas around the winding ends and clamps. Acoustic sensors are deployed at weak points marked on the weak point map, using array microphones or laser ultrasonic sensors. Secondary side voltage and current are acquired through a synchronous data acquisition device. All acquisition channels are synchronized by the same clock source to ensure phase consistency. The acquired data is processed in a windowed manner in the processor, with the time window length covering the entire process of an event's occurrence and recovery.

[0035] In terms of spatial positioning and weakness map generation, a two-dimensional near-field magnetic flux sensor array is used to obtain the leakage magnetic flux density field under rated and disturbed operating conditions. The leakage magnetic flux amplitude at each spatial location and the spatial gradient amplitude between adjacent locations are calculated. After normalization, these two values ​​are linearly weighted according to a first weight and a second weight to obtain a weakness score. Subsequently, normalization is performed with the maximum weakness score of the entire map as a reference to form a weakness map with values ​​ranging from zero to one. The calculation relationship of the weakness score is expressed as follows: In the formula: The leakage magnetic flux density measured at position (x,y) is in Tesla. For spatial gradient, It is a 2-norm; The spatial average value of leakage flux under rated operating conditions; This represents the root mean square value of the spatial gradient of the leakage magnetic flux under no-load operating conditions. As the first weight, As the second weight, and All are positive; This represents the score for weakness. Normalized weakness score; The spatial domain covered by the array. Based on the spatial distribution of the normalized weakness score, several peak locations are selected as the placement points for the acoustic sensors.

[0036] In acoustic precursor quantization, acoustic signals are acquired at selected locations, the power spectral density is calculated, and the spectral center and bandwidth are obtained within the target frequency band. Simultaneously, the ratio of the power change of the second frequency component relative to the baseline is acquired. These three quantization quantities are combined according to preset weights to form an implicit saturation precursor exponent. The target frequency band is matched with the system bandwidth, which in this case is 100 Hz–3 kHz. The spectral center and bandwidth are defined using the statistical definition of discrete spectrum, expressed as: In the formula: Let f be the power spectral density at the i-th measurement point at frequency f; The target frequency band set used for calculation; Let be the spectral center frequency of the i-th measurement point; Let be the bandwidth at the i-th measurement point. The implicit saturation precursor index uses the ratio of the change in power relative to the baseline as its component, expressed as: In the formula: The baseline value of the spectral center at the i-th measurement point; This represents the baseline bandwidth value for the i-th measurement point; Let be the power at the second frequency component of the i-th measurement point. This corresponds to the baseline power; The three components are weighted coefficients, all of which are positive and sum to one. The implicit saturation precursor index is spatially weighted and fused with the weakness map to obtain the weighted precursor index, expressed as: In the formula: K is the number of acoustic measurement points; Let be the coordinates of the i-th acoustic measurement point; This represents the normalized weakness score for that location.

[0037] In terms of voltage regulation event determination and feature extraction, the baseline value and the minimum value during the event period are estimated using a sliding window root mean square method for the secondary voltage. The relative drop is calculated and used together with the weighted precursor exponent as the triggering condition. The sliding window root mean square method preferably covers 1–3 fundamental frequency cycles. The relative drop and triggering rules are expressed as follows: In the formula: The root mean square baseline voltage before the event occurred; This represents the root mean square minimum voltage during the event. This is a relative voltage drop; Voltage threshold; The first threshold is set. When an event occurs, the time taken from the start of the event until the voltage recovers to 95% of the baseline is recorded as the recovery time, and the slope of the current rise segment within the event window is recorded as the current rise slope.

[0038] During operation, near-field magnetic flux is first collected on the dry-type transformer under rated and disturbed operating conditions to obtain the spatial distribution of the weakness score and form a weakness map. Then, acoustic sensors are deployed at the weak locations and baseline sampling is performed. Without changing the transformer's primary side wiring and load, repeatable step power and harmonic current injection are generated by controlling the energy storage converter to continuously acquire acoustic signals and electrical parameters and update the implicit saturation precursor index and weighted precursor index. Either voltage relative drop or weighted precursor index exceeding the limit is used as the voltage regulation event trigger condition. During and after each event, three features are extracted: voltage drop amplitude, recovery time, and current rise slope, and stored together with the corresponding weak location and weighted precursor index.

[0039] A weakness map is constructed using a two-dimensional near-field magnetic flux sensor array. The leakage flux amplitude and spatial gradient information are combined into a weakness score and normalized to clearly reveal the weak areas spatially, reducing the randomness caused by single-point measurements. Acoustic spectrum analysis is performed at the weak locations, and an implicit saturation precursor index is constructed using the ratio of spectral center, bandwidth, and power of the second frequency component. This index is then fused with the weakness map as spatial weights to achieve targeted spatial amplification of precursor information, thereby identifying potential saturation trends in advance before a significant voltage drop occurs. By using the weighted precursor index and voltage threshold in parallel as event judgment conditions, event triggering can reflect both macroscopic disturbances in the bus voltage and microscopic anomalies in the local magnetic circuit, improving the sensitivity and specificity of voltage regulation event capture. A unified event feature acquisition process obtains three features: voltage drop amplitude, recovery time, and current rise slope, providing direct input for subsequent degradation assessment, health scoring, and control strategies. The aforementioned effects rely on the synergistic relationship between the weakness map, the implicit saturation precursor index, and the weighted precursor index, as well as the combined use of event triggering logic, which overall improves the online observability and quantifiability of the voltage regulation performance of dry-type transformers for energy storage.

[0040] Example 2 like Figure 1 and Figure 2 As shown, this embodiment, based on Embodiment 1, performs degradation quantification and health scoring, targeting the online operation and maintenance of dry-type transformers for energy storage. It transforms the weighted precursor index, weakness map, and voltage regulation event characteristics obtained in Embodiment 1 into an accumulative, comparable, and alarm-enabled indicator system. The device configuration and data sources remain consistent with Embodiment 1. The voltage drop amplitude, recovery time, and current rise slope of voltage regulation events are acquired using the same sampling synchronization system, and the weighted precursor index and weakness map are continuously updated from Embodiment 1. This embodiment first quantifies individual events to form event severity; then, a temperature correction factor is introduced to obtain the loss of a single event; next, the event sequence is graded and accumulated through rainflow counting to form a voltage regulation degradation index; finally, a voltage regulation health score is generated based on the voltage regulation degradation index and the weighted precursor index, and alarm strategies are set.

[0041] In terms of event quantification, to ensure comparability across different operating conditions and equipment, the voltage drop amplitude, recovery time, and current rise slope are normalized and linearly combined according to preset weights to obtain a baseline value for event severity. A hazard amplification factor is then introduced by combining a weighted precursor index. The calculation relationship for event severity is as follows: In the formula: The dimensionless quantification result of the severity of the event; This is due to a relative voltage drop. V is the difference between the voltage baseline and the minimum value during the event, and V is the voltage baseline before the event; The time it takes for the voltage to recover from the start of the event to 95% of the pre-event baseline; It serves as a reference value for restoring time and is used to normalize the time dimension. This represents the average slope of the current rise segment within the event window. This is a reference value for the current slope; These are non-negative weighting coefficients of a linear combination, satisfying that the sum of the three is one, to reflect the relative contribution of the three quantities to the risk of voltage regulation; This is the hazard amplification factor, a positive value; The weighted precursor index was obtained through spatial weighting of the weakness map in Example 1. Reference value. and The data is obtained through initial or annual baseline statistics to ensure reusability under different seasons and load profiles.

[0042] In modeling the effects of temperature, considering the accelerating effect of hot spot temperature in dry-type transformer windings on insulation and voltage regulation behavior, an Arrhenius-type temperature correction factor is introduced to assign higher damage weights to voltage events of the same intensity at higher temperatures. The calculation relationship for the temperature correction factor is as follows: In the formula: This is a temperature correction factor; The apparent activation energy characterizes the thermal activation process; k is the Boltzmann constant; T is the reference temperature; T is the winding hot spot temperature during the event. Using this correction factor can eliminate the bias of seasonality and ventilation changes on degradation assessment, making the proportion of high-temperature events in the cumulative figure more consistent with the laws of physical degradation.

[0043] Regarding the determination of single-event attrition, multiplying the event severity by the temperature correction factor yields a single-event attrition that is additive with respect to lifetime depletion: In the formula: This represents the loss amount in a single event, serving as the basic unit for subsequent accumulation.

[0044] In terms of sequence accumulation, a four-point rainflow identification method is used to perform amplitude cyclic decomposition on the event sequence. Using event severity as the amplitude index, events are divided into several level intervals according to amplitude magnitude. Within each interval, the loss from a single event is summed, and the final summation yields the voltage regulation degradation index. The cumulative relationship of the voltage regulation degradation index is as follows: In the formula: Here, represents the voltage regulation degradation index; the summation index e represents each event in the event set obtained through rainflow identification. By using rainflow counting, the non-stationarity of the event sequence is transformed into standardized cyclic statistics, which preserves the differences in contributions of events of different intensities and ensures comparability across devices and time periods.

[0045] In terms of health status measurement, an exponential weight is first applied to the voltage regulation degradation index to reflect the cumulative effect of degradation. Then, a linear coefficient is used to subtract from the weighted precursor index to obtain the voltage regulation health score, which is used to intuitively describe the overall voltage regulation health of the equipment under current and recent operating conditions. The calculation relationship of the voltage regulation health score is as follows: ; In the formula: For the blood pressure health score, the value range is cropped by numerical clipping: HVR←min(1,max(0,HVR)); This is the exponential weighting coefficient for the voltage regulation degradation index, and it is a positive value. This is a linear deduction factor for the weighted precursor index, and it is a positive value. These are derived from the calculations mentioned above. The index term reflects the characteristic that "the greater the long-term accumulation, the faster the score decays," while the linear deduction term reflects the immediacy of "the stronger the precursor, the higher the current risk," taking into account both long-term and short-term factors within the same scoring framework.

[0046] During operation, the system continuously monitors the voltage regulation events and weighted precursor indices output by Implementation Example 1, and immediately calculates the severity index according to the event severity formula after each event ends. Calculate the winding hot spot temperature during the event. ,get Then write to the event queue; run rainflow identification and graded accumulation within the rolling time window, and update. Update using fixed time steps or event-triggered methods. .when When the score falls below the set minimum permissible value, the system automatically marks the spatial location with the highest normalized weakness score in the weakness map as the priority point for inspection and the target point for the control strategy. At the same time, the rising rate of the weighted precursor index and the severity of the most recent event are used for strategy tuning. For example, a slope limit command is issued to the energy storage converter to reduce the current rise slope, and an air increase command is issued to the ventilation system to reduce the winding hot spot temperature, thereby making an immediate response to the score decline at the control level.

[0047] By employing a normalized linear combination of event severity and a precursor amplification factor, a consistent event quantification caliber is obtained under different equipment, load spectra, and network-side conditions, improving cross-site comparability. Secondly, an Arrhenius-type temperature correction factor explicitly incorporates the thermal acceleration mechanism into the event loss model, ensuring reasonable weighting of voltage regulation events during high-temperature operation in the cumulative calculation, avoiding underestimation caused by relying solely on voltage indicators. Thirdly, rainflow counting transforms the non-stationarity of event sequences into standardized cyclic statistics, considering both high-amplitude, low-frequency impacts and medium-to-low-amplitude, high-frequency fatigue events, forming a voltage regulation degradation index as a cumulative representation. Fourthly, a voltage regulation health score, combining exponential decay and linear deduction, simultaneously quantifies long-term degradation and immediate risks, and is linked with a weakness map and a weighted precursor index to achieve a closed loop from "problem discovery" to "trigger control." In summary, this embodiment, without altering the primary system structure, unifies spatial precursors, single events, and long-term degradation into a single computable framework, providing an authoritative, deployable, and maintainable method for measuring voltage regulation health.

[0048] Example 3 like Figures 1 to 3 As shown, this embodiment, based on Embodiments 1 and 2, provides a device structure, installation and calibration process, and linkage control strategy with energy storage converters and ventilation actuators for field deployment. This embodiment, based on modular hardware and a data channel with a unified time reference, ensures the comparability of the two-dimensional near-field magnetic flux sensor array, acoustic sensors, and electrical parameter acquisition in the time and amplitude domains. Relying on program instructions in the processing and storage module, it generates a weakness map, weighted precursor index, voltage regulation degradation index, and voltage regulation health score in real time. Through the control interface, the health status is mapped to slope limit and airflow increase commands, thereby suppressing the occurrence of voltage regulation events and the growth of the voltage regulation degradation index.

[0049] The device consists of a near-field magnetic flux acquisition module, an acoustic acquisition module, an electrical parameter acquisition module, and a processing and storage module. A two-dimensional near-field magnetic flux sensor array is fixed to a magnetically shielded backplate and an arc-shaped guide rail to conform to the winding shape. The array's geometric pose is mechanically referenced using fixture positioning holes and a reference surface on the main body. The acoustic sensors are deployed in the weak areas indicated by the weakness map, using replaceable mounting brackets for easy maintenance. Electrical parameter acquisition employs a high-speed acquisition card in the same domain as the near-field magnetic flux and acoustic channels, and is synchronized with the same clock source. The processing and storage module incorporates a highly stable clock and time synchronization protocol stack, providing a unified timestamp for parallel processing of multiple data streams. After installation, a calibration process is executed, sequentially performing no-load baseline acquisition, step load testing, and frequency band noise injection. Unloaded baseline acquisition is used to obtain baseline values ​​for the acoustic spectrum center, bandwidth, and power of the second frequency component, and to establish the gain and zero bias of each channel of the two-dimensional near-field flux sensor array; step load testing is used to verify the consistency of delay across channels and cross-channel phase alignment; band noise injection is used to verify the bandwidth coverage and anti-aliasing capability of the acoustic channels. Time alignment of each channel adopts a unified time deviation correction model, and the time alignment relationship is expressed as: ; In the formula: The aligned timestamp; The original timestamp provided by the capture card; This refers to the relative time difference of the channel measured through synchronization messages. It operates on three channels: near-field magnetic flux, acoustics, and electrical parameters, enabling the fusion processing of multi-source data on the same time axis and ensuring the causal consistency of the weakness map, implicit saturation precursor index, weighted precursor index, and voltage regulation degradation index at the event scale.

[0050] Spatial registration uses mechanical reference points provided by the installation fixture and the distance measurement results of the array endpoints to map the physical coordinates of each sensing unit to a unified coordinate domain. Based on this, the gain of the two-dimensional near-field magnetic flux sensor array is calibrated in the amplitude domain using a known magnetic field generated by a reference coil, resulting in gain coefficients mapped from the original readings of each channel to physical quantities. In the acoustic channel, an absolute calibration of the power spectral density is used with a reference sound source to ensure the comparability of spectral center and bandwidth calculations across different devices and sites. After spatial registration, the weakness map is recalculated, and the processing and storage module refreshes the weighted precursor index and voltage regulation health score within a fixed time step. To facilitate inspection and maintenance, the system converts the normalized weakness score in the weakness map into the selection probability of weak sensing points, used to limit key locations for long-term monitoring. The conversion relationship is expressed as: In the formula: Let be the probability that the i-th position is selected. denoted as the normalized vulnerability score for that location; K represents the number of available locations. Through probabilistic allocation of sensing resources, monitoring resources are dynamically focused on vulnerable areas, balancing long-term trends with immediate risks.

[0051] The processing and storage module establishes real-time communication with the energy storage converter and the fan or duct actuator through the control interface, generating limit slope and airflow increase commands based on the current voltage regulation health score and weighted precursor index. The upper limit of the current rise slope of the energy storage converter adaptively adjusts according to the health status and precursor intensity, and the mapping relationship is expressed as: In the formula: This represents the upper limit of the current rise slope sent to the energy storage converter at time t. This represents the upper limit of the baseline slope when the patient is in good health and there are no obvious warning signs. Assess the health score for blood pressure regulation; This is a weighted precursor index; These are non-negative mapping coefficients; The upper limit of the minimum allowed slope; As a saturation function, the quantity within the parentheses is clipped between the lower and upper limits. This relationship ensures that when the health score decreases or precursors increase, the upper limit of the current rise slope decreases proportionally to suppress high [current surges / increases]. The triggered voltage drop event slows down the accumulation of the voltage regulation degradation index. clip(⋅) is a saturation operator, which can be equivalently implemented as a piecewise linear function.

[0052] The sampling and processing cycle after the control command is issued is 50–200 ms; The control input of the ventilation actuator is a weighted sum of temperature deviation and precursor intensity, limited to the actuable range. The control relationship is expressed as follows: In the formula: The ventilation control command at time t can correspond to either the fan speed or the damper opening. Baseline ventilation volume; This refers to the hot spot temperature of the winding. For reference temperature; and The coefficients are non-negative. This represents the lower and upper limits of ventilation control. This relationship automatically increases ventilation when the hotspot temperature is too high or the precursor is enhanced, reducing the hotspot temperature and saturation risk, and suppressing the weighted precursor index and pressure regulation degradation index from increasing over time.

[0053] To ensure that control strategies intervene only when necessary, the processing and storage module employs event-driven logic to generate intervention flags based on health status and early warning signs, as defined below: In the formula: g(t) is the intervention flag, which takes the value of zero or one; This is an indicator function; it takes the value 1 if the condition within the parentheses is true, and 0 otherwise. The minimum permissible value for a health score; This is the threshold for the weighted precursor index. The slope limit and ventilation increase command are issued only when g(t) = 1, to avoid unnecessary disturbances to the energy storage converter and ventilation system when they are in good health.

[0054] During operation, the system first establishes time and amplitude benchmarks for the multi-source channels through installation and calibration procedures. Then, during operation, it refreshes the weakness map, weighted precursor index, voltage regulation degradation index, and voltage regulation health score at fixed time steps. When an intervention flag is triggered, the processing and storage module sends commands to the energy storage converter and ventilation actuator in real time via the control interface, recording the changes in the weighted precursor index and relative voltage drop before and after the command, for strategy evaluation and parameter tuning. To adapt to dry-type transformers of different capacities and winding structures, when introducing new equipment, the device only needs to complete array geometry registration and channel calibration according to installation guidelines to be put into operation without changing the primary system structure.

[0055] By using modular hardware and a unified time-based data channel, the system achieves real-time output of weakness maps, weighted precursor indices, voltage regulation degradation indices, and voltage regulation health scores, improving the real-time performance and feasibility of the assessment process. Secondly, through the installation of fixtures and calibration procedures, the two-dimensional near-field magnetic flux sensor array and acoustic sensors can be repeatedly positioned and quickly replicated across different devices, improving measurement consistency and cross-device portability. Thirdly, by mapping health status and precursor intensity to the upper limit of current rise slope and ventilation control quantity, a closed loop from assessment to control is formed, enabling high-voltage... The system effectively suppresses high-temperature scenarios, thereby reducing the growth rate of the voltage regulation degradation index and decreasing the occurrence of voltage regulation events. Fourthly, by employing event protection logic, it maintains control silence when the system is in good health, avoiding unnecessary intervention in the energy storage converter and ventilation system, thus balancing equipment lifespan and energy efficiency. In summary, the system implementation provides feasible online evaluation and adaptive collaborative control capabilities for dry-type transformers used for energy storage without altering grid connection or primary system structure, enabling the stable operation of the algorithms from Examples 1 and 2 in the engineering field.

[0056] The above description is merely a preferred embodiment of the present invention. It should be understood that the present invention is not limited to the forms disclosed herein and should not be construed as excluding other embodiments. It can be used in various other combinations, modifications, and environments, and can be altered within the scope of the concept described herein through the above teachings or related technologies or knowledge. Modifications and variations made by those skilled in the art that do not depart from the spirit and scope of the present invention should be within the protection scope of the appended claims.

Claims

1. A method for evaluating the voltage regulation performance of a dry-type transformer for energy storage, characterized in that, Includes the following steps: S1. Leakage flux is collected by a two-dimensional near-field magnetic flux sensor array deployed on the outer surface of the transformer. Based on the amplitude of the leakage flux and its spatial gradient, it is weighted according to the first weight and the second weight, and then normalized to obtain a voltage regulation weakness map. The weakness map is used as the spatial weight for subsequent processing. S2. Deploy acoustic sensors at the weak locations marked on the weakness map, perform spectral analysis on the acoustic signals, and calculate the implicit saturation precursor index for each measurement point based on the changes in the spectral center, bandwidth, and power of the second frequency component relative to the baseline; then weight and fuse the precursor indices of each measurement point according to their weights in the weakness map to obtain the weighted precursor index. S3. Synchronously acquire secondary side voltage and current; when the relative voltage drop exceeds the voltage threshold or the weighted precursor index exceeds the first threshold, determine that a voltage regulation event has occurred; and obtain the voltage drop amplitude, recovery time and current rise slope of the event; S4. The voltage drop amplitude, recovery time and current rise slope are combined according to a preset weight to obtain the event severity, and the event severity is amplified by an amplification factor related to the weighted precursor index; S5. Calculate the temperature correction factor based on the winding hot spot temperature, use the Arrhenius model to characterize the effect of temperature on degradation, and take the product of the event severity and the temperature correction factor as the loss of the event; S6. Perform rainflow counting on the voltage regulation event sequence, classify and accumulate the losses of each event to obtain the voltage regulation degradation index; S7. Generate a voltage regulation health score based on the weighted precursor index and the voltage regulation degradation index, and output the location information and score of the weak area in the weakness map for operation, maintenance and control strategy formulation.

2. The method according to claim 1, characterized in that, In step S1, the first and second weights used for weighting are set to 0.3 to 0.7; the reference for the amplitude of the rated leakage flux is the average value of the leakage flux under rated operating conditions, and the reference for the spatial gradient is the root mean square value of the spatial gradient under no-load operating conditions; the weakness map is generated by bilinear interpolation to produce a two-dimensional grid heat map with a resolution of not less than 5 mm.

3. The method according to claim 1, characterized in that, The spectral analysis in step S2 uses an analysis frequency band of 100 Hz to 3 kHz, which is obtained through an FIR bandpass filter with an order of not less than 128. The first threshold is 0.6 to 0.

8. The number of measurement points is 3 to 8. The three weights of the precursor index are equal or adjusted by verification data.

4. The method according to claim 1, characterized in that, The recovery time mentioned in step S4 is the 95% recovery time, and the reference value of the recovery time is 50 ms to 200 ms. The reference value of the current rise slope is 10 A / ms to 1 kA / ms. The weights of the three linear combination terms satisfy the following: the weight of the voltage drop term is not less than 0.3, the weights of the recovery time term and the current slope term are not less than 0.2 respectively, and the sum of the three weights is 1. The coupling coefficient used for hazard amplification is 0.3 to 1.

0.

5. The method according to claim 1, characterized in that, In step S5, the activation energy of the Arrhenius model is set to 0.3–0.7 eV, and the reference temperature is 393 K–413 K. The temperature sensor uses a fiber optic grating or platinum resistance array, and the sampling period is no more than 1 s.

6. The method according to claim 1, characterized in that, In step S6, the four-point rainflow identification method is used. The severity of the event is used as the amplitude index. The event sequence is divided into no less than 8 level intervals according to the amplitude. The loss of a single event is accumulated in each interval and then summed to obtain the voltage regulation degradation index.

7. The method according to claim 1, characterized in that, The health score in step S7 adopts a combination rule of first applying an exponential weight to the voltage regulation degradation index and then deducting it with a linear coefficient according to the weighted precursor index; when the health score is lower than the set minimum permissible value, an alarm is triggered and the location with the highest score in the weakness map is recorded.

8. A voltage regulation performance evaluation system for a dry-type transformer used in energy storage to implement the method according to any one of claims 1 to 7, characterized in that, include: Near-field magnetic flux acquisition module: includes a two-dimensional near-field magnetic flux sensor array installed on the outer surface of the transformer and a matching data acquisition channel, the sampling and resolution of which can support the generation of a weakness map; Acoustic acquisition module: includes an array of microphones or a laser ultrasonic sensor deployed in the weak area and matched with a data acquisition channel, the sampling parameters of which can support the extraction of the spectral center, bandwidth and second frequency components; Electrical parameter acquisition module: used to synchronously acquire secondary side voltage and current; Processing and storage module: includes a processor and a storage medium, storing program instructions, wherein when the processor runs the program instructions, it executes the steps of any one of claims 1 to 7, and outputs a weakness map, a weighted precursor index, a voltage regulation degradation index, and a voltage regulation health score.

9. The system according to claim 8, characterized in that, The near-field magnetic flux acquisition module is equipped with a magnetic shielding backplate and a fixed arc-shaped guide rail to fit the shape of the winding. The system has a built-in calibration process, including no-load baseline acquisition, step load testing, and frequency band noise injection, to obtain the baseline values ​​of the acoustic spectrum center, bandwidth, and secondary frequency components.

10. The system according to claim 8, characterized in that, The processing and storage module includes a control interface for issuing slope limiting commands to the energy storage converter to reduce the current rise slope, and issuing ventilation gain commands to the fan or duct actuator; when the health score is lower than the set minimum permissible value or the weighted precursor index exceeds the set first threshold, the combined strategy of slope limiting and ventilation increase is automatically executed.

Images