An ultracapacitor energy storage automatic switching method and device

Abstract

A super capacitor energy storage automatic switching method and device, the method comprises the following steps: collecting the instantaneous value of three-phase voltage and current of bus at the public connection point, and calculating the effective value; when the continuous low voltage, calculate the residual voltage gradient and filter the maximum absolute value, and calculate the voltage and current harmonic entropy; based on the final residual voltage gradient and harmonic entropy, an elliptical decision equation is constructed, the suspicious island point is marked and the island confidence of the suspicious island point is calculated; combined with the current harmonic entropy, the confidence is corrected, and after window accumulation, the threshold value is determined as island; when in island state, it is divided into power surplus and insufficient scene; in the power surplus scene, first, the energy consumption resistor is put into according to the difference classification, and then the super capacitor charging and discharging control is started; in the power insufficient scene, the super capacitor is discharged. The present application realizes smooth switching of energy storage by hierarchical power regulation and super capacitor safety margin constraint, and reduces the influence of power mutation on bus voltage and frequency stability.

Description

Technical Field

[0001] This invention belongs to the field of power grid control technology and relates to a method and device for automatic switching of supercapacitor energy storage. Background Technology

[0002] The penetration rate of new energy sources such as wind power and photovoltaics in the power grid continues to rise, with the penetration rate of new energy in distribution networks exceeding 50% in some regions. The high proportion of new energy access has profoundly changed the characteristics of power grid faults, posing a severe challenge to traditional backup power automatic switching (BATS) technology.

[0003] In terms of islanding detection, traditional automatic backup power supply devices mainly rely on voltage amplitude criteria (such as voltage below 80% of the rated voltage). However, in scenarios with a high proportion of renewable energy, the low voltage ride-through characteristics of renewable energy units after a fault will inject reactive power into the bus, maintaining the bus residual voltage at 70%–90% of the rated value, causing the traditional voltage threshold criteria to completely fail. In terms of power balance control, a control strategy of parallel activation of energy storage and energy consumption devices is often adopted. In scenarios with excess power, the capacitive inrush current of supercapacitor charging and the inductive impact of resistor activation can easily cause transient overvoltage on the bus.

[0004] To address the aforementioned issues, existing solutions often employ forced disconnection of renewable energy sources or load shedding. For example, patent application CN110829580A provides a method for switching automatic transfer switches (ATS) in grid-connected substations of renewable energy power plants. This method establishes a causal relationship matrix between internal components of the grid-connected substation, a basic node state matrix, and a membership matrix between basic nodes and circuit breaker malfunctions / failures to trip. It further constructs a transformation vector to eliminate the influence of inconsistencies between circuit breaker actions and corresponding signals on islanding judgment. Furthermore, it utilizes the islanding operation state judgment matrix and threshold values ​​for fuzzy minimum calculation, achieving reliable identification of the islanding state of the grid-connected substation and reliable switching of ATS. Patent application CN115912572A provides a multi-path balancing circuit capable of automatic switching of individual energy storage units. Utilizing series-connected energy storage units, flyback transformers, bidirectional switches, and unidirectional switches, along with a control circuit containing a dual-layer MPC controller, it achieves online disconnection of failed energy storage units, online switching in of standby energy storage units, and direct energy transfer between any energy storage unit / group. Patent application CN122000937A discloses an equal-capacity AC power consumption configuration method, switching method, device, and equipment. It uses a delta-connected resistor and anti-parallel thyristors as AC power consumption branches. Through equal-capacity branch configuration and redundant group design, and by combining two methods, namely overall switching without trigger angle and phase-shifting switching with trigger angle, it achieves reliable suppression of surplus power in flexible DC systems after fault ride-through or blocking of modular multilevel converters (MMC).

[0005] In summary, the existing method redefines the technical problems to be solved by this invention, which have not yet been solved by the prior art, as follows: First, how to construct an islanding criterion that does not rely on a single voltage amplitude threshold under the condition that a high proportion of new energy is connected and the bus residual voltage is still maintained at 70% to 90% of the rated value, so as to achieve fast and reliable islanding detection; Second, how to use energy-consuming resistors and supercapacitor energy storage devices as inertia sources of different time scales for hierarchical coordinated switching, so as to avoid the superposition of resistive impact and capacitive inrush current generated by resistor connection and capacitor charging, and suppress transient overvoltage after islanding; Third, how to dynamically adjust the focus of multi-objective optimization according to the system operating status within a millisecond-level control cycle, so as to achieve adaptive variable weight rolling optimization that prioritizes voltage protection when the voltage deviation is large and takes into account power balance and economy after the voltage stabilizes. Summary of the Invention

[0006] To address the issues of excessive transient overvoltage caused by high residual bus voltage after a fault in high-proportion renewable energy access scenarios and the superposition of capacitive inrush current and resistive surges resulting from the parallel operation of supercapacitor energy storage and energy-consuming resistors, this invention provides an automatic switching method and device for supercapacitor energy storage. The method includes the following steps: collecting the instantaneous values ​​of the three-phase voltage and current of the bus at the point of common coupling and calculating the effective values; when the voltage is continuously low, calculating the residual voltage gradient and filtering to obtain the maximum absolute value, and calculating the voltage and current harmonic entropy; constructing an elliptical decision equation based on the final residual voltage gradient and harmonic entropy, marking suspected islanding points and calculating the islanding confidence of the suspected islanding points; correcting the confidence by combining the current harmonic entropy, and determining islanding if the value exceeds a threshold after window accumulation; when in an islanding state, classifying it into power surplus and power shortage scenarios; in the power surplus scenario, first activating energy-consuming resistors according to the difference in power level, and then starting the supercapacitor charging and discharging control; in the power shortage scenario, starting the supercapacitor discharging. This invention achieves smooth energy storage switching through graded power regulation and supercapacitor safety margin constraints, reducing the impact of power surges on bus voltage and frequency stability.

[0007] The present invention adopts the following technical solution: The first aspect of the present invention provides an automatic switching method for supercapacitor energy storage, comprising the following steps: S1. Collect the instantaneous values ​​of the three-phase voltage and current of the bus at the common connection point of new energy access and grid connection, and calculate the effective values ​​of voltage and current; S2. When the effective values ​​of the three-phase voltages are all lower than the predefined voltage thresholds within the predetermined duration period, calculate the residual voltage gradient and perform moving average filtering. Take the maximum absolute value of the residual voltage gradient after filtering of all phases as the final residual voltage gradient. Perform fast Fourier transform on the bus voltage and current signals collected within the predefined sliding window to calculate the voltage harmonic entropy and current harmonic entropy. S3. Based on the final residual voltage gradient and harmonic entropy, construct an elliptical decision equation and mark suspicious island points; calculate the island confidence of suspicious island points, and combine it with the current harmonic entropy to correct the island confidence when the predetermined judgment conditions are met; multiply the island confidence of each sampling time within the cumulative window length before the corresponding sampling time by the time interval and then accumulate it to obtain the cumulative confidence value; when the cumulative confidence value exceeds the set threshold, the new energy access area is in an island state at the current sampling time; S4. When in an islanded state, calculate the difference between the total output of new energy and the load power, and classify it into power surplus scenario and power shortage scenario; in the power surplus scenario, divide it into coarse adjustment and fine adjustment stages based on predetermined switching conditions; in the coarse adjustment stage, classify the levels based on the difference and put in the corresponding level of energy-consuming resistors; in the fine adjustment stage, calculate the switching safety margin based on the real-time status of the supercapacitor, and start the supercapacitor charging and discharging control when it is not less than the predefined safety margin threshold, and perform nonlinear switching of the supercapacitor charging power according to the supercapacitor start-up time; in the power shortage scenario, start the supercapacitor discharging.

[0008] Preferably, the process of constructing the elliptic decision equation in S3 is as follows: Collect grid-connected steady-state data samples, calculate the mean and standard deviation of the final residual voltage gradient of the grid-connected steady-state data samples, and use them as the mean and standard deviation of the grid-connected steady-state residual voltage gradient; calculate the mean and standard deviation of the voltage harmonic entropy of the grid-connected steady-state data samples, and use them as the mean and standard deviation of the grid-connected stable voltage harmonic entropy; use three times the standard deviation of the grid-connected steady-state residual voltage gradient as the major axis parameter of the elliptical decision boundary, and use three times the standard deviation of the grid-connected steady-state residual voltage gradient as the minor axis parameter of the elliptical decision boundary; An elliptic decision equation is constructed based on the mean of the grid-connected steady-state residual voltage gradient, the mean of the voltage harmonic entropy, and the major and minor axis parameters.

[0009] Preferably, the process of calculating the island confidence level of suspected isolated points in S3 is as follows: Substitute the final residual voltage gradient and voltage harmonic entropy calculated in real time at each sampling moment into the left side of the elliptic decision equation. When the result on the left side of the equation is not greater than 1, it is determined that the grid connection status is normal and the islanding action is not triggered. When the result on the left side of the equation is greater than 1, it is marked as a suspected island state; for any sampling time, the maximum value between 0 and the value on the left side of the equation minus 1 is taken as the island confidence level at the corresponding time.

[0010] Preferably, the process of correcting the island confidence level in S3 is as follows: The final residual voltage gradient and voltage harmonic entropy at each sampling moment are calculated in real time and substituted into the left side of the elliptic decision boundary equation. For any sampling moment, when the result on the left side of the equation at the corresponding moment is greater than 1 and the decrease of the current harmonic entropy at the current moment compared with the mean of the grid-connected steady-state current harmonic entropy exceeds the predefined decrease threshold, the product of the predefined enhancement coefficient and the island confidence at the corresponding sampling moment is updated to the island confidence at the corresponding sampling moment. Among them, the mean of the grid-connected steady-state current harmonic entropy is the mean of the current harmonic entropy of all samples in the grid-connected steady-state data sample.

[0011] Preferably, in S4, if the difference between the total output of new energy and the load power at the corresponding time is greater than 0, it belongs to the power surplus scenario; otherwise, it belongs to the power shortage scenario. The predetermined switching conditions are as follows: calculate in real time the deviation ratio of the effective voltage value of each phase relative to the rated voltage at the current moment. When the deviation ratio of any phase is greater than the predetermined voltage fluctuation rate, it is considered to be in the coarse adjustment stage; otherwise, it is in the fine adjustment stage.

[0012] Preferably, the process of classifying and activating energy-consuming resistors based on the difference in S4 is as follows: When the difference is less than the predetermined first-level threshold, the predefined first group of resistors is activated; when the difference is not less than the predetermined first-level threshold and is less than the predetermined second-level threshold, the predefined first and second groups of resistors are activated; when the difference is not less than the predetermined second-level threshold, all three groups of resistors are activated simultaneously.

[0013] Preferably, the process of calculating the switching safety margin in S4 is as follows: The remaining energy storage capacity (SOC), module temperature, terminal voltage, and measured capacity value of the supercapacitor are acquired in real time. The remaining energy storage capacity and terminal voltage are normalized to obtain the capacity availability and voltage availability, respectively. The absolute value of the difference between the module temperature and the reference temperature at the current sampling time is divided by the predefined maximum allowable temperature deviation, the negative number is taken, and 1 is added to obtain the temperature deviation factor. The ratio of the current measured capacity value of the supercapacitor to the rated capacity value is used to obtain the capacity retention rate. Multiply the current capacity availability, temperature deviation factor, voltage availability, and capacity retention rate to obtain the corresponding switching safety margin.

[0014] Preferably, the nonlinear switching process for the supercapacitor charging power in S4 is as follows: With the goal of minimizing voltage deviation, frequency deviation, power deviation, and power loss, the steady-state power reference value under the current supercapacitor startup time is solved within the feasible domain of the predefined supercapacitor power. The value of the predefined exponential function is calculated using the negative of the ratio of the supercapacitor startup time to the energy storage response time constant as the exponent. The negative of this value is added to 1 and multiplied by the steady-state power reference value to obtain the charging power reference value under the current supercapacitor startup time.

[0015] Preferably, the process of solving for the steady-state power reference value is as follows: The voltage deviation is defined as the ratio of the difference between the bus voltage and the rated reference voltage during the current supercapacitor startup time to the rated reference voltage. The frequency deviation is defined as the ratio of the difference between the frequency and the rated reference frequency during the current supercapacitor startup time to the rated reference frequency. The power tracking deviation is defined as the ratio of the difference between the charging power and the set reference power during the current supercapacitor startup time to the power reference value. The charging and discharging loss is defined as the square of the charging current at the corresponding charging power multiplied by the equivalent series resistance of the supercapacitor and then divided by the power reference value. The voltage deviation, frequency deviation, power tracking deviation, and charging and discharging loss are weighted by predefined weighting coefficients to obtain the optimization objective function. These predefined weighting coefficients are dynamically adjusted based on the real-time voltage deviation. During the solution process, two golden section points are selected within the feasible region of the supercapacitor power, and their values ​​are calculated by substituting them into the objective function. The function values ​​corresponding to the two golden section points are compared. If the function value of the first golden section point is less than that of the second golden section point, the power value corresponding to the second golden section point is used as the upper limit to update the feasible region interval. Otherwise, the power value corresponding to the first golden section point is used as the lower limit to update the feasible region interval. This process is repeated iteratively until the interval length is less than a predefined power amplitude threshold, and the midpoint of the interval is taken as the steady-state power reference value.

[0016] Preferably, the calculation process for the bus voltage and frequency under the current supercapacitor startup duration is as follows: Under the current supercapacitor startup time, the difference between the charging power and the steady-state power reference value of the previous rolling optimization cycle is multiplied by a predefined voltage-power sensitivity coefficient and then added to the steady-state value of the bus voltage of the previous rolling optimization cycle to obtain the corresponding bus voltage; the difference between the charging power and the steady-state power reference value of the previous rolling optimization cycle is multiplied by a predefined frequency-power sensitivity coefficient and then added to the steady-state value of the frequency voltage of the previous rolling optimization cycle to obtain the corresponding frequency.

[0017] A second aspect of the present invention provides a supercapacitor energy storage automatic switching device, which performs a supercapacitor energy storage automatic switching method, comprising: The data acquisition module collects the instantaneous values ​​of the three-phase voltage and current of the bus at the common connection point between the new energy access and the grid connection, and calculates the effective values ​​of the voltage and current. The feature extraction module calculates the residual voltage gradient and performs moving average filtering when the effective values ​​of the three-phase voltages are all lower than the predefined voltage thresholds within a predetermined duration. It takes the maximum absolute value of the residual voltage gradient after filtering all phases as the final residual voltage gradient. It also performs fast Fourier transform on the bus voltage and current signals collected within the predefined sliding window to calculate the voltage harmonic entropy and current harmonic entropy. The islanding state determination module constructs an elliptic decision equation and marks suspicious islanding points based on the final residual voltage gradient and harmonic entropy. It calculates the islanding confidence of the suspicious islanding points and, combined with the current harmonic entropy, corrects the islanding confidence when a predetermined determination condition is met. It multiplies the islanding confidence of each sampling time within the cumulative window length before the corresponding sampling time by the time interval and then accumulates them to obtain the cumulative confidence value. When the cumulative confidence value exceeds a set threshold, the new energy access area is in an islanding state at the current sampling time. The automatic energy storage switching module, when in an islanded state, calculates the difference between the total output of new energy sources and the load power, and classifies it into power surplus and power shortage scenarios. In the power surplus scenario, based on predetermined switching conditions, it divides the process into coarse adjustment and fine adjustment stages. In the coarse adjustment stage, it classifies the levels based on the difference and puts in the corresponding level of energy-consuming resistors. In the fine adjustment stage, based on the real-time status of the supercapacitor, it calculates the switching safety margin. When the safety margin threshold is not less than the predefined safety margin threshold, it starts the supercapacitor charging and discharging control and performs nonlinear switching of the supercapacitor charging power according to the supercapacitor start-up time. In the power shortage scenario, it starts the supercapacitor discharging.

[0018] A third aspect of the present invention provides a terminal, including a processor and a storage medium; The storage medium is used to store instructions; The processor is configured to operate according to the instructions to execute the steps of a supercapacitor energy storage automatic switching method.

[0019] A fourth aspect of the invention provides a computer-readable storage medium having a computer program stored thereon that, when executed by a processor, implements the steps of an automatic switching method for supercapacitor energy storage.

[0020] Compared with the prior art, the beneficial effects of the present invention include at least the following: 1. This invention constructs a multi-feature islanding determination mechanism by combining residual voltage gradient, voltage harmonic entropy and current harmonic entropy, and uses elliptical decision boundaries to screen abnormal operating points. Then, it corrects the confidence level by accumulating islanding confidence and changing current harmonic entropy. This can reduce misjudgments and omissions caused by fluctuations in a single electrical feature, and improve the reliability and continuous discrimination capability of islanding status identification in new energy grid-connected areas.

[0021] 2. This invention addresses the difference between the total output of new energy sources and the load power by further dividing the power surplus state into coarse adjustment and fine adjustment stages. It quickly absorbs large power surpluses by tiered deployment of energy-consuming resistors, and then uses supercapacitors for fine power regulation. At the same time, it calculates the switching safety margin by comprehensively considering the supercapacitor's SOC, terminal voltage, module temperature, and capacity retention rate. The supercapacitor is only activated when safety conditions are met, thereby reducing the operational risks caused by forced switching when the energy storage state is abnormal or the operating conditions are unsuitable.

[0022] 3. This invention determines the steady-state power reference value by simultaneously considering bus voltage deviation, frequency deviation, power tracking deviation, and supercapacitor charging and discharging losses, and optimizes it within the power feasible region using the golden section search. Furthermore, it adopts nonlinear power switching based on the supercapacitor startup time, so that the charging power gradually approaches the steady-state power reference value during the startup process, avoiding drastic fluctuations in bus voltage, frequency, and energy storage device caused by power abrupt changes, and improving the dynamic stability during automatic switching. Attached Figure Description

[0023] Figure 1 A flowchart of an automatic switching method for supercapacitor energy storage provided by the present invention; Figure 2 This is a timing diagram of the inertia layered replenishment power balance in an embodiment of the present invention; Figure 3 This is a schematic diagram of the structure of an automatic switching device for supercapacitor energy storage provided by the present invention. Detailed Implementation

[0024] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of this invention. The embodiments described in this application are merely some embodiments of this invention, and not all embodiments. Based on the spirit of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the protection scope of this invention.

[0025] Example 1 Embodiment 1 of the present invention provides an automatic switching method for supercapacitor energy storage, see reference. Figure 1 This includes the following steps: In scenarios with a high proportion of renewable energy connections, a specific coupling relationship exists between the residual voltage amplitude of the bus after a fault and the harmonic distortion rate. In grid-connected mode, the voltage gradient is small, and the harmonic entropy is high and stable (uniformly distributed); in islanded mode, the voltage gradient is large, and the harmonic entropy is low and fluctuating (concentrated in low-order harmonics). This embodiment defines this as the residual voltage-harmonic coupling effect and uses it as the physical basis for islanding detection. The discovery of this effect breaks through the limitations of traditional single electrical quantity detection and provides a theoretical basis for constructing a high-accuracy dual-dimensional detection algorithm.

[0026] S1. Collect the instantaneous values ​​of the three-phase voltage and current of the bus at the common connection point of the new energy access and grid connection, and calculate the effective values ​​of voltage and current.

[0027] The instantaneous three-phase (A, B, C phases) voltage values ​​uA(t), uB(t), uC(t) and three-phase current values ​​iA(t), iB(t), iC(t) of the 10kV bus at the common connection point between the renewable energy source and the grid side are synchronously acquired at a sampling frequency of 20kHz. High-frequency noise is removed by a second-order Butterworth low-pass filter (cutoff frequency 2kHz). The effective values ​​of voltage and current are calculated using a recursive method within a calculation window of one power frequency cycle (20ms).

[0028] Suppose the instantaneous voltage sequence of a certain phase of the bus after passing through a second-order Butterworth low-pass filter is: (n=1,2,…), sampling frequency Number of sampling points within the power frequency cycle The effective voltage value is calculated using the recursive method of moving average. Its recursive formula is: ; In the formula, This represents the effective voltage value at sampling time k. Number the current sampling time; For the first The effective value of the voltage at any given time; It is the recursive accumulation of the sum of squares of instantaneous voltage values ​​within a sliding window of length N; Let be the instantaneous voltage value at time k, belonging to any one of the phases uA(k), uB(k), and uC(k); The instantaneous voltage value before N sampling points (i.e., the data to be removed at the end of the sliding window); The number of sampling points contained in one cycle of the power frequency (N=400 in this embodiment); In this embodiment, the effective value of the three-phase current The same recursive formula is used for independent calculation; compared with the conventional point-by-point summation method, this recursive method can reduce the amount of computation for each effective value calculation from O(N) to O(1), which meets the real-time calculation requirements at a sampling rate of 20kHz.

[0029] S2. When the effective values ​​of the three-phase voltages are all lower than the predefined voltage thresholds within the predetermined duration period, calculate the residual voltage gradient and perform moving average filtering. Take the maximum absolute value of the residual voltage gradient after filtering of all phases as the final residual voltage gradient. Perform fast Fourier transform on the bus voltage and current signals collected within the predefined sliding window to calculate the voltage harmonic entropy and current harmonic entropy.

[0030] When the effective values ​​of the three-phase voltages are all lower than the predefined voltage threshold (80% of the rated voltage in this embodiment) within a predetermined duration period (300ms in this embodiment), the dual-dimensional detection process is triggered. In this embodiment, this condition is only used as a start-up criterion to reduce the amount of calculation during normal operation, and is not used as the final islanding determination basis.

[0031] Furthermore, when the dual-dimensional detection process is triggered, for each phase voltage RMS value at each sampling time, the difference between the corresponding phase voltage RMS values ​​at adjacent sampling times is calculated and divided by the sampling interval to obtain the corresponding residual voltage gradient, reflecting the instantaneous rate of voltage change; specifically expressed as: ; in, The residual pressure gradient of the phase corresponding to the k-th sampling time; This represents the effective voltage value of the phase corresponding to the k-th sampling time. The sampling interval; To suppress noise interference, a moving average is used to filter the calculation results. For each sampling time, with the corresponding time as the endpoint, the mean of the residual pressure gradients at all sampling times within a predefined moving window length is calculated as the filtered residual pressure gradient at the corresponding sampling time. Specifically, it is expressed as follows: ; In the formula, The residual pressure gradient after filtering at the k-th sampling time; The length of the sliding window is a predefined number of sampling times within the sliding window; in this embodiment, it is set to 10. The maximum absolute value of the residual pressure gradient after filtering all phases is taken as the final residual pressure gradient; this residual pressure gradient value is used in the subsequent elliptical decision boundary equation, thus making it sensitive to both single-phase and three-phase faults.

[0032] Using a predetermined sliding time interval (1024 sampling times in this embodiment) as the sliding window length, a Fast Fourier Transform (FFT) is performed on the instantaneous value sequences of three-phase voltage and three-phase current collected within the sliding window to extract the harmonic amplitude at each predetermined harmonic order (2nd to 15th in this embodiment). For the bus voltage and current signals, the energy proportion of each harmonic of different physical quantities is calculated. Specifically, it is expressed as follows: ; In the formula, , These represent the energy percentages of the j-th harmonics of the bus voltage and current signals, respectively. , These are the average bus voltage amplitude and average current amplitude of the j-th harmonic, respectively, which are the average values ​​of the three bus voltage / current amplitudes of the j-th harmonic. Based on the energy proportions of different harmonics of various physical quantities, the voltage harmonic entropy and current harmonic entropy are calculated separately; specifically expressed as follows: ; In the formula, , These are the voltage harmonic entropy and the current harmonic entropy, respectively. The larger the harmonic entropy, the more uniform the distribution of harmonic energy; the smaller the harmonic entropy, the more concentrated the harmonic energy is in certain harmonics.

[0033] S3. Based on the final residual pressure gradient and harmonic entropy, construct the elliptic decision equation and mark the suspicious island points; calculate the island confidence of the suspicious island points, and combine it with the current harmonic entropy to correct the island confidence when the predetermined judgment conditions are met; multiply the island confidence of each sampling time within the cumulative window length before the corresponding sampling time by the time interval and then accumulate it to obtain the cumulative confidence value; when the cumulative confidence value exceeds the set threshold, the new energy access area is in an island state at the current sampling time.

[0034] The feature cluster centers and major and minor axis parameters of the ellipse in grid-connected operation are calculated using offline training data; where the cluster centers are the mean of the final residual pressure gradient and harmonic entropy during grid-connected operation, the major axis parameter is the 3σ boundary in the direction of the residual pressure gradient, and the minor axis parameter is the 3σ boundary in the direction of the harmonic entropy. In this embodiment, the final residual pressure gradient is calculated at each sampling time, and its time label is the sampling time k. The harmonic entropy is calculated using an FFT sliding window, and its time label is the sampling time at the end of the FFT sliding window. When the FFT sliding window slides with a step size of 1 sampling point, the harmonic entropy is also output point by point with the same step size. Thus, the harmonic entropy and the residual pressure gradient are aligned one-to-one according to the time label to form two-dimensional feature points (g(k), Hu(k)), realizing the synchronous mapping of time-frequency dual-domain features and the unified calculation of the elliptic criterion.

[0035] Suppose that there are M groups of grid-connected steady-state data samples collected during offline training, and the final residual voltage gradient and voltage harmonic entropy of the i-th sample are respectively... , Then calculate the mean and standard deviation of the final residual pressure gradient of the grid-connected steady-state data samples, and use them as the mean and standard deviation of the grid-connected steady-state residual pressure gradient. Calculate the mean and standard deviation of voltage harmonic entropy for the grid-connected steady-state data samples, and use them as the mean and standard deviation of grid-connected stable voltage harmonic entropy. Therefore, the major axis parameters of the elliptical decision boundary are calculated and determined. (3σ boundary along the residual pressure gradient direction) and minor axis parameter (3σ boundary in the direction of harmonic entropy). According to the statistical 3σ criterion, this boundary can cover 99.73% of the grid-connected steady-state samples, and feature points outside the boundary are identified as suspicious island points.

[0036] Based on the cluster centers, residual voltage gradient, and 3σ boundary along the harmonic entropy direction during grid-connected operation, an elliptic decision equation is constructed; specifically, it is expressed as: ; In the formula, This represents the mean of the steady-state residual pressure gradient after grid connection. The mean value of harmonic entropy of grid-connected steady-state voltage; The major axis parameter represents the 3σ boundary along the residual pressure gradient direction in this embodiment. ; The standard deviation of the grid-connected steady-state residual pressure gradient; The minor axis parameter represents the 3σ boundary in the direction of harmonic entropy, as described in this embodiment. ; The standard deviation of the harmonic entropy of the grid-connected steady-state voltage; Substituting the final residual voltage gradient and voltage harmonic entropy calculated in real time at each sampling moment into the left side of the elliptic decision equation, if the result on the left side of the equation is not greater than 1, it indicates that the feature point at the corresponding sampling moment falls within the ellipse and its boundary, and the grid connection is judged to be normal, without triggering islanding action; if the result on the left side of the equation is greater than 1, it satisfies the condition. If the feature point at the corresponding sampling time falls outside the ellipse, it is considered to be in an isolated state and is marked as a suspected isolated state.

[0037] To further improve detection reliability and prevent false positives, this embodiment proposes a cumulative confidence mechanism; for any sampling time, the maximum value between 0 and the value on the left side of the equation minus 1 is taken as the island confidence level at the corresponding time; specifically expressed as: ; In the formula, Let be the island confidence score at the k-th sampling time. It reflects the degree to which the feature point deviates from the grid-connected area; the greater the deviation, the higher the confidence level. In this embodiment, the voltage harmonic entropy is used as the minor axis dimension (primary criterion) as the input to the elliptic decision boundary equation because the bus voltage harmonic distribution characteristics change most significantly when the renewable energy converter switches to voltage source mode during islanding. The current harmonic entropy is used as an auxiliary cross-validation criterion to enhance islanding confidence. Specifically: The final residual voltage gradient and voltage harmonic entropy at each sampling time are calculated in real time and substituted into the left side of the elliptic decision boundary equation. For any sampling time, when the result on the left side of the equation at the corresponding time is greater than 1 and the decrease in current harmonic entropy at the current time compared to the mean of grid-connected steady-state current harmonic entropy exceeds the predefined decrease threshold (i.e., 0.3), the product of the predefined enhancement coefficient and the island confidence at the corresponding sampling time is updated to the island confidence at the corresponding sampling time. The mean of grid-connected steady-state current harmonic entropy is the mean of the current harmonic entropy of all samples in the grid-connected steady-state data sample. The specific formula for the decrease is: ; In the formula, The decrease in current harmonic entropy at the k-th sampling time compared to the mean of grid-connected steady-state current harmonic entropy; The mean value of harmonic entropy of the grid-connected steady-state current; The current harmonic entropy at the k-th sampling time; The specific formula for updating the island confidence level is as follows: ; In the formula, The enhancement factor is predefined; in this embodiment, it is set to 1.5. Conversely, if the current harmonic entropy does not change synchronously, the original confidence level is maintained, thereby suppressing misjudgments caused by disturbances from a single harmonic source.

[0038] The island confidence score for each sampling time within the cumulative window length prior to the corresponding sampling time is multiplied by the time interval and then summed to obtain the cumulative confidence score. When the cumulative confidence score exceeds a set threshold, the island status is confirmed. Specifically, this is expressed as follows: ; in, The cumulative confidence level is M; the cumulative window length is M. The confidence accumulation threshold is set at 0.5 ms (typical value). This accumulation judgment mechanism ensures both the speed of detection (rapid accumulation when there is a serious deviation) and the reliability of detection (no false positives for brief deviations), achieving an optimal balance between detection speed and accuracy.

[0039] In this embodiment, before the system is put into operation, 72 hours of continuous operating data are collected under different operating conditions (strong / weak grid, different load rates, different renewable energy output levels). Cluster analysis is used to determine the grid connection feature center and elliptical boundary parameters. During normal grid connection operation, the feature center and boundary parameters are updated hourly using a sliding window method (window length 24 hours) to ensure that the feature library adapts to the slow changes in the grid operation mode. The grid connection feature center and elliptical boundary parameters are stored in the form of parameter vectors, which has a small data volume and is convenient for storage and updating.

[0040] In the power balancing process of a power system, different types of equipment essentially provide inertia support at different time scales. Accordingly, this embodiment abstracts the equipment in power balancing into a fast inertia layer and a fine inertia layer. The fast inertia layer is provided by energy-consuming resistors, offering fast response (sub-millisecond) and high power handling capacity, but with low adjustment accuracy and limited to power consumption. The fine inertia layer is provided by supercapacitors, offering relatively fast response (millisecond), charge and discharge capability, and high adjustment accuracy, but with high unit capacity cost. Based on this, this embodiment proposes a layered inertia-based power balancing control strategy: first, the fast inertia layer (resistors) is used to quickly smooth out major power fluctuations, and then the fine inertia layer (capacitors) is used for precise power calibration. The two work together to achieve an optimal balance between speed and accuracy.

[0041] In this embodiment, the total capacity of the energy-consuming resistors P n Based on a configuration of 30% to 50% of the maximum output of new energy sources, it is divided into three levels, as shown in Table 1.

[0042] Table 1 Total Capacity Configuration of Energy-Consuming Resistors

[0043] The three-level ratio of 20%, 30%, and 50% is based on the optimized design of the statistical characteristics of power distribution; that is, the probability of mild over-excess is the highest, so the small capacity group is used to improve the regulation accuracy; the probability of severe over-excess is low but the harm is great, so the large capacity group is used to ensure safety.

[0044] S4. When in an islanded state, calculate the difference between the total output of new energy and the load power, and classify it into power surplus scenario and power shortage scenario; in the power surplus scenario, divide it into coarse adjustment and fine adjustment stages based on predetermined switching conditions; in the coarse adjustment stage, classify the levels based on the difference and put in the corresponding level of energy-consuming resistors; in the fine adjustment stage, calculate the switching safety margin based on the real-time status of the supercapacitor, and start the supercapacitor charging and discharging control when it is not less than the predefined safety margin threshold, and perform nonlinear switching of the supercapacitor charging power according to the supercapacitor start-up time; in the power shortage scenario, start the supercapacitor discharging.

[0045] When the renewable energy access area is in an isolated state at the current sampling time, calculate the difference between the total renewable energy output and the load power at the corresponding time; specifically expressed as: ; In the formula, This represents the difference between the total output of new energy sources and the load power at the k-th sampling time. This represents the total output of new energy sources at the k-th sampling time. The load power at the k-th sampling time; If the difference between the total output of new energy sources and the load power is greater than 0 at a given time, it indicates a power surplus scenario; see reference. Figure 2 In this scenario, execution should strictly follow the sequence of fast first, then slow, and coarse first, then fine: Real-time calculation of the effective voltage value of each phase at the current moment Relative to rated voltage deviation percentage When the deviation ratio of any phase is greater than the predetermined voltage fluctuation rate (0.03 in this embodiment), it is considered to be in the coarse adjustment stage, i.e., the first stage (usually within 15ms from the moment of islanding confirmation). Based on the difference between the total output of new energy and the load power, the levels are divided and the corresponding energy-consuming resistors are put into operation. Specifically, when the difference is less than the predetermined first level threshold, it indicates that the power is slightly excessive (i.e., ΔP < 20%Pn), and only the first group of resistors is put into operation; when the difference is not less than the predetermined first level threshold and less than the predetermined second level threshold, it indicates that the power is moderately excessive (i.e., 20%Pn ≤ ΔP < 50%Pn), and the first and second groups of resistors are put into operation; when the difference is not less than the predetermined second level threshold, it indicates that the power is severely excessive (i.e., ΔP ≥ 50%Pn), and all three groups of resistors are put into operation simultaneously. In this embodiment, in the first stage, rapid inertia support is performed (t=0~15ms). t=0ms: the instant islanding is confirmed, the corresponding level of energy-consuming resistor is immediately put into operation; t=0~15ms: the resistor quickly absorbs the main power deficit, and the bus voltage drops rapidly from the overvoltage state. The core objective of this stage is to bring the voltage back to the allowable range in the shortest possible time to prevent overvoltage protection from activating.

[0046] Furthermore, when the deviation ratio of each phase is not greater than the predetermined voltage fluctuation rate (0.03 in this embodiment), the switching safety margin is calculated based on the real-time status of the supercapacitor. When the switching safety margin is less than the predefined safety margin threshold (0.3 in this embodiment), the supercapacitor is locked out and only coarse adjustment is performed by the energy-consuming resistor (i.e., it is still in the first stage). When the switching safety margin is not less than the predefined safety margin threshold, it is considered to be in the second stage (usually after t=15ms), and the supercapacitor charging and discharging control is started, smoothly entering fine adjustment according to the nonlinear trajectory. The two achieve layered coordination in terms of time scale and response characteristics, avoiding the superposition of capacitive inrush current and resistive impact. The supercapacitor charging power is nonlinearly switched according to the supercapacitor start-up time and sent to the energy storage converter as a power command. The energy storage converter adopts a dual closed-loop control structure with an inner current loop and an outer voltage loop. The outer voltage loop calculates the desired output current using the nonlinear switching result as a reference value, while the inner current loop controls the actual output current to quickly track the desired value. The actual output power of the supercapacitor continuously and accurately compensates the bus voltage under command, further stabilizing the voltage from the coarsely adjusted ±3% range to within ±1%.

[0047] In this embodiment, fine inertia adjustment is performed in the second stage (after t=15ms). Around t=15ms (when the bus voltage fluctuation rate drops to within ±3%): the supercapacitor charging and discharging control is initiated; the supercapacitor charging power is gradually increased from 0 in a ramp manner, with a slope not exceeding 0.1Pn / ms, to avoid capacitive inrush current impact; the supercapacitor is responsible for compensating for the remaining power difference after resistor absorption, precisely controlling the bus voltage near the rated value; the core objective of this stage is to achieve precise power balance and maintain stable system operation. This layered and incremental control strategy fundamentally avoids the superposition effect of capacitive inrush current and resistive impact in parallel operation mode, and the transient overvoltage can be reduced from 1.42pu to 1.08pu, a reduction of 24%.

[0048] As a preferred implementation method, the process of calculating the switching safety margin is as follows: The system acquires the supercapacitor's remaining energy storage capacity (SOC), module temperature, terminal voltage, and measured capacity value in real time. The remaining energy storage capacity and terminal voltage are normalized to obtain capacity availability and voltage availability, respectively. The temperature deviation factor is obtained by dividing the absolute value of the difference between the module temperature and the reference temperature at the current sampling moment by a predefined maximum allowable temperature deviation, taking the negative of the result, and adding 1. The capacity retention rate is obtained by calculating the ratio of the supercapacitor's current measured capacity value to its rated capacity value. The specific formula is as follows: ; In the formula, Capacity availability at sampling time k; This represents the temperature deviation factor at sampling time k. The voltage availability at sampling time k; The capacity retention rate at sampling time k; The remaining energy storage capacity at sampling time k; , In this embodiment, the upper and lower limits of the remaining energy storage capacity that can be allowed to operate are taken as 15% and 100%, respectively. The module temperature at sampling time k; For reference temperature, 25°C is used in this embodiment; The maximum allowable temperature deviation is predefined; in this embodiment, it is set to 35°C. The terminal voltage at sampling time k; , These are the upper and lower limits of the permissible terminal voltage, respectively; This represents the measured capacity value at sampling time k. This refers to the rated capacitance value of the supercapacitor. Multiply the current capacity availability, temperature deviation factor, voltage availability, and capacity retention rate to obtain the corresponding switching safety margin; the specific formula is: ; In the formula, The switching safety margin at sampling time k enables the switching decision to no longer rely on a fixed power threshold, but to dynamically adjust the degree of participation based on the real-time state of the energy storage itself.

[0049] As a preferred implementation method, the process of nonlinear switching based on the supercapacitor startup time is as follows: With the goal of minimizing voltage deviation, frequency deviation, power deviation, and power loss, the steady-state power reference value under the current supercapacitor startup time is solved within the feasible region of the predefined supercapacitor power. The value of a predefined exponential function is calculated using the negative of the ratio of the supercapacitor startup time to the energy storage response time constant as the exponent. The negative of this value is then added to 1 and multiplied by the steady-state power reference value to obtain the charging power reference value under the current supercapacitor startup time. The specific formula is as follows: ; In the formula, The charging power reference value is calculated from the start-up time of the supercapacitor over time t, so that the power of the supercapacitor increases rapidly in the early stage and gradually approaches the steady state value in the later stage, which complements the linear and rapid connection of the energy-consuming resistor. The steady-state power reference value is obtained by the rolling optimization solution; This refers to the startup time of the supercapacitor. The energy storage response time constant is fixed at 0.8ms in this embodiment, which determines the rate of power increase. the larger it is, the slower the rise).

[0050] If the difference between the total output of new energy sources and the load power at the corresponding moment is not greater than 0, it falls into a power shortage scenario, where it is unnecessary to put the energy-consuming resistors into operation, and the supercapacitor is directly started to discharge to provide inertia support; the reference value of the discharge power is: ; The discharge power of the supercapacitor is also applied in a ramp manner (i.e., it increases from 0 in a ramp manner, with a slope not exceeding 0.1Pn / ms), so as to prevent the impact of sudden current changes on the equipment; When the following conditions are met: the difference between the total output of new energy sources and the load power at each moment within a continuous predetermined period is less than the set threshold (i.e., ΔP<P0 and lasts for 200ms, where P0=5%Pn in this embodiment), the SOC of the supercapacitor reaches the upper limit SOCmax (i.e., charging cannot be continued), the backup line is successfully closed (i.e., the system resumes grid-connected operation), and the bus voltage at each sampling moment within a continuous 300ms is maintained at ±5% of the rated voltage value, the energy-consuming resistors are disconnected in stages in descending order (Group 3 → Group 2 → Group 1), with an interval of 10ms between each stage.

[0051] In this embodiment, the energy-consuming resistors are disconnected in sequence according to the order of Group 3 → Group 2 → Group 1. After each group is disconnected, wait for 10ms (about 0.5 power frequency cycle) for the bus voltage response to complete and the system to enter a new steady state before disconnecting the next group, so as to avoid closing inrush current or voltage drop caused by instantaneous power mutation; the complete disconnection process of the three groups of energy-consuming resistors takes about 30ms in total.

[0052] As an optional embodiment, when the bus voltage exceeds 1.15p.u., overvoltage protection is performed, that is, the charging of the supercapacitor is forcibly blocked, and all the energy-consuming resistors are put into operation stage by stage; when the SOC of the supercapacitor is lower than 10%, low SOC protection is performed, that is, the discharge is blocked, and only charging is allowed; when the resistor temperature exceeds 120°C, resistor overheating protection is performed, that is, the corresponding group is blocked to prevent overheating damage; when the number of switching operations of a single group of resistors exceeds 30 times per hour, frequent operation protection is performed, that is, the group is temporarily blocked for 10 minutes.

[0053] As a preferred embodiment, the optimization solving algorithm is executed cyclically at a period of 10~50ms, and the steady-state power reference value is updated in a rolling manner; To minimize voltage deviation, frequency deviation, power deviation, and power loss, an optimization objective function is constructed. The voltage deviation is defined as the ratio of the difference between the bus voltage and the rated reference voltage during the current supercapacitor startup duration to the rated reference voltage. The frequency deviation is defined as the ratio of the difference between the frequency and the rated reference frequency during the current supercapacitor startup duration to the rated reference frequency. The power tracking deviation is defined as the ratio of the difference between the charging power and the set reference power during the current supercapacitor startup duration to the power reference value. The charging and discharging loss is defined as the square of the charging current at the corresponding charging power multiplied by the equivalent series resistance of the supercapacitor and divided by the power reference value. Using predefined weighting coefficients, the voltage deviation, frequency deviation, power tracking deviation, and charging and discharging loss are weighted to obtain the optimization objective function, specifically expressed as follows: ; in, The objective function is the optimization function under the current supercapacitor startup time t; the first term is the voltage deviation penalty term, which has the highest priority and is dynamically adjusted. This is a weighting coefficient for the degree of voltage deviation; This represents the bus voltage during the current supercapacitor startup time t. The first item is the rated reference voltage; the second item is the frequency deviation penalty item, which has the second highest priority and is dynamically adjusted. The weighting coefficient for the degree of frequency deviation; The frequency is the current supercapacitor startup time t. The first item is the rated reference frequency; the second item is the power tracking deviation penalty, which has medium priority. Weighting coefficients for the degree of power tracking deviation; This represents the charging power during the current supercapacitor startup time t. The set reference power; The first item is the power reference value, used to normalize the power deviation; the fourth item is the charge / discharge loss penalty item, with the lowest priority. Weighting coefficients for the degree of charging and discharging losses; For charging current, This is the terminal voltage of the supercapacitor; This is the equivalent series resistance of the supercapacitor; Furthermore, the bus voltage and frequency under the current supercapacitor startup time change with the charging power; when the charging power increases, the power absorbed by the bus decreases and the voltage rises; conversely, the voltage decreases; within a single rolling optimization cycle, the relationship between the bus voltage and frequency and the charging power is linearized respectively. Given the current supercapacitor startup time, the difference between the charging power and the steady-state power reference value of the previous rolling optimization cycle is multiplied by a predefined voltage-power sensitivity coefficient, and then added to the steady-state bus voltage value of the previous rolling optimization cycle to obtain the corresponding bus voltage; the difference between the charging power and the steady-state power reference value of the previous rolling optimization cycle is multiplied by a predefined frequency-power sensitivity coefficient, and then added to the steady-state frequency voltage value of the previous rolling optimization cycle to obtain the corresponding frequency; specifically expressed as: ; In the formula, This is the steady-state value of the bus voltage in the previous rolling optimization cycle, that is, the bus voltage corresponding to the steady-state power reference value obtained in the previous rolling optimization cycle; The voltage-power sensitivity coefficient is determined by impedance and load characteristics and can be obtained through power flow calculation (Jacobi matrix elements in the Newton-Raphson method) or short-circuit capacity estimation (i.e., the ratio of equivalent impedance to rated reference voltage). This is the steady-state power reference value from the previous rolling optimization cycle; This is the steady-state frequency value of the previous rolling optimization cycle, that is, the frequency corresponding to the steady-state power reference value obtained in the previous rolling optimization cycle; The frequency-power sensitivity is determined by the system's total inertial time constant and damping coefficient, and is usually taken as the standard droop slope of the power grid.

[0054] As a preferred implementation, the predefined weight coefficients of each optimization objective are dynamically adjusted according to the real-time voltage deviation, achieving intelligent optimization decision-making that prioritizes voltage preservation when the voltage deviation is large, while also considering economic efficiency after the voltage stabilizes; specifically, based on the degree of voltage deviation... The dynamic adjustment formula for the weights of each objective is as follows: ; in, The target weight for voltage stability is set with a base value of 100, which increases rapidly with voltage deviation (slope 500). The target weight for frequency support is set at a base value of 50, which increases moderately with voltage deviation (slope 100). The target weight for power balance is kept constant at 10; The target weight for minimizing loss is fixed at a baseline value of 1.

[0055] As shown in Table 2, when the voltage deviation is 0 (ideal operating condition), the weighting ratio is kU:kf:kP:kloss = 100:50:10:1; when the voltage deviation is 10%, the weighting ratio becomes kU:kf:kP:kloss = 150:60:10:1, and the voltage priority is significantly improved.

[0056] Table 2 Weight distribution under typical operating conditions

[0057] This adaptive variable weight mechanism enables the optimizer to "perceive" the system state, automatically concentrate computing resources on the most urgent optimization objectives, and realize intelligent scheduling of multi-objective optimization.

[0058] Further, during the solving process, the feasible region of the supercapacitor power is determined , which is determined by the rated power of the equipment and the SOC limit; the golden section points (0.618 and 0.382 quantiles) are used to calculate the objective function value; the function values of the two quantiles are compared to narrow the interval range; the iteration is repeated until the interval length is less than the set precision (0.1% of the rated power); the midpoint of the interval is taken as the optimal power reference value as the steady-state power reference value. Specifically, two golden section points are taken in the feasible region and , P1 and P2 are respectively substituted into the optimization objective function to calculate the function values; the magnitudes of J(P1) and J(P2) are compared. If J(P1)<J(P2), the optimal point is in the interval, let ; otherwise let ; the iteration is repeated until the interval length is less than the predefined power amplitude threshold (i.e., 0.1% of the rated power), and the midpoint of the interval is taken as the optimal power reference value.

[0059] As a preferred embodiment, within a predefined detection window (i.e., 500ms) starting from the current moment, it is determined whether at each moment, the active power difference between the backup line and the busbar is not greater than a predetermined power difference threshold (i.e., |ΔP|≤5kW), the voltage deviation degree is not greater than a predetermined voltage deviation threshold (i.e., ±0.05), and the difference between the current frequency and the rated reference frequency is not greater than a predefined frequency deviation threshold (i.e., ±0.2Hz); when the conditions are met, the backup line closing process is started; Pre-synchronization control is performed before closing to adjust that the voltage amplitude difference on the backup line side (or the busbar side) is not greater than a predetermined difference threshold (i.e., 0.02), the phase difference on both sides of the line closing is not greater than a predetermined phase difference threshold (i.e., 5°), and the frequency difference on both sides is not greater than a predetermined frequency difference threshold (i.e., 0.1Hz); After the pre-synchronization conditions are met, a closing command is issued to close the BRK2 circuit breaker, and the system is switched to the backup line for power supply. Further, a single verification, that is, the closing inrush current constraint, is performed within a time window of about 100ms before closing, so as to control the closing inrush current within 5% of the rated current and realize a truly undisturbed switching.

[0060] After the closing is successful, the energy-consuming resistor and the supercapacitor are gradually withdrawn according to the hierarchical strategy, and the system returns to the normal grid-connected operation state.

[0061] This embodiment uses a 10kV distribution network test node under a 750kV substation to conduct a comprehensive field test verification of the method provided in this embodiment. The photovoltaic penetration rate of the test node is approximately 62%, and it is equipped with a 2MW / 0.5MWh supercapacitor energy storage system and a 1.5MW graded energy dissipation resistor device.

[0062] Example 2 Embodiment 2 of the present invention provides an automatic switching device for supercapacitor energy storage, see reference. Figure 3 This includes: automatic switching control cabinet, graded energy-consuming resistor cabinet and supercapacitor energy storage module cabinet; Each of the three switchgear cabinets has 10kV cable inlet and outlet terminals at the bottom, which converge via cable trays and three-core armored power cables to the cable trench above the 10kV busbar of the substation (i.e., the PCC common connection point). The automatic switching control cabinet contains a main line circuit breaker BRK1 and a standby line circuit breaker BRK2. BRK1 on the main line side is connected to the main 10kV line input via an upper primary cable, and BRK2 on the standby line side is connected to the standby 10kV line input via an upper primary cable. The lower primary cables of both circuit breakers converge into the PCC busbar. The first, second, and third groups of resistors in the graded energy dissipation resistor cabinet are connected to the PCC busbar in parallel via vacuum contactors K1, K2, and K3, respectively. The bidirectional DC / AC converter of the supercapacitor energy storage module cabinet has its AC side connected to the PCC busbar via circuit breaker BRK3 and YJV22 power cable, and its DC side connected to the supercapacitor module array via tin-plated copper busbars.

[0063] The three-sided control cabinets are connected by KVVP22-7×1.5 shielded control cables, with DB37 aviation plugs at both ends of the control cables. The automatic switching control cabinet acts as the main control node, sending switching commands to K1 / K2 / K3 in the graded energy-consuming resistor cabinet, sending power reference value commands to the bidirectional DC / AC converters in the supercapacitor energy storage module cabinet, and simultaneously receiving feedback signals such as temperature, current, SOC, and converter status in each cabinet.

[0064] BRK1 is the main power supply line circuit breaker (closes under normal operation, trips after islanding confirmation); BRK2 is the standby line circuit breaker (opens under normal operation, closes after pre-synchronization when closing without disturbance); BRK3 is the energy storage converter output circuit breaker; K1 / K2 / K3 are grouped switching vacuum contactors for energy-consuming resistors (opening / closing time ≤20ms, rated current 1250A); the solid-state switch assembly consists of high-speed IGBTs and a driver board, used for millisecond-level fast breaking; the pre-synchronization control unit is used for three-level pre-synchronization adjustment of voltage amplitude, phase, and frequency before closing; the DSP+FPGA in the integrated protection and control device undertakes all algorithm calculations and criterion decisions; the bidirectional DC / AC converter (2MW, efficiency ≥96%) undertakes the charging and discharging power conversion of the supercapacitor.

[0065] The automatic switching control cabinet is centrally located, with the graded energy-consuming resistor cabinet on one side and the supercapacitor energy storage module cabinet on the other side. These three cabinets are arranged side-by-side in the substation's secondary equipment room or local equipment room, with maintenance access of at least 800mm at both the front and back. The cabinet frame uses a welded structure of profiles and cold-rolled steel sheet skin. The bottom of the cabinet is fixed to the cable trench foundation channel steel with four M16 pre-embedded bolts, and the top is equipped with four M20 lifting rings for easy overall lifting and transportation. Cable entry and exit points are located at the top and bottom of the cabinet, with brush-sealed plates at the entrances and exits to prevent small animals from entering.

[0066] As a preferred implementation, the automatic switching control cabinet is constructed from thick cold-rolled steel plate, welded together, with an outer surface treated with electrostatic powder coating (RAL7035) and an IP41 protection rating. The cabinet features double doors at both the front and back, equipped with mechanical anti-misoperation locking mechanisms, and grounding braided straps are added to the door hinges. The cabinet body is vertically divided into three layers: upper, middle, and lower. The upper layer, the electrical instrumentation layer, houses the integrated protection and control device and the human-machine interface unit. The integrated protection and control device uses a front-panel plug-in structure, containing a digital signal processing board, a programmable logic device board, an analog signal acquisition board (16-channel synchronous sampling), and a communication interface board (IEC61850, Modbus-RTU). All boards are connected via a backplane VPX bus. The human-machine interface unit, consisting of a color LCD touchscreen, status indicator lights, and operation buttons, is installed on the front panel of the cabinet.

[0067] The method described in Example 1 operates entirely within an automatic switching control cabinet and is implemented through a comprehensive protection and control device; specifically, it includes: a data acquisition module, a feature extraction module, an islanding status determination module, and an automatic energy storage switching module; The data acquisition module collects the instantaneous values ​​of the three-phase voltage and current of the bus at the common connection point between the new energy access and the grid connection, and calculates the effective values ​​of the voltage and current. The feature extraction module calculates the residual voltage gradient and performs moving average filtering when the effective values ​​of the three-phase voltages are all lower than the predefined voltage thresholds within a predetermined duration. It takes the maximum absolute value of the residual voltage gradient after filtering all phases as the final residual voltage gradient. It also performs fast Fourier transform on the bus voltage and current signals collected within the predefined sliding window to calculate the voltage harmonic entropy and current harmonic entropy. The islanding state determination module constructs an elliptic decision equation and marks suspicious islanding points based on the final residual voltage gradient and harmonic entropy. It calculates the islanding confidence of the suspicious islanding points and, combined with the current harmonic entropy, corrects the islanding confidence when a predetermined determination condition is met. It multiplies the islanding confidence of each sampling time within the cumulative window length before the corresponding sampling time by the time interval and then accumulates them to obtain the cumulative confidence value. When the cumulative confidence value exceeds a set threshold, the new energy access area is in an islanding state at the current sampling time. The automatic energy storage switching module, when in an islanded state, calculates the difference between the total output of new energy sources and the load power, and classifies it into power surplus and power shortage scenarios. In the power surplus scenario, based on predetermined switching conditions, it divides the process into coarse adjustment and fine adjustment stages. In the coarse adjustment stage, it classifies the levels based on the difference and puts in the corresponding level of energy-consuming resistors. In the fine adjustment stage, based on the real-time status of the supercapacitor, it calculates the switching safety margin. When the safety margin threshold is not less than the predefined safety margin threshold, it starts the supercapacitor charging and discharging control and performs nonlinear switching of the supercapacitor charging power according to the supercapacitor start-up time. In the power shortage scenario, it starts the supercapacitor discharging.

[0068] The color LCD touch screen is used for displaying operating parameters, querying event logs, and setting parameters; the lower terminal wiring layer provides external wiring for all high-voltage, low-voltage, and communication terminals.

[0069] The middle layer is the execution unit layer, housing solid-state switching components and a pre-synchronization control unit. The solid-state switching components consist of high-speed IGBT modules (rated voltage 1700V, rated current 1200A, inherent response time ≤1ms), a gate driver board, an aluminum alloy heat sink, and a forced-air cooling system. The IGBT modules are press-fitted onto the heat sink with thermally conductive silicone grease, and two independent centrifugal fans are located below the heat sink for forced cooling. The pre-synchronization control unit uses a separate relay-protected chassis and is connected to the integrated protection and control device via a backplane fiber optic cable.

[0070] The lower layer is the terminal wiring layer, which houses the high-voltage power input / output terminal blocks, low-voltage signal terminal blocks, communication terminal blocks, and grounding copper busbars. The high-voltage terminal blocks use UK10N type rail-mounted terminals, while the low-voltage terminal blocks use UK3N type, arranged along DIN rails. A main copper grounding busbar is located at the bottom of the cabinet, connected to the substation grounding grid via copper conductors. Four sets of axial flow fans are installed at the top of the cabinet, forming a forced ventilation channel with the top air inlet / outlet grilles.

[0071] The material and protection level of the tiered energy-consuming resistor cabinet are consistent with those of the automatic switching control cabinet. The cabinet is vertically divided into three independent resistor compartments by two thick steel plate partitions, housing the first, second, and third groups of alloy resistor elements from bottom to top. The capacities of the three groups are 20%, 30%, and 50% of the total rated power consumption, respectively (typical configurations are 300kW, 450kW, and 750kW). Each resistor compartment contains a ceramic tube skeleton supported by a stainless steel frame, with Cr20Ni80 alloy resistance wire evenly wound on the surface of the ceramic tube, allowing for short-term overload of 1.5 times / 5 minutes. Each resistor compartment is independently equipped with one vacuum contactor (rated current 1250A, opening and closing time ≤20ms), one set of KTY series temperature sensing elements, and one temperature control switch. The three contactors are connected to the signal terminals of the solid-state switch assembly in the middle layer of the automatic switching control cabinet via control cables inside the cabinet. The cabinet is equipped with four sets of axial flow exhaust fans on the top and an air inlet grille at the bottom. When the temperature inside the resistor compartment reaches 120℃, the corresponding set is activated by the temperature control switch.

[0072] The supercapacitor energy storage module cabinet adopts a drawer-type plug-in structure, arranged in 8 layers, with 6 independent drawers on each layer. Each drawer accommodates 4 strings of supercapacitor modules. The rated voltage of a single module is 2.7~3.8V, and the modules are connected in series and parallel to form a 750V DC energy storage unit. The front panel of the drawers is equipped with status indicator lights, emergency stop buttons, and latches. Aluminum alloy guide rails and limit latches are provided between the drawers to facilitate hot-swappable replacement of individual drawers during operation and maintenance. The modules are connected in series and parallel via tin-plated copper busbars, and M10 stainless steel bolts are used for fastening. An air inlet grille is located at the bottom of the cabinet, and two centrifugal fans are located at the top, forming a forced air cooling channel from bottom to top. A bidirectional DC / AC converter (2MW, efficiency ≥96%) is installed on the side and rear of the module cabinet. The converter is connected to the module DC busbar via an independent copper busbar and to the 10kV side via a YJV22 type power cable.

[0073] The automatic switching control cabinet is connected to the graded energy-consuming resistor cabinet via a YJV22-8.7 / 15-3×95 power cable, handling the main circuit power for resistor switching; it is bridged with the supercapacitor energy storage module cabinet via a tin-plated copper busbar, handling the main circuit for energy storage charging and discharging. Low-voltage signals between the cabinets are connected via KVVP22-7×1.5 shielded control cables. The connectors at both ends of the control cables use DB37 aviation plugs, providing protection against mis-insertion and quick-connect / remove functionality. Each of the three cabinets has a copper main grounding busbar at its bottom, which is then connected via copper conductors to the main grounding busbar in the substation's secondary equipment room, and finally to the substation's grounding grid.

[0074] This disclosure can be a system, method, and / or computer program product. A computer program product may include a computer-readable storage medium having computer-readable program instructions loaded thereon for causing a processor to implement various aspects of this disclosure.

[0075] Computer-readable storage media can be tangible devices capable of holding and storing instructions for use by an instruction execution device. Computer-readable storage media can be, for example—but not limited to—electrical storage devices, magnetic storage devices, optical storage devices, electromagnetic storage devices, semiconductor storage devices, or any suitable combination of the foregoing. More specific examples (a non-exhaustive list) of computer-readable storage media include: portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), static random access memory (SRAM), portable compact disc read-only memory (CD-ROM), digital multifunction disc (DVD), memory sticks, floppy disks, mechanical encoding devices, such as punch cards or recessed protrusions storing instructions thereon, and any suitable combination of the foregoing. The computer-readable storage media used herein are not to be construed as transient signals themselves, such as radio waves or other freely propagating electromagnetic waves, electromagnetic waves propagating through waveguides or other transmission media (e.g., light pulses through fiber optic cables), or electrical signals transmitted through wires.

[0076] The computer-readable program instructions described herein can be downloaded from computer-readable storage media to various computing / processing devices, or downloaded via a network, such as the Internet, local area network, wide area network, and / or wireless network, to an external computer or external storage device. The network may include copper transmission cables, fiber optic transmission, wireless transmission, routers, firewalls, switches, gateway computers, and / or edge servers. A network adapter card or network interface in each computing / processing device receives the computer-readable program instructions from the network and forwards them to the computer-readable storage media in the respective computing / processing device.

[0077] Computer program instructions used to perform the operations of this disclosure may be assembly instructions, instruction set architecture (ISA) instructions, machine instructions, machine-dependent instructions, microcode, firmware instructions, status setting data, or source code or object code written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Smalltalk, C++, etc., and conventional procedural programming languages ​​such as the "C" language or similar programming languages. The computer-readable program instructions may execute entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving a remote computer, the remote computer may be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or may be connected to an external computer (e.g., via the Internet using an Internet service provider). In some embodiments, electronic circuitry, such as programmable logic circuitry, field-programmable gate arrays (FPGAs), or programmable logic arrays (PLAs), is personalized by utilizing the status information of the computer-readable program instructions to implement various aspects of this disclosure.

[0078] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the protection scope of the claims of the present invention.

Claims

1. A method for automatic switching of supercapacitor energy storage, characterized in that, Includes the following steps: S1. Collect the instantaneous values ​​of the three-phase voltage and current of the bus at the common connection point of new energy access and grid connection, and calculate the effective values ​​of voltage and current; S2. When the effective values ​​of the three-phase voltages are all lower than the predefined voltage thresholds within the predetermined duration period, calculate the residual voltage gradient and perform moving average filtering. Take the maximum absolute value of the residual voltage gradient after filtering of all phases as the final residual voltage gradient. Perform fast Fourier transform on the bus voltage and current signals collected within the predefined sliding window to calculate the voltage harmonic entropy and current harmonic entropy. S3. Based on the final residual pressure gradient and harmonic entropy, construct the elliptic decision equation and mark the suspicious island points; Calculate the island confidence level of suspected island points, and combine it with the current harmonic entropy. When the predetermined judgment conditions are met, the island confidence level is corrected. Multiply the island confidence level of each sampling time within the cumulative window length before the corresponding sampling time by the time interval and accumulate them to obtain the cumulative confidence value. When the cumulative confidence value exceeds the set threshold, the new energy access area is in an island state at the current sampling time. S4. When in an isolated state, calculate the difference between the total output of new energy and the load power, and divide it into power surplus scenario and power shortage scenario; in the power surplus scenario, divide it into coarse adjustment and fine adjustment stages based on the predetermined switching conditions; in the coarse adjustment stage, classify the levels based on the difference and put in the corresponding level of energy-consuming resistors. During the fine-tuning phase, the switching safety margin is calculated based on the real-time status of the supercapacitor. When the safety margin threshold is not less than the predefined threshold, the supercapacitor charging and discharging control is initiated, and the supercapacitor charging power is nonlinearly switched according to the supercapacitor startup time. In the case of insufficient power, the supercapacitor discharge is initiated.

2. The automatic switching method for supercapacitor energy storage according to claim 1, characterized in that: The process of constructing the elliptic decision equation in S3 is as follows: Collect grid-connected steady-state data samples, and calculate the mean and standard deviation of the final residual pressure gradient of the grid-connected steady-state data samples, which are used as the mean and standard deviation of the grid-connected steady-state residual pressure gradient. Calculate the mean and standard deviation of voltage harmonic entropy of the grid-connected steady-state data sample, and use them as the mean and standard deviation of grid-connected stable voltage harmonic entropy; use three times the standard deviation of the grid-connected steady-state residual voltage gradient as the major axis parameter of the elliptical decision boundary, and use three times the standard deviation of the grid-connected steady-state residual voltage gradient as the minor axis parameter of the elliptical decision boundary. An elliptic decision equation is constructed based on the mean of the grid-connected steady-state residual voltage gradient, the mean of the voltage harmonic entropy, and the major and minor axis parameters.

3. The automatic switching method for supercapacitor energy storage according to claim 1, characterized in that: The process for calculating the island confidence score of a suspected isolated point in S3 is as follows: Substitute the final residual voltage gradient and voltage harmonic entropy calculated in real time at each sampling moment into the left side of the elliptic decision equation. When the result on the left side of the equation is not greater than 1, it is determined that the grid connection status is normal and the islanding action is not triggered. When the result on the left side of the equation is greater than 1, it is marked as a suspected island state; for any sampling time, the maximum value between 0 and the value on the left side of the equation minus 1 is taken as the island confidence level at the corresponding time.

4. The automatic switching method for supercapacitor energy storage according to claim 1, characterized in that: The process of correcting the island confidence level in S3 is as follows: The final residual voltage gradient and voltage harmonic entropy at each sampling moment are calculated in real time and substituted into the left side of the elliptic decision boundary equation. For any sampling moment, when the result on the left side of the equation at the corresponding moment is greater than 1 and the decrease of the current harmonic entropy at the current moment compared with the mean of the grid-connected steady-state current harmonic entropy exceeds the predefined decrease threshold, the product of the predefined enhancement coefficient and the island confidence at the corresponding sampling moment is updated to the island confidence at the corresponding sampling moment. Among them, the mean of the grid-connected steady-state current harmonic entropy is the mean of the current harmonic entropy of all samples in the grid-connected steady-state data sample.

5. The automatic switching method for supercapacitor energy storage according to claim 1, characterized in that: In S4, if the difference between the total output of new energy sources and the load power at the corresponding time is greater than 0, it belongs to the power surplus scenario; otherwise, it belongs to the power shortage scenario. The predetermined switching conditions are as follows: calculate in real time the deviation ratio of the effective voltage value of each phase relative to the rated voltage at the current moment. When the deviation ratio of any phase is greater than the predetermined voltage fluctuation rate, it is considered to be in the coarse adjustment stage; otherwise, it is in the fine adjustment stage.

6. The automatic switching method for supercapacitor energy storage according to claim 1, characterized in that: The process of classifying levels based on differences and installing energy-consuming resistors in S4 is as follows: When the difference is less than the predetermined first-level threshold, the predefined first group of resistors is activated; when the difference is not less than the predetermined first-level threshold and is less than the predetermined second-level threshold, the predefined first and second groups of resistors are activated; when the difference is not less than the predetermined second-level threshold, all three groups of resistors are activated simultaneously.

7. The automatic switching method for supercapacitor energy storage according to claim 1, characterized in that: The process of calculating the switching safety margin in S4 is as follows: The remaining energy storage capacity (SOC), module temperature, terminal voltage, and measured capacity value of the supercapacitor are acquired in real time. The remaining energy storage capacity and terminal voltage are normalized to obtain the capacity availability and voltage availability, respectively. The absolute value of the difference between the module temperature and the reference temperature at the current sampling time is divided by the predefined maximum allowable temperature deviation, the negative number is taken, and 1 is added to obtain the temperature deviation factor. The ratio of the current measured capacity value of the supercapacitor to the rated capacity value is used to obtain the capacity retention rate. Multiply the current capacity availability, temperature deviation factor, voltage availability, and capacity retention rate to obtain the corresponding switching safety margin.

8. The automatic switching method for supercapacitor energy storage according to claim 1, characterized in that: The nonlinear switching process for supercapacitor charging power in S4 is as follows: With the goal of minimizing voltage deviation, frequency deviation, power deviation, and power loss, the steady-state power reference value under the current supercapacitor startup time is solved within the feasible domain of the predefined supercapacitor power. The value of the predefined exponential function is calculated using the negative of the ratio of the supercapacitor startup time to the energy storage response time constant as the exponent. The negative of this value is added to 1 and multiplied by the steady-state power reference value to obtain the charging power reference value under the current supercapacitor startup time.

9. The automatic switching method for supercapacitor energy storage according to claim 8, characterized in that: The process of solving for the steady-state power reference value is as follows: The voltage deviation is defined as the ratio of the difference between the bus voltage and the rated reference voltage during the current supercapacitor startup time to the rated reference voltage. The frequency deviation is defined as the ratio of the difference between the frequency and the rated reference frequency during the current supercapacitor startup time to the rated reference frequency. The power tracking deviation is defined as the ratio of the difference between the charging power and the set reference power during the current supercapacitor startup time to the power reference value. The charging and discharging loss is defined as the square of the charging current at the corresponding charging power multiplied by the equivalent series resistance of the supercapacitor and then divided by the power reference value. The voltage deviation, frequency deviation, power tracking deviation, and charging and discharging loss are weighted by predefined weighting coefficients to obtain the optimization objective function. These predefined weighting coefficients are dynamically adjusted based on the real-time voltage deviation. During the solution process, two golden section points are selected within the feasible region of the supercapacitor power, and their values ​​are calculated by substituting them into the objective function. The function values ​​corresponding to the two golden section points are compared. If the function value of the first golden section point is less than that of the second golden section point, the power value corresponding to the second golden section point is used as the upper limit to update the feasible region interval. Otherwise, the power value corresponding to the first golden section point is used as the lower limit to update the feasible region interval. This process is repeated iteratively until the interval length is less than a predefined power amplitude threshold, and the midpoint of the interval is taken as the steady-state power reference value.

10. The automatic switching method for supercapacitor energy storage according to claim 9, characterized in that: The calculation process for the bus voltage and frequency under the current supercapacitor startup time is as follows: Under the current supercapacitor startup time, the difference between the charging power and the steady-state power reference value of the previous rolling optimization cycle is multiplied by a predefined voltage-power sensitivity coefficient and then added to the steady-state value of the bus voltage of the previous rolling optimization cycle to obtain the corresponding bus voltage; the difference between the charging power and the steady-state power reference value of the previous rolling optimization cycle is multiplied by a predefined frequency-power sensitivity coefficient and then added to the steady-state value of the frequency voltage of the previous rolling optimization cycle to obtain the corresponding frequency.

11. An automatic switching device for supercapacitor energy storage, comprising the method described in any one of claims 1-10, characterized in that, include: The data acquisition module collects the instantaneous values ​​of the three-phase voltage and current of the bus at the common connection point between the new energy access and the grid connection, and calculates the effective values ​​of the voltage and current. The feature extraction module calculates the residual voltage gradient and performs moving average filtering when the effective values ​​of the three-phase voltages are all lower than the predefined voltage thresholds within a predetermined duration. It takes the maximum absolute value of the residual voltage gradient after filtering all phases as the final residual voltage gradient. It also performs fast Fourier transform on the bus voltage and current signals collected within the predefined sliding window to calculate the voltage harmonic entropy and current harmonic entropy. The island state determination module constructs an elliptic decision equation and marks suspicious island points based on the final residual pressure gradient and harmonic entropy. Calculate the island confidence level of suspected island points, and combine it with the current harmonic entropy. When the predetermined judgment conditions are met, the island confidence level is corrected. Multiply the island confidence level of each sampling time within the cumulative window length before the corresponding sampling time by the time interval and accumulate them to obtain the cumulative confidence value. When the cumulative confidence value exceeds the set threshold, the new energy access area is in an island state at the current sampling time. When the energy storage automatic switching module is in an islanded state, it calculates the difference between the total output of new energy sources and the load power, and divides it into power surplus scenario and power shortage scenario. In the power surplus scenario, it divides the process into coarse adjustment and fine adjustment stages based on predetermined switching conditions. In the coarse adjustment stage, it classifies the levels based on the difference and puts in the corresponding level of energy-consuming resistors. During the fine-tuning phase, the switching safety margin is calculated based on the real-time status of the supercapacitor. When the safety margin threshold is not less than the predefined threshold, the supercapacitor charging and discharging control is initiated, and the supercapacitor charging power is nonlinearly switched according to the supercapacitor startup time. In the case of insufficient power, the supercapacitor discharge is initiated.

12. A terminal, comprising a processor and a storage medium; characterized in that: The storage medium is used to store instructions; The processor is configured to operate according to the instructions to perform the steps of the method according to any one of claims 1-10.

13. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the program implements the steps of the method according to any one of claims 1-10.

Images