A novel collaborative method for bidirectional power flow adaptive protection of power system switchgear for high-proportion energy storage integration

By acquiring transient electrical quantities of the energy storage system and constructing adaptive start-up criteria, and by removing fault components, the problem of false tripping or failure to tripping in high-proportion energy storage power systems by traditional protection methods is solved, thereby improving the reliability and sensitivity of the protection device.

CN123073673APending Publication Date: 2026-11-10NANJING ZHENGRUI POWER TECH CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611582735.9
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-10-13
Publication Date
2026-11-10

AI Technical Summary

Technical Problem

In power systems with a high proportion of energy storage, traditional protection methods are not adaptable enough to transient scenarios with frequent reversals of bidirectional power flow, leading to maloperation or failure of protection devices to operate, making it difficult to effectively isolate fault areas.

Method used

By acquiring transient electrical quantities at the energy storage grid connection point, latching the steady-state baseline before switching and recording the negative-sequence transient buffer sequence, an adaptive start-up criterion is constructed based on the expected trajectory of the positive-sequence current and the negative-sequence transient baseline. Fault components are removed, a weighted comprehensive directional index is generated, and the tripping of switching equipment is decided collaboratively.

Benefits of technology

It reduces the polarization interference of power flow reversal transients on the protection directional elements, and improves the reliability and sensitivity of the protection device under complex operating conditions.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123073673A_ABST
    Figure CN123073673A_ABST
Patent Text Reader

Abstract

This invention discloses a novel bidirectional power flow adaptive protection collaborative method for power system switchgear with high-proportion energy storage access. The method includes: acquiring transient electrical quantities at the energy storage grid connection point; latching a steady-state reference and recording a negative-sequence transient buffer sequence when a charging / discharging switching transient process is detected; determining a fault based on an adaptive start-up criterion during the transient period; predicting the expected trajectory of the positive-sequence current based on the switching progress analysis to remove the positive-sequence fault component; and performing data-driven extrapolation based on the buffer sequence to remove the negative-sequence fault component; reconstructing a polarization reference to eliminate phase angle offset using the separated fault components; and generating a comprehensive directional index based on energy information using a reliability-weighted fusion method; and coordinating with an overcurrent criterion to control the switchgear tripping. This invention reduces the polarization interference of power flow reversal transients on protection directional elements, reduces boundary jitter in hard switching logic, and improves the reliability and sensitivity of the protection device under complex operating conditions.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of power system relay protection technology, and in particular to a novel bidirectional power flow adaptive protection cooperative method for power system switchgear oriented towards high proportion of energy storage access. Background Technology

[0002] With the advancement of new power system construction, high-proportion energy storage systems are widely integrated into distribution and transmission networks, becoming key equipment for smoothing load fluctuations and providing grid support. In actual operation, energy storage systems need to frequently respond to dispatch commands and switch between charging and discharging modes. This rapid change in operating state leads to continuous reversals in the power flow direction at the grid connection point and surrounding lines. During the transient process of power flow reversal, underlying electrical quantities such as the system bus voltage phase angle and sequence component characteristics will experience dynamic shifts and distortions, placing higher demands on relay protection devices that ensure the safe and stable operation of the power grid.

[0003] Currently, line protection devices widely used in power grids mainly rely on steady-state abrupt changes in power frequency electrical quantities and memory polarization technology. In the direction determination stage, the conventional approach is to extract a snapshot of the bus voltage over a period of time before the fault occurs as a reference, and then compare its phase angle with the measured current after the fault. Simultaneously, for different types of short-circuit faults, fixed positive-sequence or negative-sequence directional element priority switching logic is typically pre-set. When a short circuit occurs under steady-state operating conditions, the aforementioned protection method based on steady-state reference and fixed logic can isolate the fault area.

[0004] However, traditional protection methods are insufficiently adaptable in transient scenarios with frequent reversals of bidirectional power flow. Because electrical quantities evolve continuously during transient processes, a fixed steady-state reference often corresponds to the system state under the old power flow direction, leading to polarization reference distortion. Furthermore, under complex transient conditions, the signal-to-noise ratio of each sequence electrical characteristic fluctuates significantly, and fixed criterion priority logic cannot cover all boundary cases, easily causing protection maloperation or failure to operate in the sensitivity blind zone. Therefore, it is necessary to study a protection method that can isolate interference signals and achieve smooth decision-making under complex transient conditions to improve the robustness and reliability of new power system protection devices. Summary of the Invention

[0005] To solve the above-mentioned technical problems, the present invention adopts the following technical solution:

[0006] A novel bidirectional power flow adaptive protection coordination method for power system switchgear oriented towards high-proportion energy storage integration includes:

[0007] Acquire transient electrical quantities at the energy storage grid connection point;

[0008] Active power calculation and transient identification are performed based on transient electrical quantities. When the power converter enters the charging and discharging switching transient process, the steady-state reference before switching is latched and the negative sequence transient buffer sequence is recorded.

[0009] During the charge-discharge switching transient process, the expected trajectory of the positive sequence current is predicted in real time based on the steady-state reference and transient electrical quantities before the switching, and the negative sequence transient baseline is constructed by data-driven extrapolation based on the negative sequence transient buffer sequence.

[0010] Fault determination is based on an adaptive start-up criterion constructed from the expected trajectory of positive-sequence current and the negative-sequence transient baseline.

[0011] When a fault is detected, the positive-sequence fault component is removed using the expected trajectory of the positive-sequence current, and the negative-sequence fault component is removed using the negative-sequence transient baseline.

[0012] Using the separated positive-sequence fault components and negative-sequence fault components, the positive and negative sequence direction discrimination polarization benchmarks that eliminate phase angle shift are reconstructed respectively. Based on the energy information contained in the positive and negative sequence fault components, the reliability adaptive weighted fusion is performed to generate a weighted comprehensive direction index.

[0013] The weighted composite directional index is used in conjunction with the pre-configured overcurrent criterion for overcurrent protection based on the current current amplitude extracted from transient electrical quantities to make a coordinated decision, control the switching equipment to trip, and restore the normal protection operation mode after exiting the charging and discharging switching transient process.

[0014] For example, the positive-sequence current full reversal increment is adaptively calculated in the following manner:

[0015] By combining the amplitude ratio of the pre-configured target switching power to the active power before switching, and the pre-configured power factor angle difference under charging and discharging conditions, the phase angle of the positive sequence current phasor in the steady-state reference before switching is reversed and the amplitude is scaled to estimate the target positive sequence current phasor.

[0016] The difference between the target positive-sequence current phasor and the positive-sequence current phasor before switching is used to obtain the positive-sequence current full reversal increment that is aligned with the current actual operating point reference.

[0017] For example, in the process of calculating the positive order confidence weight and the negative order confidence weight, a very small positive number with the same dimensions as the apparent power is introduced into the normalized denominator to prevent numerical overflow. The very small positive number is based on the pre-configured rated apparent power setting.

[0018] When generating the weighted composite directional index, the polarity symbols of the positive and negative order directional discrimination polarization benchmarks obtained from the reconstruction are extracted as directional indicator symbols. The positive order confidence weights and negative order confidence weights are combined with the directional indicator symbols and weighted summation is performed to generate the weighted composite directional index.

[0019] For example, in the step of predicting the expected trajectory of the positive-sequence voltage, the system operating scenario satisfies a preset voltage sensitivity constraint:

[0020] The ratio of the energy storage system's access capacity to the short-circuit capacity of the connected power grid does not exceed a preset proportional threshold, so that the bus voltage change during the transient process meets the linear superposition approximation requirement of the expected trajectory of the positive sequence voltage.

[0021] The beneficial effects of this invention are:

[0022] This invention reduces the polarization interference of power flow reversal transients on protection directional elements, reduces boundary jitter of hard switching logic, and improves the reliability and sensitivity of protection devices under complex operating conditions. Attached Figure Description

[0023] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0024] Figure 1 The flowchart of the novel bidirectional power flow adaptive protection cooperative method for power system switchgear with high proportion of energy storage access is shown in this application.

[0025] Figure 2 This application shows a flowchart of recording the steady-state baseline and negative-order transient buffer sequence before latch switching;

[0026] Figure 3 This document illustrates a flowchart illustrating the adaptive start-up criterion used in this application for fault determination.

[0027] Figure 4 The flowchart illustrating the process of stripping the positive-sequence fault component in this application is shown;

[0028] Figure 5 The flowchart illustrating the stripping of negative sequence fault components in this application is shown;

[0029] Figure 6 The flowcharts of the positive and negative sequence direction discrimination polarization references for eliminating phase angle shifts in this application are shown.

[0030] Figure 7 The flowchart of the credibility adaptive weighted fusion method in this application is shown;

[0031] Figure 8 The flowchart illustrates the process of resuming normal protection operation mode after exiting the charge / discharge switching transient process. Detailed Implementation

[0032] Example 1: This example provides a novel bidirectional power flow adaptive protection coordination method for power system switchgear oriented towards high-proportion energy storage access, such as... Figure 1As shown, the method includes the following steps 101-105:

[0033] In applications where a high proportion of energy storage is integrated into the distribution or transmission network, frequent charging and discharging switching of the energy storage system can lead to bidirectional power flow reversal. Traditional protection devices, during transient processes, are prone to systemic misjudgments of directional elements due to phase angle shifts in the polarization reference.

[0034] Step 101: Obtain the transient electrical quantities of the energy storage grid connection point.

[0035] Specifically, high-frequency instantaneous data of three-phase voltage and three-phase current, i.e., transient electrical quantities, or high-frequency electrical quantities, are collected in real time by instrument transformers installed at the grid connection point of the energy storage. These continuously collected raw electrical quantities are the basic data source for subsequent state identification, transient evolution tracking, and fault analysis.

[0036] Step 102: Perform active power calculation and transient identification based on transient electrical quantities. When the power converter is identified to enter the charging and discharging switching transient process, latch the steady-state reference before switching and record the negative sequence transient buffer sequence.

[0037] Based on this, the collected high-frequency electrical quantities are preprocessed to extract the active power and its changing trend, which reflect the power flow direction. When a sharp, monotonic change in active power consistent with charge-discharge switching characteristics is detected, the system determines that the power converter has initiated a switching action for its operating mode. At this time, the steady-state electrical characteristics prior to the switching action are immediately frozen and saved as the initial physical reference point for subsequent trajectory prediction. Simultaneously, a short-term data caching mechanism for negative-sequence components is activated to record the dynamic evolution of negative-sequence components during transient periods caused by imperfect decoupling of the power converter control loop.

[0038] Step 103: During the charge-discharge switching transient process, the expected trajectory of the positive-sequence current is predicted in real time based on the steady-state reference and transient electrical quantities before the switching, and the negative-sequence transient buffer sequence is used to perform data-driven extrapolation to construct the negative-sequence transient baseline; the positive-sequence current expected trajectory and the negative-sequence transient baseline are used to form an adaptive start-up criterion for fault determination; when a fault is determined to occur, the positive-sequence fault component is removed using the positive-sequence current expected trajectory, and the negative-sequence fault component is removed using the negative-sequence transient baseline.

[0039] Furthermore, after confirming entry into the transient tracking window, conventional start-up elements based on steady-state snapshots will fail due to the loss of the reference. Therefore, a dynamically updated start-up criterion is used to monitor in real time whether a short-circuit fault occurs. Once a fault is confirmed, the method will perform signal stripping processing in two independent dimensions: positive sequence and negative sequence. For the positive sequence network, the expected evolution path of the positive sequence component in a fault-free state is predicted by combining circuit physics laws and real-time switching progress, and then this part of the non-fault transient component caused by control actions is removed from the measured total. For the negative sequence network, short-term trend extrapolation is performed using the aforementioned cached historical transient data to establish a dynamic negative sequence transient baseline, and then the negative sequence fault component triggered by the short-circuit fault is extracted.

[0040] Step 104: Using the separated positive-sequence fault components and negative-sequence fault components, the positive and negative sequence direction discrimination polarization benchmarks that eliminate phase angle shift are reconstructed respectively, and the credibility adaptive weighted fusion is performed based on the energy information contained in the positive and negative sequence fault components to generate a weighted comprehensive direction index.

[0041] Accordingly, after extracting the fault components, they are converted into polarization reference and operational quantities for determining the fault direction. Since these components have already been stripped of phase offset interference caused by power flow reversal at the underlying signal processing stage, the constructed direction criterion can restore the intrinsic characteristics of the fault. Considering the significant fluctuations in the signal-to-noise ratio of positive and negative sequence signals under different fault types and switching progress, a dynamic allocation mechanism based on the apparent power of the signals is introduced. By evaluating the energy carried by each of the two sequence components, their weights in the final decision are automatically assigned, outputting a comprehensive direction determination result that takes into account all boundary conditions.

[0042] Step 105: The weighted comprehensive directional index and the pre-configured overcurrent criterion for overcurrent protection in the same direction based on the current current amplitude extracted from transient electrical quantities are used to make a coordinated decision, control the switching equipment to trip, and restore the normal protection operation mode after exiting the charging and discharging switching transient process.

[0043] Simultaneously, after obtaining the comprehensive direction determination result, it is logically coordinated with the traditional current amplitude over-limit judgment criterion. When it is confirmed that the fault is located in the positive direction of the protection and the current reaches the action threshold, a trip command is issued to the switching equipment to isolate the fault; if it is determined to be in the reverse direction, the output on this side is blocked to avoid non-selective maloperation during power flow reversal transients. When the charging and discharging switching process is detected to be over and the power has returned to a stable state, the transient buffer data is automatically cleared and the relevant status bits are reset, so that the protection logic smoothly transitions back to the normal steady-state operation monitoring mode.

[0044] This embodiment solves the problem of polarization reference distortion in power flow reversal transients by introducing a positive-sequence current expectation trajectory based on analytical prediction and a negative-sequence transient baseline based on data-driven short-time extrapolation. This can eliminate non-fault transient interference caused by power converter control actions and reduce the misleading effect of phase angle offset on directional elements.

[0045] Example 2: In one possible implementation, active power calculation and transient identification are performed based on transient electrical quantities. When the power converter enters a charging / discharging switching transient process, the steady-state reference before the switching is latched and the negative-sequence transient buffer sequence is recorded, such as... Figure 2 As shown, the specific steps include 201-203:

[0046] Step 201: Calculate the instantaneous active power based on transient electrical quantities, and extract the sliding window average active power and the rate of change of active power.

[0047] To obtain high-fidelity underlying electrical characteristics, the sampling frequency configured in this embodiment is 4000 Hz for the data acquisition stage. Those skilled in the art will understand that the sampling frequency should meet the Nyquist criterion requirements and ensure the accuracy of the symmetrical component calculation; the specific value can be determined based on the actual hardware conditions of the protection device.

[0048] Optionally, after obtaining the instantaneous values ​​of the three-phase voltage and the three-phase current, a symmetrical component transformation matrix is ​​used to decouple the three-phase coupling quantities, and output the positive-sequence voltage phasor, positive-sequence current phasor, negative-sequence voltage phasor, and negative-sequence current phasor.

[0049] Furthermore, the instantaneous active power at the energy storage grid connection point is calculated using the instantaneous values ​​of the measured three-phase instantaneous values, based on the instantaneous power theory calculation formula. On this basis, a sliding window mean filter is applied to the continuously acquired instantaneous active power with the power frequency frequency as the time window width to generate the sliding window average active power. Based on the time interval between adjacent sliding window centers, the rate of change of active power is calculated for each sampling point, providing basic numerical values ​​for subsequent transient feature identification.

[0050] Step 202: Compare the absolute value of the active power change rate with the pre-configured power change rate threshold, and make a comprehensive judgment based on the monotonicity characteristics of the sliding window average active power. When the triggering condition is met, confirm that the charging and discharging switching transient process has been entered.

[0051] Specifically, dual feature verification is used to isolate normal operation fluctuations from transient processes caused by charge / discharge switching. During execution, the following different logical decision branches exist:

[0052] Under the first judgment condition, when the absolute value of the active power change rate continuously exceeds the pre-configured power change rate threshold for half a power frequency cycle, and the sliding window average active power shows a monotonic trend towards zero, it is confirmed that the power converter is performing a charging / discharging mode switch. The first sampling moment that meets the above joint conditions is marked as the switching start moment, and subsequent transient tracking and latching logic is triggered.

[0053] Under the second judgment condition, if the active power change rate does not reach the pre-configured power change rate threshold, or if the sliding window average active power does not show a monotonic evolution characteristic toward zero, then the current system is determined to be in normal operating mode or experiencing general load fluctuations. In this case, the normal protection and monitoring status is maintained, and the transient tracking mechanism is not activated.

[0054] Step 203: Latch the positive-sequence voltage phasor, positive-sequence current phasor, and active power before the switch in the transient electrical quantities as the steady-state reference before the switch, and write the negative-sequence voltage phasor and negative-sequence current phasor in the transient electrical quantities into the negative-sequence transient buffer sequence period by period.

[0055] Furthermore, after confirming the occurrence of a switching event and marking the start time of the switching, historical data of the positive-sequence components within the power frequency cycle prior to that time are extracted and frozen. Specifically, the frozen objects include the positive-sequence voltage phasor, the positive-sequence current phasor, and the active power before the switching. This type of data collectively constitutes the steady-state reference before the switching, which is used to characterize the initial physical state of the system before the power flow reversal occurs.

[0056] Simultaneously, a dynamic data caching mechanism is established from the start of the switching process. The calculated negative-sequence voltage and current phasors are extracted periodically and written into the negative-sequence transient cache sequence in chronological order. This sequence records the negative-sequence oscillation characteristics generated during charge / discharge switching due to imperfect transient decoupling of the power converter control loop, providing a sample source for subsequent data-driven extrapolation.

[0057] According to one aspect of this application, the setting of the pre-configured power change rate threshold satisfies the following constraints: the lower limit of the pre-configured power change rate threshold is higher than the natural change rate of active power under normal operating conditions; the upper limit of the pre-configured power change rate threshold is lower than the minimum transient power change rate, which is jointly calculated and determined by the rated power of the power converter corresponding to the energy storage grid connection point, the rated power ramp-up time, and the pre-configured reliability coefficient.

[0058] To avoid false triggering and ensure effective capture of slow switching processes, the pre-configured power change rate threshold needs to be set with upper and lower boundaries. The lower limit constraint is used to shield normal power fluctuations caused by grid frequency offsets or regular load switching. The upper limit constraint limits the physical execution capability of the power converter, and its calculation formula is as follows:

[0059] dP / dt min =K rel ×(2×P N ) / T ramp ;

[0060] In the formula, dP / dt min K is the calculated minimum transient power change rate, which is the lower bound of the power change rate corresponding to the energy storage power converter switching at a rate lower than the rated rate. rel For the pre-configured reliability coefficient, P N T represents the rated active power of the power converter. ramp This refers to the rated power ramp-up time of the power converter.

[0061] The pre-configured reliability coefficient must meet the following constraints: its lower limit should ensure that the calculated rate of change of active power still exceeds the set threshold when the energy storage converter executes the switching command at the lowest operating rate; its upper limit should avoid being too large, which could cause normal power fluctuations to be falsely triggered. In this embodiment, the reliability coefficient is set to 0.4, and those skilled in the art can determine a suitable value based on the control bandwidth characteristics of the actual power converter.

[0062] Example 3: In one possible implementation, an adaptive start-up criterion is constructed based on the expected trajectory of the positive-sequence current and the negative-sequence transient baseline to determine the fault, specifically including the following steps 301-304:

[0063] Step 301: Extract the current measured positive sequence current and the current measured negative sequence current from the transient electrical quantities.

[0064] As an optional implementation, during the activation of the charge / discharge switching transient tracking window, the protection device extracts the current sequence component data from the electrical quantities after symmetrical component decoupling preprocessing on a sampling cycle basis, obtaining the current measured positive sequence current and the current measured negative sequence current. This type of real-time extracted sequence current data provides basic observation values ​​for subsequent dynamic deviation verification, replacing the fixed steady-state reference values ​​that fail during transient processes.

[0065] Step 302: Calculate the deviation of the positive sequence trajectory using the analytically predicted expected trajectory of the positive sequence current and the current measured positive sequence current.

[0066] Potential fault mutations are identified by comparing measured values ​​with theoretical predictions. The vector difference magnitude between the current measured positive-sequence current and the expected value corresponding to the expected trajectory of the positive-sequence current is calculated to generate an index reflecting anomalies in the positive-sequence network. During normal, fault-free switching, this deviation only reflects model approximation errors and remains at a low baseline level; however, during short-circuit faults, it exhibits a step-like increase due to fault current injection. The specific calculation expression is as follows:

[0067] dI 1,err =|I 1,meas -I 1,exp |;

[0068] In the formula, dI 1,err I represents the positive sequence trajectory deviation. 1,meas I is the currently measured positive sequence current phasor. 1,exp Let |...| represent the expected phasor corresponding to the expected trajectory of the positive sequence current, and |...| represent the calculation of the magnitude of the complex phasor.

[0069] Step 303: Calculate the sudden increment of the negative sequence baseline using the negative sequence transient baseline obtained by extrapolating the negative sequence transient buffer sequence and the current measured negative sequence current.

[0070] In potential asymmetric fault scenarios, an independent verification mechanism for the negative sequence dimension is introduced. A negative sequence transient baseline, constructed using short-term extrapolation from historical data, is extracted. The magnitude of the vector difference between this baseline and the current measured negative sequence current is calculated to generate a sudden increment in the negative sequence baseline. This increment is used to remove background negative sequence oscillations caused by imperfect decoupling of the power converter control loop, thus highlighting the abrupt negative sequence changes induced by asymmetric short circuits. The specific calculation is as follows:

[0071] dI 2,err =|I 2,meas -I 2,base |;

[0072] In the formula, dI 2,err For negative-order baseline burst increments, I 2,meas I is the currently measured negative sequence current phasor. 2,base This is the phasor corresponding to the negative-order transient baseline.

[0073] Step 304: When the positive sequence trajectory deviation exceeds the pre-configured positive sequence start threshold, or the negative sequence baseline sudden increase exceeds the pre-configured negative sequence start threshold, and the over-limit state is maintained for a preset time interval, a fault is confirmed to have occurred.

[0074] The fault initiation decision is made by combining the dynamic deviations from the two dimensions mentioned above. In actual execution, the following parallel logical decision branches exist:

[0075] Under the first starting condition, if the positive sequence trajectory deviation exceeds the pre-configured positive sequence starting threshold, and this over-limit state continues for one sampling interval, a suspected fault is determined to have occurred. In this embodiment, the pre-configured positive sequence starting threshold is set to 0.2 times the rated current. This value range can avoid the maximum model deviation during normal switching transients, while ensuring the response capability to faults at the end of the minimum protection range.

[0076] Under the second starting condition, if the sudden increase in the negative sequence baseline exceeds the pre-configured negative sequence starting threshold, and the over-limit state is also maintained continuously for one sampling interval, then a suspected asymmetrical fault is determined to have occurred. In this embodiment, the pre-configured negative sequence starting threshold is set to 0.05 times the rated current to ensure the sensitivity for detecting single-phase grounding and two-phase short-circuit faults.

[0077] In the third operating condition, if the increments of the above two dimensions do not reach their respective start-up thresholds, or if the over-limit state fails to maintain the required number of continuous sampling points, the current transient tracking state is maintained, and it is determined that no fault has occurred.

[0078] Once the first or second startup condition is met, the system marks the current moment as the fault determination moment and simultaneously freezes the switching progress parameter values ​​at this moment. The constraint of continuously maintaining a preset time interval is used to filter out high-frequency noise during the transient process and prevent malfunctions of the starting components. The establishment of the fault determination moment marks the formal triggering of the subsequent multi-sequence transient component adaptive separation process.

[0079] Example 4: In one possible implementation, the expected trajectory of the positive-sequence current is predicted analytically based on the switching progress to remove the positive-sequence fault component, such as... Figure 4 As shown, the specific steps include 401-404:

[0080] Step 401: Calculate the normalized switching progress parameters based on the pre-switching active power in the pre-switching steady-state reference, the pre-configured switching target power, and the sliding window average active power calculated in real time based on transient electrical quantities.

[0081] Optionally, the pre-configured switching target power is determined based on the scheduling instructions received by the energy storage system. For example, when the system switches from a charging state to a discharging state, the target power is the rated discharging power; otherwise, it is the rated charging power. A dimensionless parameter reflecting the physical switching process is constructed by calculating the offset ratio of real-time active power relative to the initial and target states. To prevent data overflow, the calculation results are limited to the range of 0 to 1. The specific calculation formula is as follows:

[0082] k prog =limit 0_1 ((P avg -P pre ) / (Pset -P pre ));

[0083] In the formula, k prog For the normalized switching progress parameter, P avg P is the average active power calculated in real time using a sliding window. pre P is the active power before the switch. set For the pre-configured target switching power, limit 0_1 This indicates a limit operation that restricts the calculation result to between 0 and 1.

[0084] Step 402: Construct the expected trajectory of the positive sequence current based on the switching progress parameters and the positive sequence current full reversal increment calculated based on the steady-state reference before switching and the pre-configured switching target power;

[0085] Accordingly, the switching progress parameter is used as a linear interpolation coefficient, multiplied by the positive-sequence current full reversal increment, and then superimposed onto the pre-switching steady-state reference. This generates a time-varying current prediction sequence, characterizing the physical trajectory of the positive-sequence current during an ideal switching process without a short-circuit fault. The specific calculation formula is as follows:

[0086] I 1,exp =I 1,pre +k prog ×dI 1,set ;

[0087] In the formula, I 1,exp I is the desired phasor corresponding to the desired trajectory of the positive-sequence current. 1,pre k is the positive sequence current phasor before switching. prog For the normalized switching progress parameter, dI 1,set This is the increment of the positive sequence current in full reversal.

[0088] Step 403: Combining the pre-configured system equivalent positive sequence impedance and the expected positive sequence current trajectory, predict the expected positive sequence voltage trajectory based on the positive sequence voltage phasor in the pre-switching steady-state reference.

[0089] Simultaneously, based on the Thevenin equivalent circuit principle, the voltage drop increment caused by the predicted positive-sequence current change flowing through the system impedance is superimposed on the bus voltage before switching to construct a model for the expected evolution of the bus voltage. The specific calculation expression is as follows:

[0090] U 1,exp =U 1,pre -(I 1,exp -I 1,pre )×Z1;

[0091] In the formula, U 1,expU is the desired phasor corresponding to the desired trajectory of the positive-sequence voltage. 1,pre For the positive sequence voltage phasor before switching, I 1,exp I is the desired phasor corresponding to the desired trajectory of the positive-sequence current. 1,pre Z1 represents the positive-sequence current phasor before switching, and Z2 represents the pre-configured equivalent positive-sequence impedance of the system. This formula reflects that the system-side electromotive force is approximately constant during the transient period, and the change in bus voltage is determined only by the voltage drop increment caused by the change in positive-sequence current flowing through the equivalent impedance of the system.

[0092] In some optional implementations, to satisfy the linear superposition approximation condition of the expected positive-sequence voltage trajectory, the system operating scenario must meet a preset voltage sensitivity constraint. Specifically, the ratio of the energy storage system's connected capacity to the short-circuit capacity of the connected power grid does not exceed a preset proportional threshold, which is typically set to 10%. Under this constraint, the change in bus voltage is much smaller than the rated voltage, ensuring the estimation accuracy of the linear impedance voltage drop model.

[0093] Step 404: When a fault is determined to have occurred, the expected values ​​corresponding to the current measured positive sequence electrical quantity and the expected trajectory of positive sequence current and positive sequence voltage are vectored and the positive sequence fault component is extracted.

[0094] Once the starting element is confirmed to be faulty, the switching progress parameters corresponding to the current moment are extracted and substituted into the aforementioned trajectory model to obtain the expected voltage and expected current under that progress. Further, in the complex domain, the measured phasor is subtracted from the expected phasor, and through a dynamic reference stripping mechanism, non-fault transient interference caused by power flow reversal is eliminated, outputting the fault component that purely reflects the short-circuit characteristics. The specific calculation formula is as follows:

[0095] dU1=U 1,meas -U 1,exp ;

[0096] dI1=I 1,meas -I 1,exp ;

[0097] In the formula, dU1 is the extracted positive-sequence voltage fault component, U 1,meas U is the currently measured positive-sequence voltage phasor. 1,exp dI1 is the desired positive-sequence voltage phasor; dI1 is the extracted positive-sequence current fault component; I 1,meas I is the currently measured positive sequence current phasor. 1,exp Let be the desired positive sequence current phasor.

[0098] According to one aspect of this application, the positive-sequence current full reversal increment is adaptively calculated in the following manner:

[0099] ① Combining the amplitude ratio of the pre-configured target switching power and the active power before switching, as well as the pre-configured power factor angle difference under charging and discharging conditions, the phase angle of the positive sequence current phasor in the steady-state reference before switching is reversed and the amplitude is scaled to estimate the target positive sequence current phasor.

[0100] Accordingly, to ensure that the baseline for trajectory prediction aligns with the actual operating point, it is necessary to extrapolate the target current state based on the measured current state before the switching occurs. The scaling factor of the current amplitude is determined by calculating the absolute ratio of the target power to the power before the switching; simultaneously, a 180-degree phase reversal is introduced to characterize the change in power flow direction, and the power factor angle difference under both charging and discharging conditions is superimposed to correct the phase. The specific extrapolation expression is as follows:

[0101] I 1,set =|P set / P pre |×|I 1,pre |×e (j×(θ_I+180+d_φ)) ;

[0102] In the formula, I 1,set To estimate the target positive-sequence current phasor, |P set / P pre | represents the power amplitude ratio, |I 1,pre | represents the magnitude of the positive sequence current phasor before switching, e is the base of the natural logarithm, j is the imaginary unit, θ_I is the phase angle of the positive sequence current before switching, 180 represents the phase angle reversal caused by power flow reversal, and d_φ is the difference between the pre-configured power factor angles of the discharge and charging conditions.

[0103] The pre-configured power factor angle difference under charging and discharging conditions can be determined based on the reactive power control setpoints of the energy storage power converter in both charging and discharging operating modes. Specifically, the power factor angle under charging conditions is calculated from the active power and the corresponding reactive power command, and the power factor angle under discharging conditions is calculated similarly; the difference between the two is the power factor angle difference. When the energy storage system operates at a unity power factor in both charging and discharging conditions, this difference is zero. Those skilled in the art can determine this parameter based on the actual reactive power control strategy of the energy storage system.

[0104] ② Calculate the difference between the target positive sequence current phasor and the positive sequence current phasor before switching to obtain the positive sequence current full reversal increment that is aligned with the current actual operating point reference.

[0105] Optionally, subtraction is performed in the complex domain to obtain the total current change required to transition from the initial state to the target state. Furthermore, adaptive calculation replaces the estimation method using fixed rated parameters, eliminating potential trajectory start-end deviations that may occur when the energy storage system triggers switching under partial load operation. The specific calculation expression is as follows:

[0106] dI 1,set =I 1,set -I 1,pre ;

[0107] In the formula, dI 1,set For the positive sequence current full reversal increment, I 1,set For the target positive sequence current phasor, I 1,pre This is the positive sequence current phasor before switching.

[0108] According to one aspect of this application, the pre-configured system equivalent positive-sequence impedance is obtained by:

[0109] I. Obtain the short-circuit capacity at the grid connection point, system impedance angle, and rated line voltage at the grid connection point from the power grid dispatching department.

[0110] Network topology parameters issued by the superior power grid dispatching system are read from the static parameter configuration library. These parameters reflect the equivalent grid strength and electrical characteristics behind the energy storage system's access point and are a necessary input source for calculating the Thevenin equivalent impedance.

[0111] II. Based on the ratio of the square of the rated line voltage at the grid connection point to the short-circuit capacity at the grid connection point, and combined with the system impedance angle, the pre-configured system equivalent positive sequence impedance is calculated.

[0112] Furthermore, using the obtained static network parameters, the equivalent impedance viewed from the grid connection point to the system side is evaluated in the setting calculation module of the protection device. The specific calculation formula is as follows:

[0113] Z1=(U N 2 / S k )×e (j×θ_S) ;

[0114] In the formula, Z1 is the pre-configured system equivalent positive sequence impedance, U N S is the rated line voltage at the grid connection point. k Here, e is the short-circuit capacity at the grid connection point, j is the base of the natural logarithm, θ_S is the system impedance angle.

[0115] Example 5: In one possible implementation, data-driven extrapolation is performed based on the negative-order transient buffer sequence to remove negative-order fault components, such as... Figure 5 As shown, the specific steps include 501-504:

[0116] Step 501: Extract the negative-order transient buffer sequence within the preset sampling window before the fault determination time.

[0117] As an optional implementation, after the system locks onto the moment of the fault, it backtracks and retrieves previously established dynamically cached data. To balance data freshness and statistical significance in regression calculations, a finite-length negative-order data sample is extracted immediately preceding the determination moment. The preset sampling window width is typically set to 1 / 4 of the power frequency, corresponding to 20 discrete sampling points at a sampling rate of 4000 Hz. The aforementioned data records the non-fault negative-order evolution trend in the short period before the short circuit occurs.

[0118] Step 502: Apply least squares linear regression to the negative-order electrical quantities in the negative-order transient buffer sequence within the preset sampling window to obtain the negative-order transient change rate and the negative-order mean within the window.

[0119] Accordingly, in the extracted discrete complex samples, a linear fitting operation is performed in the complex domain to extract the slope of macroscopic changes hidden in micro oscillations. By minimizing the sum of squared residuals between the data at each sampling point and the fitted line, the slope parameter characterizing the evolution trend of the negative-sequence current is calculated. The specific calculation formula is as follows:

[0120] K 2I =Σ((t i -t avg )×(I 2_i -I 2_avg )) / Σ((t i -t avg ) 2 );

[0121] In the formula, K 2I For negative-order transient rate of change, Σ represents the summation operation over all sampling points within the preset sampling window, and t i For the i-th sampling time, t avg I is the arithmetic mean of all sampling times within the window. 2_i Let I be the negative-sequence current phasor at the i-th sampling time. 2_avg This represents the average value of the negative sequence current phasors within the window.

[0122] Step 503: Using the negative-order transient change rate and the negative-order mean within the window, short-time linear extrapolation is performed based on the time difference between the fault determination time and the center time of the preset sampling window to construct the negative-order transient baseline at the fault determination time.

[0123] Furthermore, after obtaining the historical slope of change, it is extended forward along the time axis to the current fault determination time to predict the negative-sequence state that the system would have exhibited had a short-circuit fault not occurred. The effectiveness of this extrapolation mechanism is based on the physical foundation of the power converter control loop response characteristics. Specifically, when the power converter adopts a control architecture dominated by a proportional-integral regulator, its closed-loop time constant is usually much larger than the extrapolation time span, causing the negative-sequence component to maintain an approximately constant rate of change within a short prediction interval. The specific calculation formula is as follows:

[0124] I 2,base =I 2_avg +K 2I ×(t f -t avg );

[0125] In the formula, I 2,base For the phasor corresponding to the constructed negative-order transient baseline, I 2_avg K represents the mean value of the negative sequence current phasors within the window. 2I For negative order transient rate of change, t f At the fault determination time, t avg This is the arithmetic mean of all sampling times within the window.

[0126] Step 504: Extract the negative sequence fault component by vector difference between the measured negative sequence electrical quantity at the fault determination time and the negative sequence transient baseline.

[0127] Simultaneously, after establishing a dynamic reference benchmark, the actual total negative sequence quantity collected at the current moment is subtracted from the extrapolated expected benchmark quantity. Through vector subtraction in the complex domain, the background oscillation component caused by imperfect control decoupling is removed, obtaining the negative sequence response signal triggered by an asymmetric short-circuit fault. The same extraction, regression, extrapolation, and subtraction process is also used to process the negative sequence voltage independently. The specific calculation expression is as follows:

[0128] dI2=I 2,meas -I 2,base ;

[0129] In the formula, dI2 is the extracted negative sequence current fault component, and I 2,meas I is the measured negative sequence current phasor at the fault determination time. 2,base This is the phasor corresponding to the negative-order transient baseline.

[0130] Example 6: In one possible implementation, the positive and negative sequence direction discrimination polarization references for eliminating phase angle shift are reconstructed respectively, such as... Figure 6 As shown, the specific steps include 601-603:

[0131] Step 601: Using the voltage component in the positive sequence fault component as the polarization reference, the current component in the positive sequence fault component is rotated according to the pre-configured line positive sequence impedance angle and then projected onto the polarization reference direction to calculate the positive sequence direction discrimination torque.

[0132] Optionally, to mitigate phase angle shift interference caused by power flow reversal, this embodiment utilizes the separated pure positive-sequence fault component to construct a criterion. In the complex plane, the positive-sequence current fault component is rotated according to the impedance angle of the protected line to align it with the line impedance direction. Further, the real part of the complex conjugate dot product of the rotated current phasor and the positive-sequence voltage fault component is calculated to extract the active directional energy related to the fault location. The specific calculation formula is as follows:

[0133] T pos =Re[dU1×(dI1×e (j×θ_Z) ) * ];

[0134] In the formula, T pos For the positive sequence direction to determine the torque, Re[...] indicates taking the real part of the complex number, dU1 is the positive sequence voltage fault component, dI1 is the positive sequence current fault component, e is the base of the natural logarithm, j is the imaginary unit, θ_Z is the pre-configured line positive sequence impedance angle, and * indicates taking the complex conjugate.

[0135] After obtaining the calculation result, its absolute value is compared with a pre-configured positive order sensitivity threshold. If the absolute value does not exceed the threshold, the system outputs a direction pending flag to prevent misjudgment caused by weak signals due to background noise or model residuals.

[0136] Step 602: Check whether the amplitude of the current component in the negative sequence fault component meets the pre-configured negative sequence sensitivity threshold. If it does, use the voltage component in the negative sequence fault component as the polarization reference, rotate the current component in the negative sequence fault component according to the positive sequence impedance angle of the line, and project it to the polarization reference direction to calculate the negative sequence direction discrimination torque. If it does not meet the requirements, set the negative sequence direction discrimination torque to zero.

[0137] Meanwhile, considering that three-phase symmetrical faults do not generate significant negative-order excitation, an availability check needs to be performed before calculating the negative-order criterion. In practice, the following parallel logical branches exist:

[0138] In the first operating condition, if the amplitude of the negative sequence current fault component is less than the pre-configured negative sequence current sensitivity threshold, the current fault type is determined to be a symmetrical fault or a weak negative sequence excitation, and the negative sequence direction criterion is marked as unavailable.

[0139] In the second operating condition, if the amplitude reaches or exceeds the threshold, the calculation of the negative-sequence directional discrimination torque is activated. Its mathematical structure is symmetrical to the positive-sequence criterion; it obtains the negative-sequence active energy by rotating the negative-sequence current fault component and projecting it onto the direction of the negative-sequence voltage fault component. The specific calculation formula is as follows:

[0140] T neg =Re[dU2×(dI2×e (j×θ_Z) ) * ];

[0141] In the formula, T neg dU2 is the negative sequence direction discrimination torque, dI2 is the negative sequence voltage fault component, dI2 is the negative sequence current fault component, and θ_Z is the pre-configured positive sequence line impedance angle.

[0142] Similarly, if the absolute value of the calculated negative sequence direction discrimination torque does not exceed the pre-configured negative sequence sensitivity threshold, then the negative sequence network outputs a direction undetermined identifier.

[0143] Step 603: Use the positive sequence direction discrimination torque and the negative sequence direction discrimination torque as the positive and negative sequence direction discrimination polarization references.

[0144] Specifically, the polarization reference refers to the direction discrimination reference index based on the reconstruction of pure fault components and the elimination of transient phase angle offset. Its physical meaning is reflected by the sign (positive / negative) of the direction discrimination torque, which indicates the direction of the fault relative to the protection installation location.

[0145] After the parallel processing described above, the system generates a polarization reference index that is free from transient power flow interference. The two torques are complementary in physical mechanism: when a three-phase symmetrical fault occurs, the direction determination is dominated by the positive-sequence torque; while when an asymmetrical fault coincides with a power reversal zero-crossing condition that causes a decrease in the positive-sequence signal-to-noise ratio, the negative-sequence torque takes over the direction determination task.

[0146] To ensure the reliability of the criteria, the pre-configured positive-sequence sensitivity threshold is set to 1% of the rated apparent power in this embodiment; the pre-configured negative-sequence sensitivity threshold can be appropriately lowered due to the low background noise, and is set to 0.5% of the rated apparent power in this embodiment; the pre-configured negative-sequence current sensitivity threshold is set to 0.05 times the rated current in this embodiment. These threshold settings provide boundary constraints for subsequent weighted fusion.

[0147] Example 7: In one possible implementation, based on the energy information contained in the positive and negative sequence fault components, a confidence-adaptive weighted fusion is performed, such as... Figure 7 As shown, the specific steps include 701-703b:

[0148] Step 701: Calculate the product of the magnitude of the voltage component and the magnitude of the current component in the positive sequence fault component to obtain the positive sequence apparent power characterizing the positive sequence energy information.

[0149] As an optional implementation, to evaluate the strength of fault information that can be utilized by the positive sequence criterion, the magnitudes of the separated pure positive sequence voltage fault component and positive sequence current fault component are extracted. Multiplying these magnitudes yields the apparent power generated by the fault in the positive sequence network. This value reflects the signal-to-noise ratio level of the positive sequence signal under the current system operating conditions; when a three-phase symmetrical fault occurs, this value tends to the global maximum value.

[0150] Step 702: Calculate the product of the magnitude of the voltage component and the magnitude of the current component in the negative sequence fault component to obtain the negative sequence apparent power characterizing the negative sequence energy information.

[0151] Accordingly, the magnitudes of the pure negative-sequence voltage fault component and the negative-sequence current fault component are extracted and their product is calculated. This calculation result characterizes the amount of energy injected into the negative-sequence network by the asymmetric fault. When the asymmetric fault is exactly superimposed on the zero-crossing power reversal condition, the positive-sequence component tends to zero, while the negative-sequence apparent power becomes prominent and becomes the main information carrier for direction determination.

[0152] Step 703: Based on the proportions of positive-order apparent power and negative-order apparent power in the total fault apparent power, calculate the positive-order confidence weight and negative-order confidence weight respectively.

[0153] To achieve smooth transitions under multiple operating conditions, replacing traditional hard switching or fixed priority logic, the physical complementarity of the two sets of criteria is represented as dynamic normalized weights. By calculating the ratio of the apparent power of a single sequence component to the sum of the apparent powers of the two sequences, the weights can be automatically adjusted as the fault conditions continuously change. When the fault component of a certain sequence approaches zero, the corresponding weight naturally approaches zero, and the system automatically degenerates into a single-criterion operating mode without the need for additional exit logic.

[0154] In step 703a, during the calculation of positive and negative confidence weights, a minimal positive number with the same dimensions as apparent power is introduced into the normalized denominator to prevent numerical overflow. The minimal positive number is based on the pre-configured rated apparent power setting.

[0155] When performing normalization operations, the risk of data overflow due to the denominator approaching zero must be considered. A constant term is added to the denominator to maintain computational stability. To ensure consistency of physical dimensions and prevent deviations in the weight distribution under normal fault conditions, this minimum positive number is set to be on the order of one ten-thousandth of the rated apparent power. The specific calculation formula is as follows:

[0156] w pos=(|dU1|×|dI1|) / (|dU1|×|dI1|+|dU2|×|dI2|+ε);

[0157] w neg =(|dU2|×|dI2|) / (|dU1|×|dI1|+|dU2|×|dI2|+ε);

[0158] In the formula, w pos For the positive order confidence weight, w neg The negative sequence confidence weights are represented by |dU1| and |dI1|, which are the magnitudes of the positive sequence voltage and current fault components, respectively, and |dU2| and |dI2|, which are the magnitudes of the negative sequence voltage and current fault components, respectively. ε is a very small positive number with a value of 10. -4 ×U N ×I N U N I is the rated line voltage at the grid connection point. N This is the rated current of the energy storage system.

[0159] Step 703b: When generating the weighted composite direction index, the polarity symbols of the positive and negative order direction discrimination polarization benchmarks obtained from the reconstruction are extracted as direction indicator symbols. The positive order confidence weight and the negative order confidence weight are combined with the direction indicator symbols and weighted summation is performed to generate the weighted composite direction index.

[0160] Accordingly, the polarity characteristics of the positive-sequence and negative-sequence discriminative torques are extracted and linearly combined with the previously calculated dynamic weights. The value range of this comprehensive index is constrained to between -1 and 1; the larger its absolute value, the more consistent the directional indications of the two sets of criteria and the higher the reliability. The specific calculation expression is as follows:

[0161] D mix =w pos ×sgn(T pos )+w neg ×sgn(T neg );

[0162] In the formula, D mix As a weighted composite directional indicator, w pos with w neg These represent the positive and negative confidence weights, respectively, sgn(...) is the sign function, and T... pos With T neg The torque is determined by its positive and negative direction.

[0163] When specifically executing the mapping of the symbol function, sgn takes the value of 1 when the corresponding discrimination torque is greater than 0; sgn takes the value of -1 when it is less than 0; and sgn takes the value of 0 when the corresponding torque outputs an undetermined direction in the test, thus eliminating the interference of unreliable criteria on the final decision.

[0164] Example 8: In one possible implementation, the weighted composite directional index is used in conjunction with the pre-configured overcurrent criterion for in-direction overcurrent protection based on the current current amplitude extracted from transient electrical quantities to make a coordinated decision, control the switching equipment to trip, and restore the normal protection operation mode after exiting the charging / discharging switching transient process, such as... Figure 8 As shown, the specific steps include 801-803:

[0165] Step 801: When the weighted composite direction index indicates a positive direction and the current amplitude extracted based on transient electrical quantities meets the pre-configured overcurrent criterion, a trip permission signal is sent to the switchgear.

[0166] Specifically, the weighted composite directional index generated by the preceding calculation is obtained and compared with a pre-configured composite directional judgment threshold. If the index is greater than the set positive threshold, the current fault is confirmed to be located in the positive direction of the line. Simultaneously with directional confirmation, the real-time amplitude of the grid connection point current is extracted and compared with a pre-set overcurrent action value. When both the directional and overcurrent conditions are met, the collaborative decision-making logic issues a trip command to the corresponding switching equipment to isolate the fault area.

[0167] Correspondingly, since the energy storage system mainly experiences a reversal of the current phase angle when switching between charging and discharging modes, and its rated current amplitude is approximately equal in both operating conditions, the pre-configured overcurrent criterion does not need to be adaptively modified with the transient process. The conventional setting strategy based on the line rated load and the minimum short-circuit current at the end can maintain reliability.

[0168] Step 802: When the weighted composite direction index indicates the opposite direction, the protection exit is locked; when the weighted composite direction index is between the positive and negative direction determination thresholds, the protection exit is kept locked and the update results of subsequent sampling cycles are awaited.

[0169] In another judgment condition, if the weighted composite direction index is less than the negative of the pre-configured composite direction judgment threshold, the system confirms that the fault occurred in the reverse direction of the protection. In this state, even if the measured current amplitude has exceeded the overcurrent action threshold, the collaborative decision module will still intercept the trip permission signal, maintain the blocking state of the protection output, and avoid non-selective maloperation caused by reverse through-fault or transient current reversal.

[0170] Furthermore, when the value of the weighted composite direction index is between the positive and negative thresholds, it indicates that the consistency of the positive and negative order criteria is weak, the system output direction is in an undetermined state, the exit is locked and a short-term delay waiting mechanism is activated, and the cumulative calculation results of subsequent continuous sampling cycles are used for secondary direction confirmation.

[0171] Step 803: When the switching progress parameter reaches the upper limit value and the active power change rate falls below the pre-configured power change rate threshold based on the transient electrical quantity monitoring, and it is confirmed that the charging and discharging switching transient process is exited, the data in the negative sequence transient buffer sequence is cleared, the transient tracking state is reset and the normal protection operation mode is restored.

[0172] Based on this, the lifecycle of the transient tracking window is managed by monitoring the evolution of the switching progress parameters in real time. When the calculated normalized switching progress parameter reaches the upper limit of 1, and the active power change rate falls below the pre-configured power change rate threshold, it is determined that the charging and discharging switching action of the energy storage power converter has been physically completed, and the power grid re-enters steady-state operation. At this time, a memory cleanup operation is performed to release the storage space occupied by the negative-sequence transient cache sequence and reset the transient identification flag. Along with this reset operation, the positive-sequence expected evolution trajectory automatically degenerates into a fixed steady-state reference benchmark, the negative-sequence transient baseline is returned to zero, and the protection logic smoothly switches back to the conventional directional overcurrent protection mode that does not rely on transient analysis, achieving seamless connection across all operating conditions.

[0173] Example 9: When the power converter corresponding to the energy storage grid connection point adopts a non-standard control architecture, this example provides a method for constructing a negative sequence transient baseline alternative for the non-standard control architecture.

[0174] In practical engineering applications, some power converters do not use standard proportional-integral (PI) controllers, but instead employ proportional-resonant controllers or high-bandwidth feedforward decoupling structures. These non-standard control architectures have shorter negative-sequence closed-loop response time constants, and their negative-sequence transient components contain oscillating elements with frequencies higher than the power frequency. Under these conditions, if a long regression window is maintained, the first-order linear extrapolation model will exhibit significant tracking errors due to its inability to follow high-frequency curvature changes. To accommodate such non-standard control architectures, this embodiment provides two parallel alternative strategies to reconstruct the negative-sequence transient baseline.

[0175] In the first alternative strategy, a shortened regression window is used to reduce the data span. The preset sampling window width for extracting the negative-order transient buffer sequence is shortened from one-quarter of the power frequency cycle to one-eighth of the power frequency cycle. At a sampling frequency of 4000 Hz, this window corresponds to 10 discrete sampling points. By reducing the time length of the observation interval, the evolution trajectory of the negative-order component within this shorter interval approximates the linear condition again. After extracting the data within this shortened window, least-squares linear regression and short-time linear extrapolation steps are performed to obtain the negative-order transient baseline at the fault determination time.

[0176] In the second alternative strategy, the preset sampling window width of one-quarter of the power frequency cycle is maintained, but a second-order polynomial regression algorithm is used instead of a first-order linear regression algorithm. In the complex domain, a second-order fitting equation is constructed with sampling time as the independent variable and the negative-sequence current phasor as the dependent variable, containing quadratic, linear, and constant terms. The equation coefficients are solved by minimizing the sum of squared residuals between the data at each sampling point and the fitted curve. The specific fitting formula is as follows:

[0177] I 2_fit =A×(t i -t avg ) 2 +B×(t i -t avg )+C;

[0178] In the formula, I 2_fit For the fitted negative-sequence current phasor, A is the complex coefficient of the quadratic term characterizing the evolution curvature, B is the complex coefficient of the linear term characterizing the evolution slope, and C is the complex coefficient of the constant term. i For the i-th sampling time within the preset sampling window, t avg This is the arithmetic mean of all sampling times within the window.

[0179] Furthermore, after obtaining coefficients A, B, and C using the least squares method, the current fault determination time is substituted into the aforementioned second-order polynomial equation to calculate and output the negative-sequence transient baseline corresponding to that fault determination time. Introducing quadratic coefficients can track the curvature changes during the evolution of the negative-sequence component, reducing the fitting residuals for nonlinear high-frequency oscillations and maintaining the separation accuracy of the negative-sequence fault component under non-standard control architecture conditions. The aforementioned second-order polynomial regression substitution logic is also applicable to the construction of the negative-sequence voltage transient baseline.

[0180] In this application, an adaptive start-up criterion based on dynamic trajectory deviation is constructed to avoid the risk of false start-up or failure to operate during transient processes. Furthermore, in the face of large signal-to-noise ratio fluctuations caused by zero-power ride-through of different fault types, the limitations of fixed-priority hard switching are improved by proposing a credibility-weighted fusion mechanism based on the apparent power of fault components. This allows the contribution weights of positive and negative order criteria to be smoothly and continuously adjusted according to fault conditions, reducing the sensitivity blind zone of a single criterion, reducing decision jitter under boundary conditions, and improving the robustness and operational reliability of the protection device in complex bidirectional power flow switching scenarios.

[0181] Furthermore, the aforementioned effects can be verified by deploying the method of this invention in a protection device at the energy storage grid connection point, under simulated conditions of transient charging and discharging switching superimposed with line short-circuit faults. Unlike traditional protection methods that use a fixed steady-state polarization reference and fixed positive and negative sequence priority switching, the method of this invention can maintain the correctness of direction determination during power flow reversal transients, avoiding misjudgment of direction due to polarization reference distortion. Those skilled in the art will understand that specific protection performance indicators may vary depending on the grid topology, the proportion of energy storage capacity, and the type of fault.

[0182] The sequence numbers of the above embodiments of the present invention are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.

[0183] In the above embodiments of the present invention, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0184] The preferred embodiments of the present invention have been described in detail above. However, the present invention is not limited to the specific details in the above embodiments. Within the scope of the technical concept of the present invention, various equivalent transformations can be made to the technical solutions of the present invention, and these equivalent transformations all fall within the protection scope of the present invention.

Claims

1. A novel bidirectional power flow adaptive protection cooperative method for power system switchgear oriented towards high-proportion energy storage access, characterized in that, include: Acquire transient electrical quantities at the energy storage grid connection point; Active power calculation and transient identification are performed based on transient electrical quantities. When the power converter enters the charging and discharging switching transient process, the steady-state reference before switching is latched and the negative sequence transient buffer sequence is recorded. During the charge-discharge switching transient process, the expected trajectory of the positive sequence current is predicted in real time based on the steady-state reference and transient electrical quantities before the switching, and the negative sequence transient baseline is constructed by data-driven extrapolation based on the negative sequence transient buffer sequence. Fault determination is based on an adaptive start-up criterion constructed from the expected trajectory of positive-sequence current and the negative-sequence transient baseline. When a fault is detected, the positive-sequence fault component is removed using the expected trajectory of the positive-sequence current, and the negative-sequence fault component is removed using the negative-sequence transient baseline. Using the separated positive-sequence fault components and negative-sequence fault components, the positive and negative sequence direction discrimination polarization benchmarks that eliminate phase angle shift are reconstructed respectively. Based on the energy information contained in the positive and negative sequence fault components, the reliability adaptive weighted fusion is performed to generate a weighted comprehensive direction index. The weighted composite directional index is used in conjunction with the pre-configured overcurrent criterion for overcurrent protection based on the current current amplitude extracted from transient electrical quantities to make a coordinated decision, control the switching equipment to trip, and restore the normal protection operation mode after exiting the charging and discharging switching transient process.

2. The method according to claim 1, characterized in that, Active power calculation and transient identification are performed based on transient electrical quantities. When the power converter enters the charging / discharging switching transient process, the steady-state reference before the switching is latched and the negative-sequence transient buffer sequence is recorded. Specifically, this includes: Instantaneous active power is calculated based on transient electrical quantities, and the sliding window average active power and active power change rate are extracted. The absolute value of the active power change rate is compared with the pre-configured power change rate threshold, and a comprehensive judgment is made in combination with the monotonicity characteristics of the sliding window average active power. When the triggering condition is met, the charging and discharging switching transient process is confirmed to be entered. The positive-sequence voltage phasor, positive-sequence current phasor, and active power before the switching in the transient electrical quantities are latched as the steady-state reference before the switching, and the negative-sequence voltage phasor and negative-sequence current phasor in the transient electrical quantities are written into the negative-sequence transient buffer sequence periodically.

3. The method according to claim 1, characterized in that, Fault determination is based on an adaptive starting criterion constructed from the expected trajectory of positive-sequence current and the negative-sequence transient baseline, specifically including: Extract the current measured positive sequence current and the current measured negative sequence current from the transient electrical quantities; The deviation of the positive sequence trajectory is calculated by comparing the expected trajectory of the positive sequence current predicted analytically with the current measured positive sequence current. The sudden increment of the negative sequence baseline is calculated by extrapolating the negative sequence transient buffer sequence and comparing it with the current measured negative sequence current. When the deviation of the positive sequence trajectory exceeds the pre-configured positive sequence start threshold, or the sudden increase of the negative sequence baseline exceeds the pre-configured negative sequence start threshold, and the over-limit state is maintained for a preset time interval, a fault is confirmed to have occurred.

4. The method according to claim 1, characterized in that, Based on the analysis of switching progress, the expected trajectory of positive-sequence current is predicted to remove the positive-sequence fault component, specifically including: Based on the pre-switching active power in the pre-switching steady-state reference, the pre-configured switching target power, and the sliding window average active power calculated in real time according to transient electrical quantities, the normalized switching progress parameters are calculated. Based on the switching progress parameter and the positive sequence current full reversal increment, construct the expected trajectory of the positive sequence current; By combining the pre-configured system equivalent positive sequence impedance and the expected positive sequence current trajectory, the expected positive sequence voltage trajectory is predicted based on the positive sequence voltage phasor in the pre-switching steady-state reference. When a fault is determined, the expected values ​​corresponding to the current measured positive sequence electrical quantity and the expected trajectory of positive sequence current and positive sequence voltage are vectored to extract the positive sequence fault component.

5. The method according to claim 1, characterized in that, Data-driven extrapolation based on negative-order transient buffer sequences to remove negative-order fault components specifically includes: Extract the negative-order transient buffer sequence within the preset sampling window before the fault determination time; Apply least squares linear regression to the negative-sequence electrical quantities within the preset sampling window to obtain the negative-sequence transient rate of change. Short-time linear extrapolation is performed using the negative-sequence transient rate of change to construct the negative-sequence transient baseline at the fault determination time. The negative sequence fault component is extracted by vector difference between the measured negative sequence electrical quantity at the fault determination time and the negative sequence transient baseline.

6. The method according to claim 1, characterized in that, The positive and negative sequence direction discrimination polarization references for eliminating phase angle shift are reconstructed separately, specifically including: Using the voltage component in the positive sequence fault component as the polarization reference, the current component in the positive sequence fault component is rotated by phase angle according to the pre-configured positive sequence impedance angle and projected onto the polarization reference direction to calculate the positive sequence direction discrimination torque. Check whether the amplitude of the current component in the negative sequence fault component meets the pre-configured negative sequence sensitivity threshold. If it does, use the voltage component in the negative sequence fault component as the polarization reference, rotate the current component in the negative sequence fault component according to the phase angle of the line positive sequence impedance angle and project it to the polarization reference direction to calculate the negative sequence direction discrimination torque. The positive-sequence direction discrimination torque and the negative-sequence direction discrimination torque are used as the polarization references for positive and negative sequence direction discrimination.

7. The method according to claim 1, characterized in that, Based on the energy information contained in the positive and negative sequence fault components, a reliability-adaptive weighted fusion is performed, specifically including: The positive-sequence apparent power, which characterizes the positive-sequence energy information, is obtained by multiplying the magnitude of the voltage component and the magnitude of the current component in the positive-sequence fault component. The negative-sequence apparent power, which characterizes negative-sequence energy information, is obtained by multiplying the magnitude of the voltage component and the magnitude of the current component in the negative-sequence fault component. Based on the proportions of positive-order apparent power and negative-order apparent power in the total fault apparent power, the positive-order confidence weight and negative-order confidence weight are calculated respectively.

8. The method according to claim 1, characterized in that, The weighted composite directional index is used in conjunction with the pre-configured overcurrent criterion for in-direction overcurrent protection based on the current current amplitude extracted from transient electrical quantities to make a coordinated decision, control the switching equipment to trip, and restore the normal protection operation mode after exiting the charging and discharging switching transient process. Specifically, this includes: When the weighted composite direction index indicates a positive direction, and the current amplitude extracted based on transient electrical quantities meets the pre-configured overcurrent criterion, a trip permission signal is sent to the switchgear. When the weighted composite direction indicator points in the opposite direction, the exit is locked to protect it. When it is confirmed that the charging / discharging switching transient process has been exited, the data in the negative sequence transient buffer sequence is cleared, the transient tracking state is reset, and the normal protection operation mode is restored.

9. The method according to claim 4, characterized in that, The pre-configured system equivalent positive sequence impedance is obtained in the following way: Obtain the short-circuit capacity at the grid connection point, system impedance angle, and rated line voltage at the grid connection point from the power grid dispatching department; Based on the ratio of the square of the rated line voltage at the grid connection point to the short-circuit capacity at the grid connection point, and combined with the system impedance angle, the pre-configured system equivalent positive sequence impedance is calculated.

10. The method according to claim 2, characterized in that, The setting of the pre-configured power change rate threshold satisfies the following constraints: The lower limit of the pre-configured power change rate threshold is higher than the natural change rate of active power under normal operating conditions; The upper limit of the pre-configured power change rate threshold is lower than the minimum transient power change rate. The minimum transient power change rate is determined by the rated power of the power converter corresponding to the energy storage grid connection point, the rated power ramp-up time, and the pre-configured reliability coefficient.