A method and system for three-phase imbalance treatment of a distribution box

Abstract

This invention discloses a method and system for managing three-phase imbalance in distribution boxes. It tracks the residual current of the three phases and identifies occasional correlation patterns during periods when aggregated indicators meet standards, locating deep-seated unbalanced phases that cannot be reflected at the aggregation level. It determines the switching impact level by combining branch zero-crossing dwell characteristics and allocates commutation resources based on the historical operating status and aging degree of contacts, taking into account equipment lifespan losses. After switching is performed from low to high impact level, the actual connected phase of the branch is verified by instantaneous interruption recovery waveform, and the original wiring records are corrected, avoiding reliance on unverified command states for continuous judgment. Furthermore, it verifies the operating sequence between distribution boxes from the same source and disperses the switching direction to avoid the risk of local superimposed impacts and predicts the management effect under the most unfavorable scenario. Finally, it outputs a dispatch command only when the management benefit can cover the switching loss, reducing the impact of commutation management on equipment lifespan and the stability of power supply in the distribution area.

Description

Technical Field

[0001] This invention relates to the field of power distribution network operation monitoring technology, and in particular to a method and system for managing three-phase imbalance in a distribution box. Background Technology

[0002] In actual operation, low-voltage distribution substations often experience challenges in maintaining perfect three-phase load balance due to the randomness and volatility of user load connections. Long-term phase-to-phase current imbalance not only increases neutral line current and line losses but also accelerates localized overheating and aging of distribution transformers, directly impacting the lifespan of distribution equipment and the reliability of power supply. Current imbalance mitigation methods typically rely solely on whether the phase-to-phase imbalance at the aggregation level exceeds the standard. Once this indicator falls within the compliant range, investigations cease. However, aggregation indicators themselves have blind spots. Structural deviations where one phase is consistently heavier than the other two can be diluted by averaging effects, appearing normal on the surface, thus preventing the true problem from being detected for a long time.

[0003] In addition, existing methods often only focus on whether the command is issued when performing phase switching operations, lacking assessment of the impact of switch switching and contact life loss, and failing to verify whether the load is actually connected to the target phase after the command is issued; when multiple distribution boxes in the transformer area perform phase switching operations at the same time, there is also a general lack of unified timing coordination, which can easily cause short-term superimposed impacts in some areas. Summary of the Invention

[0004] This invention discloses a method and system for managing three-phase imbalance in distribution boxes. The aim is to continue tracking the residual current of the three phases within the range where the aggregated indicators meet the standards, identify sporadic correlation patterns to locate the deep-seated unbalanced phases, allocate management resources to switchable branches by combining switching impact calculations and contact life status, verify the actual connected phases and correct wiring records after execution, and conduct overall verification of the action sequence and switching direction of distribution boxes from the same source. Execution is only initiated when the management benefits cover equipment losses, achieving refined management that balances identification depth, equipment safety, and multi-device collaboration.

[0005] The first aspect of this invention proposes a method for mitigating three-phase imbalance in a distribution box, comprising: The operation data of the transformer substation is obtained, and the phase-to-phase and neutral line current imbalance is calculated based on the operation data to form the imbalance characteristics of the transformer substation. The abnormality of the standard period is screened through the imbalance characteristics of the transformer substation to determine the hidden imbalance phase. Based on the implicit unbalanced phase, the switchable branch collection is performed to obtain the branch set to be switched. The switching impact is measured on the zero-crossing dwell short-short branch in the branch set to be switched to determine the branch impact level. Based on the branch impact level, the life quota allocation is performed to determine the commutation number quota. Based on the commutation frequency quota, a phase switching sequence is generated by switching in an ascending order of impact. The measured phase deviation is obtained by verifying the ownership of the transient recovery waveform in the phase switching sequence. The wiring record is corrected according to the measured phase deviation to form a corrected wiring record. Based on the revised wiring record, phase-switching planning is performed to form a phase-switching scheme. The risk of overlapping distribution boxes from the same source in the phase-switching scheme is verified on the transformer substation side to generate a switching direction limit. Based on the switching direction limit, the most unfavorable simultaneous rate transformer substation imbalance is estimated to obtain the transformer substation side impact result. The gain reduction is performed by using the disturbance loss in the affected area results to obtain effective governance benefits. The quota comparison and verification is performed based on the effective governance benefits to determine the commutation start criterion. The phase-by-phase action is planned and the commutation scheduling command is output based on the commutation start criterion.

[0006] A second aspect of this invention provides a three-phase imbalance mitigation system for a distribution box, comprising: The balance identification module is used to acquire the operating data of the transformer area, calculate the phase-to-phase and neutral line current imbalance based on the operating data of the transformer area to form the imbalance characteristics of the transformer area, and screen the abnormality of the standard period through the imbalance characteristics of the transformer area to determine the hidden imbalance phase. The quota allocation module is used to perform phase-switching branch aggregation based on the implicit unbalanced phase to obtain the phase-to-switch branch set, measure the switching impact of the zero-crossing dwell short-short branch in the phase-to-switch branch set to determine the branch impact level, and perform lifetime quota allocation based on the branch impact level to determine the switching number quota. The execution verification module is used to generate a phase switching sequence based on the phase switching quota in an ascending order of impact, use the transient recovery waveform in the phase switching sequence to verify the attribution and obtain the measured phase deviation, and correct the wiring record for the measured phase deviation to form a corrected wiring record. The superposition and coordination module is used to execute phase-by-phase switching planning based on the corrected wiring record to form a phase-by-phase switching scheme, to perform area-side verification of the risk of superposition of the same source distribution box in the phase-by-phase switching scheme to generate a switching direction limit, and to estimate the most unfavorable simultaneous rate of area imbalance based on the switching direction limit to obtain the area-side impact result. The instruction generation module is used to reduce the gain by applying the disturbance loss in the influence results of the transformer area to obtain effective governance benefits, and to verify the quota comparison based on the effective governance benefits to determine the commutation start criterion. Based on the commutation start criterion, it plans phase-by-phase actions and outputs commutation scheduling instructions.

[0007] The beneficial effects of this invention are reflected in the following points: 1. Meeting the aggregated index for phase imbalance does not equate to true three-phase load balance. A situation where one phase is consistently heavier than the other two may be masked by the averaging effect. This invention further tracks the regular deviations of the residual current in each phase within the compliance period, identifying deep-seated imbalance phases that cannot be reflected by the aggregated index. Combined with the zero-crossing dwell characteristics of the branch and the historical contact operation status, it rationally allocates management resources to branches with commutation capabilities. 2. After completing the switching action, this invention further verifies the actual phase connected to the branch by restoring the waveform and synchronously updates the verification result to the system record. Compared to relying solely on issuing commands to confirm the execution result, this verification step can promptly detect situations where commands do not match reality, such as switch jamming or wiring deviations. This ensures that the wiring log always reflects the true operating status of the equipment, preventing erroneous states from being continuously used as correct records for extended periods. 3. If the commutation decision for a single branch is not considered from the perspective of the entire distribution area, it can easily lead to new local risks and resource waste: simultaneous switching of multiple distribution boxes to the same phase may cause short-term cumulative impacts, while solutions with limited benefits are difficult to compensate for equipment losses after implementation. This invention performs unified verification and directional dispersion of the operation sequence of distribution boxes from the same source, and compares the governance benefits and switching losses before execution, issuing dispatch instructions only for solutions where the benefits can cover the losses. Attached Figure Description

[0008] Figure 1 This is a flowchart of a method for managing three-phase imbalance in a distribution box according to the present invention.

[0009] Figure 2 This is a schematic diagram of the three-phase load distribution and phase commutation topology of the same source distribution box in the transformer area of ​​the present invention.

[0010] Figure 3 This is a structural block diagram of a three-phase imbalance mitigation system for a distribution box according to the present invention.

[0011] Wherein: 1-Distribution transformer; 2-Three-phase busbar; 3-Grouping of same-source distribution boxes; 4-Distribution box; 5-Branch; 6-Switching switch; 7-Currently connected phase; 8-Target phase; 9-Switching path. Detailed Implementation

[0012] In the following description, specific details such as particular system architectures and techniques are set forth for illustrative purposes and not for limitation, in order to provide a thorough understanding of the embodiments of this application. However, those skilled in the art will understand that this application may also be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, apparatuses, circuits, and methods have been omitted so as not to obscure the description of this application with unnecessary detail.

[0013] References to "one embodiment" or "some embodiments" as described in this specification mean that one or more embodiments of this application include a specific feature, structure, or characteristic described in connection with that embodiment. Therefore, the phrases "in one embodiment," "in some embodiments," "in other embodiments," "in still other embodiments," etc., appearing in different parts of this specification do not necessarily refer to the same embodiment, but rather mean "one or more, but not all, embodiments," unless otherwise specifically emphasized. The terms "comprising," "including," "having," and variations thereof mean "including but not limited to," unless otherwise specifically emphasized.

[0014] The technical solutions of the embodiments of this application will be described below.

[0015] like Figure 1 As shown, this embodiment of the invention provides a method for mitigating three-phase imbalance in a distribution box, including the following steps S101-S105: Step S101: Obtain the operating data of the transformer area, calculate the phase-to-phase and neutral line current imbalance based on the operating data of the transformer area to form the imbalance characteristics of the transformer area, and identify the hidden unbalanced phases by screening the abnormal periods of compliance through the imbalance characteristics of the transformer area.

[0016] Specifically, the system acquires transformer area operation data. Within the transformer area, the integrated terminal collects the three-phase currents Ia, Ib, and Ic, and the neutral current In from the main meter of the transformer area at a unified cycle. Smart meters synchronously collect the phase currents of their respective branches. The collection cycle is once every 15 minutes. The sampling times of all terminals within the same transformer area are aligned using a unified clock to avoid data misalignment within the same time period due to local clock drift. The integrated terminal is responsible for current collection from the main meter of the transformer area, while the smart meters are responsible for current collection from each branch metering point. Together, they constitute a complete source of transformer area operation data. The transformer area operation data records the four values ​​of the three-phase current and neutral current collected in each moment, and archives them in chronological order of collection time. If a current is missed due to communication interruption, a missing measurement flag is recorded at that moment. If a missing measurement exceeds four consecutive collection cycles, the entire segment is skipped without interpolation. In the extreme case of simultaneous three-phase current loss, the entire transformer substation's operational data record is marked as all phases missing at that moment, without further distinction of which phase is missing. The neutral current In is measured and connected to the same acquisition circuit via their respective current transformers, maintaining the same level of measurement accuracy. Transformer substation operational data is archived according to substation number. Data reported by multiple acquisition terminals within the same substation undergoes timestamp alignment verification before archiving. Data failing verification is individually marked as timestamp abnormal and is not forcibly spliced ​​with data from other terminals. The communication link between the converged terminal and the smart meter uses a unified carrier or wireless channel. In the event of a single communication failure, the terminal locally caches the acquisition result and retransmits it in the next communication cycle. Successfully retransmitted records are archived at the original acquisition time. The number of acquisition terminals in the same substation within the operational data varies with the actual number of connected branch metering points. When adding or removing terminals, previously archived records are not retrospectively unified.

[0017] The imbalance characteristics of a transformer substation are calculated based on the phase-to-phase and neutral current imbalances measured using substation operation data. The phase-to-phase imbalance is calculated using the formula ε(t) = (I_max(t) - I_min(t)) / I_avg(t), where I_max(t), I_min(t), and I_avg(t) are the maximum, minimum, and average values ​​of the three-phase currents at that moment, respectively, all taken from the substation operation data records at the same moment. ε(t) represents the phase-to-phase imbalance at that moment. The substation imbalance characteristics consist of two values: ε(t) at that moment and the neutral current In(t) at the same moment. These two values ​​are recorded in pairs for each moment according to the same acquisition cycle of the substation operation data. For any equivalent missing measurement of the three-phase current, the substation imbalance characteristics for that moment are not calculated; insufficient data is directly noted without estimation or filling. Similarly, for moments in the substation operation data where all phases are missing, the substation imbalance characteristics are not calculated, and the processing method is the same as for single-phase missing measurements. The neutral current In(t) itself is not involved in the calculation of ε(t). It is only archived as an independent dimension along with the transformer area's imbalance characteristics for future reference. It does not participate in subsequent calculations and judgments such as compliance determination and residual tracking, nor is it weighted and merged into a single indicator, retaining its independent recording value. The transformer area's imbalance characteristics are archived sequentially time-by-time without cross-time-time moving average processing, preserving the original fluctuation details as a basis for judging whether there are hidden deviations within the compliance period. In extreme cases where the value of one phase current is significantly higher than that of the other two phases by several times, it is judged as a suspected short circuit rather than a regular imbalance. This moment is not included in the regular statistics of the transformer area's imbalance characteristics and is separately marked as a suspected short circuit. The calculation of ε(t) strictly uses the measured values ​​of the three-phase currents at the same moment, without using the moving window average as a substitute, ensuring that the transformer area's imbalance characteristics reflect the true current distribution at that moment.

[0018] In some embodiments, the step of determining latent imbalance phases by screening for abnormalities in the target period through the imbalance characteristics of the transformer area includes: obtaining a target period sequence by screening for target periods based on the imbalance characteristics of the transformer area; generating target period residual features by performing residual tracking based on the target period sequence; identifying occasional associated phase features using the target period residual features to determine occasional associated phases; and verifying the latent asymmetry attribution based on the occasional associated phases to determine latent imbalance phases.

[0019] The qualified time period sequence is obtained by screening the unbalance characteristics of the distribution transformer area. The phase-to-phase unbalance degree ε(t) in the unbalance characteristics of the distribution transformer area is compared with the standard limit corresponding to the voltage level of the distribution transformer area. The standard limit is the upper limit of phase-to-phase unbalance degree given in the operation procedure of the distribution transformer of that type. Different voltage levels correspond to different limits. For 10 kV distribution transformer areas, it is usually 15%, and for low-voltage distribution transformer areas, it can be relaxed to 20%. The specific value is determined by referring to the table according to the actual connected voltage level of the distribution transformer area. The moment when ε(t) does not exceed the standard limit is judged as a qualified moment. The boundary case where ε(t) is exactly equal to the standard limit is also judged as qualified, without additional judgment of choosing the lower limit. The interval of consecutive qualified moments is merged into one qualified time period. If a non-qualified moment is inserted between two qualified moments, it is regarded as the boundary between the two qualified time periods, and no merging is performed. The boundary treatment ensures that the qualified time period sequence is continuously qualified throughout the entire time period without interruption. The compliance time sequence records the start and end times of all compliance times for the transformer substation within the current assessment period, arranged in chronological order. Times with insufficient imbalance characteristic data for a transformer substation are not included in the compliance determination and do not interrupt the continuity determination of adjacent compliance times. If both times before and after the time with insufficient data are compliance times, they are considered as the same uninterrupted compliance time period. A single compliance time period lasting less than 30 minutes is considered an occasional compliance occurrence and is not included in the compliance time sequence; only officially compliance times are counted. For transformer substations that do not meet the standard at any time during the entire assessment period, the compliance time sequence is empty, and the substation is determined to have a persistent and significant imbalance, which is not within the scope of the implicit problems addressed by this method. Standard limits are revised annually in conjunction with the latest distribution network operation standards. The revised limits only apply to compliance determinations after the revision takes effect and do not retrospectively reassess compliance time sequences generated before the revision.

[0020] Residual characteristics for the compliance period are generated by residual tracking based on the compliance period sequence. Within each compliance period covered by the compliance period sequence, the residuals of the three-phase currents relative to the mean are calculated for each sampling point. The residuals are calculated using the formula ΔIx(t) = Ix(t) - I_avg(t), where x is one of the three phases a, b, and c, Ix(t) is the measured current at that corresponding moment, and I_avg(t) is the mean of the three phases. ΔIx(t) is the residual at that corresponding moment; a positive value indicates that the current of that phase is higher than the mean, and a negative value indicates that it is lower than the mean. The residual characteristics for the compliance period record the residual sequence phase by phase and moment by moment. The three-phase residuals satisfy the identity relationship that the sum of the three is zero at the same moment. When recording, all three phase values ​​are retained instead of storing only two phases and then calculating the third phase. The identity relationship serves as an implicit check for data integrity. If the sum of the three at a certain moment deviates significantly from zero, the data at that moment is judged to be abnormal and is not included in the residual characteristics for the compliance period. When the proportion of residuals with the same sign for a certain phase within the same compliance period reaches more than 80% of the total duration of that period, the compliance period residual characteristics are associated with the continuous bias indicator of that phase. This proportion is taken from the statistical experience value of the residual distribution of historical compliance periods and is used to distinguish between true structural bias and random fluctuations. Residuals are not calculated for non-compliance intervals between compliance periods. The compliance period residual characteristics only cover the compliance periods, and the residual sequences between two compliance periods are independent of each other. The number of sampling points in the residual sequence corresponds one-to-one with the number of sampling points in the corresponding compliance period's transformer area operation data, ensuring that the temporal resolution of residual tracking is consistent with the original acquisition cycle. In the case of a continuous absence of measurement for a certain phase, the transformer area imbalance characteristics at the corresponding time cannot be calculated, resulting in an empty compliance period sequence within the evaluation cycle of that transformer area. In this case, the compliance period residual characteristics are also empty, and the transformer area is treated as the case of continuous significant imbalance described above, and this step is not performed separately.

[0021] The residual characteristics of the compliance period are used to identify and determine the characteristics of intermittently correlated phases. For phases with a persistent bias in the residual characteristics of the compliance period, further verification is conducted to determine whether the time points of increased residuals for that phase exhibit regular repetition across multiple natural days. Specifically, each day is divided into 96 time slots, with the slot width corresponding to a 15-minute data collection period. The frequency of time slots where the residual of a phase exceeds 1.5 times its own standard deviation on different dates is statistically analyzed. The standard deviation, rather than the mean, is used as the benchmark because the residual fluctuates around zero, and using the mean as a multiple benchmark would easily lose its discriminative significance. If the frequency of a certain time slot within the statistical period (the most recent 30 days) reaches more than 60%, it is determined that the phase has an intermittent correlation characteristic in that time slot. The intermittent correlation characteristic records the time slot number and corresponding frequency. This intermittent correlation phenomenon often corresponds to the concentrated start-up and shutdown habits of a certain type of load during specific periods, causing the phase current to deviate regularly from its own fluctuation range during these fixed periods, while returning to stability at other times. This periodic regular deviation is precisely the signal that the intermittent correlation characteristic aims to capture. When the same phase meets the hit condition in multiple time slots, the occasionally associated phases are recorded separately for each time slot and not merged into a single interval, because different time slots may correspond to different load triggers and should not be mixed. The occasionally associated phase record records the phases that hit the occasionally associated characteristics and their corresponding time slot distribution within the target achievement period sequence. If two of the three phases simultaneously hit the occasionally associated characteristics, the occasionally associated phase record records the time slot distribution of each of the two phases separately, without retaining only one phase due to the coexistence of two phases. Phases whose residual characteristics during the target achievement period do not carry a persistent bias indicator are not included in the time slot statistics of this sub-step; it is directly determined that there are no occasionally associated phases for that phase.

[0022] The latent imbalance phases are determined by verifying the latent asymmetric attribution based on the occasional correlation phase identification. The distribution of hit time slots in the occasional correlation phases is checked, and it is verified whether the corresponding frequency reaches 60% or higher in both the first and last 15 days of the most recent 30-day statistical window. If both sub-windows meet the standard, it is considered a continuous recurrence, and the attribution is confirmed. If only one sub-window meets the standard, and the other sub-window is below 60%, it is considered an occasional fluctuation within that window and is not confirmed. The 15-day division is to verify that the occasional correlation phenomenon has stability across a half-month timescale, rather than being a short-term coincidence concentrated in a few days. If the transformer area has been in operation for less than 30 days and has not yet accumulated a complete statistical window, the attribution verification is changed to a 50 / 50 division based on the actual number of days already in operation. The latent imbalance phase record includes the phases whose attribution has been confirmed, the corresponding time slot distribution, and the hit frequency in each of the two sub-windows. In cases where both phases in an occasional associated phase match are correctly identified and their attribution is confirmed, the confirmation results for each latent imbalance phase are recorded separately without prioritization, and the time slot distributions of the two phases are archived independently. Phases that fail attribution confirmation are not counted as latent imbalance phases, and the assessment within the time frame of that compliant period is considered as having no latent imbalance detected. The attribution confirmation criterion relies solely on the frequency data carried by the occasional associated phases, without introducing any additional external information to assist in the judgment, ensuring that the input source for the judgment process is singular and traceable. Once a latent imbalance phase is confirmed, it is considered the final conclusion of that assessment cycle, and will not be re-verified within the same assessment cycle. Newly collected data will be included in the statistics of the next assessment cycle.

[0023] Step S102: Based on the implicit unbalanced phase, perform phase switching branch aggregation to obtain the set of phase switching branches, measure the switching impact of the zero-crossing dwell short branches in the set of phase switching branches to determine the branch impact level, and perform life quota allocation to determine the commutation number quota based on the branch impact level.

[0024] Specifically, the set of branches to be commutated is obtained by performing phase-switching branch aggregation based on the implicit unbalanced phases. For example... Figure 2As shown, branch 5 within the transformer area with commutation capability refers to branch 5 with a relay group commutation switch 6 installed on the load side. This type of switch can arbitrarily switch and connect the branch load in the three-phase busbar 2. The switch is installed downstream of the branch main switch. The list of branches with this capability is registered with the transformer area equipment ledger. When two phases of latent imbalance are simultaneously affected, both phases are included in the current collection scope. Subsequent collection and impact assessments are carried out separately for each phase without prioritization to avoid delaying the investigation of the other phase. The selection of the two target phases is coordinated in the subsequent phase-by-phase commutation planning stage to avoid concentrating on the same phase. The latent imbalance phase is only marked with the phase and the time pattern of the occurrence, without carrying the specific load amplitude. This step additionally verifies the actual load current level of the branch and includes it in the internal record of the branch set to be commutated, as the basic data for measuring the extent to which the imbalance can be improved after the branch is commutated. Equipment replacement on-site or oversights in record-keeping can cause discrepancies between the connected phases recorded in the record and the actual phases of the branches. The current connected phases of the branches recorded in the inventory must be compared item by item with the actual connected phases verified during the most recent on-site inspection. If there are discrepancies, the on-site verification result shall prevail, and the branch group awaiting phase replacement shall be reassessed accordingly to determine whether the branch is truly connected to an overloaded phase. If a branch in the branch group awaiting phase replacement is currently locked under other maintenance or commissioning / decommissioning operations, it shall not be included in this collection to avoid the same switch being operated by two different work processes simultaneously. Branches in the branch group awaiting phase replacement are grouped by connected phase, and within each group, they are further arranged according to their rated capacity from largest to smallest. Branches with larger capacities contribute more significantly to improving phase imbalance.

[0025] In some embodiments, determining the branch impact level by measuring the switching impact of the short-staying branches with zero-crossing dwell time in the set of branches to be switched includes: statistically analyzing the zero-crossing dwell time of each branch in the set of branches to be switched to form a dwell time distribution; screening short-staying branches based on the dwell time distribution to generate a short-staying branch set; calculating the switching impact amplitude by using the short-staying branch set; and verifying the amplitude classification by the switching impact amplitude to determine the branch impact level.

[0026] For each branch in the set of branches to be commutated, the dwell time distribution at zero crossing is statistically analyzed. For each branch in the set of branches to be commutated, a power quality monitoring terminal integrated with the commutation switch collects the current waveform in real time within the most recent complete power frequency cycle (corresponding to 20 milliseconds of a 50 Hz power grid frequency). The sampling density is 200 points per cycle, which is the standard configuration for this type of terminal and is sufficient to distinguish millisecond-level waveform details near the zero crossing. A window is taken before and after each zero crossing; the duration during which the absolute value of the current is lower than 5% of the rated current within the window is recorded as the dwell time of that zero crossing. For branches dominated by resistive loads, the current waveform is close to an ideal sine wave, and the dwell time is stable and close to a fixed proportion of the cycle. For branches dominated by inductive loads, as long as the current waveform remains close to a sine wave, the dwell time is similar to that of resistive loads and does not shift independently due to phase lag compared to voltage. For branches containing nonlinear loads such as switching power supplies, the current waveform is prone to multiple small fluctuations or even multi-peak phenomena near the zero-crossing point. The dwell time distribution of these branches is significantly shortened and fluctuates more. The dwell time distribution is generated by taking the most recent complete cycle rather than averaging historical multiple cycles because the waveform shape of the same branch can change at different times of the day. Multi-cycle averaging would smooth out these time-varying differences and mask the actual switching risk at the current moment. When there are many branches waiting to be commutated, the waveform acquisition and dwell time calculation for each branch are completed sequentially. The processing time for a single branch is usually on the order of hundreds of milliseconds, and the batch processing will not put significant load pressure on the real-time monitoring system of the distribution area.

[0027] A short-dwelling branch set is generated by screening short-dwelling branches based on the dwell time distribution. All branches in the dwell time distribution are sorted from shortest to longest dwell time. Branches within the shortest 20% of the sorted range are selected as preliminary candidates. This proportion corresponds to the quantile where the inrush current risk significantly deviates from the statistical characteristics of the normal waveform group in field switching cases of this type of commutator switch. Among the preliminary candidates, branches with an absolute dwell time value less than 5% of the power frequency cycle are judged as short-dwelling and included in the short-dwelling branch set. In actual transformer substation scenarios, most branches are connected to near-linear loads such as lighting and conventional household appliances, with waveforms close to sine waves. The dwell time values ​​of these branches are generally sufficient. Those that truly fall into the short-dwelling branch set are usually branches connected to non-linear power electronic loads such as frequency converters, switching power supplies, and LED drivers. These types of loads introduce harmonics in their design, making the current waveform steeper near the zero-crossing point. Branches in the short-staying branch set are also labeled with load type. The load type information is taken from the registration record of the terminal equipment connected to the branch in the equipment ledger. If the registration record is missing, it is marked as a missing load type. Branches with a lower dwell time ranking but still higher than the threshold often appear in scenarios where the overall load level of the distribution area is high and the dwell time distribution of most branches is relatively tight. In this case, the two criteria of relative ranking and absolute threshold are inconsistent, indicating that a comprehensive judgment should be made in conjunction with the overall load level of the distribution area. In addition to the steep zero-crossing point, the current waveform of inverter loads is usually accompanied by obvious high-order harmonic components. The harmonic components are not within the calculation range of this sub-step, but the phenomenon of shortened dwell time is positively correlated with harmonic content, which is one of the indirect evidences for determining whether the branch should be included in the short-staying branch set.

[0028] The inrush current amplitude during commutation is obtained by measuring the inrush current using a short-dwelling branch set. For each branch in the short-dwelling branch set, the maximum rate of change (dI / dt)_max of its current waveform near the zero-crossing point is taken. The rate of change is calculated point-by-point using the first-order difference of the current value within the dwell window, and the maximum value is taken. The inrush current amplitude is calculated using the formula I_inrush=k×(dI / dt)_max×L_eq / R_eq, where k is the dimensionless coupling coefficient of this type of commutation switch, L_eq is the equivalent inductance of this branch, R_eq is the equivalent resistance of this branch, and I_inrush is the measured inrush current amplitude. This relationship reflects the hindering effect of the branch inductance on the sudden change in current at the instant the commutation switch contacts close. The more drastic the current change and the larger the branch inductance, the greater the instantaneous impact on the contacts. k here serves as an empirical conversion coefficient from the inductance hindering effect to the inrush current, rather than a single physical quantity in a strict sense. The coupling coefficient k is obtained through back-calibration using measured inrush current data of this type of commutator switch, and its value incorporates multiple influences such as contact structure and magnetic circuit. For short-staying branches and concentrated branches where the corresponding load type is missing or a typical inductance value cannot be found, L_eq is replaced by the median inductance within the applicable load range of this type of commutator switch. The switching impact amplitude calculation result is associated with the inductance substitution indicator to indicate that the accuracy of this value is lower than the result obtained based on the measured inductance. The equivalent inductance of variable frequency equipment loads fluctuates with operating conditions. For these branches, L_eq is taken from the rated operating condition inductance value marked on the equipment nameplate, without real-time operating condition tracking calculation. Therefore, the calculated switching impact amplitude does not reflect instantaneous changes in operating conditions. The coupling coefficient k of this type of commutator switch reflects the suppression or amplification effect of the internal contact gap and magnetic circuit structure on inrush current. Therefore, the k value cannot be universally applied across models.

[0029] The branch surge level is determined by verifying the surge amplitude of the switching surge. The I_inrush value in the surge amplitude is compared with the rated surge current rating table of this type of commutator switch. The rating limit is taken from the corresponding surge current threshold given in the switch's factory test report. This rating table was developed with reference to the destructive test results of the same batch of switches in a laboratory environment. The surge current amplitude is gradually increased until visible ablation marks appear on the contacts. The upper limit of the severe rating is determined by leaving a safety margin below this critical value. The physical consequences of the three surge levels are clearly different: mild surge usually only causes slight heating of the contact surface and does not affect normal conduction; moderate surge may accelerate the wear rate of the contact surface coating; severe surge has the risk of local ablation or even instantaneous adhesion of the contact surface. Once adhesion occurs, the switch may not be able to disconnect normally and requires on-site manual handling. Branches with I_inrush below the mild threshold are considered to have negligible switching risk, and the branch surge level is directly taken as mild and included in the subsequent switching queue in the highest priority order. For branches where I_inrush falls within the three-level range, the branch impact level is determined by the corresponding level. For branches with switching impact amplitude associated with the inductor replacement indicator, the branch impact level determination result is also associated with this indicator. If the severe level is calculated based on the replacement inductor, it is recommended to perform an on-site test verification before performing commutation. When different switching impact amplitudes are obtained from multiple calculations for the same branch, the branch impact level is taken from the highest level, because the physical damage to the contacts is cumulative. For branches corresponding to moderate and severe impact levels, commutation should be performed during periods of relatively low load to further reduce the actual impact during switching.

[0030] In some embodiments, the step of determining the commutation frequency quota based on the branch impact level by performing lifetime quota allocation includes: performing historical action analysis based on the branch impact level to obtain the cumulative number of actions; calculating the contact idle oxidation based on the cumulative number of actions to obtain the oxidation correction lifetime; using the oxidation correction lifetime to allocate quota gradient to obtain the commutation quota gradient; and verifying the commutation frequency quota by extending the commutation quota gradient to determine the commutation frequency quota.

[0031] The cumulative number of actions is obtained by analyzing historical actions based on the branch impact level. For branches with a determined impact level, the system queries all historical commutation action records of the branch's commutator since commissioning, and associates each action record with the time of occurrence and the corresponding branch impact level at that time. Different impact levels do not have equal impact on the actual wear of the switch life. The wear caused to the contacts by a single severe impact may be equivalent to the cumulative effect of multiple mild impacts. Therefore, the cumulative number of actions is counted separately according to the impact level. The total number of actions used in the nominal remaining life calculation below is the equivalent number after weighting the statistics by each level, rather than a simple arithmetic sum without distinguishing levels. The weight values ​​are specified according to the switch model. The action records do not distinguish whether the action was ultimately successfully executed, because regardless of whether the closing is successfully completed, the contacts have already experienced one mechanical contact at the moment of action, and the wear has already occurred at the moment the action occurs, which is irrelevant to the final result of the action. For branches that are commissioned for the first time and have no historical action records, the cumulative number of actions for each level starts from zero. If a branch switch body has been replaced due to a fault, its operation records before replacement are stored in a separate ledger for the old body. The cumulative operation count for the new body restarts from the replacement time, as the physical wear condition of the new contacts is completely unrelated to the old contacts. If a branch switch with a severe impact level has occurred more than three times consecutively in the cumulative operation count, the system will additionally indicate that the contacts of that branch switch are showing an accelerated aging trend, prompting priority for equipment inspection. For some older models of phase-changing switches, their historical operation records may be stored in different log formats; these format-different records are also included in the cumulative operation count.

[0032] The oxidation correction life is obtained by calculating the contact oxidation during idle periods based on the cumulative number of operations. The total number of cumulative operations is compared with the rated base life L_base (unit: cycles, i.e., the rated total number of commutations) specified by the manufacturer for this type of commutator to obtain the nominal remaining life of the branch switch. When the commutator switch is not operated for a long time, an oxide film will slowly oxidize on the contact surface due to oxygen, resulting in poor conductivity. The longer it is idle, the thicker the oxide film becomes, increasing the contact resistance and making it prone to local overheating or even micro-arc discharge at the moment of closure. This additional loss caused by idle periods cannot be reflected by simply counting the number of cumulative operations, so it is necessary to introduce idle time for correction. The oxidation correction lifetime is calculated using the formula L_ox = L_rem × (1 - β × T_idle), where L_rem is the nominal remaining lifetime of the branch switch (L_rem = L_base - N, where N is the total number of cumulative operations), T_idle is the idle time of the branch switch since its last actual operation (in days), and β is the oxidation decay coefficient (in days). -1The oxidation correction life is calculated based on the oxidation rate of the switch contact material measured in accelerated aging tests. L_ox represents the life loss due to both mechanical wear and oxidation during idle periods. When T_idle exceeds the maximum safe idle time specified for this switch model, the oxidation correction life calculation result is capped at zero. Before commutation, this branch switch must undergo a continuity test to confirm that the contact conductivity has not failed due to excessive oxide film thickness. For branches where the total number of cumulative operations has reached or exceeded L_base, the mechanical wear of the contacts has reached the design limit, and the oxidation correction life is no longer calculated; the switch is directly transferred to equipment replacement evaluation. The β coefficient of this switch model is calibrated in conjunction with the anti-sulfurization performance of the contact material; the calibration result directly affects the accuracy of the oxidation correction life calculation.

[0033] The commutation quota gradient is obtained by using the oxidation correction life allocation quota gradient. The oxidation correction life value is converted into a quota base according to different tiers: branches with an oxidation correction life value exceeding 70% of the basic life of the switch model have contacts in near-new condition, and a higher commutation quota gradient is applied, allowing for more commutations within the current evaluation cycle; branches between 30% and 70% have accumulated some contact wear, and a middle commutation quota gradient is applied; branches below 30% have contacts nearing the end of their lifespan, and a lower commutation quota gradient is applied. Even if the current evaluation determines that the branch is connected to an overloaded phase, the frequency of commutation needs to be limited to delay equipment degradation. When the oxidation correction life tier of the same branch changes compared to the previous evaluation cycle, the commutation quota gradient is recalculated based on the latest tier. Since contact condition is a continuously changing process, the quota allocation needs to be adjusted in real time to keep pace with this change. The quota base is further adjusted in tiers based on the impact level of the branch: For branches with a severe impact level, the actual consumption of relative lifespan is greater with each switching operation, and the switching quota gradient is lowered from the original level to avoid the quota allocated based on nominal lifespan masking the additional losses actually caused by a single operation. Branches with an oxidation correction lifespan reduction ratio (i.e., β×T_idle) exceeding 50% are classified as severely oxidized. For branches with contacts classified as severely oxidized, the switching quota gradient is lowered by one level, and those already at the lowest level are not lowered further because oxidation increases the risk of each operation. For branches with contacts in the middle level, if the impact level of this switching operation is determined to be severe, the quota base will be lowered by one level from the middle level; if the branch is also classified as severely oxidized, it will be lowered by another level. When both downgrade conditions are combined, the cumulatively lowered quota will be applied.

[0034] The phase switching quota is determined by extending the phase switching frequency quota based on the phase switching quota gradient verification. The quota base given by the phase switching quota gradient is compared with the actual number of phase switching operations used in the previous assessment cycle for that branch. For branches with a quota surplus in the previous cycle, the surplus is incorporated into the phase switching frequency quota at a 50% conversion ratio. This conversion, rather than full carry-over, is to prevent branches that have not triggered phase switching demand for a long time from accumulating quotas and causing a large number of actions to be released in a concentrated manner once a phase switching demand is triggered in a certain assessment cycle. The impact of such concentrated release on the contacts is no less than that of continuous high-frequency daily switching. For branches newly included in this phase switching scope that have no quota usage records to compare in the previous cycle, the phase switching frequency quota is directly executed according to the quota base of this cycle, without any carry-over conversion. For branches that have been determined to be transferred to replacement assessment and have not participated in the phase switching quota gradient allocation, their phase switching capabilities are in a pending state. This sub-step also does not calculate the phase switching frequency quota for these branches, and the actual phase switching actions are handled separately by the equipment replacement process. If the actual number of operations on a certain branch line exceeds the quota in a given cycle, this is mostly due to temporary over-quota operations caused by emergency on-site fault handling. Such over-quota records do not affect the calculation method of the current commutation quota base; the over-quota is recorded as an independent event in the branch line's operation history. The remaining operations are discounted by 50% and incorporated into the current commutation quota, meaning the corresponding discounted number of operations is added to the original quota base. When the remaining number of operations is odd, the discounted result is rounded down to avoid values ​​like half-quota, which have no practical operational significance. Each time the commutation quota is included in a commutation scheduling instruction, it is considered consumed once. The quota balance is updated in real-time by subtracting the consumed operations in the current assessment cycle from the quota base.

[0035] Step S103: Based on the commutation frequency quota, a phase switching sequence is generated by switching in an ascending order of impact. The phase switching sequence is used to verify the attribution of the transient recovery waveform to obtain the measured phase deviation. The wiring record is corrected to form a corrected wiring record based on the measured phase deviation.

[0036] Specifically, a phase switching sequence is generated based on the commutation frequency quota and in ascending order of impact. For branches with a commutation frequency quota greater than zero, their corresponding impact levels are extracted one by one, and the switching queue is arranged in order of impact level from mild to severe, with mild branches at the front of the queue for first execution. When there are multiple branches within the same impact level, they are arranged in descending order of oxidation correction lifetime values, with branches with higher values ​​being executed first. This allows contacts in poor condition to be involved in the phase switching sequence as late as possible. This situation is common in scenarios where branches in the same batch have similar years of operation but different contact wear levels due to load fluctuations. Branches with less wear are tested first, while branches in worse condition are scheduled to be involved in the current switching later. The phase switching sequence determines a specific target phase and a scheduled execution time for each branch. The target phase is selected as the phase with the lowest current load level, excluding the currently overloaded phase. The scheduled execution times are arranged in a queue order with intervals of at least 5 minutes between adjacent branches to allow sufficient time to observe the actual effect of the previous switching. Before the scheduled execution time arrives, the system pushes an execution notice to the dispatch terminal. The action is only officially issued after the on-duty personnel confirm that there are no abnormalities, avoiding the risk of overlapping operations caused by rashly executing when other parallel operations are in the transformer area. The ascending order is considered to execute low-impact switching actions first, and the system can monitor the operating status after these actions are completed in real time. Once an abnormality is detected, the high-impact actions that follow can be suspended in time, avoiding the continuous and intensive execution of multiple high-risk operations. This arrangement logic draws on the common practice of first testing low-risk operations and then following up with high-risk operations in the graded commissioning of power equipment.

[0037] In some embodiments, the step of verifying the attribution of the transient interruption recovery waveform in the phase switching sequence to obtain the measured phase deviation includes: acquiring transient interruption recordings based on the phase switching sequence to form a transient interruption recovery waveform; performing waveform glitch tracking based on the transient interruption recovery waveform to obtain recovery glitch features; verifying the attribution branch information based on the recovery glitch features to generate an attribution branch identifier; and tracing back the phase deviation based on the attribution branch identifier to obtain the measured phase deviation.

[0038] The transient recovery waveform is formed by acquiring transient interruption waveforms based on the phase switching sequence. When each branch in the phase switching sequence reaches its predetermined execution time, the phase-switching switch first disconnects the original phase connection, then closes to the target phase after a safety gap. The safety gap is typically controlled between 20 and 50 milliseconds. A gap that is too short may cause a short circuit due to instantaneous parallel connection between the old and new phase contacts; a gap that is too long will prolong the load loss time. The specific value is factory-set for this model of phase-switching switch and is not adjusted on-site. The entire process from disconnection to closing is recorded by the waveform recording device of this branch. The waveform recording device and the phase-switching switch are integrated and installed in the same equipment box, sharing the same set of current transformers to acquire the original signal, avoiding additional errors introduced by inconsistent sensors. The device is directly linked to the acquisition accuracy of the transient recovery waveform and is uniformly calibrated at the factory. The recorded content includes the steady-state current before disconnection, the current drop at the moment of disconnection, the residual signal during the power outage gap, and the complete process of the current recovering to a steady state after closing. These four segments together constitute a complete transient recovery waveform record. The time span of the transient interruption recovery waveform recording covers 100 milliseconds before the interruption to 500 milliseconds after the reconnection. The first segment is used to confirm the normal operating state before the interruption as a reference, and the second segment is used to observe whether the recovery process is smooth. The waveform sampling rate is 1000 points per cycle, which is much higher than the sampling density of conventional monitoring, in order to capture the rapid changes in details during the transient interruption recovery process. After the waveform data is temporarily stored locally, it is uploaded to the main monitoring station of the distribution area in the next communication window without occupying real-time communication bandwidth. If the execution interval between two adjacent branches in the phase switching sequence is compressed to less than 5 minutes, the transient interruption recovery waveform acquisition range may contain part of the influence of the switching action of the adjacent branch. Such cases are handled separately in the recovery glitch feature extraction stage.

[0039] Based on the transient recovery waveform, waveform glitches are tracked to obtain recovery glitch characteristics. The ideal process of the current recovering to steady state after closing in the transient recovery waveform should be a smooth exponential approximation curve. If a short-term abrupt change deviates from this curve during the actual recovery process, with an amplitude exceeding the theoretical curve prediction value by more than 10% and a duration not exceeding one power frequency cycle, it is determined as a glitch. The theoretical curve is obtained by fitting the typical time constant corresponding to the load type of this branch and serves as a comparison benchmark. If the load type information for this branch is missing, the theoretical curve is replaced by the typical time constant within the applicable load range of this type of phase-changing switch. Each glitch is recorded with three characteristic quantities: its occurrence time, amplitude, and duration. These three characteristics together constitute a complete characterization of each glitch in the recovery glitch feature. The source of the glitch may be its own transient caused by closing the circuit, or it may be external interference such as load changes in other branches within the same transformer area or voltage fluctuations on the grid side. The phase characteristics of the two types of sources are different: transient glitches usually appear immediately after the closing time of the transient recovery waveform, while glitches caused by external interference may appear at any time during the recovery process. Each recovery spike is recorded individually without merging or compression. Even if multiple spikes occur in the same recovery process, each spike retains its independent time, amplitude, and duration. The recovery spike characteristics generated by a typical commutation operation are usually between 1 and 3. A higher number of recovery processes often indicates that the load fluctuations in the distribution area within that switching window are relatively frequent. This is unrelated to whether there is any ambiguity in the true attribution and only reflects the background noise level of the power grid at that time.

[0040] For example, the step of verifying the belonging branch information based on the recovered burr features to generate the belonging branch identifier includes: performing simultaneous switching actions based on the recovered burr features to establish a coupled switching list; identifying the burr difference features of each branch according to the coupled switching list; verifying the uniqueness of the branch using the burr difference features to obtain a unique belonging candidate; and generating the belonging branch identifier by marking the unique belonging candidate.

[0041] Based on the recovery glitch characteristics, a coupled switching list is established by performing synchronous switching actions. The time of occurrence of each glitch in the recovery glitch characteristics is compared line by line with the switching records of other branches within the same transformer substation area to which this transient interruption recovery window belongs. These switching records are taken from the unified operation log of the transformer substation, which records the precise time of disconnection and closing for each phase-switching action of each branch in real time, with millisecond-level accuracy, sufficient to support the time comparison requirements of this sub-step. If the disconnection or closing action of other branches falls within 200 milliseconds before or after the time of occurrence of this glitch, it is identified as a potential source of coupling and included in the candidate range of this coupled switching list. This 200-millisecond range is determined with reference to the typical duration of the closing transient process of this type of phase-switching switch. Actions of other branches exceeding this range usually have sufficiently attenuated transient effects and are no longer included in the investigation. If two adjacent branch cabinets in the transformer substation topology share the same incoming cable, the action of any branch on that cable may leave traces in the recovery glitch characteristics through the shared line. Such shared incoming line relationships are pre-registered with the transformer substation drawings and are checked first during the investigation. Each identified potential coupling source branch and its corresponding action time are included in the coupling switching list. Within the same evaluation period, when the overall commutation actions of the transformer area are performed intensively, the number of branches hit in the coupling switching list increases accordingly. In transformer areas with sparse commutation actions and large time intervals between the switching of most branches, the coupling switching list typically only includes the target branch itself, and is almost unaffected by other actions. If other non-commutation switching operations occur simultaneously within the transformer area, these operations are also included in the investigation scope. As long as the action time falls within a 200-millisecond window, they are also included in the coupling switching list, not limited to the actions of the commutation switches themselves.

[0042] The glitch difference characteristics of each branch are identified based on the coupling switching list. For each potential coupling source branch in the coupling switching list, typical transient characteristics recorded in its historical commutation operations are retrieved, including the attenuation rate of the closing inrush current and the transient duration. These are compared item by item with the similar historical characteristics recorded in the target branch. The historical transient characteristics are averaged from the transient records of the branch's most recent commutation operations to avoid the impact of random fluctuations in a single record on the stability of the glitch difference characteristic comparison benchmark. If the branch has fewer than 5 historical operations, all existing records are averaged. Different branches often exhibit identifiable differences in transient attenuation rates due to differences in the type of load connected and the line impedance. Branches with heavier inductive loads attenuate more slowly, while branches with predominantly resistive loads attenuate more quickly. The physical root of this difference lies in the different characteristics of load absorption and release of transient energy, which is also the core basis for distinguishing different branches based on glitch difference characteristics. Transient duration is also related to line impedance; the longer the branch and the higher the impedance, the longer it usually takes for the transient to decay to steady state. The differences between each candidate branch and the target branch in terms of decay rate and duration constitute the glitch difference characteristics. The smaller the difference, the closer the transient characteristics of the candidate branch are to the measured glitch. If the coupling switching list only includes the target branch itself, the glitch difference characteristics directly determine that the glitch belongs to the target branch, and the difference items are all recorded as zero. If a candidate branch in the coupling switching list is also being put into operation for the first time and lacks historical transient records for comparison, the difference of this candidate branch is temporarily marked as uncalculated and is not included in this uniqueness determination.

[0043] The uniqueness of the branch is verified using the glitch difference characteristics to obtain a unique candidate. The attenuation rate difference and duration difference of each candidate branch in the glitch difference characteristics are verified item by item against the normal measurement error range of the transient characteristics of this model of commutator switch. The error range is taken from the transient measurement repeatability accuracy given in the manufacturer's test report; typically, the attenuation rate difference is allowed to be within 5%, and the duration difference is allowed to be within 10%. Candidate branches whose differences in both glitch difference characteristics fall within their respective error ranges are considered to be highly consistent with the current glitch characteristic; candidate branches whose only difference falls within the error range and whose other difference is significantly exceeded are not considered consistent, because the two indicators reflect different physical characteristics, and both must be consistent to be convincing. In cases where multiple candidate branches in the glitch difference characteristics simultaneously fall within the error range and cannot be uniquely distinguished, the time interval between the glitch occurrence time and the action time of each candidate branch is further compared. The candidate branch with the shortest interval is determined as the unique candidate, because the strength of the transient signal decays over time, and actions with shorter intervals generally contribute more to the current glitch. If only one candidate branch falls within the error range, that candidate is directly identified as the sole candidate, and time interval comparison is no longer required. If the differences between two values ​​of multiple candidate branches simultaneously fall within the error range and are difficult to distinguish, the interval between the action time of each candidate branch and the time of the glitch occurrence is compared. The candidate branch with the shorter interval is identified as having a more significant contribution due to insufficient attenuation of the transient signal, and the glitch is assigned to it. If the differences of all candidate branches significantly exceed the error range, the glitch is identified as a transient event of the target branch itself, and the sole candidate is the target branch itself.

[0044] The attribution branch identifier is generated by labeling using a unique attribution candidate. When the unique attribution candidate is the target branch itself, the attribution branch identifier is labeled as its own transient state. The time and amplitude information corresponding to this spike are retained as part of the target branch's transient characteristics, used to gradually accumulate and calibrate the branch's own typical transient benchmark. This continuously accumulated benchmark data is the source of historical transient records retrieved in the spike difference feature comparison stage. The more times it is labeled as its own transient state, the more stable and reliable the subsequent comparison benchmark becomes. When the unique attribution candidate points to another branch, the attribution branch identifier is labeled as external coupling, and associated with the external branch's number. This spike is not included in the analysis scope of the target branch's transient characteristics, but is retained in its historical record as a supplementary observation of the external branch's transient characteristics. When multiple spikes exist in the recovered spike features, each corresponding to a different unique attribution candidate, the attribution branch identifier is labeled separately for each spike. The labeling results are independent and do not affect each other due to multiple attributions occurring in the same recovery process. If multiple consecutive spikes belong to the same external branch during the same switching process, it indicates that the external branch may be in a state of continuous load fluctuation rather than a single, occasional action. In such cases, a continuous disturbance flag is additionally associated to distinguish it from a single, occasional external coupling. After the labeling is completed, the branch identification, along with the original time and amplitude information of the corresponding spike, is stored in the transient feature file of that branch. The labeling results of the branch identification do not require manual review because the entire decision chain, from recovering the spike feature extraction to determining the unique candidate, is based on reproducible quantitative comparison at each step.

[0045] The measured phase deviation is obtained by tracing back the phase deviation based on the assigned branch identification. The assigned branch identification marks the switching action to which the transient spike belongs. The current waveform after recovery to steady state is taken and compared with the phase relationship of the three-phase voltage waveform of the corresponding transformer area. The phase difference between each phase of the three-phase voltage is 120 degrees. The comparison method is to calculate the angle between the branch current phase and the phase of each phase of the three-phase voltage. The phase with the angle closest to 0 degrees is the actual phase connected to that branch (this method requires the power factor angle of the branch load to not exceed 60 degrees; this prerequisite is met for both conventional resistive and inductive loads). This measured result is compared with the originally set target phase in the phase switching sequence. If they match, the phase switching is deemed truly effective, and the measured phase deviation is recorded as zero. If they do not match, the phase switching is deemed not to have taken effect as expected, possibly due to internal contacts of the switch being stuck in their original positions, or a long-term deviation between the branch wiring and the system record. The measured phase deviation records the specific difference between the target phase and the measured phase. If the branch load power factor angle is close to the 60-degree upper limit, causing the angles of two phases to fall within 30 degrees simultaneously and be so close that the boundary of the smallest angle corresponding to the phase cannot be uniquely determined, it is judged as interference in the measurement data and marked as requiring retesting. This is not directly included in the formal judgment result of the measured phase deviation. If the branch identification is marked as an external coupling glitch, it is not used for phase backtracking in this sub-step; only the switching action corresponding to the glitch marked as transient is compared. If the same branch performs multiple switching operations within this evaluation cycle, the measured phase deviation is verified separately for each operation, and each switching operation is judged independently for effectiveness.

[0046] Corrected wiring records are generated based on the measured phase deviations. For example... Figure 2As shown, for branches with a measured phase deviation of zero, the current connected phase 7 field in the system ledger is updated to the target phase 8 set in this phase switching sequence, consistent with the measured result. The update is directly confirmed, and the update is completed along the phase switching path 9. The update action is simultaneously synchronized to the transformer area drawing management system to maintain consistency between the ledger record and the drawing annotation. For branches with a measured phase deviation that is not zero, the ledger update is based on the measured phase, not the originally set target phase, because the measured phase reflects the actual electrical wiring status. If the original planned target phase is still recorded, it will be out of sync with the actual operating status, and the imbalance calculation based on this record will continue to show systematic deviations. The corrected wiring record records the actual connected phase of the branch after the update, and the change trajectory is added in chronological order of occurrence, without overwriting previous historical records. For branches where the measured phase deviation is not zero and similar deviations have occurred in previous evaluation cycles, the cause of the deviation is more likely to be a fault in the switch itself. After correction, the wiring record, while updating the phase field in the ledger, will also be associated with an equipment anomaly flag, indicating that the switch itself in that branch needs on-site inspection, rather than just a simple record-level information correction. If the same branch exhibits a non-zero measured phase deviation for two consecutive evaluation cycles, even if the ledger correction is completed each time, a recurring deviation flag will still be added, because recurring deviations are more likely to indicate a persistent mechanical fault in the switch itself than a single, occasional occurrence.

[0047] Step S104: Based on the corrected wiring record, perform phase-by-phase switching planning to form a phase-by-phase switching scheme. Verify the risk of overlapping distribution boxes from the same source in the phase-by-phase switching scheme on the transformer substation side to generate switching direction limits. Based on the switching direction limits, estimate the most unfavorable simultaneous rate transformer substation imbalance to obtain the transformer substation side impact result.

[0048] Specifically, phase-by-phase switching planning is performed based on the revised wiring records to form a phase-by-phase switching scheme. Branches whose connected phases match the original target phases in the revised wiring records are considered to have been successfully switched in this round and will not be included in the new round of planning. Branches whose connected phases do not match the original target, or whose switching has not yet been scheduled for this evaluation period, are re-included in this step's planning scope. During planning, all branches to be planned are grouped according to the target phase, and the number of branches planned to be put into that phase and the total load capacity (converted from the actual load current level of the centralized records of the branches to be switched) are counted phase by phase. The sum of the planned capacities of each of the three phases is superimposed with the current three-phase load status of the transformer area to form the expected load distribution after the planning is completed. The recalculation of the target phases is not a mechanical application of the phases previously given in the revised wiring records, but a recalculation based on the latest three-phase load status of the transformer area, because the load distribution of the transformer area may have changed, and the original target phases may no longer be the optimal choice. The phase-by-phase switching scheme records the current connected phases, the proposed target phases, and the load capacity of each branch to be planned, grouped according to the proposed target phases. If a branch is marked with an equipment anomaly in the revised wiring record, it will still be included in the planning scope in this step, but the proposed target phase corresponding to that branch in the phase-by-phase switching scheme will be temporarily recorded as the same as the current connected phases. The actual switching target will be replanned after the equipment maintenance is completed. During the planning process, if the total planned capacity of a phase significantly exceeds its current carrying capacity due to a large influx of branches, the system will simultaneously check the upper limit of the phase's capacity. The portion exceeding the upper limit will not be included in the phase-by-phase switching scheme and will be temporarily suspended from inclusion in this round of schemes to avoid excessive single-phase planning causing new local overloads.

[0049] In some embodiments, the step of verifying the overlapping risks of the same-source distribution boxes in the phase-switching scheme at the transformer substation side to generate commutation direction limits includes: statistically analyzing the affiliation of the same-source distribution boxes for the phase-switching scheme to obtain a group list of the same-source distribution boxes; verifying the timing overlap of the actions based on the same-source distribution box group list to obtain the degree of timing overlap; calculating the transformer substation side risk based on the degree of timing overlap to determine the overlapping risk level; and generating commutation direction limits by planning the commutation direction dispersion based on the overlapping risk level.

[0050] For phase-switching schemes, statistical analysis of the attribution of distribution boxes from the same source is conducted to obtain a group list of these distribution boxes. For example... Figure 2As shown, in the phase-switching scheme, each branch to be planned is assigned to a distribution box (4) based on the equipment ledger of the transformer substation. Whether two distribution boxes are from the same source depends on whether they share the same low-voltage outgoing line of the same distribution transformer (1) or the same upstream low-voltage trunk line. Distribution boxes from the same source are ultimately grouped into distribution box group 3. The relationship of the same source is pre-marked according to the power supply topology diagram of the transformer substation and does not need to be verified on-site each time. The reason why the relationship of the same source is pre-marked instead of being verified on-site each time is because the power supply topology connection between distribution boxes is a fixed physical wiring relationship. Once the transformer substation is built and put into operation, it will remain unchanged for a long time unless the line is modified. For distribution boxes that share the same low-voltage outgoing line or trunk line, the load current is transmitted through the same line. The current change caused by the switching action of any branch in the distribution box will be coupled to other distribution boxes in the same group through the impedance of the shared line. This is the physical basis for the subsequent verification of the action sequence within the group. Distribution boxes within the same transformer substation are grouped into several groups based on their shared origin. Each group shares a specific segment of upstream power supply resources, but the groups are independent and do not share resources. A typical transformer substation is usually divided into two to four shared-origin groups, each containing three to eight distribution boxes, the exact number depending on the substation size and power supply topology design. The distribution box numbers within each group, along with a list of branches within that group involving phase-switching schemes, together constitute the shared-origin distribution box group list. An isolated distribution box with only one box, not sharing an origin with any other distribution boxes, constitutes a separate group containing only itself. It is treated the same as other groups, except that the number of distribution boxes within the group is only one.

[0051] The degree of timing overlap is determined by verifying the timing of actions based on the same-source distribution box group list. Within each group of the same-source distribution box group list, each branch within the group is compared against the predetermined execution time arranged in the phase switching sequence. If the execution time interval between any two branches is less than 30 minutes, it is determined as a timing overlap. The 30-minute interval is determined with reference to the time required for the current in the distribution area to recover to a steady state after a load change. This time is derived from statistical observation of the current decline curve after previous load step disturbances in this type of distribution area. In most scenarios, the current fluctuation caused by the step disturbance has basically subsided within 30 minutes. The overlap count is verified one by one in pairs of branches within the same-source distribution box group list. If three branches in a group are all less than 30 minutes apart, it is counted as three pairwise overlaps rather than being counted as one in general. This pairwise recording rather than merging statistics is to allow subsequent verification steps to trace the risk contribution of each pair of overlapping branches separately, rather than mixing the overlap of the entire group together. When there are many branches within a group, the number of verifications for each pair increases quadratically with the number of branches, resulting in a significant increase in verification computation. However, since a single comparison is simply a time difference calculation, the overall computation time remains within an acceptable range. The temporal overlap level records the number of overlaps within the current evaluation period for that group, the specific branch pairs involved, and their respective execution time intervals. If a group contains only one branch to be switched and cannot form a pairwise comparison, it is determined to have no overlap risk. The single action within this group is executed independently and is not constrained by the same-origin superposition logic.

[0052] For example, the step of determining the superimposed risk level by calculating the risk on the transformer area side based on the degree of temporal overlap includes: assessing the transformer area's load-bearing status based on the degree of temporal overlap to obtain the real-time margin of the transformer area; dynamically calculating the risk tolerance based on the real-time margin of the transformer area to form a dynamic risk tolerance; performing a critical interval verification on the dynamic risk tolerance to generate a tolerance exceedance result; and determining the superimposed risk level by classifying the risk level based on the tolerance exceedance result.

[0053] The real-time margin of a distribution area is obtained by assessing the load-bearing status of the distribution area based on the degree of time overlap. The time overlap is recorded in groups with the number of overlaps. The current real-time load of the distribution transformer to which this group belongs is taken. The load data comes from the real-time monitoring terminal of the transformer itself, and the sampling period is consistent with the distribution area's operating data. The monitoring terminal is integrated with the transformer itself, directly measuring the three-phase current at the low-voltage side outlet of the transformer and converting it. The real-time margin of the distribution area is calculated using the formula M=(P_rated-P_current) / P_rated, where P_rated is the rated capacity of the distribution transformer, P_current is the current real-time load, and M is the margin ratio calculated in this study, typically between 0 and 1. If the distribution transformer is currently operating under short-term overload (P_current exceeds P_rated), M is negative. The closer the ratio is to 1, the further it is from full load, and the greater the additional impact space it can withstand. If the distribution transformer corresponding to the same group is in a special operating state such as tap changer adjustment, the real-time margin calculation of the distribution area uses the actual rated capacity after adjustment, instead of the nameplate nominal value. In some distribution areas, two distribution transformers operate in parallel, jointly supplying power to the same batch of distribution boxes. For these areas, the real-time margin is calculated by combining the total rated capacity and total real-time load of both transformers, treating them as a single load unit, and not splitting them into two independent margin values ​​for separate calculation. The calculation of the real-time margin is based on the actual execution time of this procedure, rather than using the historical time used for time series overlap verification, because load levels continuously change over time, and the margin assessment needs to reflect the closest possible current actual load condition.

[0054] Dynamic risk tolerance is calculated based on the real-time margin of the distribution area. A smaller real-time margin value indicates that the area is closer to full load, and the space for overlapping impacts from similar distribution boxes is correspondingly smaller. Therefore, the risk tolerance needs to be dynamically tightened along with the margin. The dynamic risk tolerance is calculated using the formula T_dynamic=M×k_impact, where M is the margin ratio calculated from the real-time margin of the distribution area, and k_impact is the baseline impact coefficient, equal to the upper limit of the short-term overload capacity allowed by the transformer model (the absolute capacity value corresponding to 10% to 15% of the rated capacity). The specific value is given by the manufacturer's technical parameters for that transformer model. T_dynamic is the dynamic risk tolerance calculated in this instance. In the extreme case where the real-time margin of the distribution area is close to zero, the dynamic risk tolerance tightens synchronously to near zero, meaning that almost no overlapping switching operations of similar distribution boxes are allowed. Conversely, in the case of a more lenient margin where M is close to 1, the tolerance does not exceed k_impact itself, because this coefficient is already the upper limit of the transformer's allowed short-term overload. In situations where multiple distribution boxes within the same power supply range of a single distribution transformer require simultaneous tolerance calculation, each group independently calculates the same dynamic risk tolerance value based on the same real-time load margin for the distribution area. The tolerance is not further divided and allocated among multiple groups because different groups involve different overlapping branches, and determining whether an exceedance has occurred is an independent comparison process. The benchmark impact factor is fixed according to the transformer model and is not adjusted with real-time load fluctuations in the distribution area. This ensures that the tightening or loosening of the dynamic risk tolerance is entirely driven by the real-time load margin of the distribution area, avoiding the difficulty in tracing the cause of tolerance changes due to simultaneous fluctuations in multiple variables.

[0055] The dynamic risk tolerance is checked for critical intervals to generate tolerance exceedance results. The sum of the rated capacities of the overlapping branches in the time-series overlap is compared with the allowable superimposed impact capacity given by the dynamic risk tolerance. If multiple branches complete phase switching simultaneously within the same time window, their newly added rated capacities to the target phase will superimpose and appear within a short period. The calculation of the sum of rated capacities involves directly adding the rated capacities of the two or more overlapping branches without further conversion of power factor or simultaneity rate. The superposition effect is estimated in the most conservative way to ensure that the verification results do not underestimate the actual risk. If the sum of the rated capacities of the overlapping branches exceeds the dynamic risk tolerance, it is determined to be outside the critical interval, meaning that if these branches are switched on and off simultaneously as originally planned, the short-term superimposed load may cause local overload of the distribution transformer; if it does not exceed the limit, it is determined to be within the safe range, and the original timing sequence can be followed without additional intervention. The tolerance exceedance results are summarized by comparing the exceedance magnitude with the corresponding list of overlapping branches. The exceedance magnitude is the difference between the sum of the rated capacities and the dynamic risk tolerance. If no exceedance is found, this difference is negative, and this negative number is also retained in its entirety without absolute value processing. Its specific magnitude reflects how much safety margin the overlapping combination has from the critical point. When there are multiple different pairwise overlapping combinations within the same group, the tolerance exceedance results are verified group by group. The group with the largest exceedance magnitude is taken as the representative result of this group's assessment, and the verification results of the other groups are also retained separately. The risk contribution of each pair of overlapping branches is thus clearly visible.

[0056] The risk level is determined by categorizing the results of exceeding the tolerance limit. Groups with negative exceedances are classified as low-risk, allowing for phase switching as planned without additional restrictions. Groups with positive exceedances not exceeding 20% ​​of the dynamic risk tolerance limit are classified as medium-risk; those exceeding 20% ​​are classified as high-risk. This 20% threshold is determined based on statistically significant data from short-term overload capacity tests of this type of distribution transformer, where failure rates increase significantly when overload exceeds the rated margin by 20%. When the sum of the rated capacities of overlapping branches exceeds the dynamic risk tolerance limit, and the exceedance percentage reaches this threshold, the risk level transitions from medium to high-risk. A higher exceedance percentage indicates a more pronounced short-term impact on the transformer area relative to the current carrying capacity. The risk level is recorded by including the categorization result of this assessment and the original value of the corresponding exceedance. When dynamic risk tolerance is tightened to near zero due to extremely small real-time margins in the transformer area, even if the absolute value of the exceedance is not large, the relative proportion can easily exceed the 20% grading threshold. In such cases, the grading rules are applied accurately, without any relaxation of standards due to the small tolerance value. If multiple pairs of overlapping combinations exist within the same group, each corresponding to different tolerance exceedance results, the highest-grade combination among them is taken as the representative grade for that group, avoiding the use of lighter combinations to mask truly concerning overlapping risks. The three-tier grading system here maintains the same granularity in name as the three-tier grading system used for branch impact and branch life grading, but the two three-tier systems independently assess different physical objects without mutual conversion or merging.

[0057] The risk level is overlaid to plan the direction of phase switching and generate switching direction limits. For groups with a low risk level, no additional direction restrictions are imposed, and each branch within the group continues to operate according to the original target phase of the phase switching scheme. For groups with a medium risk level, branches originally planned to be concentrated on the same phase are instead distributed to two different phases. The specific distribution method involves alternating the distribution of load capacity from largest to smallest among the two target phases. Branches with larger capacities are given priority to occupy the phase with the lighter load after distribution. The two target phases are selected from the two phases with relatively lighter current loads in the transformer area and are not the phases with the heavier load indicated by the implicit imbalance, to avoid creating new local concentrations on the new target phases after distribution. For groups with a high risk level, further staggered execution of phase commutation actions within the group is required, with the execution interval between different branches extended to over 30 minutes. This ensures that actions that would otherwise overlap completely avoid the same timing window. Directional dispersion addresses the long-term target phase load concentration problem after switching, while timing staggering addresses the short-term superimposed impact problem at the moment of switching. The physical risks they address are different; therefore, both methods are used simultaneously for high-risk groups, rather than one replacing the other. The phase commutation direction limit record for this group is used to record the final determined direction allocation scheme or timing staggered arrangement, serving as a constraint correction to the phase-by-phase commutation scheme. If the original target phase changes due to directional dispersion adjustments within the same group, this change only takes effect at the phase commutation direction limit level; the original record of the phase-by-phase commutation scheme is not overwritten or modified, and both records are kept to preserve the complete adjustment trajectory.

[0058] The impact on the transformer substation side is obtained by estimating the most unfavorable simultaneity rate of the commutation direction limit. The final direction allocation scheme determined by the commutation direction limit takes the target phase and rated capacity of each branch (this rated capacity is taken from the branch nameplate nominal value, distinct from the actual load current level recorded in the phase-by-phase commutation scheme, to match the full-load assumption). Assuming all branches to be commutated simultaneously reach their respective maximum loads—the most unfavorable simultaneity scenario—the estimated maximum loads for each of the three phases are obtained by summing the target phases. This assumption is a deliberately conservative estimate; in reality, the probability of all branches simultaneously reaching maximum load is very low. However, this most unfavorable assumption ensures that the scheme's performance in actual operation will not be worse than the estimated result. The estimated three-phase maximum loads are substituted into the phase-to-phase imbalance formula to calculate the estimated phase-to-phase imbalance under the most unfavorable scenario after the scheme is implemented. This is compared with the currently measured phase-to-phase imbalance of the transformer substation to obtain the improvement margin, determining whether this round of commutation scheme can still improve the imbalance level in the worst-case scenario. The forecast results show that under the most unfavorable simultaneity scenario, the imbalance of a certain scheme increases instead of decreasing. The affected area-side scheme is marked as requiring further review, indicating that the phase-switching scheme needs manual re-verification of the direction allocation before formal implementation. When multiple independent groups within the same transformer area generate their own commutation direction limits during this assessment period, the affected area-side results are summarized from the forecasts of all groups and listed separately by group. Different groups from the same source are connected to different upstream power resources, and the forecast calculations of the maximum three-phase load are carried out independently by group.

[0059] Step S105: Use the disturbance loss in the influence results of the transformer area to reduce the gain and obtain effective governance benefits. Use the effective governance benefits to verify the quota comparison and determine the commutation start criterion. Plan the phase-separated actions according to the commutation start criterion and output the commutation scheduling command.

[0060] In some embodiments, the step of using the disturbance loss in the transformer area side impact results to reduce the gain and obtain effective governance benefits includes: analyzing historical losses based on the transformer area side impact results to form disturbance loss; reducing the theoretical gain based on the disturbance loss to obtain a reduced gain value; performing a base-down verification based on the retention period based on the reduced gain value to obtain a net gain result; and calculating and merging the dilution effect based on the net gain result to obtain effective governance benefits.

[0061] The analysis of historical losses based on the impact results of the transformer substation side forms disturbance losses. The estimated maximum load distribution and the imbalance of the most unfavorable scenario corresponding to each group in the impact results of the transformer substation side are used to match commutation cases with similar load characteristics and similar switching scales in the history of this type of transformer substation. For each case, two data points are extracted: the actual instantaneous power loss and the transient additional loss. The instantaneous power loss is estimated by multiplying the time from the switching interruption to the completion of the reconnection by the average load of the branch during the power outage. The transient additional loss is the energy obtained by integrating the product of the current exceeding the steady-state value during the recovery process after switching and the operating voltage of the branch over time. The disturbance loss records three items: the number of historical cases matched this time, the mean and standard deviation of the loss data for each case. The mean serves as a representative estimate of the current loss, and the standard deviation reflects the degree of fluctuation between historical cases. The disturbance loss consists of two parts: fixed costs and variable costs that vary with the number of branches. Fixed costs correspond to the costs incurred each time the waveform recording equipment is activated and the switching coordination is performed. When there are fewer than 5 historical cases, the disturbance loss is conservatively estimated using the theoretical upper limit of the loss given in the factory test report of the switching switch model, and the source is clearly indicated to distinguish it from the estimate based on historical statistics. When multiple branches are switched on and off simultaneously in the same group, the disturbance loss is estimated separately for each branch according to the historical cases matched, and then summed, instead of a rough calculation by simply multiplying by the number of branches. In addition to load characteristics and switching scale, the matching conditions for historical cases also require that the capacity of the distribution transformer in the transformer area to which the case belongs is similar to that of the transformer area in this case. Historical cases with large capacity differences may not be suitable for direct use in the current loss estimation, even if the load characteristics given by the transformer area side are similar.

[0062] The theoretical gain is calculated by subtracting the disturbance loss from the theoretical gain to obtain the reduced gain value. The theoretical gain is the improvement rate recorded in the impact results of the distribution area compared to the current measured value. It is converted into equivalent power benefit according to the loss reduction benefit coefficient corresponding to the unit imbalance improvement of the distribution area. The loss reduction benefit coefficient is the line loss increment corresponding to the unit imbalance given in the technical parameters of the distribution transformer model. That is, for every percentage point improvement in the imbalance of the distribution area, the additional losses caused by the three-phase current asymmetry of the distribution transformer and low-voltage line are reduced accordingly. This coefficient is directly related to the equipment capacity and line impedance and is pre-calibrated according to the model. This improvement rate is an ideal estimate assuming that the treatment process does not generate any additional losses. Actual implementation will inevitably be accompanied by disturbance losses, which need to be deducted from the theoretical gain to reflect the true net effect. The reduced gain value is calculated by subtracting the average disturbance loss from the theoretical gain. The reduced gain value may be positive or negative. A positive value indicates that the benefit from improving the imbalance can still cover the switching losses. A negative value indicates that the treatment plan is not worthwhile after considering the overall losses. In this case, the corresponding commutation plan is not economically reasonable to implement in this evaluation period. When disturbance losses are marked as being from a conservative estimate, an additional two standard deviations are deducted as a safety margin when calculating the reduced gain value. Two standard deviations correspond to approximately a 95% confidence interval under a normal distribution, preventing overestimation of actual benefits due to inaccurate loss estimates. For branches with negative reduced gain values, their corresponding commutation schemes remain fully recorded within this evaluation period and are not deleted from the original schemes simply because the reduction result is negative.

[0063] The net gain result is obtained by performing a baseline verification based on the reduced gain value during the retention period. For the same group, if the reduced gain value fails to exceed the baseline during each evaluation period, a retention period is recorded. The retention period count N_delay is accumulated according to the consecutive evaluation periods that fail to meet the standard. Once a period is determined to meet the standard, N_delay is reset to zero. If the evaluation of the group is temporarily suspended due to a fault in the commutator switch, this suspension is not included in the count of consecutive periods that fail to meet the standard, so as to avoid misjudging equipment failure as continuous failure of the solution. The baseline used for verification is calculated using the formula B_dyn=B_0×max(1-γ×N_delay,φ), where max(1-γ×N_delay,φ) represents the larger of 1-γ×N_delay and φ. B_0 is the preset original baseline for this type of transformer area (using the same equivalent power efficiency unit as the reduced gain value), γ is the dimensionless stagnation attenuation coefficient (ranging from 0 to 1), and φ is the dimensionless baseline lower limit ratio (ranging from 0 to 1). B_dyn is the dynamic baseline that actually takes effect in this verification, and its value is no less than φ times the original baseline. Each time the number of stagnation cycles increases, the baseline continues to decrease proportionally to γ. If the baseline fails to meet the standard for several consecutive evaluation cycles, it will gradually decrease with the number of stagnation cycles until it reaches the lower limit set by φ and then stops decreasing. The difference between the reduced gain value and B_dyn is denoted as G. G is the net gain result obtained in this verification: a positive G indicates positive governance value; a negative or zero G indicates insufficient governance value, and the retention period N_delay for that group is incremented by one. The net gain result records the relationship between the dynamic baseline B_dyn used in this verification, the corresponding retention period N_delay, and the reduced gain value. The retention attenuation coefficient γ and the lower limit ratio φ are uniformly determined according to the power supply area to which the transformer belongs.

[0064] Effective governance benefits are obtained by merging and diluting the net gain results. The net gain results represent multiple groups that fail to meet the standards. If the commutation actions involved in each group are feasible to be merged into the same scheduling window in terms of execution timing, it is verified whether these groups can share the fixed overhead portion of disturbance losses through merged execution. Fixed overhead is not repeatedly generated as the number of merged branches increases; this is precisely the portion that is repeatedly deducted when the net gain results are calculated independently, but only deducted once after merging. The feasibility of timing merging is determined by whether the interval between the scheduled execution times of each group falls within the same scheduling window. The merging and diluting benefits are calculated using the formula ΔG=(n-1)×C_fixed, where n is the number of groups to be merged, C_fixed is the fixed overhead of a single scheduling (the unit is the same as the net gain result, representing the equivalent energy benefit unit), and ΔG is the total fixed overhead saved after merging, which is directly included in the sum of the merged net gain results. Effective governance benefits are calculated using the formula E=ΣG_i+ΔG, where G_i is the net gain of the i-th group participating in the merger, ΣG_i is the sum of the net gains of all groups, and E is the effective governance benefit after merging and dilution. If E turns positive, the merged groups are considered to have positive governance value. The list of groups participating in this summation is also retained for subsequent group-based processing. Groups that are not feasible for merging are still judged based on their independent net gain results and are not forcibly included in the merger calculation. There is no upper limit to the number of groups n participating in the merger; the larger the value of n, the more fully the fixed costs are diluted. However, actual implementation is still constrained by the number of on-site personnel and equipment. The final value of n is based on the available on-site resources.

[0065] The commutation start-up criteria are determined by comparing and verifying quotas based on effective governance benefits. For each group included in the summary of effective governance benefits, the commutation frequency quota consumed by each branch actually participating in governance within that group is calculated. The group's share of the effective governance benefits (for groups not included in the consolidation, it's their own net gain result; for groups included in the consolidation, the effective governance benefits after consolidation are distributed according to the absolute value of their respective net gain results) is divided by the quota consumption to obtain the unit quota benefit for that group. For example, if a group's net gain result is 300 equivalent power benefit units and it consumes quota 3 times, the unit quota benefit is 100 units per time. The higher the unit quota benefit, the better the imbalance improvement effect achieved by consuming the same number of commutation frequency quota resources, directly reflecting the efficiency of quota utilization, a scarce resource. Groups with unit quota benefits lower than the preset minimum efficiency threshold for that type of transformer area are deemed not to meet the commutation start-up criteria, even if their net gain result is positive; groups with unit quota benefits reaching or exceeding the threshold are deemed to meet the start-up conditions. In cases where the effective governance benefit is zero, the commutation start-up criterion directly determines that no group meets the start-up conditions during the current assessment period, and the calculation of unit quota revenue is no longer performed. The calculation of unit quota revenue is the ratio of the group's net gain to the quota consumption. The dimension of the ratio is the amount of benefit improvement brought about by each unit of commutation. The specific value of the efficiency threshold is dynamically adjusted according to the quota scarcity of the power supply area to which the transformer belongs. The more scarce the quota resources in the area, the higher the threshold is set, prompting the limited quota to be prioritized for groups where the commutation start-up criterion is easier to meet.

[0066] The planning process outputs commutation scheduling instructions based on the commutation initiation criteria. During the planning phase, all groups meeting the initiation conditions are selected from the commutation initiation criteria; groups not meeting the conditions are temporarily suspended, and no corresponding scheduling instructions are generated for this evaluation cycle. For each group meeting the conditions, the target phase and predetermined execution time (using the original settings for that branch in the phase switching sequence, without recalculation), branch impact level (using the level determined in the impact grading stage), and the commutation frequency quota consumed this time (recorded as one consumption when included in the scheduling instruction) are extracted for each branch. The core of the planning is to reorder the originally independent branch lists belonging to different groups according to impact level from light to heavy, merging them into a single cross-group execution queue for issuance, rather than issuing separate instructions to each group. For example, in this assessment period, both distribution boxes No. 2 and No. 5, which share the same source, met the startup conditions. Group No. 2 includes branch 11 and branch 14, and group No. 5 includes branch 23. After the three branches are merged and rearranged, the phase switching dispatch instructions are as follows: Branch 11 switches from phase A to phase C, executed at 08:15, with a minor impact level and consuming one quota; Branch 23 switches from phase B to phase A, executed at 08:20, with a moderate impact level and consuming one quota; Branch 14 switches from phase A to phase B, executed at 08:35, with a severe impact level and consuming one quota. Before issuing the instructions, the system checks the quota reserve for each branch, and branches with insufficient reserves are removed from the instructions. This assessment cycle begins with the collection of transformer area operation data, and goes through the identification of hidden imbalance phases, branch aggregation, impact level determination, lifespan quota allocation, phase switching, wiring record correction, scheme planning verification, and finally the output of dispatch instructions, completing a complete closed loop. When the next assessment cycle arrives, transformer area operation data will be collected again to start a new round of assessment.

[0067] To execute the methods described in the above embodiments and achieve the corresponding functions and technical effects, this application also provides the following system. See [link to system]. Figure 3 , Figure 3 This application provides a structural block diagram of a three-phase imbalance mitigation system 300 for a distribution box, comprising: The balance identification module 301 is used to acquire the operating data of the transformer area, calculate the phase-to-phase and neutral line current imbalance based on the operating data of the transformer area to form the imbalance characteristics of the transformer area, and screen the abnormality of the standard period through the imbalance characteristics of the transformer area to determine the hidden imbalance phase. The quota allocation module 302 is used to perform phase-switching branch aggregation based on the implicit unbalanced phase to obtain the set of phase-switching branches, measure the switching impact of the zero-crossing dwell short branches in the set of phase-switching branches to determine the branch impact level, and perform lifetime quota allocation based on the branch impact level to determine the commutation number quota. The execution verification module 303 is used to generate a phase switching sequence based on the phase switching quota in an ascending order of impact, use the transient recovery waveform in the phase switching sequence to verify the attribution to obtain the measured phase deviation, and correct the wiring record for the measured phase deviation to form a corrected wiring record. The superposition and coordination module 304 is used to execute phase-switching planning based on the corrected wiring record to form a phase-switching scheme, perform area-side verification of the risk of superposition of the same source distribution box in the phase-switching scheme to generate a switching direction limit, and estimate the most unfavorable simultaneous rate of area imbalance based on the switching direction limit to obtain the area-side impact result. The instruction generation module 305 is used to apply the disturbance loss in the influence result of the transformer area to reduce the gain and obtain effective governance benefits, and to verify the quota comparison through the effective governance benefits to determine the commutation start criterion. Based on the commutation start criterion, it plans the phase-by-phase action and outputs the commutation scheduling instruction.

[0068] The aforementioned three-phase imbalance mitigation system 300 for a distribution box can implement one of the three-phase imbalance mitigation methods for a distribution box as described in the above-described method embodiments. The options in the above method embodiments are also applicable to this embodiment and will not be detailed here. The remaining contents of this application's embodiments can be referred to the contents of the above method embodiments, and will not be repeated in this embodiment.

[0069] The purpose of the above embodiments is to reproduce and derive the technical solution of the present invention by way of example, and to fully describe the technical solution, purpose and effect of the present invention. The purpose is to enable the public to have a more thorough and comprehensive understanding of the disclosure of the present invention, and not to limit the scope of protection of the present invention.

Claims

1. A method for mitigating three-phase imbalance in a distribution box, characterized in that, include: The operation data of the transformer substation is obtained, and the phase-to-phase and neutral line current imbalance is calculated based on the operation data to form the imbalance characteristics of the transformer substation. The abnormality of the standard period is screened through the imbalance characteristics of the transformer substation to determine the hidden imbalance phase. Based on the implicit unbalanced phase, the switchable branch collection is performed to obtain the branch set to be switched. The switching impact is measured on the zero-crossing dwell short-short branch in the branch set to be switched to determine the branch impact level. Based on the branch impact level, the life quota allocation is performed to determine the commutation number quota. Based on the commutation frequency quota, a phase switching sequence is generated by switching in an ascending order of impact. The measured phase deviation is obtained by verifying the ownership of the transient recovery waveform in the phase switching sequence. The wiring record is corrected according to the measured phase deviation to form a corrected wiring record. Based on the revised wiring record, phase-switching planning is performed to form a phase-switching scheme. The risk of overlapping distribution boxes from the same source in the phase-switching scheme is verified on the transformer substation side to generate a switching direction limit. Based on the switching direction limit, the most unfavorable simultaneous rate transformer substation imbalance is estimated to obtain the transformer substation side impact result. The gain reduction is performed by using the disturbance loss in the affected area results to obtain effective governance benefits. The quota comparison and verification is performed based on the effective governance benefits to determine the commutation start criterion. The phase-by-phase action is planned and the commutation scheduling command is output based on the commutation start criterion.

2. The method according to claim 1, characterized in that, The step of identifying latent imbalance phases by screening for abnormalities in the time period of the imbalance characteristics of the transformer area includes: Based on the aforementioned imbalance characteristics of the transformer substations, the qualified time periods are screened to obtain the qualified time period sequence; Residual features for the target period are generated by residual tracking based on the target period sequence. The residual characteristics during the target period are used to identify the characteristics of occasional associated phases and determine the occasional associated phases; The latent unbalanced phases are determined by verifying the latent asymmetric attribution based on the occasional associated phases.

3. The method according to claim 1, characterized in that, The determination of the switching impact level for the concentrated zero-crossing dwell short-short branch of the phase-to-be-swapped branch includes: For the set of branches to be switched, the dwell time distribution is formed by statistically analyzing the zero-crossing dwell time of each branch; Based on the residence duration distribution, a short residence branch set is generated by screening short residence branches. The switching impact amplitude is obtained by calculating the commutation closing inrush current using the aforementioned short-staying branch set; The branch impact level is determined by verifying the amplitude of the impact by the cutting.

4. The method according to claim 1, characterized in that, The determination of the commutation frequency quota for the lifespan quota allocation based on the branch impact level includes: Based on the impact level of the branch, historical action analysis is performed to obtain the cumulative number of actions; The oxidation correction lifetime is obtained by calculating the contact idle oxidation based on the cumulative number of actions. The commutation quota gradient is obtained by using the oxidation correction lifetime allocation quota gradient; The phase-switching quota is determined by sequentially verifying the phase-switching quota gradient.

5. The method according to claim 1, characterized in that, The method of verifying the attribution of the phase shift sequence using the transient recovery waveform to obtain the measured phase deviation includes: Based on the phase switching sequence, the instantaneous interruption recording waveform is acquired to form the instantaneous interruption recovery waveform; Based on the transient recovery waveform, waveform glitch tracking is performed to obtain recovery glitch features; The restored burr feature is used to verify the attribution branch information and generate an attribution branch identifier; The measured phase deviation is obtained by tracing back the phase deviation based on the assigned branch identifier.

6. The method according to claim 1, characterized in that, The step of verifying the overlapping risks of the same source distribution boxes in the phase-switching scheme at the transformer substation side and generating commutation direction limits includes: For the aforementioned phase-switching scheme, the affiliation of the same-source distribution boxes is statistically analyzed to obtain a group list of the same-source distribution boxes; The degree of timing overlap is obtained by verifying the timing overlap of actions based on the group list of the same source distribution boxes; The risk level of superimposed risks is determined by calculating the degree of temporal overlap at the transformer substation side. The commutation direction limit is generated by the superimposed risk level planning direction dispersion.

7. The method according to claim 1, characterized in that, The method of using the disturbance loss in the transformer area side impact results to reduce gain and obtain effective mitigation benefits includes: Based on the analysis of the influence results on the transformer substation side, historical losses are formed as disturbance losses. The reduced gain value is obtained by subtracting the theoretical gain based on the disturbance loss. The net gain result is obtained by performing a base-down verification of the residence period based on the reduced gain value. The net gain results were used to calculate and consolidate the dilution effect to obtain effective governance benefits.

8. The method according to claim 5, characterized in that, The step of verifying the attribution branch information of the recovered burr features and generating the attribution branch identifier includes: Based on the restored burr characteristics, perform simultaneous cutting actions to identify and establish a coupled cutting list; Based on the aforementioned coupling switching list, identify the burr difference characteristics of each branch; The uniqueness of the branch is verified using the burr difference characteristics to obtain a unique candidate for assignment; The unique candidate for attribution is used to generate the attribution branch identifier.

9. The method according to claim 6, characterized in that, The method of determining the superimposed risk level by calculating the risk on the transformer substation side based on the degree of temporal overlap includes: The real-time margin of the transformer area is obtained by assessing the load-bearing status of the transformer area based on the degree of temporal overlap. Dynamic risk tolerance is formed by dynamically calculating the risk tolerance based on the real-time margin of the transformer area; Perform a critical interval verification on the dynamic risk tolerance to generate a tolerance exceedance result; The risk level is determined by classifying the results based on the tolerance limit.

10. A three-phase imbalance mitigation system for a distribution box, characterized in that, include: The balance identification module is used to acquire the operating data of the transformer area, calculate the phase-to-phase and neutral line current imbalance based on the operating data of the transformer area to form the imbalance characteristics of the transformer area, and screen the abnormality of the standard period through the imbalance characteristics of the transformer area to determine the hidden imbalance phase. The quota allocation module is used to perform phase-switching branch aggregation based on the implicit unbalanced phase to obtain the phase-to-switch branch set, measure the switching impact of the zero-crossing dwell short-short branch in the phase-to-switch branch set to determine the branch impact level, and perform lifetime quota allocation based on the branch impact level to determine the switching number quota. The execution verification module is used to generate a phase switching sequence based on the phase switching quota in an ascending order of impact, use the transient recovery waveform in the phase switching sequence to verify the attribution and obtain the measured phase deviation, and correct the wiring record for the measured phase deviation to form a corrected wiring record. The superposition and coordination module is used to execute phase-by-phase switching planning based on the corrected wiring record to form a phase-by-phase switching scheme, to perform area-side verification of the risk of superposition of the same source distribution box in the phase-by-phase switching scheme to generate a switching direction limit, and to estimate the most unfavorable simultaneous rate of area imbalance based on the switching direction limit to obtain the area-side impact result. The instruction generation module is used to reduce the gain by applying the disturbance loss in the influence results of the transformer area to obtain effective governance benefits, and to verify the quota comparison based on the effective governance benefits to determine the commutation start criterion. Based on the commutation start criterion, it plans phase-by-phase actions and outputs commutation scheduling instructions.

Images