A data center power grid power supply intelligent control system

Abstract

The application discloses a kind of data center power grid power supply intelligent control systems, it is related to power supply and distribution control technical field, the present application, through generating repeated disturbance sequence identification in the same disturbance process, and combining reclosing instruction signal, DC bus state, inverter output state and shunt state, the current power supply event is typed, in the static switching unit current power path is kept under the maintenance of transient drop event and pending event, battery charging branch is prohibited to be put into and maintains load support, only in the sequence of execution regression control of reclosing process in the same repeated disturbance sequence end and city power recovery stable is kept, target path is transferred, rectification input is recovered, battery charging branch is recovered and feeder restriction is removed, multiple voltage drop and recovery in the same reclosing process are no longer handled according to mutually independent power failure event.

Description

Technical Field

[0001] This invention relates to the field of power supply and distribution control technology, and in particular to an intelligent control system for power supply to a data center power grid. Background Technology

[0002] Data centers are load scenarios highly sensitive to power supply continuity and power quality. Their power supply links typically include mains power access, backup power switching, uninterruptible power supply (UPS), and downstream feeder power distribution. Existing data center power supply control systems usually determine whether a power outage has occurred based on measurements such as mains voltage, current, and frequency, combined with fixed thresholds and delay logic, and execute power switching, load support, and power restoration control accordingly. This type of control can meet the power switching requirements under general continuous power outage conditions, but its control object is mainly focused on single power outage events, lacking targeted collaborative judgment and linkage control for short-term voltage drops caused by transient faults in upstream power lines and repeated recovery processes.

[0003] When a transient fault occurs in the upstream power supply line and is accompanied by multiple reclosings, the voltage at the mains connection point will drop and recover multiple times within a short period of time. If the existing control system still handles each drop separately as a single voltage limit exceedance or a fixed delay, it is easy to misjudge repeated voltage drops in the same disturbance process as continuous power loss, thus prematurely releasing the original power supply path holding state, triggering the static switching unit to switch over, or causing the uninterruptible power supply unit to repeatedly switch between support and recovery. This not only causes frequent changes in the power supply path, but also lacks consistency between the inverter support state, the rectifier recovery state, and the downstream feeder state, causing different power supply units within the same data center to enter different operating states during the same disturbance process.

[0004] Furthermore, after a brief restoration of mains voltage, existing control systems typically restore power to the rectifier input, charging branch, and feeder according to the individual recovery rules of each device. If the recovery process is not uniformly correlated with the aforementioned disturbance process, the rectifier input and charging branch may recover prematurely before the repeated disturbance ends, and downstream feeders may enter different recovery stages sequentially. When a voltage drop occurs again, the system will re-enter the support or switching process, causing the power supply path to change back and forth, the bus support state to switch repeatedly, and the feeder recovery sequence to become disordered. This makes it difficult to guarantee continuous power supply during repeated disturbances and stable return after the disturbance ends.

[0005] Therefore, how to correlate and distinguish repeated voltage drops and recovery during the same disturbance process under the condition of transient faults in the upstream power supply line accompanied by multiple reclosing, and coordinate the power supply path maintenance, uninterrupted power supply support and sequential return control during the recovery phase accordingly, has become a technical problem that needs to be solved. Summary of the Invention

[0006] This application provides an intelligent control system for data center power grid supply, which solves the problems of power supply erroneous switching, repeated switching, and recovery oscillation in data center under multiple reclosing conditions.

[0007] This invention provides an intelligent control system for data center grid power supply, including a mains power access unit, a backup power supply unit, a static switching unit, an uninterruptible power supply unit, a feeder power supply unit, a status acquisition unit, and a central control unit. The mains power access unit and the backup power supply unit are respectively connected to the input terminal of the static switching unit. The uninterruptible power supply unit includes a rectifier input terminal, an inverter output terminal and a battery charging branch. The output terminal of the static switching unit is connected to the rectifier input terminal, and the inverter output terminal is connected to the feeder power supply unit. The status acquisition unit acquires the voltage, current, phase angle, reclosing indication signal, DC bus status, inverter output status, and branch status of the mains access point. The central control unit generates a repeating disturbance sequence identifier within the same disturbance process, and determines the current power supply event as an event type based on the repeating disturbance sequence identifier, the reclosing indication signal, the DC bus status, the inverter output status, and the branch status. The event types include temporary sag holding events, continuous fault events, and pending events. When the event type is a temporary sag event or a pending event, the current power path of the static switching unit is maintained, the battery charging branch is prohibited from being connected, and the uninterruptible power supply unit is controlled to maintain load support. When the event type is a persistent fault event, the holding of the current power path is released; after the repetitive disturbance sequence ends and the mains power recovers and stabilizes, the input path of the static switching unit corresponding to the mains power access unit is set as the target path, and regression control is executed in the order of target path switching, rectifier input recovery, battery charging branch recovery and feeder restriction release. The central control unit includes a disturbance memory module. The disturbance memory module saves the power path holding state of the static switching unit, the battery charging branch prohibition state, the feeder power supply unit restriction state, and the uninterruptible power supply unit support state when the first voltage drop corresponding to each repeated disturbance sequence identifier begins. In subsequent voltage drops under the same repeated disturbance sequence identifier, the corresponding stored control state is directly loaded.

[0008] In some embodiments, the status acquisition unit includes a reclosing signal interface, which is used to receive status quantities output by the upstream power supply line protection device, station control device, or dispatch communication link, and generate the reclosing indication signal.

[0009] In some embodiments, the central control unit determines the change in voltage at the mains access point from the normal zone to the drop zone, from the drop zone to the recovery zone, and the change in the state of the reclosing indication signal as boundary events, and merges multiple voltage drops and recoveries with a time interval between adjacent boundary events that is less than the duration of the boundary event merging into the same repeating disturbance sequence identifier.

[0010] In some embodiments, when the repetitive disturbance sequence identifier is valid, the reclosing indication signal is within the reclosing execution related time period, the inverter output state is continuous and the DC bus state is higher than the support threshold, the temporary sag holding event is output. When the voltage at the mains access point is lower than the power failure threshold and continues to exceed the fault confirmation time, the continuous fault event is output. When the DC bus status is below the support release threshold, the continuous fault event is output; When the phase angle of the mains access point changes beyond the limit relative to the previous recovery state, the continuous fault event is output. If the aforementioned conditions are not met, output the event to be judged.

[0011] In some embodiments, under the temporary hold event or the pending event, the central control unit outputs a transfer suppression command to the static switching unit and pauses the current transfer delay count; under the persistent fault event, the transfer suppression command is canceled and the current transfer delay count is restarted.

[0012] In some embodiments, the feeder power supply unit includes critical feeder groups and non-critical feeder groups, and the central control unit stores a feeder priority table; under the temporary sag hold event or the pending event, the critical feeder groups are continuously powered. When the pending event continues for longer than the power supply confirmation time, the non-critical feeder group is sequentially subjected to power supply limitation or power outage according to the feeder priority table, and the current power path holding is allowed to be released only after the sequential execution is completed.

[0013] In some embodiments, the central control unit establishes mains return conditions when the repetitive disturbance sequence identifier fails, the mains access point voltage is in the recovery zone and continues to exceed the stable confirmation time, the inverter output state is continuous and no new tripping occurs in the branch state. After the mains power return condition is met, the input path of the static switching unit corresponding to the mains power access unit is set as the target path, the static switching unit is controlled to switch to the target path, and after the switch is completed, current ramp-up control is performed on the rectifier input terminal.

[0014] In some embodiments, during the recovery control process after the mains power return condition is met, the central control unit restores the battery charging branch within a preset recharge limit determined by combining the current load status, the DC bus status, and the power supply status of the key feeder group, and releases the feeder restrictions level by level according to the recovery order in the feeder priority table. An observation period is maintained between adjacent recovery levels. If no new repeating disturbance sequence identifier is generated during the observation period and no new tripping occurs in the branch status, the next recovery level is executed.

[0015] In some embodiments, during the regression control process corresponding to the mains power regression condition, if a new repeating disturbance sequence identifier is generated, or the mains power access point voltage leaves the recovery band, the currently unfinished regression step is terminated, the current power path of the static switching unit is restored, the battery charging branch is disabled again, and the feeder state that was completed before the termination is maintained.

[0016] In some embodiments, when the reclosing signal interface does not receive a valid status value, the central control unit generates an estimated reclosing flag based on the interval between the start times of two adjacent voltage drops, the duration of the previous voltage recovery, and the direction of phase angle change, and replaces the reclosing indication signal with the estimated reclosing flag.

[0017] Through the above technical solution, the present invention can achieve at least the following beneficial effects: This invention generates a repeating disturbance sequence identifier within the same disturbance process and classifies the current power supply event by combining the reclosing indication signal, DC bus status, inverter output status, and branch status. Under temporary sag holding events and pending events, the current power path of the static switching unit is maintained, the battery charging branch is prohibited from being connected, and load support is maintained. The maintenance of the current power path is only released under continuous fault events. After the repeating disturbance sequence ends and the mains power recovers and stabilizes, regression control is executed in the order of target path switching, rectifier input recovery, battery charging branch recovery, and feeder restriction release. This prevents multiple voltage drops and recoveries within the same reclosing process from being treated as independent power loss events, thereby avoiding repeated switching of the static switching unit, back-and-forth switching of the uninterruptible power supply unit between support and recovery, and the splitting of the feeder power supply status between the disturbance and recovery phases.

[0018] By defining the transition of the mains connection point voltage from the normal range to the drop range, from the drop range to the recovery range, and the change in the reclosing indication signal status as boundary events, and merging multiple voltage drops and recoveries with time intervals between adjacent boundary events shorter than the merging duration of the boundary events into the same repetitive disturbance sequence identifier, the repetitive disturbance process has a unified temporal boundary and sequence affiliation. This allows for the differentiation between a current voltage recovery belonging to a short-term recovery within the same disturbance process and a new power supply phase, thus providing a unified basis for subsequent event classification and regression timing judgment.

[0019] By outputting a temporary sag holding event when the repetitive disturbance sequence identifier is valid, the reclosing execution is within the relevant period, the inverter output status is continuous, and the DC bus status is higher than the support threshold, and outputting a continuous fault event when the mains access point voltage is continuously lower than the power failure threshold, the DC bus status is lower than the support release threshold, or the phase angle exceeds the limit jump relative to the previous recovery status, and outputting a transfer suppression command to the static switching unit and pausing the current transfer delay count under the temporary sag holding event or pending event, the backup power supply switching is based on continuous fault confirmation rather than on a single temporary sag exceeding the limit, thereby reducing erroneous switching and control jitter during repetitive disturbances.

[0020] By implementing current ramp-up control at the rectifier input during the regression control process after the mains power return condition is met, the battery charging branch is restored within the preset recharge upper limit. The feeder restrictions are released step by step according to the recovery order in the feeder priority table. The incomplete regression steps are stopped when a new recurring disturbance sequence is detected or the mains access point voltage leaves the recovery zone. This ensures that the rectifier takeover, recharge recovery, and feeder recovery are carried out in a controlled manner in the same regression chain. This avoids the sudden increase in recharge, chaotic recovery order, and half-recovery state oscillation caused by the disconnection between the rectifier input, charging branch, and feeder recovery after a short-term mains power recovery.

[0021] By directly loading the stored control state into the subsequent voltage drop under the same repeated disturbance sequence identifier, the power path holding state, battery charging branch prohibition state, feeder power supply unit restriction state and uninterruptible power supply unit support state that have been verified in the previous drop can be directly reused in the subsequent repeated drops. This reduces the inconsistency of actions caused by repeated calculation and repeated allocation in the same disturbance process, and makes the subsequent drops preferentially return to the previously verified control combination. Attached Figure Description

[0022] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments will be briefly described below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation on the scope of this application.

[0023] Figure 1This is a framework diagram of the intelligent control system for data center power grid supply in the embodiment. Detailed Implementation

[0024] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0025] All terms used in this application (including technical and scientific terms) have the meanings commonly understood by those skilled in the art, unless otherwise defined. It should be noted that the terms used herein should be interpreted in a manner consistent with the context of this specification, and not in an idealized or overly rigid way.

[0026] To facilitate understanding, the relevant terms and concepts involved in the embodiments of this application will be introduced below: The reclosing indication signal refers to the state quantity output by the upstream power supply line protection device, station control device, or dispatch communication link, which characterizes the reclosing execution status; the repeated disturbance sequence identifier refers to the sequence number after associating multiple voltage drops and recoveries within the same disturbance process; the current power supply path refers to the power supply path from the input power supply to the output terminal that the static switching unit keeps connected at the current moment; the DC bus status refers to the set of states characterizing the DC side support capability of the uninterruptible power supply unit, including at least the bus voltage status, discharge status, and remaining support capability status; the branch status refers to the set of states consisting of the closing status, tripping status, and recovery permission status of each branch of the feeder power supply unit; the drop zone refers to the voltage range where the mains access point voltage is lower than the normal operating range but has not reached the power failure state; the recovery zone refers to the voltage range where the mains access point voltage returns from the drop zone and enters the return judgment range; the mains return condition refers to the combination of conditions that allow the transition from the support power supply state to the normal power supply state after the repeated disturbance sequence ends.

[0027] Furthermore, the normal range refers to the voltage range where the mains access point voltage is within the rated allowable deviation and has not triggered a drop state. This serves as a reference range for identifying the transition from normal power supply to a disturbance process. The engineering standard can be set to 90%~110% of the rated voltage and can be adjusted in conjunction with the allowable deviation of the upstream power supply. The judgment window is the time interval within which the central control unit performs alignment and event classification on the voltage, current, phase angle, DC bus status, inverter output status, and branch status of the same batch of mains access points. The default value is 10~40ms, which can be adjusted from 5~100ms. The control cycle is the execution cycle for the central control unit to complete one status reception, event classification, and control output. The default value is 5~20ms, which can be adjusted from 2~50ms. The target path is the input path of the static switching unit corresponding to the mains access unit, which serves as the priority access path for regression control after the mains return condition is met. All the above time parameters are stored in the current control configuration and remain unchanged during the effective period of the same repetitive disturbance sequence to avoid drift in the event classification and regression control standards.

[0028] In some embodiments, the central control unit organizes coordinated control of the static switching unit, uninterruptible power supply unit, and feeder power supply unit around the recurring disturbance sequence identifier. During the disturbance phase, the central control unit classifies the current power supply event based on the voltage drop and recovery process of the mains access point, the reclosing indication signal, the DC bus status, the inverter output status, and the branch status. Under the conditions of a temporary sag holding event or a pending event, the central control unit maintains the current power path, prohibits the battery charging branch from being connected, and maintains load support. Under the condition of a continuous fault event, the central control unit releases the holding of the current power path and allows the static switching unit to complete the switching according to the input conditions. After the recurring disturbance sequence ends, the central control unit establishes the mains return conditions and executes the return control in the order of target path switching, rectifier input recovery, battery charging branch recovery, and feeder restriction release.

[0029] Example 1: like Figure 1 As shown, this embodiment provides an intelligent control system for data center grid power supply. In this embodiment, the central control unit sequentially executes the following steps for repeated voltage drops and recovery within the same disturbance process: generating a repeated disturbance sequence identifier, determining the event type, maintaining the current power path, providing uninterrupted power supply support, and performing sequential return control after mains power recovery. This ensures that the static switching unit, uninterrupted power supply unit, and feeder power supply unit maintain coordinated operation during the same disturbance process. This embodiment adopts the following technical solution, which includes a mains power access unit, a backup power supply unit, a static switching unit, an uninterrupted power supply unit, a feeder power supply unit, a status acquisition unit, and a central control unit. The mains power access unit and the backup power supply unit are respectively connected to the input terminal of the static switching unit. The uninterruptible power supply unit includes a rectifier input terminal, an inverter output terminal and a battery charging branch. The output terminal of the static switching unit is connected to the rectifier input terminal and the inverter output terminal is connected to the feeder power supply unit. The status acquisition unit collects the voltage, current, phase angle, reclosing indication signal, DC bus status, inverter output status, and branch status of the mains access point; The central control unit generates a repeating disturbance sequence identifier within the same disturbance process, and determines the current power supply event as an event type based on the repeating disturbance sequence identifier, reclosing indication signal, DC bus status, inverter output status, and branch status. The event types include temporary sag holding event, continuous fault event, and pending event. In the event of a transient hold-up, the current power path is maintained, the battery charging branch is disabled, and the uninterruptible power supply unit is controlled to maintain load support; in the event of a pending event, the current power path is maintained, the battery charging branch is disabled, and the uninterruptible power supply unit is controlled to maintain load support; in the event of a persistent fault, the hold-up of the current power path is released. After the repeated disturbance sequence ends and the mains power stabilizes, the input path of the static switching unit corresponding to the mains power access unit is set as the target path, and regression control is executed in the following order: first switch to the target path, then restore the rectifier input terminal, then restore the battery charging branch, and finally remove the feeder power supply unit restriction. The central control unit includes a disturbance memory module. The disturbance memory module saves the power path holding state, battery charging branch prohibition state, feeder power supply unit restriction state and uninterruptible power supply unit support state of the static switching unit at the beginning of the first voltage drop corresponding to each repeated disturbance sequence identifier, and directly loads the corresponding stored control state in subsequent voltage drops under the same repeated disturbance sequence identifier. In this embodiment, the status acquisition unit synchronously acquires the mains access point voltage, mains access point current, and phase angle change status, and adds a unified time stamp to the DC bus status, inverter output status, and branch status before sending them to the central control unit. The mains access point voltage includes at least three-phase RMS status, dip status, recovery status, and power failure status; the mains access point current includes at least three-phase RMS status, impulse status, and continuous current status; the inverter output status includes at least continuous output status, output cut-off status, and output recovery status. The central control unit performs correlation processing on the above statuses according to the same decision window, and simultaneously outputs control commands to the static switching unit, uninterruptible power supply unit, and feeder power supply unit based on the same control status to maintain the timing consistency between event classification, path holding, support power supply, and return control.

[0030] In one example, the state acquisition unit generates a state snapshot record for the state set formed within each control cycle. The state snapshot record is a set of state records summarized according to a unified time stamp, including at least the sampling time, state source, state category, state value, validity flag, and the identifier of the repeating disturbance sequence to which it belongs. The mains access point voltage, mains access point current, and phase angle change states are sampled in the same batch. The DC bus state, inverter output state, and branch state are aligned to the sampling batch with the most recent valid update value. If any state quantity is missing, exceeds its range, or its time deviation exceeds one control cycle, the central control unit marks the state quantity as invalid and prioritizes using the corresponding valid state from the previous control cycle, without directly triggering an event type switch based on the invalid state. When multiple state quantities in the same state snapshot record conflict simultaneously, the central control unit first determines whether to enter a conservative hold state based on the mains access point voltage and DC bus state, and then determines whether to allow the continuation of the original regression step by combining the inverter output state and branch state.

[0031] The restoration of stable mains power is confirmed by maintaining the recovery range without any new voltage drop events. A new voltage drop event is confirmed by the mains connection point voltage re-entering the voltage drop range or a power outage state. When executing return control, the central control unit first confirms that the transition conditions between the current power path and the target path are met, then restores the rectifier input, followed by the battery charging branch, and finally removes the feeder power supply unit restriction; subsequent steps remain prohibited until any of these steps are completed.

[0032] In this embodiment, at the start of the first voltage drop corresponding to each repeating disturbance sequence identifier, the disturbance memory module saves the power path holding state of the static switching unit, the battery charging branch prohibition state, the feeder power supply unit restriction state, and the uninterruptible power supply unit support state, and associates the control state confirmed by feedback with the corresponding repeating disturbance sequence identifier. For subsequent voltage drops under the same repeating disturbance sequence identifier, after determining that the repeating disturbance sequence identifier is still valid, the central control unit directly loads the stored control state corresponding to the repeating disturbance sequence identifier, so that the static switching unit, battery charging branch, feeder power supply unit, and uninterruptible power supply unit are preferentially restored to the previously verified control combination. If, during the loading process, the actual state of the static switching unit, the inverter output state, or the branch state is found to be inconsistent with the stored control state, the loading of the inconsistent parts is stopped, and the currently executed conservative holding state is maintained.

[0033] In one example, the disturbance memory module writes the stored control state corresponding to the same repeating disturbance sequence identifier into the disturbance memory record table. The disturbance memory record table is a data record table used for indexing and reading back stored control states, and includes at least the sequence number, writing time, power path hold status, battery charging branch prohibition status, feeder power supply unit restriction status, uninterruptible power supply unit support status, memory reference voltage, memory reference phase angle, memory reference bus voltage, level validity flag, level failure indicator, and version binding index. The version binding index is a data index key used to identify the tuning version used in the current control cycle. After the first voltage drop begins, the central control unit writes the stored control state of the corresponding level when the corresponding control state is stable and continues for more than the memory confirmation time. The memory confirmation time is set to 50~200ms by default and can be adjusted from 20~500ms. The feeder power supply unit restriction status is written after each level of power restriction is completed and confirmed by branch status readback. If a boundary event re-entry, branch status change unconfirmed, or version binding index changes at the current writing time, the current writing is canceled and the previous valid record is used. After a repeating perturbation sequence fails, the central control unit can either convert the corresponding record to a failed state or put it into an archived state. In the archived state, it is only used for runtime backtracking and will no longer be involved in direct retrieval.

[0034] In this embodiment, the status acquisition unit includes a reclosing signal interface, which is used to receive status quantities output by the upstream power supply line protection device, station control device or dispatch communication link, and generate a reclosing indication signal. The reclosing indication signal includes execution start state, execution hold state, execution end state, lockout state, invalid state, and communication interruption state. The reclosing execution related time period refers to the time interval within the preset hold period corresponding to the reclosing indication signal being in the execution start state, execution hold state, or execution end state. After receiving relevant status quantities from the upstream power supply line, the status acquisition unit retains the time of status change and forms the most recent status change record. The most recent status change record includes at least the status source, status category, change time, and validity flag. When the reclosing indication signal is in an invalid state or a communication interruption state, the central control unit does not directly use this status quantity as a continuous fault criterion, but instead continues to perform event classification in conjunction with the voltage drop and recovery process at the mains connection point.

[0035] In this embodiment, the central control unit determines the change in voltage at the mains access point from the normal zone to the drop zone, from the drop zone to the recovery zone, and the change in the state of the reclosing indication signal as boundary events. Multiple voltage drops and recoveries with a time interval between adjacent boundary events that is less than the duration of the boundary event merging are grouped into the same repeating disturbance sequence identifier. The sequence record corresponding to the repeating disturbance sequence identifier includes at least the sequence number, the start time of the first drop, the start time of the most recent recovery, the boundary event type, the cumulative number of drops, the cumulative number of recoveries, the current validity flag, and the end flag. The central control unit establishes a new sequence record upon detecting the first boundary event. When subsequent boundary events arrive, if the time interval between the boundary event and the most recent boundary event is less than the boundary event merging duration, the boundary event is added to the current sequence record; if the time interval is not less than the boundary event merging duration, the current sequence record is closed and a new sequence record is established. The end of the repeating disturbance sequence is confirmed when no new boundary events are received during the observation period and the mains access point voltage remains within the recovery band.

[0036] Specifically, boundary events transitioning from the normal zone to the drop zone can be confirmed by ensuring that the drop condition is met for 2-5 consecutive sampling points. Boundary events transitioning from the drop zone to the recovery zone can be confirmed by ensuring that the recovery condition is met for 1-3 consecutive power frequency cycles, thus suppressing the impact of single-point noise on boundary event identification. The boundary event merging duration is used to determine whether adjacent drops and recoveries belong to the same disturbance process, with a default value of 100-1500ms, which can be adjusted from 50-3000ms. The observation period is used to confirm the end of the current repetitive disturbance sequence, with a default value of 50-1000ms, which can be adjusted from 20-3000ms. When the reclosing indication signal changes during the effective period of the current repetitive disturbance sequence, the central control unit writes the time of its state change into the current sequence record and uses this time of state change as an auxiliary basis for updating the time of the most recent boundary event. When the boundary event is triggered only by a change in the reclosing indication signal and the mains access point voltage has not left the recovery zone, the central control unit maintains the current event type and only updates the effective duration of the repetitive disturbance sequence identifier.

[0037] In this embodiment, when the repetitive disturbance sequence identifier is valid, the reclosing indication signal is within the reclosing execution related time period, the inverter output state is continuous, and the DC bus state is higher than the support threshold, the output sag holding event is triggered. When the voltage at the mains access point is lower than the power failure threshold and continues to exceed the fault confirmation time, a continuous fault event is output. When the DC bus condition is below the support release threshold, a continuous fault event is output; When the phase angle of the mains access point changes beyond the limit relative to the previous recovery state, a continuous fault event is output. If the aforementioned conditions are not met, output the event to be judged; The support threshold and support release threshold are determined based on the DC bus status, current load status, and battery support capacity status. When the DC bus status is below the support release threshold, the central control unit determines that the uninterruptible power supply unit cannot continue to maintain the current support. An out-of-limit jump in phase angle relative to the previous recovery state is confirmed by both the direction and magnitude of the change in phase angle at the current recovery time relative to the reference phase angle of the previous stable cycle. The preset hold period begins timing when the reclosing indication signal enters the execution end state. If a voltage drop occurs again within this preset hold period, the central control unit maintains the current recurring disturbance sequence identifier valid.

[0038] The priority of persistent fault events is higher than that of temporary drop hold events, and the priority of temporary drop hold events is higher than that of pending events. When multiple judgment conditions are met simultaneously within the same judgment window, the central control unit outputs a single event type according to the above priority.

[0039] For example, the support threshold and support release threshold are set using a hysteresis-coordinated approach, with the support threshold being higher than the support release threshold to avoid event type jitter caused by the DC bus state fluctuating around the threshold. The difference between the two can be adjusted to 2%~10% of the DC bus rated value and can be corrected based on historical sample quantiles, validation set tuning results, and the equipment's allowable fluctuation range. The fault confirmation duration is set to 20~300ms by default and can be adjusted to 10~1000ms; the preset hold period is set to 100~1500ms by default and can be adjusted to 50~3000ms. Continuous fault events should be confirmed when the same continuous fault condition is met for two consecutive judgment windows, and temporary sag hold events should be confirmed when the current judgment window meets the condition and the inverter output state is continuous. The pending event is treated as a hold event type when the aforementioned conditions are not met. The above threshold, duration, and priority relationships are not rewritten online during the validity period of the same repetitive disturbance sequence, and can only be switched to new tuning values ​​after the repetitive disturbance sequence fails.

[0040] In this embodiment, under a temporary hold event or a pending event, the central control unit outputs a transfer suppression command to the static switching unit and pauses the current transfer delay count; under a continuous fault event, the transfer suppression command is canceled and the current transfer delay count is restarted. Pausing the current transfer delay count means freezing the transfer waiting state already entered by the static switching unit and prohibiting the issuance of transfer execution commands; restarting the current transfer delay count means that after canceling the transfer suppression command, the static switching unit restarts the timing according to the preset delay start point. The central control unit continuously verifies the accessibility status of the backup power supply unit under a temporary sag hold event or a pending event. The accessibility status of the backup power supply unit includes at least the input valid status, phase enabled status, and output enabled status. When a persistent fault event is established and the accessibility status of the backup power supply unit is established, the static switching unit performs a power path transfer.

[0041] It is understood that the confirmation of the current power path is based on at least the input selection readback status of the static switching unit, the valid status of the selected input side voltage, and the continuous output status of the uninterruptible power supply unit (UPS). Only when all three are consistent will the central control unit confirm the input path as the current power path. The accessibility status of the backup power supply unit can be confirmed by the input valid status, phase allowable status, and output allowable status. The phase allowable status can be set according to the phase angle tolerance allowed by the static switching unit, which is set to 5°~20° by default and can be adjusted to 3°~30°. If a persistent fault event has been established but the accessibility status of the backup power supply unit has not yet been established, the central control unit will maintain the current power path holding result unchanged, continue to prohibit the battery charging branch from being put into operation, and retain the continuous power supply of the critical feeder group and the limited power supply result of the non-critical feeder group until the accessibility status of the backup power supply unit is established or the current recurring disturbance sequence ends.

[0042] In this embodiment, the feeder power supply unit includes critical feeder groups and non-critical feeder groups, and the central control unit stores the feeder priority table; it maintains continuous power supply to the critical feeder groups under temporary sag events or pending events. When the pending event continues for longer than the power supply confirmation time, power supply limitation or power outage shall be executed sequentially for non-critical feeder groups according to the feeder priority table, and the current power path holding shall be released only after the sequential execution is completed. Critical feeder groups refer to the collection of feeders carrying core computing loads, core storage loads, core network loads, or necessary environmental protection loads. Non-critical feeder groups refer to the collection of feeders carrying loads that can be delayed and restored. The feeder priority table includes at least the feeder number, feeder group, load level, power limiting permission status, power outage permission status, restoration order, and most recent branch status. During the duration of the pending event, the central control unit performs power limiting or power outage on non-critical feeder groups according to the feeder priority table from low to high. Power limiting includes keeping the feeder connected and suppressing new load input; power outage includes disconnecting the power supply path of the corresponding feeder. Critical feeder groups maintain continuous power supply within the limits of current support capacity.

[0043] Furthermore, the feeder priority table is generated based on load level, branch status, and operating configuration during initial operation, and managed in subsequent operations using the version bound index corresponding to the current control configuration. The currently effective content of the feeder priority table includes at least the feeder number, feeder group, load level, current restriction status, most recent action time, recovery sequence, and validity flag. After the central control unit performs power limiting or power withdrawal on a non-critical feeder group, it confirms the effectiveness of the action by reviewing the branch status. If the branch status does not return to a state consistent with the target action within 1-3 control cycles, the central control unit maintains the branch in its current restriction status and records the unconfirmed flag, and does not include the branch in subsequent recovery levels. During the effective period of the same recurring disturbance sequence, the feeder priority table does not immediately replace the currently effective content with manually rewritten results. The manually rewritten results participate in subsequent regression control only after the current recurring disturbance sequence fails or the mains power regression conditions are re-established.

[0044] In this embodiment, when the reclosing signal interface does not receive a valid status quantity, the central control unit generates an estimated reclosing flag based on the interval between the start times of two adjacent voltage drops, the duration of the previous voltage recovery, and the direction of phase angle change, and replaces the reclosing indication signal with the estimated reclosing flag. The estimated reclosing flag includes estimated start state, estimated hold state, estimated end state, and estimated invalid state. When the central control unit (CCU) does not receive a valid state value for the reclosing indication signal, it generates an estimated reclosing flag based on whether there is a short recovery phase between the start times of two adjacent voltage dips, whether the duration of the previous recovery falls within the reclosing observation window, and whether the phase angle change direction is continuous with the previous recovery direction. The CCU sets an independent valid flag for the estimated reclosing flag; when a valid reclosing indication signal is subsequently received, the valid reclosing indication signal replaces the estimated reclosing flag, and the associated state of the current repetitive disturbance sequence identifier is updated.

[0045] For example, the reclosing observation window is a time window used to determine whether two adjacent voltage drops are related to the same reclosing process. The default value is 100-1500ms, which can be adjusted from 50-3000ms. When the duration of the previous recovery falls within this reclosing observation window and the phase angle change direction remains continuous, the central control unit allows the generation of an estimated reclosing flag. If the interval between the start times of two adjacent voltage drops is missing, the duration of the previous recovery cannot be confirmed, or the phase angle change direction is inconsistent in two consecutive sampling batches, the central control unit sets the estimated reclosing flag to an invalid state and performs event classification only based on the mains connection point voltage, DC bus status, inverter output status, and branch status, without triggering event type switching solely based on the estimated reclosing flag. After a valid reclosing indication signal is restored, the central control unit overwrites the estimated reclosing flag with this valid state quantity and uses the overwriting time as the associated state update time for the current recurring disturbance sequence identifier.

[0046] In a preferred embodiment of Example 1, for subsequent voltage drops under the same repetitive disturbance sequence identifier, the central control unit directly recalls the existing control state corresponding to the repetitive disturbance sequence identifier as follows: At the start of the first voltage drop corresponding to each repeated disturbance sequence identifier, the central control unit establishes a hierarchical memory set corresponding to the sequence via the disturbance memory module, and stores the executed and verified control states separately according to the path holding layer, the uninterruptible power supply support layer, and the feeder limitation layer.

[0047] Hierarchical memory sets can be represented as: , in, The repeating perturbation sequence is identified as A hierarchical memory set; To maintain the existing control state of the path layer; The existing control status of the uninterruptible power supply support layer; This refers to the existing control state of the feeder constraint layer; , and These are valid markers for the path preservation layer, uninterruptible power supply support layer, and feeder constraint layer, respectively. This is the memory reference voltage corresponding to the repetitive perturbation sequence; This is the memory reference phase angle corresponding to the repeated perturbation sequence; The memory reference bus voltage corresponding to this repetitive perturbation sequence; An index is bound to the version corresponding to the recurring disturbance sequence. The path holding layer records the current power path, path holding state, transfer suppression state, and current transfer delay freeze state of the static switching unit; the uninterruptible power supply support layer records the rectifier input terminal latching state, battery charging branch prohibited state, inverter support state, and corresponding support maintenance level; the feeder restriction layer records the restriction mask for each branch to limit or exit power supply, the order of executed restrictions, and the current allowed recovery boundary; the current allowed recovery boundary is the feeder priority boundary for currently allowed to release restrictions; subsequent calls are no longer to a single state bit, but to a set of control states that have been verified within the same recurring disturbance sequence.

[0048] This is only used to bind the matching weight, call threshold, feeder priority table, and regression parameter version corresponding to the writing to this hierarchical memory set. The version binding index corresponding to the current control cycle is... In case of inconsistency, the central control unit does not perform direct loading of the recurring disturbance sequence and maintains the current conservative hold state.

[0049] When a subsequent voltage drop occurs under the same repeating disturbance sequence identifier, the central control unit does not recalculate the control allocation results obtained at the time of the first disturbance. Instead, it first matches the current acquisition state with the memory reference state corresponding to the sequence to obtain the memory recall matching degree. , in, The repeating perturbation sequence is identified as At the present moment The degree of matching of memory retrieval; As the voltage deviation weight; The phase angle deviation weight; The weight is the bus deviation. This represents the current voltage at the mains power connection point. This is the memory reference voltage corresponding to the stored control state of the repetitive perturbation sequence; Calibrate the voltage deviation; The phase angle of the mains power connection point at the current moment; The memory reference phase angle for the existing control state corresponding to this repetitive perturbation sequence; Calibrate the phase angle deviation; This represents the current DC bus voltage. The memory reference bus voltage corresponding to the stored control state of this repetitive disturbance sequence; The bus deviation calibration values ​​are set as follows: voltage deviation calibration value is set at 2%~4% of the rated voltage, phase angle deviation calibration value is set at 3°~6°, and bus deviation calibration value is set at 1%~2% of the rated value of the DC bus; the weights are given using an offline calibration method to meet the requirements. Within the same repeating perturbation sequence, the data is not rewritten in real time to avoid caliber drift. The matching degree is updated once upon arrival of each boundary event and can be checked once within the subsequent 1-2 power frequency cycles. The smaller the value, the closer the current disturbance state is to the verified control state of the sequence.

[0050] Used to calculate memory retrieval matching degree , and If the same sampling batch is used after the same boundary event; if any input quantity is missing, exceeds the range, or the sampling time is misaligned for more than one control cycle, or the corresponding memory reference quantity has not been written, the memory call matching degree at the current moment will not participate in the hierarchical call determination, and the call enable energy of the corresponding level will be set to zero.

[0051] After obtaining the matching degree, the central control unit performs the call determination in the order of path maintenance layer, uninterruptible power supply support layer, and feeder restriction layer, and generates call enable energy for each layer: , in, The repeating perturbation sequence is identified as At the present moment hierarchical The call to energy; For hierarchical identification, take , or ; The aforementioned definition will be used; hierarchical The corresponding matching degree call threshold; hierarchical Valid marker; hierarchical The failure indication value at the current moment. The call threshold value of the path retention layer is higher than that of the uninterruptible power supply support layer, and the call threshold value of the uninterruptible power supply support layer is higher than that of the feeder restriction layer, so that the path retention layer is given priority in the call, and the feeder restriction layer is only allowed to replay under more stringent conditions. The failure conditions of the path retention layer include the unavailability of the source side corresponding to the stored power path and the inconsistency between the actual state of the static switching unit and the state of the stored path; the failure conditions of the uninterruptible power supply support layer include the DC bus state being lower than the support release threshold and the inverter output state being discontinuous; the failure conditions of the feeder restriction layer include the occurrence of a new trip in the branch state, the inconsistency between the feeder topology and the memory time, and the manual rewriting of the restricted branch. The call priority and failure conditions are both attached to the same disturbance memory module, without introducing additional functional units.

[0052] Once a failure indicator is set to valid, it will not be immediately cleared by a single sampling recovery before the identifier of the same repeating disturbance sequence fails. Only when the corresponding anomaly has not appeared in two consecutive matching degree verification windows, and the version binding index corresponding to the current control cycle is consistent with the version binding index when the existing control state of this layer was written, is the failure indicator of that layer allowed to be reset and reinstated for use. If a boundary event re-entry, feeder topology change, or manual rewriting occurs within the memory confirmation period, the current write will be cancelled, and the original existing control state will be used.

[0053] When a call to a certain layer causes the energy to be 1, the central control unit directly uses the existing control state of that layer as the current control output; when the call causes the energy to be 0, the current conservative control state of that layer remains unchanged. Its call output can be represented as: , in, The current time level The actual output control state; hierarchical The call to energy; The repeating perturbation sequence is identified as Time level The existing control state; The current time level The conservative maintenance state. When the path maintenance layer is invoked, the central control unit directly restores the current power path maintenance state, transition suppression state, and transition delay freeze state corresponding to the recurring disturbance sequence; when the uninterruptible power supply support layer is invoked, the central control unit directly restores the rectifier input terminal prohibited state, the battery charging branch prohibited state, and the inverter support maintenance state; when the feeder restriction layer is invoked, the central control unit directly restores the previously verified feeder restriction mask and the executed restriction sequence of the recurring disturbance sequence, so that the power limiting steps are no longer recalculated from the starting point of the feeder priority table, but the subsequent actions are shortened to a replay of a verified control state.

[0054] When executed in the above manner, during the first voltage drop in the same repetitive disturbance sequence, the central control unit still performs judgment and control according to the event judgment logic, path holding control logic, uninterruptible power supply support logic, and feeder limiting logic corresponding to the first voltage drop in the repetitive disturbance sequence. Only after the control state stabilizes and continues for more than the memory confirmation time will the disturbance memory module write the current state of the corresponding layer into the hierarchical memory set. The memory confirmation time is set to 50ms~200ms. The path holding layer and uninterruptible power supply support layer are written when they continuously and stably reach this time after a boundary event. The feeder limiting layer is written after each level of power limiting is completed and confirmed by branch feedback. The object written is not the instantaneous state, but the already verified control result.

[0055] In subsequent voltage drops under the same recurring disturbance sequence, after recognizing that the sequence identifier is still valid, the central control unit first calls the path holding layer within the current control cycle, causing the static switching unit to immediately return to the previously verified path holding state. Subsequently, in the next control cycle, the uninterruptible power supply support layer is called to restore the previously verified combination of rectifier blocking, battery charging branch prohibition, and inverter support. If no new tripping occurs in the branch state and the feeder topology remains unchanged, the feeder restriction layer is called to replay the previously executed power restriction results according to the original mask. If the upper layer has been called but the lower layer does not meet the calling conditions, only the called results of the upper layer are maintained, and the lower layer is kept in the current conservative state, without cross-layer forced linkage.

[0056] When a bus fluctuation amplifies, inverter output state jumps, static switching unit state readbacks are inconsistent, or a new trip occurs in the branch state within the same recurring disturbance sequence, the central control unit sets the corresponding layer's failure indication to valid and stops the re-calling of existing control states at that layer. Successfully invoked upper-layer states are not revoked but frozen at the current conservative level. If the anomaly disappears during subsequent verification and continues beyond the failure resolution time, the path maintenance layer and uninterruptible power supply support layer can be allowed to participate in the re-calling process again. The feeder restriction layer will still be verified under stricter conditions, and will continue to maintain the feeder states already executed before the termination until it passes verification. This avoids reapplying mismatched historical control states to the current system due to local state drift within the same recurring disturbance sequence.

[0057] Example 2: Based on Example 1, this example further provides a regression control method for the mains power recovery phase. In this example, after the repetitive disturbance sequence ends, the central control unit sequentially executes target path switching, rectifier input recovery, battery charging branch restriction recovery, and feeder restriction removal, ensuring a continuous connection between the regression process and the aforementioned disturbance control process. This example adopts the following technical solution: The central control unit establishes mains return conditions when the repeating disturbance sequence identifier fails, the mains access point voltage is in the recovery zone and continues to exceed the stable confirmation time, the inverter output status is continuous and no new tripping occurs in the branch status. After the mains power return condition is met, the input path of the static switching unit corresponding to the mains power access unit is set as the target path, the static switching unit is controlled to switch to the target path, and the current ramp-up control is performed on the rectifier input terminal after the switch is completed. During the recovery control process after the mains power return condition is met, the central control unit restores the battery charging branch within the preset recharge limit determined by combining the current load status, DC bus status and key feeder group power supply status, and releases the feeder restrictions level by level according to the recovery order in the feeder priority table. An observation period is maintained between adjacent recovery levels. If no new repeated disturbance sequence identifier is generated during the observation period and no new tripping occurs in the branch status, the next recovery level is executed. In this embodiment, the central control unit establishes a mains return condition when the repetitive disturbance sequence identifier fails, the mains access point voltage is in the recovery zone and continues to exceed the stability confirmation time, the inverter output state remains continuous, and no new tripping occurs in the branch state. After the mains return condition is established, the central control unit sets the input path of the static switching unit corresponding to the mains access unit as the target path and controls the static switching unit to switch to the target path. After confirming that the static switching unit has completed the target path switch, the central control unit performs current ramp-up control on the rectifier input terminal, so that the rectifier input current gradually recovers according to the predetermined ramp-up process, and restores the battery charging branch within the preset recharge limit. The preset recharge limit is determined in combination with the current load state, DC bus state, and key feeder group power supply state, and is constrained by prioritizing the continuous power supply of the key feeder group and the stability of the DC bus. The feeder restriction is lifted in the order of recovery in the feeder priority table, with a preset observation period between adjacent recovery levels. If no new repeating disturbance sequence identifier is generated during the observation period, the mains access point voltage does not leave the recovery zone, and no new tripping occurs in the branch status, the next recovery level is executed.

[0058] If a new repetitive disturbance sequence identifier is generated during the regression control process corresponding to the mains power return condition, or if the mains power access point voltage leaves the recovery zone, the current unfinished regression step is stopped, the current power path of the static switching unit is restored, the battery charging branch is disabled again, and the feeder state that was completed before the stop is maintained. The central control unit sequentially records the current steps of the regression control, including at least the current step identifier, step start time, step completion flag, and abort flag. When a new recurring disturbance sequence is generated during regression control or the mains connection point voltage leaves the recovery zone, the central control unit marks the incomplete step as aborted and re-enters the control states of current power path hold, charging branch prohibition, and feeder restriction hold. Completed feeder states remain in their current state, while incomplete feeder states do not proceed to the next recovery stage. After a new recurring disturbance sequence ends, the central control unit re-establishes the mains regression conditions and continues regression control from the previous stable state corresponding to the aborted step.

[0059] In a preferred embodiment of Example 2, the specific method for performing current ramp-up control on the rectifier input terminal and restoring the battery charging branch within a preset recharge limit is as follows: After the mains power return conditions are met and the static switching unit has switched to the target path, the central control unit divides the return phase into a rectifier takeover segment, a limited recharge segment, and a coordinated release segment. In the rectifier takeover segment, the battery charging branch remains disabled, and the rectifier input gradually takes over the load current borne by the inverter side. In the limited recharge segment, the battery charging branch is restored without changing the current power path, but the recharge current is not a fixed value; instead, it is segmented and limited based on the mains power recovery quality and the return progress. In the coordinated release segment, the recharge upper limit continues to increase as the feeder restriction is lifted, ensuring that the rectifier input, load recovery, and battery recharge maintain a gradual, unidirectional increase.

[0060] The quality recovery closed loop is calculated using the following formula: , in, The recovery quality factor for the current assessment period; The effective value of the mains access point voltage during the current assessment period; To restore the reference voltage to the mains power supply; Calibrate the voltage deviation; The phase angle of the mains access point during the current assessment period; The reference phase angle recorded when establishing the mains return condition; Quantitative analysis of phase angle jump; This represents the DC bus fluctuation amplitude during the current assessment period. Calibrate the busbar fluctuation; , and These are the voltage deviation weight, phase angle deviation weight, and bus fluctuation weight, respectively. , and The sum is taken as 1. The voltage deviation calibration value is set at 2%~5% of the rated voltage, the phase angle jump calibration value is set at 3°~8°, and the bus fluctuation calibration value is set at 1%~3% of the rated bus voltage. The quality closed-loop recovery is updated once every 2~5 consecutive power frequency cycles as a rolling evaluation window; when... During descent, the rectified climb step size decreases, and the recharge upper limit is lowered; when When the release threshold for restoring quality is lowered, the recharging is prohibited.

[0061] The average effective value of the mains connection point voltage within the current rolling evaluation window. Take the average phase angle within the same rolling evaluation window corresponding to the effective voltage value. The difference between the maximum and minimum DC bus voltage values ​​within the same rolling evaluation window is used; all three are aligned using the same timestamp. If any input quantity is missing, exceeds its range, or has a time deviation exceeding one control cycle, the recovery quality closed loop of the current evaluation cycle is deemed invalid, the central control unit maintains the rectified input target current of the previous evaluation cycle without further adjustment, and sets the dynamic recharge upper limit to zero.

[0062] The regression progress closed loop is calculated using the following formula: , in, The regression progress factor for the current evaluation period; This refers to the load margin of the uninterruptible power supply unit. Quantitatively define the load margin; This represents the number of feeders currently in a stable power supply state. This represents the total number of feeders planned for restoration. and These are the margin weight and the feeder progress weight, respectively. and The sum is taken as 1. The load margin of the uninterruptible power supply unit is calculated based on the difference between the inverter's rated output capacity and the current inverter output power. The load margin calibration value is set at 10%~20% of the inverter's rated output capacity. The regression progress closed loop is updated every 200~500ms, or once after each primary feeder restriction is lifted; when At lower levels, only rectified input is allowed to cover the existing load; recharge current is not allowed to enter the high-level range. At time 0, it is counted as 0; greater than 0. Press at time count; Only the number of feeders that remain closed and have not experienced any new trips during the current regression progress refresh cycle is counted; when there are no feeders with restrictions to be lifted during this regression process, the feeder recovery item is treated as already satisfied; when the feeder topology changes or the feeder priority table is manually rewritten, the regression progress factor for the current evaluation cycle will no longer be increased, but will only maintain the result of the previous evaluation cycle or revert to the previous level.

[0063] The current ramp-up step size at the rectifier input is given by the following formula: , in, The ramp-up step size for the rectified input current in the current evaluation cycle; Minimum climb step size; This represents the maximum climbing step length. and Using the aforementioned definitions, the minimum ramp-up step size is set to 1% to 3% of the rectified rated input current, and the maximum ramp-up step size is set to 5% to 10% of the rectified rated input current. When fluctuations occur in the recovery quality closed loop or the return progress closed loop, the step size automatically converges to the minimum ramp-up step size, allowing the rectified input terminal to take over the load with a conservative slope.

[0064] The target current at the rectifier input is updated using the following formula: , in, Set the rectified input target current for the current evaluation cycle; The rectified input target current for the previous evaluation cycle; The aforementioned definition will be used; The upper limit of the rectified input current is allowed. The target current for battery charging is the one used in the previous evaluation cycle. This allows the rise in rectified input current and the recovery of charging current to share the same remaining capacity constraint. When the rectifier margin is insufficient, priority is given to maintaining load power supply before compressing the charging current. Within the same evaluation cycle, the central control unit first updates the target rectified input current based on the target battery charging current of the previous evaluation cycle, and then updates the target battery charging current for the current evaluation cycle based on the dynamic recharge limit.

[0065] The dynamic recharge limit of the battery charging branch is given by the following formula: , in, This represents the dynamic recharge limit for the current assessment period; Recharge release threshold; This is the mid-range recharge threshold; This is the full recharge threshold; Release threshold to restore quality; To restore the high-quality threshold; and These are the primary recharge ratio coefficient and the secondary recharge ratio coefficient, respectively. This is a preset maximum recharge limit. It is acceptable The value is 0.2~0.4. The threshold is 0.5~0.8. The recharge release threshold, mid-stage recharge threshold, and full recharge threshold are set in an increasing order, with the recovery quality release threshold lower than the recovery quality high threshold. After the battery charging branch is restored, Take no higher than Furthermore, the value does not encroach on the margin of the rectifier load connection, and its ramp rate is set at 0.3 to 0.6 times the ramp rate of the rectifier input current, so that the recharge recovery is slower than that of the rectifier connection.

[0066] During the same mains power return control process, , , , , , , , , , , , , , , , , , and All settings are bound to a set of offline settings corresponding to the establishment of the current mains power return conditions, and will not be rewritten online before the current return control ends. Switching to the next set of settings is only allowed when a new repeating disturbance sequence identifier is generated, the mains power return conditions are re-established, or the current return control ends. Different wiring methods, different feeder grouping methods, or different capacity levels correspond to different setting versions, and they should not be mixed across different scenarios.

[0067] Furthermore, the version binding index is used to identify the tuning version used in the current control cycle, and is at least associated with the feeder priority table version, threshold tuning version, regression parameter version, and disturbance memory recall threshold version. When establishing the mains power regression conditions, the central control unit writes the version binding index that is currently in effect into the current regression process, and keeps this version binding index unchanged until the current regression process ends. When the operator modifies the feeder priority table, threshold tuning value, or regression parameter, the modification result only generates a new candidate version and does not directly overwrite the effective version corresponding to the current regression process. When the version binding index corresponding to the current control cycle is inconsistent with the version binding index corresponding to the control state already stored in the disturbance memory module, the central control unit maintains the current conservative hold state and does not execute the direct loading of the corresponding layer to ensure that the stored control state is consistent with the current control caliber.

[0068] When executed in the above manner, in the initial stage after the static switching unit switches to the target path... Typically lower, Keeping it at zero, the central control unit only rises. This allows the rectifier input to initially handle the existing load on the inverter output side. When and At this time, the recharge limit enters the first level, and the battery charging branch resumes conduction; when subsequent feeder restrictions are gradually lifted and the uninterruptible power supply unit load margin remains above the rated range, As it increases, the dynamic recharge limit rises from the first level to the second level and finally to the preset recharge limit. The recharge limit is updated in segments as the regression process progresses, and is no longer fixed as a single value.

[0069] Upgrading the dynamic recharge limit is based on simultaneously meeting the entry conditions for the corresponding level in two consecutive recharge progress refresh cycles; downgrading is triggered by a single refresh cycle. and If the entry conditions for the same gear are not met consecutively within two adjacent refresh cycles, the central control unit maintains the current gear and does not perform a gear upgrade. If any cycle experiences an invalid recovery quality closed loop, a change in feeder topology, a manual rewriting of the feeder priority table, an increase in the evaluation window for two consecutive bus fluctuations, a discontinuity in inverter output status, or a new tripping in branch status, it is only allowed to maintain the current gear or return to the previous gear.

[0070] When, during the recovery process, the bus fluctuation increases for two consecutive evaluation windows, the inverter output status becomes discontinuous, or a new trip occurs in the branch status, and no new recurring disturbance sequence identifier is generated and the mains connection point voltage remains within the recovery band, the central control unit will... Maintain at the level of the previous cycle, and Return to the previous gear; if Further lower Then The voltage is immediately set to zero, pausing the battery charging branch from continuing to rise. If a new repeating disturbance sequence is generated, or the mains access point voltage leaves the recovery zone, the currently incomplete regression step is aborted, the current power path of the static switching unit is restored, the battery charging branch is disabled again, and the feeder state that was completed before the abort is maintained.

[0071] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features therein. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of this application.

[0072] Furthermore, those skilled in the art will understand that although some embodiments herein include certain features included in other embodiments but not others, combinations of features from different embodiments are meant to be within the scope of this application and form different embodiments. For example, all the embodiments above can be used in any combination. The information disclosed in this background section is intended only to enhance the understanding of the general background of this application and should not be construed as an admission or in any way implying that such information constitutes prior art known to those skilled in the art.

Claims

1. A smart control system for data center power grid supply, characterized in that, It includes a mains power access unit, a backup power supply unit, a static switching unit, an uninterruptible power supply unit, a feeder power supply unit, a status acquisition unit, and a central control unit; The mains power access unit and the backup power supply unit are respectively connected to the input terminal of the static switching unit. The uninterruptible power supply unit includes a rectifier input terminal, an inverter output terminal and a battery charging branch. The output terminal of the static switching unit is connected to the rectifier input terminal, and the inverter output terminal is connected to the feeder power supply unit. The status acquisition unit acquires the voltage, current, phase angle, reclosing indication signal, DC bus status, inverter output status, and branch status of the mains access point. The central control unit generates a repeating disturbance sequence identifier within the same disturbance process, and determines the current power supply event as an event type based on the repeating disturbance sequence identifier, the reclosing indication signal, the DC bus status, the inverter output status, and the branch status. The event types include temporary sag holding events, continuous fault events, and pending events. When the event type is a temporary sag event or a pending event, the current power path of the static switching unit is maintained, the battery charging branch is prohibited from being connected, and the uninterruptible power supply unit is controlled to maintain load support. When the event type is a persistent fault event, the holding of the current power path is released; after the repetitive disturbance sequence ends and the mains power recovers and stabilizes, the input path of the static switching unit corresponding to the mains power access unit is set as the target path, and regression control is executed in the order of target path switching, rectifier input recovery, battery charging branch recovery and feeder restriction release. The central control unit includes a disturbance memory module. The disturbance memory module saves the power path holding state of the static switching unit, the battery charging branch prohibition state, the feeder power supply unit restriction state, and the uninterruptible power supply unit support state when the first voltage drop corresponding to each repeated disturbance sequence identifier begins. In subsequent voltage drops under the same repeated disturbance sequence identifier, the corresponding stored control state is directly loaded.

2. The intelligent control system for data center power grid supply according to claim 1, characterized in that, The status acquisition unit includes a reclosing signal interface, which is used to receive status quantities output by the upstream power supply line protection device, station control device, or dispatch communication link, and generate the reclosing indication signal.

3. The intelligent control system for data center power grid supply according to claim 1, characterized in that, The central control unit identifies the voltage at the mains access point as transitioning from the normal band to the drop band, from the drop band to the recovery band, and changes in the state of the reclosing indication signal as boundary events. It also merges multiple voltage drops and recoveries with time intervals between adjacent boundary events that are less than the duration of boundary event merging into the same repeating disturbance sequence identifier.

4. The intelligent control system for data center power grid supply according to claim 1, characterized in that, When the repeating disturbance sequence identifier is valid, the reclosing indication signal is within the reclosing execution related time period, the inverter output state is continuous, and the DC bus state is higher than the support threshold, the temporary sag hold event is output. When the voltage at the mains access point is lower than the power failure threshold and continues to exceed the fault confirmation time, the continuous fault event is output. When the DC bus status is below the support release threshold, the continuous fault event is output; When the phase angle of the mains access point changes beyond the limit relative to the previous recovery state, the continuous fault event is output. If the aforementioned conditions are not met, output the event to be judged.

5. The intelligent control system for data center power grid supply according to claim 1, characterized in that, In the event of the temporary stop holding event or the pending event, the central control unit outputs a transfer suppression command to the static switching unit and pauses the current transfer delay count; in the event of the continuous fault event, the transfer suppression command is canceled and the current transfer delay count is restarted.

6. The intelligent control system for data center power grid supply according to claim 1, characterized in that, The feeder power supply unit includes critical feeder groups and non-critical feeder groups. The central control unit stores a feeder priority table. Under the temporary sag hold event or the pending event, the critical feeder groups are continuously powered. When the pending event continues for longer than the power supply confirmation time, the non-critical feeder group is sequentially subjected to power supply limitation or power outage according to the feeder priority table, and the current power path holding is allowed to be released only after the sequential execution is completed.

7. The intelligent control system for data center power grid supply according to claim 6, characterized in that, The central control unit establishes mains return conditions when the repeating disturbance sequence identifier fails, the mains access point voltage is in the recovery zone and continues to exceed the stable confirmation time, the inverter output state is continuous and no new tripping occurs in the branch state. After the mains power return condition is met, the input path of the static switching unit corresponding to the mains power access unit is set as the target path, the static switching unit is controlled to switch to the target path, and after the switch is completed, current ramp-up control is performed on the rectifier input terminal.

8. The intelligent control system for data center power grid supply according to claim 7, characterized in that, During the recovery control process after the mains power return condition is met, the central control unit restores the battery charging branch within the preset recharge limit determined by combining the current load status, the DC bus status, and the power supply status of the key feeder group. The feeder restrictions are released level by level according to the recovery order in the feeder priority table. An observation period is maintained between adjacent recovery levels. If no new repeating disturbance sequence identifier is generated during the observation period and no new tripping occurs in the branch status, the next recovery level is executed.

9. The intelligent control system for data center power grid supply according to claim 8, characterized in that, During the regression control process corresponding to the mains power regression condition, if a new repeating disturbance sequence identifier is generated, or the mains power access point voltage leaves the recovery band, the currently unfinished regression step is stopped, the current power path of the static switching unit is restored, the battery charging branch is disabled again, and the feeder state that was completed before the stoppage is maintained.

10. The intelligent control system for data center power grid supply according to claim 2, characterized in that, When no valid status value is received at the reclosing signal interface, the central control unit generates an estimated reclosing flag based on the interval between the start times of two adjacent voltage drops, the duration of the previous voltage recovery, and the direction of phase angle change, and replaces the reclosing indication signal with the estimated reclosing flag.

Images