Modular data center dry-type transformer and ups coordination system

Abstract

The application discloses a dry-type transformer and UPS collaborative system for a modular data center, and particularly relates to the technical field of power supply and distribution collaborative control of a modular data center, and is used for solving the problems of insufficient cooperation between a dry-type transformer and an uninterruptible power supply in the existing modular data center, and local overload and insufficient utilization of redundancy during maintenance switching and capacity expansion. By constructing power distribution topology modeling, state modeling, working condition acquisition, scene identification, capacity envelope and collaborative execution integrated control link in the collaborative control device, the dry-type transformer thermal time domain margin, the uninterruptible power supply redundancy capability and the power distribution topology and operation scene are uniformly included in capacity constraint and switching evaluation logic, so that the technical effects of automatically planning and checking load migration path during maintenance switching, fault switching and capacity expansion process, avoiding over-limit operation of the transformer and the uninterruptible power supply, and guaranteeing continuous power supply and collaborative capacity management of the modular data center are achieved.

Description

Technical Field

[0001] This invention relates to the field of modular data center power supply and distribution coordination control technology, specifically a modular data center dry-type transformer and UPS coordination system. Background Technology

[0002] In existing modular data centers, dry-type transformers and uninterruptible power supplies (UPS) are typically designed and configured separately in a hierarchical manner: the transformer side performs capacity verification based on nameplate capacity, short-time overload curves, and empirical safety factors; the UPS side uses a monitoring system to collect output power, power factor, and the number of redundant modules in real time, and judges whether the load exceeds limits based on simple thresholds. During operation, maintenance personnel rely heavily on static design calculations, on-site experience, and instantaneous power values ​​on monitoring screens to manually plan the operation sequence of maintenance, fault switching, and capacity expansion, at most combining a small amount of offline simulation or simple load statistics. They do not link the thermal accumulation state of the transformer, the redundancy capacity of the UPS, the power distribution topology, and the specific switching steps under a unified logic.

[0003] For scenarios requiring frequent adjustments to power supply paths within a short period, such as maintenance and transfer, uninterruptible power supply (UPS) fault switching, and rack expansion, existing technologies typically only check whether the apparent power at a given moment is less than the transformer's rated capacity or whether the UPS load is below the configured percentage limit. They lack time-based capacity evolution constraints and load trajectory prediction based on the complete switching sequence. As a result, on the one hand, to avoid risks, overly conservative capacity margins and operating procedures are often adopted, leading to transformers and UPS operating at low utilization rates for extended periods, making it difficult to fully release the overall capacity of modular data centers. On the other hand, when load fluctuations are large or multiple scenarios overlap, relying solely on static thresholds and human experience can easily overlook the short-term cumulative effects of certain switching moments, causing local transformers or UPS to operate close to or even beyond their capacity for a certain period, increasing the risk of overload, tripping, and cascading failures.

[0004] In summary, the core problem with existing technologies is the lack of a mechanism that can simultaneously consider the thermal time-domain margin of dry-type transformers, the redundancy capability of uninterruptible power supplies, and the distribution topology and typical operating scenarios within a collaborative control device, and conduct unified capacity constraints and switching assessments for operations such as maintenance and transfer, fault switching, and capacity expansion in the time dimension. This makes it difficult to achieve a precise balance between capacity utilization and operational safety. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a modular data center dry-type transformer and UPS collaborative system to solve the problems mentioned in the background section.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a modular data center dry-type transformer and UPS collaborative system, comprising: S1, Distribution Topology Modeling Module, is used to obtain the wiring relationship between the dry-type transformer uninterruptible power supply and the load branch, and generate a distribution topology diagram; S2, Transformer State Modeling Module, is used to collect current, voltage, winding temperature, harmonics, and ambient temperature, and to calculate the thermal time-domain margin fingerprint based on the preset thermal model and short-time overload curve. S3, Uninterruptible Power Supply Operating Condition Acquisition Module, is used to acquire the output power, power factor, harmonics, operating mode, and redundancy capacity of the uninterruptible power supply to form an operating condition vector. S4, Scene Event Recognition Module, is used to identify maintenance and fault switching / expansion access power switching scenarios based on operation and maintenance requests and protection actions, and generate load change vectors and switching sequences. S5, Capacity Envelope Construction Module, is used to calculate the power-time capacity envelope based on thermal time-domain margin fingerprint, base load and operating condition vector, and apply the switching sequence to the distribution topology diagram to pre-simulate the load of each transformer and determine whether the capacity envelope boundary is exceeded. S6, the collaborative execution correction module, is used to generate a collaborative operation plan based on feasible switching sequences, issue load migration instructions and switch control instructions, update the hot time domain margin fingerprint during execution, adjust subsequent operation steps and record operation evidence data when the safety margin is lower than the preset lower limit.

[0007] Furthermore, S1 includes: Extract equipment names, equipment identifiers, voltage levels, and circuit names from electrical construction drawings, cable lists, and switchgear nameplates to form timestamped intermediate records; Based on the results of a power-on test, the intermediate records are checked for continuity and disconnection, and the circuits that fail the continuity and disconnection check are marked as pending confirmation. After removing equipment records marked as retired, a configuration record is generated containing upstream equipment identifier, downstream equipment identifier, connection type, voltage level, and rated current value. The configuration record is then written to non-volatile memory and assigned a topology configuration version number to form a power distribution configuration evidence chain.

[0008] Furthermore, S2 includes: Under unified time synchronization, the transformer condition modeling module collects three-phase current values, bus voltage values, winding hot spot temperatures, harmonic distortion rates, and ambient temperatures according to a preset observation window. Abnormal measurement points are identified based on physical limits. Missing measurement records and measurement records with time deviations are weighted down and then aligned to a unified time axis.

[0009] Furthermore, the transformer condition modeling module converts the aligned three-phase current values ​​and bus voltage values ​​into apparent power and active power in each rhythm cycle, and combines them with harmonic distortion rate and ambient temperature to form an equivalent heat load. The additional apparent power upper limit of the preset time span set is calculated based on the thermal model parameters and short-time overload curve parameters, and the corresponding thermal time-domain margin fingerprint is generated.

[0010] Furthermore, S3 includes: The uninterruptible power supply (UPS) status acquisition module receives status messages containing device identifiers, time stamps, power and operating status under unified time synchronization conditions. Qualified messages are selected based on the acquisition rhythm and allowable deviation. Messages that exceed the numerical range are marked as abnormal messages and are not included in the generation of working condition vectors during the rhythm cycle. After deduplicating duplicate messages according to the idempotency flag within the same rhythm cycle, the uninterruptible power supply (UPS) condition vector is generated by summarizing the data according to the device identifier and written into the condition record area for use by the capacity envelope construction module and the scene event recognition module.

[0011] Furthermore, S4 includes: Under unified time synchronization, the scene event recognition module obtains operation records and protection records with equipment identification and time tags from the duty officer's operation interface, the operation and maintenance management system and protection devices. Within the scene observation window, records are matched according to the operation type, equipment category, and sequence relationship in the operating procedure to identify scene events that change the power supply structure and load structure; Based on the power distribution topology diagram and the uninterruptible power supply condition vector, a load change vector containing load branch identifiers and apparent power increase / decrease amounts, as well as a candidate switching sequence with sequence identifiers, are generated.

[0012] Furthermore, S5 includes: The capacity envelope construction module reads the thermal time-domain margin fingerprint from the transformer condition modeling module, reads the uninterruptible power supply condition vector from the uninterruptible power supply condition acquisition module, and calculates the basic load level of each dry-type transformer in combination with the distribution topology diagram. A power time capacity envelope is generated over a preset time span. The time span identifier, additional apparent power limit, base load level, thermal time domain margin fingerprint version information, and capacity envelope model version information are associated with the transformer identifier and stored in the capacity envelope record area of ​​the collaborative control device.

[0013] Furthermore, based on the power-time capacity envelope, the capacity envelope construction module calculates the expected apparent power of each dry-type transformer at each time point within the observation window for the candidate switching sequence given by the scene event recognition module. It is compared with the maximum allowable apparent power limit at the corresponding time point in the power time capacity envelope. When the expected apparent power at a time point exceeds the limit, the candidate switching sequence is marked as infeasible in the planning record area. Candidate switching sequences are marked as feasible when the expected power does not exceed the upper limit at all time points, and a comprehensive score is formed based on the heat margin consumption balance and the number of switching operations, and written into the planning record area.

[0014] Furthermore, S6 includes: The collaborative execution correction module generates a collaborative operation plan in the collaborative control device based on the feasible switching sequence and comprehensive score given by the capacity envelope construction module; Write the load migration command and switch control command, along with the plan identifier and idempotency key, into the execution queue; Under unified time synchronization, control messages are sent and status feedback messages are received. Calculate the safety margin by reading the thermal time-domain margin fingerprint, the uninterruptible power supply condition vector, and the actual load value; When the safety margin is lower than the preset lower limit, the unexecuted instructions are adjusted according to the preset correction strategy and the running evidence data is written to the evidence chain record area.

[0015] Compared with the prior art, the present invention has the following beneficial effects: 1. By sequentially completing power distribution topology modeling, transformer status modeling, uninterruptible power supply (UPS) condition acquisition, scene event recognition, capacity envelope construction, and collaborative execution correction within the collaborative control device, the thermal time domain margin of dry-type transformers, UPS redundancy capabilities, power distribution topology, and typical operating scenarios are uniformly incorporated into the same set of capacity constraints and switching evaluation logic. This achieves the technical effect of automatically planning and verifying load migration paths during maintenance and transfer, fault switching, and capacity expansion operations, preventing transformers and UPS from operating beyond limits, and ensuring continuous power supply and collaborative capacity management of modular data centers in multiple scenarios.

[0016] 2. By implementing version locking, idempotent control, and evidence chain recording for power distribution configuration, thermal time-domain margin fingerprint, uninterruptible power supply condition vector, candidate switching sequence, and collaborative operation plan, and dynamically adjusting the load migration step size and operation rhythm based on the safety margin monitoring results during plan execution, the system achieves traceability of operation process, iterative optimization of parameters, and significant reduction in the risk of misoperation and switching impact, thus enabling the system to have good engineering feasibility and long-term operational stability. Attached Figure Description

[0017] Figure 1 This is a schematic diagram of the structure of a modular data center dry-type transformer and UPS collaborative system according to the present invention. Detailed Implementation

[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] Example: Figure 1 A schematic diagram of the structure of a modular data center dry-type transformer and UPS collaborative system is provided. The modular data center dry-type transformer and UPS collaborative system includes: S1, Distribution Topology Modeling Module, is used to obtain the wiring relationships between the dry-type transformer uninterruptible power supply and load branches, and generate a distribution topology diagram. Specifically, it is implemented as follows: In a modular data center, the power distribution topology modeling module is primarily used to establish the wiring relationships between dry-type transformers, uninterruptible power supplies (UPS), and load branches, providing a unified, complete, and traceable power distribution topology diagram for subsequent collaborative control. In this embodiment, a modular data center site refers to a power supply and distribution unit consisting of at least one dry-type transformer, at least one set of UPS, and multiple rack load branches, typically located within the same substation or computer room. The power distribution topology diagram is an abstract representation of the actual electrical connections between incoming circuit breakers, the low-voltage busbars of the dry-type transformer, the input and output terminals of the UPS, the outgoing ports of the distribution cabinets, and the rack load branches. Each device or connection point is considered a node, and each circuit breaker, disconnector, or cable connection is considered a connection relationship, reflecting the power supply path between nodes.

[0020] The collaborative control device refers to the control unit deployed at the station control layer or the computer room monitoring layer. It can be an industrial control computer, a programmable controller, or an embedded device with storage and communication capabilities, used to run various functional modules and save configuration information; its non-volatile memory can be a solid-state drive, flash memory, or other storage media with power-loss retention capabilities. The power distribution configuration evidence chain refers to the various versions of the power distribution topology diagram formed at different construction and operation stages and their generation process, including configuration content, version identifier, generation time, and source information, used to reconstruct the original power distribution structure during subsequent audits and problem investigations.

[0021] The power distribution topology modeling module first obtains basic information from electrical construction drawings, cable lists, and switchgear nameplates. The electrical construction drawings determine the high-voltage and low-voltage wiring circuits of each dry-type transformer, the incoming and outgoing ports of the uninterruptible power supply (UPS), and the assigned circuits of the load branches in the cabinets. The cable list determines the start and end points and circuit numbers of each cable segment. The switchgear nameplates confirm the equipment model, voltage level, and rated current. During the information acquisition phase, the module adopts a unified equipment naming rule, assigning unique equipment identifiers to dry-type transformers, UPS, switchgear, and load branches. Preferably, the equipment identifier can be a combination of a region identifier, cabinet number, and circuit sequence number, separated by a connector, ensuring that the same equipment corresponds to the same identifier in different documents.

[0022] The module will extract the equipment name, equipment identifier, voltage level, rated capacity or rated current value and corresponding circuit name from the above data and organize them into intermediate records. Each intermediate record will be appended with a timestamp. The timestamp is preferably the time given by a unified time system, and an allowable deviation range can be set, for example, it can be set to tens of seconds to several minutes, as determined by the engineering design or operation and maintenance standards. When data is imported multiple times or drawings are revised multiple times, the module will record the import batch time as construction stage information to distinguish the configuration source of different stages.

[0023] To ensure the authenticity of the configuration, the power distribution topology modeling module verifies the continuity of intermediate records by combining the results of a power-on test before construction is completed or before phased commissioning. The power-on test refers to closing the corresponding circuit breakers and disconnect switches on-site according to the planned circuits, and confirming whether the circuit is connected by measuring voltage, current and indicator light status under no-load or test load conditions. The module can be set to obtain the closed and energized status of each circuit from the monitoring and control device or intelligent terminal during the test, and compare the actual connected circuits with the circuits marked on the construction drawings one by one. When a circuit marked as "should be connected" on the construction drawings is found but not detected as connected in the test, the circuit is marked as pending confirmation and is not directly written into the configuration record of the current effective version, but only retained in the intermediate status table. It can be imported into the new version after subsequent manual or retest confirmation.

[0024] When the same equipment identifier and circuit number appear multiple times in the imported data, the module prioritizes the latest batch of records based on construction phase information and timestamps, marking older records as historical versions for later traceability. During the record cleaning phase, the module screens records with the same equipment identifier but marked as decommissioned or dismantled based on the markings of equipment decommissioning, removal, or standby disconnection in the as-built documentation or maintenance system, removing them from the current version of the topology and retaining only historical versions. For records with the same name but different equipment identifiers and located in different cabinets, the module considers them as different equipment and retains them separately according to their numbers.

[0025] In timestamp processing, the module uses a pre-configured time deviation range as the basis for judgment. When the difference between the timestamp of a record and the current system time exceeds this range and there is no corresponding construction stage description, the record is classified as abnormal data and will not be included in the generation of the current version by default. If its validity is subsequently confirmed manually, the record can be referenced when importing a new configuration version. For judgments involving physical limits, the module refers to the rated voltage, rated current, and primary system design parameters on the equipment nameplate, and combines them with the safety factor given by the project design to set a reasonable range. Preferably, the physical limit reference value can be set as the rated value multiplied by a safety factor range pre-given by the design unit. Values ​​outside this range are considered obviously unreasonable and are used to help identify abnormal data. After the above verification and screening, the power distribution topology modeling module organizes the electrical connections between the low-voltage side bus of each dry-type transformer and its downstream uninterruptible power supply (UPS) input terminal, as well as the connections between the UPS output terminal and its downstream distribution cabinet and cabinet load branches, into a final configuration record. Each configuration record includes at least the upstream equipment identifier, downstream equipment identifier, connection type, voltage level, and rated current value. The connection type can be set to one of the following limited sets: bus to transformer, transformer to UPS, UPS to distribution cabinet, and distribution cabinet to load branch, which is used to indicate the electrical level to which it belongs.

[0026] All configuration records are stored in a structured manner in the non-volatile memory of the collaborative control device. Preferably, an index can be created based on the device identifier, and each configuration record can be stored as a set of fields to facilitate quick lookup by device identifier and version number. A topology configuration version number is assigned to this set of configuration records as a whole. When a module generates a new version, it can be configured to first organize and verify the configuration records in the internal cache. In a single write operation, the entire set of configuration records and the corresponding version number are written to the non-volatile memory. After the write is completed, the version number is marked as the currently effective version to avoid version inconsistencies caused by partial record updates. At the same time, the version number, generation time, construction stage information, and source data type are saved together to form a power distribution configuration evidence chain. Any subsequent module that references the power distribution topology diagram must specify the version number to ensure consistency with the operating structure at that time.

[0027] Preferably, in a system comprising a dry-type transformer with a rated capacity of several kilovolt-amperes and two uninterruptible power supply (UPS) units, the distribution topology modeling module establishes a one-to-many relationship between the low-voltage busbar of the dry-type transformer and the input terminals of the two UPS units. In the corresponding configuration record, the upstream device identifier is the busbar identifier, and the downstream device identifier is the identifier of the two UPS input terminals. On the UPS output side, a many-to-many relationship is established between each UPS output terminal and several cabinet load branches. In the corresponding configuration record, the upstream device identifier is the UPS output terminal identifier, and the downstream device identifier is the identifier of each cabinet load branch, so as to fully reflect the power supply path under normal operating conditions.

[0028] To facilitate queries by other modules, the power distribution topology modeling module provides a topology configuration reading interface in the collaborative control device. This interface requires the device identifier and topology configuration version number to be included as query conditions when called. When the caller does not include a device identifier or version number, the module returns a parameter missing flag. When the version number carried by the caller does not exist in the local storage or has been marked as obsolete, the module returns a version not found flag. When the device identifier carried by the caller is not found in the specified topology configuration version, the module returns a device not found flag, which is used to prompt the caller to verify the device identifier or change the query conditions.

[0029] The missing parameter, version non-existent, and device non-existent flags mentioned above can be set to predefined status codes or text markers to guide the caller to complete the conditions or switch to an available version. When calling the same version of the topology configuration multiple times, the module directly reads the corresponding configuration record from non-volatile memory based on the device identifier. Multiple reads with identical content are not repeatedly written to the log record, ensuring that the reading behavior of the same version remains semantically idempotent. When the power distribution structure needs to be adjusted, such as adding a load branch or replacing an uninterruptible power supply, the maintenance personnel enter the new connection relationship and generate a new configuration version through the power distribution topology modeling module. The module retains the old version completely and marks the new version as the currently effective version. The power distribution configuration evidence chain allows for the retrospective analysis of the power distribution structure at any time.

[0030] Through the above process, after obtaining the construction drawings, inventory, nameplates, and the results of the first power-on test, engineers in this field do not need to rely on uncertain experience judgments. They only need to follow the order of data entry, on / off verification, record cleaning, and version confirmation to establish a power distribution topology diagram consistent with the actual site in the collaborative control device. This provides clear and stable upstream structural information for subsequent modules such as transformer status modeling, uninterruptible power supply condition acquisition, and capacity envelope construction, thereby ensuring that the implementation of the entire collaborative system has a clear physical basis and reproducibility.

[0031] S2, Transformer State Modeling Module, is used to collect current, voltage, winding temperature, harmonics, and ambient temperature. Based on a pre-set thermal model and short-time overload curve, it calculates the thermal time-domain margin fingerprint. Specifically, it is implemented as follows: The transformer condition modeling module operates under unified time synchronization conditions. It continuously assesses the thermal state and allowable capacity of each dry-type transformer in the modular data center, generating thermal time-domain margin fingerprints with a fixed rhythm. This provides directly usable constraint information for the capacity envelope construction module and the collaborative execution correction module. Unified time synchronization conditions refer to the station control layer or collaborative control device broadcasting a time reference to each acquisition unit through the time synchronization system. This ensures that time labels from different measurement points remain consistent within the allowable error range, facilitating subsequent alignment of measured values ​​along the time axis.

[0032] Three-phase current refers to the instantaneous or short-time average current measurement of each phase winding on the low-voltage side of a dry-type transformer. Bus voltage refers to the voltage measurement between each phase and the neutral point on the low-voltage side bus. Both are provided by current transformers, voltage transformers, and measurement units within the distribution cabinet. Winding hot spot temperature refers to the representative highest temperature measured by a temperature sensor embedded in the winding or core. Ambient temperature refers to the air temperature measured by a temperature probe installed around the transformer. Harmonic distortion rate refers to the total distortion degree calculated by the harmonic measurement unit based on the relationship between the amplitudes of each harmonic component and the fundamental component.

[0033] The transformer condition modeling module can be set with a preset observation window to store the measurement sequence of the above physical quantities in a recent period. The length of the observation window is determined by the engineers based on the data update frequency and load fluctuation characteristics. Preferably, it can be set to several minutes. Within this time period, the module aligns the three-phase current value, bus voltage value, winding hot spot temperature, harmonic distortion rate and ambient temperature to a unified time axis according to the time tag.

[0034] During alignment, if some measurement values ​​are missing at a certain moment or the time stamp deviates from the unified time synchronization system by more than the allowable deviation range, the module can be set to use the most recent valid measurement for interpolation, or mark the record corresponding to that moment as a record with low credibility and give it a lower weight in subsequent calculations. For example, the weight of the record in calculating the heat load can be reduced to a certain proportion of the normal record. Alternatively, if a record is judged to have low credibility multiple times in a row, it can be removed from the thermal state estimation.

[0035] Physical limits refer to the reasonable range defined by the rated voltage and rated current indicated on the equipment nameplate, as well as the maximum allowable temperature rise and safety factor given by the design unit. After alignment, the module checks the three-phase current, bus voltage, winding hot spot temperature and harmonic distortion rate at each time point. When a certain measured value significantly exceeds the upper limit determined by the combination of the rated value and the safety factor or is significantly lower than the lower limit close to zero, the module regards the measured point as an abnormal measured point and participates in subsequent calculations according to the weighting or elimination rules. The weighting rule can be set to reduce the contribution of the measured point to a part of the normal value when calculating the equivalent heat load in this cycle. The elimination rule can be set to use the effective measured value at the adjacent time to replace the measured point, so as to ensure that the thermal state estimation is not affected by isolated abnormal values.

[0036] Thermal model parameters refer to a set of parameters used to describe the temperature change characteristics of the windings and core of a dry-type transformer under different loads and ambient temperatures. These include the thermal inertia time constant, thermal resistance coefficient, and coefficients relating temperature rise to load. Short-time overload curve parameters refer to a set of time-to-allowable overload ratio correspondences obtained by discretizing the short-time overload capacity curve provided by the manufacturer. Both parameters are entered into the collaborative control device by the design or commissioning personnel before on-site commissioning and managed through version numbers. Within each rhythm cycle, the transformer state modeling module selects data from several recent time points from the observation window. Preferably, it can be set to calculate the equivalent thermal load within that time period in chronological order, converting the three-phase current and bus voltage at each time point into apparent power and active power. Then, it estimates the additional losses based on the harmonic distortion rate, converts the additional losses into an equivalent load ratio, and uses this, along with the current load level, as the thermal load input.

[0037] The module can update the current thermal accumulation state using a discrete accumulation method. That is, it uses the thermal state calculated from the previous cycle as a starting point, and updates the estimated winding hot spot temperature based on the equivalent heat load and ambient temperature changes observed in the current cycle. This ensures that the estimated hot spot temperature trend is consistent with the hot spot temperature measured by the sensor, and fine-tunes the thermal model parameters when the temperature exceeds a preset deviation range. In this way, without providing specific formulas, those skilled in the art can apply the thermal model parameters and short-time overload curve parameters to the field based on commonly used electrothermal equivalent models. After the current thermal accumulation state is determined, the transformer state modeling module calculates the allowable additional apparent power limit over several preset time spans. These time spans can include short, medium, and long time spans, such as one minute, five minutes, and fifteen minutes.

[0038] During the calculation process, the module starts with the currently estimated winding hot spot temperature and ambient temperature, and uses different additional apparent power assumptions as trial values. It simulates the temperature rise within the corresponding time span in the thermal model, while also referencing the allowable overload ratio for that time span in the short-time overload curve parameters. When a certain trial additional apparent power causes the temperature to rise close to the allowable temperature rise limit while the overload ratio does not exceed the specified range, the module uses the additional apparent power near that trial value as the upper limit of the additional apparent power for that time span. To account for additional losses caused by harmonics, the module can set a correction coefficient related to the harmonic distortion rate to reduce the upper limit of apparent power within a certain proportion. The correction coefficient can change monotonically decreasing with the total harmonic distortion rate. For example, a higher coefficient is maintained when the total harmonic distortion rate is in a low range, and the coefficient is reduced to a smaller proportion when the total harmonic distortion rate is close to the preset upper limit. The specific relationship can be achieved using a piecewise linear relationship or a lookup table method, configured by engineers according to on-site standards, so that the higher the harmonic level, the lower the upper limit of the additional apparent power.

[0039] After obtaining the upper limits of the additional apparent power for each time span, the transformer state modeling module combines these upper limits in chronological order to form a thermal time-domain margin fingerprint. The thermal time-domain margin fingerprint is an ordered set representing the allowable additional apparent power for each time span under the current thermal state and harmonic level. This set includes the transformer identifier, the start and end times of the observation window, the identifiers of each time span, the corresponding upper limit of the additional apparent power, the version number of the thermal model parameters and the version number of the short-time overload curve parameters, as well as the fingerprint generation time and fingerprint version number, used for version locking and evidence chain recording during subsequent calls. The aforementioned thermal time-domain margin fingerprint is written to the state recording area in the collaborative control device. The state recording area can be indexed in non-volatile memory according to the transformer identifier, and a sequence number is generated for each fingerprint record, allowing subsequent modules to read the most recent fingerprint or the fingerprint within a specified time range based on the transformer identifier.

[0040] Preferably, the transformer state modeling module can be set to update the three-phase current value and bus voltage value once per second, update the winding hot spot temperature and ambient temperature once every ten seconds, and maintain the latest data for several rhythm cycles within a five-minute observation window. Within each ten-second rhythm cycle, it completes a thermal accumulation state estimation and thermal time-domain margin fingerprint update. Under this setting, the module can provide an upper limit of additional apparent power at three time spans: one minute, five minutes, and fifteen minutes. The upper limit of additional apparent power at each time span can be set as a certain percentage range of the transformer's rated capacity, determined by the designer based on the equipment's short-term overload capacity and life requirements. For example, when the temperature is low and the load is low, it is allowed that the additional apparent power is close to a certain higher percentage of the rated capacity for a short period of time, while when the temperature is close to the upper limit or the harmonic level is high, the additional apparent power is controlled at a lower percentage.

[0041] In actual operation, when the current and temperature collected over a certain period indicate that the transformer is operating under low load and low ambient temperature for a long time, the module will give an upper limit of the additional apparent power for each time span. The subsequent capacity envelope construction module can arrange short-term load superposition based on this. When it is observed that the three-phase current value is close to the rated current, the winding hot spot temperature is close to the upper limit of the allowable temperature rise, and the harmonic distortion rate is high, the module will obtain a higher temperature estimate after updating the thermal accumulation state. At this time, the upper limit of the additional apparent power over a longer time span will be compressed or even close to zero, so as to indicate that the subsequent module should not superimpose new loads on the transformer.

[0042] Through the continuous process of measurement, time alignment, outlier handling, thermal accumulation estimation, and additional apparent power upper limit calculation described above, engineers in this field can configure thermal model parameters and short-time overload curve parameters according to the model of the dry-type transformer and the acquisition capability of the field equipment in actual engineering projects, and generate thermal time-domain margin fingerprints using the same logic as in this embodiment. This provides a clear physical basis for subsequent capacity envelope construction and load scheduling decisions based on thermal time-domain margin fingerprints, and truly realizes collaborative capacity management constrained by transformer thermal state.

[0043] S3, Uninterruptible Power Supply (UPS) Operating Condition Acquisition Module, is used to acquire UPS output power, power factor, harmonics, operating mode, and redundancy capacity to form an operating condition vector. Specifically, it is implemented as follows: The uninterruptible power supply (UPS) status acquisition module is deployed in the collaborative control device. It continuously acquires the operating status of each UPS within the modular data center under unified timing conditions and constructs a UPS status vector under a fixed rhythm. This provides directly usable operating status information to the capacity envelope construction module and the scene event recognition module. Here, an UPS refers to a power supply device that provides backup power to the rack load. A single UPS unit can be composed of one or more power modules connected in parallel. The UPS status vector refers to a set of state variables obtained for a single UPS within a certain rhythm cycle, including active power, power factor corresponding to apparent power, output harmonic content, operating mode, and redundant capacity that can be carried under a given redundancy level. Active power refers to the effective power actually transmitted by an uninterruptible power supply (UPS) to a load. Apparent power can be calculated by the UPS monitoring unit based on the effective values ​​of the output voltage and current. The power factor is the ratio of active power to apparent power. Output harmonic content refers to the total distortion degree obtained by converting each harmonic component in the output voltage or current of the UPS to the fundamental component.

[0044] Operating mode describes the discrete state of the current working mode of the uninterruptible power supply (UPS). It can include inverter power supply, mains bypass, and energy-saving modes. Inverter power supply means that power is supplied to the load through the inverter while maintaining battery charging. Mains bypass means that the load is directly powered from the mains through a bypass circuit. Energy-saving mode means that some power modules are offline to improve efficiency but still meet load requirements. Redundancy capacity is the range of load that can be increased under the current operating mode and the number of modules online, without reducing the predetermined backup capacity and reliability level. It is usually calculated and provided by the UPS monitoring unit based on rated capacity, current load, and preset redundancy strategy. To construct the UPS operating condition vector, the operating condition acquisition module periodically receives status messages from each UPS monitoring unit via a communication link. The status message includes device identifier, time stamp, active power, apparent power or power factor, output harmonic content, operating mode, redundancy capacity, and an idempotent identifier. The device identifier uniquely identifies a UPS, and the idempotent identifier distinguishes between new and old messages within the same cycle.

[0045] Under unified time synchronization, the collaborative control device broadcasts the time reference to the uninterruptible power supply monitoring unit. When the uninterruptible power supply monitoring unit sends a status message, it adds the current time synchronization time, so that the time tags of different devices can correspond to the same time axis within the allowable error range.

[0046] The operating condition acquisition module can be set to an acquisition rhythm, dividing the status messages from multiple uninterruptible power supply monitoring units into periods according to this rhythm. Within each period, messages whose timestamps fall within the time range of that period and whose difference from the center time of the period does not exceed the allowable deviation are considered as candidate messages for that period. The allowable deviation can be set to several seconds, which is determined by engineers based on network latency and sampling capabilities. When the timestamp of a status message deviates from the center time of the current period by more than the allowable deviation, the module can mark the message as a time-invalid message and not use it as a source of operating condition vectors in this period. During the numerical range check, the module can set a reasonable numerical range in the configuration based on the rated active power and rated apparent power of the uninterruptible power supply. For example, the active power and apparent power should not significantly exceed the rated capacity multiplied by the design safety factor, the power factor should not exceed the closed interval between zero and one, and the redundant capacity should not be negative and should not exceed the rated capacity. Based on this, messages that exceed the reasonable range are marked as numerically abnormal messages and are discarded or downgraded according to rules when constructing the operating condition vector in the current cycle. Downgrading can be manifested as only accepting the operating mode in the message and ignoring its power data, while discarding means that the message does not participate in the generation of any operating condition vector in this cycle.

[0047] Within the same acquisition cycle, for duplicate status messages from the same device identifier and carrying the same idempotency identifier, the module retains only the last arriving message through an idempotency control strategy, treating previous messages with the same idempotency identifier as duplicate messages to avoid the same status being counted multiple times due to network retransmission or re-transmission. The idempotency identifier can be generated by the uninterruptible power supply (UPS) monitoring unit using a combination of time and counters, and remains unique within an observation window. For valid messages that pass the time and value checks, the module summarizes them by device identifier, combining the active power, apparent power (or apparent power derived from the power factor), output harmonic content, current operating mode, and redundancy capacity of each UPS within that cycle into a UPS operating condition vector. The module then appends the device identifier, operating condition vector version number, and generation time to this vector. The operating condition vector version number can be generated according to the global sequence of the collaborative control device, used to lock the corresponding version when the module is called in subsequent sessions.

[0048] When no valid status message from a specified device identifier that passes time and value checks is received within a certain acquisition cycle, the operating condition acquisition module can be configured not to generate a new operating condition vector for that device. Instead, it retains the operating condition vector from the previous cycle in the operating condition record area as the most recent valid record, and records that cycle as an acquisition missing event for subsequent analysis of the reliability of the acquisition link. The generated uninterruptible power supply (UPS) operating condition vectors are stored in the operating condition record area of ​​the collaborative control device. The operating condition record area can be located in non-volatile storage media and indexed by device identifier and generation time, allowing the capacity envelope construction module and the scene event identification module to query the most recent one or several operating condition records by device and time range. The operating condition record area can provide services externally through the operating condition query interface, carrying the device identifier and an optional time range when requesting. When the device identifier is not carried, the module returns a parameter missing flag; when the specified device identifier does not exist in the record area, it returns a device non-existent flag; when there are no operating condition vectors within the specified time range, it returns a record non-existent flag. This is used by the caller to distinguish between different situations such as device configuration errors, long-term device inactivity, and short-term acquisition interruption.

[0049] Preferably, in a modular data center containing several uninterruptible power supplies (UPS), the UPS operating condition acquisition module can be configured to acquire active power, power factor, and output harmonic content from each UPS monitoring unit once per second, and operating mode and redundant capacity once every ten seconds. A 60-second operating mode observation window is maintained in the collaborative control device. The number of operating mode changes for each UPS within this observation window is counted. When the number of operating mode changes for a certain UPS within the observation window exceeds the pre-configured upper limit, this situation is recorded as an operating stability alarm, and a stability flag is added to the corresponding operating condition vector to prompt the capacity envelope construction module to impose additional constraints on the operating condition of the UPS when constructing the capacity envelope and evaluating the switching sequence.

[0050] In actual operation scenarios, when an uninterruptible power supply (UPS) is in inverter power supply mode for a long time and its power factor is close to one, its output harmonic content is at a low level, and its redundancy capacity is stable within a predetermined range, the operating condition acquisition module will continuously generate operating condition vectors with stable operating modes and stable power levels. The subsequent capacity envelope construction module can then consider the UPS as suitable for load migration and maintenance switching. When a UPS frequently switches between inverter power supply and mains bypass in a short period of time, or its operating mode frequently switches between energy-saving mode and inverter power supply, and the number of changes in the operating mode observation window exceeds the upper limit, the operating condition acquisition module will mark the UPS with an operating stability alarm and store this mark along with the operating condition vector in the operating condition record area, so that subsequent modules can avoid selecting the UPS to undertake additional loads.

[0051] Through the aforementioned status message reception, time alignment, numerical rationality check, idempotent deduplication, retention of the most recent valid operating condition during the missing report period, operating condition vector construction, operational stability statistics, and interface-based query and error identification mechanisms, engineers in this field can implement an uninterruptible power supply (UPS) operating condition acquisition module in a collaborative control device. This module can stably output a UPS operating condition vector containing active power, power factor, output harmonic content, operating mode, redundancy capacity, and stability markers. It can also provide accurate and traceable UPS operating condition information for the capacity envelope construction module and the scene event identification module, thereby supporting collaborative capacity management and scene scheduling decisions under the conditions of redundancy constraints and operational stability constraints.

[0052] S4, Scene Event Recognition Module, is used to identify maintenance and fault switching / expansion access power switching scenarios based on operation and maintenance requests and protection actions, and to generate load change vectors and switching sequences. Specifically, it is implemented as follows: The scenario event recognition module is deployed in the collaborative control device. Under unified timing conditions, it identifies key behaviors within the modular data center that alter the power supply or load structure. For each key behavior, it generates a load change vector and a set of candidate switching sequences, allowing the capacity envelope construction module and the collaborative execution correction module to verify and schedule them within a clear scenario framework. Scenario events here refer to actions during daily data center operations, such as maintenance switching, uninterruptible power supply (UPS) failure switching, rack or module expansion, and transformer-side power switching, triggered by operator requests or protection device actions. These actions share the common characteristic of changing the load and power supply path of certain UPS or transformers within a short period.

[0053] The load change vector is a set of expected load increases or decreases calculated based on the current power distribution topology and uninterruptible power supply (UPS) operating condition vector for a specific scenario event, according to the load branch dimension. Each record contains at least the load branch identifier, the current load level, and the expected increase or decrease in apparent power for this action, used to describe the change at the load level. The candidate switching sequence is a time sequence list of switch opening / closing operations and UPS operating mode adjustment actions that need to be executed sequentially to achieve a specific scenario event. Each item contains the device identifier, target status, and relative time, used to guide the subsequent collaborative execution correction module to issue control commands in a controlled order.

[0054] The scene event recognition module collects several basic information related to operation from the operator's interface, the operation and maintenance management system, and the protection device: the operator's interface provides manually issued closing commands, opening commands, uninterruptible power supply operation mode adjustment requests, and maintenance application tags; the operation and maintenance management system provides work numbers and planned times related to work tickets, maintenance plans, and capacity expansion plans; and the protection device provides overcurrent, short circuit, and other protection action signals, as well as corresponding equipment identifiers and action times.

[0055] The above information is given a unified time tag under the unified time synchronization condition of the collaborative control device. The unified time synchronization condition means that the collaborative control device periodically broadcasts the time reference to each data source through the time synchronization system. Each data source adds the current time synchronization time when issuing operation records or protection action records, so that the scene event recognition module can arrange the records from different systems into a time sequence on a unified time axis.

[0056] The module can be set up with a scene observation window to aggregate related operations and protection actions that occur within the same time period. The length of the observation window is determined by the operation and maintenance strategy, and can preferably be set to several minutes to cover the operation time required for a maintenance switch, an uninterruptible power supply fault switch, or a capacity expansion access operation. Within this observation window, the module classifies and groups the records according to the operation source, device identification, and operation mode.

[0057] Specifically, when a maintenance request mark is detected in a certain observation window for a specific uninterruptible power supply (UPS) output switch or its downstream distribution cabinet outgoing switch, accompanied by several opening and closing commands for the same distribution section, and the sequence of these commands matches the step sequence in the pre-recorded maintenance and transfer operation procedure, the module merges the relevant operations in that window into a maintenance and transfer scenario event; when a protection action record of a UPS appears within a very short time (such as overcurrent protection action causing inverter shutdown), and immediately followed by an operation record of another UPS being put into inverter mode or switched to bypass mode, the module identifies it... The module identifies uninterruptible power supply (UPS) failure switching scenario events as follows: When a new cabinet or load module access plan is marked as started in the operation and maintenance management system, and a new branch appears in the topology configuration version provided by the distribution topology modeling module, and a closing operation for that branch appears in the operator's operation interface record, the module identifies it as a capacity expansion access scenario event; When the transformer protection device records an upstream power switching signal or the dispatching system issues a power switching command, and the corresponding transformer high-voltage side or low-voltage side circuit breaker performs a specific combination of closing and opening operations within a short period of time, the module identifies it as a transformer-side power switching scenario event.

[0058] To ensure the reproducibility of the judgment process, the module imports the corresponding operating procedures for maintenance and transfer, uninterruptible power supply (UPS) fault switching, capacity expansion, and power switching from the operation and maintenance management system during the initial configuration phase. Each procedure is abstracted into a set of pattern templates consisting of "operation type + device category + sequence relationship." During runtime, the actual operation sequence in the observation window is matched against these pattern templates. When the matching degree exceeds a preset threshold and the corresponding device identifier is consistent, it is determined to be a scenario event of the corresponding type. For confirmed scenario events, the scenario event identification module extracts the UPS, distribution cabinet, and load branch relationships related to this event from the topology configuration records provided by the distribution topology modeling module, based on the currently effective distribution topology configuration version. Combining this with the current operating condition vector provided by the UPS operating condition acquisition module, and under the assumed switch states and operating modes before and after the scenario event, the power supply path on each load branch is recalculated to calculate the load increase or decrease caused by this action, thereby constructing a load change vector.

[0059] During the construction process, the module uses load branches as the granularity to determine the current uninterruptible power supply (UPS) or transformer to which each branch belongs, as well as the UPS or transformer to which it is to belong after the candidate behavior is implemented. It estimates the current apparent power based on the branch's current, voltage, and power factor to obtain a current apparent power value. Then, using the target load ratio given in the migration rules or capacity expansion plan as a reference, it obtains the expected increase or decrease in apparent power for this behavior, which is used as the expected load increase or decrease for that branch in the load change vector. The load change vector obtained in this way reflects whether each branch "migrates out," "migrates in," or remains unchanged under a certain scenario event, and the corresponding capacity change magnitude. After obtaining the load change vector, the scenario event identification module generates a set of candidate switching sequences for each scenario event according to the operating procedures. The candidate switching sequences include at least the switch opening and closing operation sequence and the UPS operating mode adjustment sequence. During generation, the mechanical constraints of the switching mechanism, protection coordination requirements, and the UPS's ability to withstand sudden load fluctuations must be considered.

[0060] The module can be configured to first generate a standard switching sequence based on the operating procedures, and then derive one or two backup switching sequences by adjusting the order of some operations or splitting or merging certain operation steps. For example, in a maintenance and switching scenario, it can generate a switching method of "closing the second uninterruptible power supply output switch first and then disconnecting the first uninterruptible power supply output switch," or it can generate a switching method of "adjusting the load ratio of the two uninterruptible power supplies first and then operating the output switches in sequence," so that the capacity envelope construction module can select the sequence with lower risk under the capacity envelope constraints. Each candidate switching sequence is assigned a unique sequence identifier when it is generated. This identifier can be formed by combining the scenario event identifier and the sequence number, and is used for sequential control and idempotency control during the subsequent collaborative execution correction module execution process, so as to avoid the same sequence being executed multiple times by mistake during communication retries or system switching.

[0061] Preferably, in a typical maintenance and transfer scenario, the operation and maintenance management system has a maintenance plan entry for the "first uninterruptible power supply" (UPS). The plan specifies the maintenance time period and the allowable load relocation ratio. The scenario event recognition module can set the load planned to be relocated from the first UPS to a certain percentage of the UPS's rated capacity, such as a fraction, and combine this with the current active power, apparent power, and redundancy capacity provided by the UPS operating condition acquisition module to calculate the current upper limit of the load that can be relocated from the first UPS. When the operator issues an operation request on the operation interface to switch a batch of cabinet loads from the first UPS to the second UPS, and sequentially issues the second UPS output switch closing command and the first UPS output command in the observation window, the system can handle this scenario. When a switch tripping command is given, the module identifies it as a maintenance and transfer scenario event. In the constructed load change vector, the branches corresponding to these cabinets are marked as moving out of and into the second uninterruptible power supply (UPS) and the corresponding apparent power increase or decrease values ​​are filled in. Regarding the candidate switching sequence, the module can construct one switching method: first close the output switch of the second UPS and then disconnect the output switch of the first UPS. In the sequence, a step is set for the second UPS to adjust the load ratio in advance to buffer the power change at the moment of switching. Another switching method is to first adjust the operating mode to make the two UPSs share the load to a certain target ratio, and then execute the switch opening and closing operations in sequence to reduce the step load borne by one UPS at the moment of switching.

[0062] Through the aforementioned time-series-based record merging, operation procedure-driven pattern matching, load change vector calculation supported by topology and operating condition information, and candidate switching sequence generation process, those skilled in the art can implement a scenario event recognition module in the collaborative control device. This module can automatically identify the impact of key behaviors on each load branch in typical scenarios such as maintenance and switching, uninterruptible power supply fault switching, capacity expansion and access, and power switching. It can also provide clear, orderly, and traceable switching schemes, providing a stable scenario foundation for subsequent collaborative verification and execution based on capacity envelope and thermal time-domain margin fingerprint.

[0063] S5, the capacity envelope construction module, is used to calculate the power-time capacity envelope based on the thermal time-domain margin fingerprint, base load, and operating condition vector, and apply the switching sequence to the distribution topology diagram to pre-simulate the load of each transformer to determine whether the capacity envelope boundary is exceeded. The specific implementation is as follows: The capacity envelope construction module is deployed in the collaborative control device and operates under unified timing conditions. It is used to establish a power-time capacity envelope for each dry-type transformer, taking into account the thermal time-domain margin fingerprint of the dry-type transformer, the current base load level, and the uninterruptible power supply (UPS) operating condition vector. It also performs feasibility assessment and optimization of candidate switching sequences provided by the scenario event recognition module. The power-time capacity envelope describes the capacity constraints of a single dry-type transformer over several preset time spans. Specifically, it refers to the acceptable upper limit curve of additional apparent power for the transformer over different time spans, provided that the winding temperature does not exceed the allowable limit, the harmonic additional loss and phase-to-phase imbalance do not exceed the configuration range, and the predetermined redundancy level is not reduced. The upper limit of additional apparent power refers to the additional apparent power that can be added while keeping the current base load level unchanged.

[0064] The base load level refers to the total apparent power borne by the primary or secondary side of each transformer under the current operating conditions. This is calculated by the module using the distribution topology diagram provided by the distribution topology modeling module and the uninterruptible power supply (UPS) condition vector provided by the UPS condition acquisition module. Specifically, based on the active power, power factor, and corresponding load branch affiliation of each UPS, the apparent power of the downstream branches of each UPS is accumulated along the topology path to the upstream transformer nodes, thus obtaining the current base apparent power of each transformer. The thermal time-domain margin fingerprint is generated by the transformer state modeling module. It represents the upper limit of additional apparent power allowed across multiple time spans under the current thermal accumulation state and harmonic level. This fingerprint includes the time span identifier, the corresponding upper limit of additional apparent power, the transformer identifier, and the version number of the thermal model and short-time overload curve.

[0065] When constructing the power-time capacity envelope, the capacity envelope construction module first reads the most recent thermal time-domain margin fingerprint record of each dry-type transformer from the transformer state modeling module, reads the uninterruptible power supply (UPS) condition vector related to the downstream of the transformer from the UPS condition acquisition module, and then determines the set of UPS and load branches electrically connected to the transformer through the distribution topology diagram. Based on this, the current basic load level of the transformer and the additional apparent power space at different time spans are calculated respectively. To ensure sufficient constraints, when utilizing the additional apparent power limit given in the thermal time-domain margin fingerprint, the module comprehensively considers the redundancy capacity and output harmonic content reflected in the uninterruptible power supply (UPS) operating condition vector. When the UPS redundancy capacity downstream of a transformer is insufficient or the output harmonic content is close to the configuration limit, the module can be set to tighten the additional apparent power limit over the corresponding time span by a certain proportion. This is to avoid an increase in overall operational risk due to insufficient downstream UPS capacity when there is still thermal margin on the transformer side. At the same time, for transformers with obvious load imbalance or a high proportion of single-phase load, the module can refer to the distribution topology and load type information when calculating the additional apparent power limit and appropriately reduce the allowable additional apparent power. This ensures that the power-time-capacity envelope not only reflects thermal constraints but also takes into account harmonic, imbalance, and redundancy constraints.

[0066] The power time capacity envelope can be implemented as a set of records organized according to a preset time span. Each record contains at least a time span identifier, the additional apparent power limit within the corresponding time span, the current basic load level, the thermal time domain margin fingerprint version number, and the capacity envelope model version number. The module writes these records, along with the transformer identifier and the envelope generation time, into the non-volatile storage medium of the collaborative control device. The version locking information in the capacity envelope record indicates the topology version, thermal model version, and uninterruptible power supply condition version on which the envelope is based, so that the constraints at that time can be completely restored during subsequent backtracking and re-evaluation.

[0067] After completing the capacity envelope construction, the capacity envelope construction module evaluates the candidate switching sequences provided by the scenario event identification module for each scenario. Under unified timing conditions, the switching sequence is applied to the power distribution topology and current operating conditions according to the relative times of each operation in the switching sequence, generating the expected load trajectory of each transformer within the entire scenario execution observation window. Specifically, the module first determines, based on the load change vector given by the scenario event identification module, which load branches will be moved out or in by which uninterruptible power supply (UPS) or which branch at which time in the scenario event. Then, according to the order of switch opening and closing operations and UPS operating mode adjustments in the switching sequence, several key moments are divided on the time axis, with each key moment corresponding to a set of switch states and UPS operating mode combinations.

[0068] For each critical moment, the capacity envelope construction module updates the power distribution topology to simulate the state after the switching operation at that moment. Combining the uninterruptible power supply (UPS) condition vector and load change vector, it recalculates the apparent power attribution of each UPS and each load branch. Following the same rules as the calculation of the base load level, it accumulates the branch apparent power to the transformer nodes to obtain the expected apparent power value of each transformer at that moment. If the impact of a switching action on the load between two critical moments can be regarded as a gradual and smooth change, the module can be set to use linear interpolation or conservative estimation based on the larger expected load side between adjacent critical moments to cover the expected load at any moment during the scenario execution.

[0069] Subsequently, within an observation window used to evaluate the scenario, the capacity envelope construction module compares the expected apparent power of each transformer at each time point with the sum of the additional apparent power limit plus the base load at the corresponding time span or the most recent time point in the transformer's power-time capacity envelope, using a preset time span in the thermal time-domain margin fingerprint as a benchmark or a finer time step. When the expected apparent power of a transformer at any time point exceeds the maximum permissible apparent power represented by the capacity envelope limit at that time point, the module immediately marks the current candidate switching sequence as infeasible and records the sequence identifier, the out-of-bounds transformer identifier, and the time of the out-of-bounds occurrence in the planning record area for subsequent analysis and procedure optimization. When the expected apparent power of all transformers within the entire observation window does not exceed the upper limit of their respective power-time capacity envelopes, the candidate switching sequence is marked as feasible.

[0070] For candidate switching sequences marked as feasible, the capacity envelope construction module also forms a comprehensive score based on the degree of heat margin consumption balance and the number of switching operations. The degree of heat margin consumption balance can be measured by statistically analyzing the distribution of the ratio of the expected apparent power of each transformer to its respective capacity envelope upper limit throughout the entire observation window. For example, it can be set as the difference between the maximum and minimum utilization rates of all transformers at all time points or the difference between the average utilization rates of each transformer. The closer the utilization rates are and the smaller the difference is, the more balanced the load distribution among the transformers is and the more balanced the heat margin consumption is. The number of switching operations is obtained by statistically analyzing the number of switching opening and closing actions and uninterruptible power supply operation mode adjustment actions in the switching sequence. The fewer the number of actions, the more beneficial it is to improve system reliability and reduce the risk of misoperation without violating safety constraints.

[0071] The module can be configured with a comprehensive scoring rule, weighting the inverse indicator of heat margin consumption balance with the number of switching operations according to pre-configured weights to obtain a comprehensive score for a candidate switching sequence. A higher score indicates that the sequence is superior in terms of thermal safety and operational economy. For example, the comprehensive score can be calculated as follows: normalize the difference between the maximum and minimum utilization rates of each transformer within the observation window to obtain a balance index; normalize the number of switching operations and the number of operating mode adjustments included in the switching sequence to obtain an operational complexity index; and linearly combine these two indicators according to preset weight coefficients to obtain a comprehensive score. This embodiment does not limit the specific weight values. The feasibility markers, boundary violation information (if any), and comprehensive scores of all candidate switching sequences are written into the planning record area of ​​the collaborative control device. The planning record area is indexed by scenario event identifiers and switching sequence identifiers. When the collaborative execution correction module needs to select an execution plan, it can query the planning record area, read all feasible sequences under a certain scenario, sort them according to the comprehensive score, and select the sequence with the highest score or that meets specific constraints as the actual execution plan.

[0072] The capacity envelope construction module can provide a switching sequence evaluation interface and external query capabilities. When the caller carries a scenario event identifier and a switching sequence identifier in the request, the module retrieves the corresponding record in the planning record area and returns the feasibility mark, comprehensive score, and, if infeasible, the cross-boundary transformer identifier and cross-boundary time. When the call does not carry a switching sequence identifier or a scenario event identifier, the module returns a parameter missing identifier. When the carried scenario event identifier or switching sequence identifier does not exist in the planning record area, a sequence non-existent identifier is returned. When the capacity envelope version or topology configuration version specified by the caller is not in the version set of the current collaborative control device, a version non-existent identifier is returned to help the caller distinguish between different situations such as parameter errors, missing historical records, and version inconsistencies.

[0073] Preferably, in the aforementioned maintenance and switching scenario, for a dry-type transformer with a rated capacity of several kVA, the thermal time-domain margin fingerprint provided by the transformer state modeling module, with the additional apparent power upper limit at three time spans of one minute, five minutes, and fifteen minutes, is respectively within a certain percentage range of the rated capacity. After the capacity envelope construction module constructs the power-time-capacity envelope in conjunction with the current base load level, it calculates two candidate switching sequences. In the first method, which mainly involves "closing the second uninterruptible power supply output switch first and then disconnecting the first uninterruptible power supply output switch," under a certain... At the moment of switching, the expected apparent power of the transformer is close to the value corresponding to the upper limit of the capacity envelope for a short period of time. Although no over-limit occurs, the utilization rate is close to 1, and the heat margin consumption is highly concentrated. In the second method, which is mainly based on "adjusting the load ratio of the two uninterruptible power supplies first and then operating the switches in sequence", the expected apparent power trajectory calculated by the capacity envelope construction module is always lower than the upper limit of the capacity envelope by a certain margin throughout the entire observation window. The utilization rate in each time span is relatively smooth, and the comprehensive score is significantly higher than that of the first method. Therefore, the second candidate switching sequence is marked as feasible and given a higher score.

[0074] In scenarios where time resolution requirements are not high, the capacity envelope construction module can also discretize the capacity envelope using a hierarchical interval approach. This involves dividing the apparent power range corresponding to each time span into several safety levels, such as a safety level, an alarm level, and a prohibition level. The upper limit of the capacity envelope is then mapped to the upper boundary of the safety level or the boundary between the safety level and the alarm level. When generating the operation plan, the collaborative execution correction module only selects the switching sequence that causes the expected apparent power to fall into the safety level. This hierarchical interval approach can be implemented through simple comparison and table lookup, making it relatively easy to implement. It can be considered an equivalent implementation of the power-time capacity envelope, which is beneficial for reproducing the capacity constraint and switching sequence evaluation process in different engineering projects using the same logic. This allows the collaborative capacity management scheme proposed in this invention to be reliably implemented while ensuring thermal safety, redundancy safety, and operational stability.

[0075] S6, the collaborative execution correction module, is used to generate a collaborative operation plan based on feasible switching sequences, issue load migration instructions and switching control instructions, update the hot temporal margin fingerprint during execution, adjust subsequent operation steps and record operational evidence data when the safety margin falls below a preset lower limit. Specifically, it is implemented as follows: The collaborative execution correction module is also deployed in the collaborative control device. After the capacity envelope construction module provides feasible switching sequences and their comprehensive scores, it transforms the collaborative control scheme evaluated by the capacity envelope construction module into a collaborative operation plan that can be gradually executed on-site and dynamically corrected during execution. The collaborative operation plan is a complete action arrangement for a specific scenario event within a certain observation window. It consists of several load migration instructions and several switch control instructions in chronological order. The load migration instructions are load adjustment commands issued to the uninterruptible power supply (UPS) monitoring unit to smoothly transfer a portion of the load from one UPS to another within a specified rhythm and time span, while maintaining the load voltage level and wiring structure unchanged. The switch control instructions are closing or opening commands issued to the switch cabinet or distribution device control unit to change the power supply path at a specified time, so that the distribution topology gradually converges to the target state according to the candidate switching sequence selected by the capacity envelope construction module.

[0076] The collaborative execution correction module reads all candidate switching sequences marked as feasible and their corresponding comprehensive scores from the planning record area for a specific scenario event. Based on a pre-configured selection strategy, it selects one or more sequences as the basis for the collaborative operation plan. The selection strategy can be set to prioritize the sequence with the highest comprehensive score, or it can take into account the number of switching operations, the running time window, and the importance level of upstream services while meeting safety constraints. After selecting a collaborative operation plan, an idempotent key and a sequence number are generated for the plan. The idempotent key is used to uniquely identify this plan instance in the entire system, and the sequence number is used to determine the order of the plans in the execution queue. Both are written into the execution queue along with the plan content. The execution queue can store collaborative operation plans in states such as "pending execution", "in execution" or "completed". The collaborative execution correction module uses a unified time synchronization condition as a time reference, and retrieves the instructions that are about to expire from the execution queue according to the execution time or relative time attached to each instruction in the plan. It then sends control messages to the corresponding uninterruptible power supply monitoring unit and switch control unit through the station control network. The control messages contain the target equipment identifier, target status, expected execution time, and associated plan identifier and idempotent key. After the field equipment completes the action, it sends back the execution result and actual execution time through status feedback messages.

[0077] To prevent the same scenario from being executed multiple times, the collaborative execution correction module checks the idempotency key carried in the request when it receives a new plan distribution request from the upstream system or operation and maintenance management system. When it finds that the idempotency key has been registered as "confirmed plan" in the execution queue or running evidence data, it will not enqueue it again, but will return a duplicate request identifier. The caller can use this to determine that the plan has already been executed or completed and does not need to be distributed again.

[0078] During the execution of the plan, the collaborative execution correction module does not simply assume that the previous capacity envelope assessment results remain valid. Instead, it obtains the latest thermal time-domain margin fingerprints of each dry-type transformer from the transformer condition modeling module at a fixed rhythm, and obtains the latest uninterruptible power supply (UPS) condition vectors and actual load values ​​of key branches from the UPS condition acquisition module and field measurement devices. Under unified time synchronization, these measured values ​​are aligned with the original power-time capacity envelope. For each monitoring moment, the difference or ratio between the upper limit of the capacity envelope and the actual apparent power of each transformer at the current time point is calculated, and this difference or ratio is used as the safety margin at that moment. The safety margin can be defined as the margin value obtained by subtracting the actual apparent power from the upper limit of the capacity envelope, or it can be defined as the reverse indicator of the utilization rate of dividing the actual apparent power by the upper limit of the capacity envelope. The module sets a lower limit of the safety margin for each transformer in the configuration. When the safety margin of any transformer is found to be lower than the lower limit within a certain monitoring cycle, it is considered that the current execution progress is too fast or the load fluctuation exceeds expectations, and subsequent operation steps need to be corrected.

[0079] The correction strategy is configured by engineers before the system is put into operation. It may include reducing the magnitude of subsequent load migration, extending the time interval between the next load migration and the switching operation, temporarily suspending subsequent unnecessary migration steps, and activating backup load paths or backup uninterruptible power supplies in advance. When the collaborative execution correction module detects that the safety margin is lower than the lower limit, it selects one or more strategies in a preset priority order and applies them to the instructions that have not yet been executed. For example, it may prioritize reducing the magnitude of the next load migration by a certain proportion and shift the execution time of the next switching control instruction backward by a few seconds to allow room for recovery of transformer thermal state and uninterruptible power supply conditions. If the safety margin still cannot be restored to above the lower limit within several consecutive monitoring cycles, the module can further execute stronger strategies, such as canceling some subsequent migration steps or immediately switching to the backup collaborative operation plan with the second-best score but lower peak load in the planning record area, while sending alarm information to the operation and maintenance management system.

[0080] Key data throughout the entire plan execution process will be compiled into operational evidence data and written into the evidence chain record area. The operational evidence data includes at least the plan identifier, associated idempotent key, comprehensive score when the plan was selected, start and end times of plan execution, safety margin trajectory of each transformer during execution, thermal time-domain margin fingerprint version number and capacity envelope version number used for evaluation, adjustment strategy type and effective time triggered during execution, issuance and confirmation times of each load migration command and switch control command, and whether the load distribution at the end of execution is consistent with the original plan. Afterwards, maintenance personnel can adjust the lower limit of safety margin, evaluation observation window length, and uninterruptible power supply redundancy capacity configuration based on the operational evidence data to make the parameters closer to field experience.

[0081] The collaborative execution correction module provides a plan distribution interface and a plan status query interface. The plan distribution interface requires the caller to provide a scenario event identifier, a selected switching sequence identifier, and a desired execution time window. If any of the above required parameters are not provided, a parameter missing identifier is returned. If the scenario event identifier or switching sequence identifier specified by the caller does not exist in the planning record area, a plan non-existent identifier is returned. If the current execution queue has reached the pre-configured capacity limit or the number of available execution threads is lower than the safety lower limit, a resource shortage identifier is returned, prompting the caller to extend the retry interval or use a backup plan. The plan status query interface can query the current plan status by plan identifier or idempotent key, including statuses such as "pending execution," "in execution," "completed," and "aborted," as well as the result of the most recent safety margin assessment. If the query parameters are incomplete or the corresponding record cannot be found in both the evidence chain record area and the execution queue, a parameter missing identifier or a plan non-existent identifier is also returned.

[0082] Preferably, in the aforementioned maintenance and transfer scenario, for the collaborative operation plan with the highest comprehensive score after evaluation by the capacity envelope construction module, the collaborative execution correction module can be set to migrate the load from the first uninterruptible power supply to the second uninterruptible power supply in two steps within a time window of about ten seconds, with an interval of several seconds between each migration step. Clear load migration instructions and corresponding switching control instructions are issued for each migration step. Throughout the migration process, the module monitors the safety margin changes of each transformer at a rhythm of one second or several seconds. When the safety margin is consistently higher than the configuration lower limit, and the safety margin curve in the operational evidence data is smooth and does not approach the capacity envelope upper limit for an extended period, the maintenance personnel can adjust the safety margin lower limit appropriately, fine-tune the evaluation observation window length according to the actual switching time, and revise the uninterruptible power supply redundant capacity configuration. This ensures that subsequent similar maintenance and transfer operations shorten execution time and reduce capacity waste caused by conservative limits while guaranteeing safety.

[0083] Based on the above description, those skilled in the art can, under existing station control systems and monitoring network conditions, implement a collaborative execution correction module using conventional queue management, timing control, condition judgment, and record storage methods. This module will form a closed loop with the aforementioned power distribution topology modeling module, transformer status modeling module, uninterruptible power supply (UPS) condition acquisition module, scene event recognition module, and capacity envelope construction module. This will enable the dry-type transformer and UPS of the modular data center to perform pre-planned collaborative operations in actual operation, and to automatically reduce the load migration rate, reduce the load migration amplitude, or switch to a backup collaborative operation scheme when the safety margin is insufficient. This will achieve the collaborative capacity management and operational risk control effects required by this invention.

[0084] In the operating scenario shown in this embodiment: The modular data center is configured with one 1600 kVA dry-type transformer, two 800 kVA uninterruptible power supply (UPS) units, and several rack load branches. The low-voltage busbar of the transformer supplies power to the two UPS units through a distribution cabinet, and each UPS unit powers several rack rows. After the project is completed and passes a power-on test, the power distribution topology modeling module, based on the construction drawings, cable list, switchgear nameplates, and power-on test results, forms the currently effective power distribution topology configuration version in the collaborative control device. The one-to-many relationship between the low-voltage busbar of the dry-type transformer and the input terminals of the two UPS units, as well as the many-to-many relationship between the output terminals of the two UPS units and the load branches of each rack, are solidified in the form of configuration records and written into the power distribution configuration evidence chain.

[0085] During normal operation, the transformer condition modeling module collects three-phase current, bus voltage, winding hot spot temperature, ambient temperature, and harmonic distortion rate rhythmically under unified timing conditions. After downweighting or eliminating abnormal measurement points, it estimates the current thermal accumulation state based on pre-entered thermal model parameters and short-time overload curve parameters. It calculates the allowable additional apparent power limit at three time spans: one minute, five minutes, and fifteen minutes, and generates a thermal time-domain margin fingerprint with transformer identification, time span identification, and version number, which is written to the status record area. The uninterruptible power supply (UPS) condition acquisition module receives status messages from two UPS monitoring units at a rhythm of one second. After time alignment, numerical rationality checks, and idempotent deduplication, it constructs a condition vector for each UPS in each acquisition cycle, recording the active power, power factor, output harmonic content, operating mode, redundant capacity, and operating stability markers for that cycle. These are stored in the condition record area according to the device identification and time index, providing continuously updated operating condition information for subsequent capacity calculations.

[0086] On the morning of a certain workday, the operation and maintenance management system had already entered the work plan for "maintenance of the first uninterruptible power supply". The plan specified the maintenance time period and the proportion of load that could be moved out of the first uninterruptible power supply, such as not exceeding a certain proportion of its rated capacity. At this time, the two uninterruptible power supplies were roughly carrying similar proportions of the total load. The thermal time-domain margin fingerprint given by the transformer state modeling module showed that there was still a certain amount of additional apparent power space within a short time span.

[0087] Before the maintenance begins, the operator issues an operation request on the operating interface to "migrate from the first uninterruptible power supply to the second uninterruptible power supply" for several rack rows. Around the scheduled maintenance time, the operator issues the closing command for the output switch of the second uninterruptible power supply and the opening command for the output switch of the first uninterruptible power supply in sequence. At the same time, the corresponding maintenance plan is marked as "start execution" in the operation and maintenance management system.

[0088] The scene event recognition module receives the above operation requests and switch operation records within the set scene observation window. Combined with the fact that the protection device did not perform any protection action, it matches this series of records with the pre-imported maintenance and transfer operation procedure template. It finds that the similarity between the operation type, equipment category, and sequence relationship and the maintenance and transfer template exceeds the preset threshold, and the equipment identification involved is consistent with the first uninterruptible power supply and the second uninterruptible power supply in the maintenance plan. Therefore, the time period is grouped into a maintenance and transfer scene event.

[0089] The module then calls the currently effective topology configuration version provided by the power distribution topology modeling module to re-organize the power supply path for the relevant cabinet load branches. Combined with the operating condition vector currently provided by the uninterruptible power supply (UPS) operating condition acquisition module, it calculates the apparent power attribution of each branch under the two operating modes of "before migration" and "after migration". It also counts the total apparent power migrated from the first UPS and the total apparent power migrated into the second UPS at the branch granularity. In this way, a load change vector is constructed to clarify the current load level of each relevant load branch and the expected increase or decrease in apparent power during this maintenance and transfer.

[0090] Based on this, the scene event recognition module generates at least two candidate switching sequences according to the maintenance and operation procedures. One sequence adopts the step sequence of "first closing the second uninterruptible power supply output switch and then disconnecting the first uninterruptible power supply output switch". The other sequence adds the action of "adjusting the load ratio of the two uninterruptible power supplies in advance" to make the two uninterruptible power supplies share the load to the target ratio before operating the output switches in sequence. Each of these two sequences is assigned a unique sequence identifier.

[0091] After receiving the load change vector and candidate switching sequence set for the scenario event, the capacity envelope construction module first reads the latest thermal time-domain margin fingerprint of the transformer from the status record area, and reads the uninterruptible power supply (UPS) condition vector related to the downstream of the transformer from the condition record area. Based on the current topology configuration, it calculates the basic load level of the transformer, and then comprehensively considers the redundant capacity and output harmonic content of the downstream UPS. If necessary, it tightens the additional apparent power upper limit given by the thermal time-domain margin fingerprint, constructs the power-time-capacity envelope within the current maintenance window, and writes the topology configuration version, thermal model parameter version, and condition vector version as version locking information into the envelope record.

[0092] Subsequently, the capacity envelope construction module applies the two candidate switching sequences to the power distribution topology diagram. Under unified timing conditions, it adjusts the timing along the time axis according to their respective switching operations and operating modes to simulate the entire maintenance and switching process. For each critical moment, it updates the switch status and uninterruptible power supply operating mode, and recalculates the apparent power attribution of each cabinet branch in combination with the load change vector. The apparent power of the branch is accumulated to the transformer node to obtain the expected apparent power trajectory. Within the predetermined maintenance observation window, the expected apparent power is compared with the sum of the upper limit of the power-time capacity envelope and the basic load at each time point. It is found that the expected apparent power of the first sequence is close to the maximum allowable apparent power represented by the upper limit of the capacity envelope at a certain moment close to the switching, while the expected apparent power of the second sequence is always lower than the upper limit of the capacity envelope by a certain margin throughout the window. Therefore, the first sequence is marked as feasible but with high thermal margin utilization, while the second sequence is marked as feasible and with relatively balanced thermal margin utilization. A comprehensive score is calculated based on the load balancing degree and the number of switching operations. In the end, the second sequence has a higher comprehensive score.

[0093] The capacity envelope construction module writes the feasibility markers, out-of-bounds information (if any), and comprehensive scores of the two candidate switching sequences into the planning record area. Before the maintenance begins, the collaborative execution correction module queries the planning record corresponding to the scenario event, reads the score results of the two feasible sequences, selects the second sequence with the highest comprehensive score according to the pre-configured selection strategy to generate a collaborative operation plan, assigns an idempotent key and sequence number to the plan, writes it into the execution queue, and returns the plan identifier and idempotent key to the operation and maintenance management system in the plan issuance interface response.

[0094] After maintenance is initiated, the collaborative execution correction module, using unified timing conditions as a time reference, retrieves load migration instructions and switch control instructions sequentially from the execution queue according to the planned execution times. It then sends a "increase load ratio" load migration instruction to the second uninterruptible power supply monitoring unit via the station control network, issues closing or opening commands to relevant distribution cabinets and output switches, and updates the planned execution status upon receiving status feedback messages. During planned execution, as the transformer status modeling module and the uninterruptible power supply operating condition acquisition module continuously provide the collaborative execution correction module with the latest thermal time-domain margin fingerprint, operating condition vector, and actual load value, the collaborative execution correction module calculates the safety margin at the current moment under the set monitoring rhythm, and adjusts the actual load value accordingly. Using the difference between the power and capacity envelope upper limit or the utilization rate as the basis for judgment, if the safety margin is found to be close to the configuration lower limit at a certain moment, the module can fine-tune the subsequent steps according to the pre-configured correction strategy. For example, the load amplitude of the second migration action can be slightly reduced and its execution time can be delayed by a few seconds to buffer temporary load fluctuations. When the entire maintenance and transfer process is completed and the safety margin is always higher than the configuration lower limit, the collaborative execution correction module organizes the plan identifier, idempotent key, execution start and end time, safety margin trajectory at each monitoring moment, the thermal time domain margin fingerprint version and capacity envelope version on which it is based, and whether the correction strategy was actually triggered during the process into operational evidence data and writes it into the evidence chain record area.

[0095] During routine analysis after maintenance, operation and maintenance personnel can retrieve operational evidence data for the maintenance and switching scenario through the planned status query interface. They can see the utilization of safety margins and actual switching times for each time span under the current configuration parameters. After comparing this with field experience, they can appropriately adjust the lower limit of safety margin, the length of the evaluation observation window, and the configuration of uninterruptible power supply redundancy capacity. This ensures that subsequent similar maintenance and switching operations do not exceed the safety boundaries of the power-time-capacity envelope, while also reducing capacity waste caused by overly conservative reservations. In this way, collaborative capacity management and operational risk control of dry-type transformers and uninterruptible power supplies can be achieved in a real engineering environment.

[0096] All calculations involved in the embodiments are dimensionless numerical calculations, and the preset parameters and thresholds in the calculations are set by those skilled in the art according to the actual situation.

[0097] It should be noted that this invention can be deployed on the device itself to realize embedded applications, or it can run on a PC or other terminal with a user interface, thereby meeting various hardware environments and usage requirements.

[0098] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, as a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wireless or wired transmission; wired transmission methods include optical fiber, twisted pair, coaxial cable, etc.; wireless transmission includes infrared, microwave, etc. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center containing one or more sets of available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium. A semiconductor medium can be a solid-state drive.

[0099] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and modules described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0100] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple modules or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or modules may be electrical, mechanical, or other forms.

[0101] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical modules; they may be located in one place or distributed across multiple network modules. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0102] In addition, the functional modules in the various embodiments of this application can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module.

[0103] If the aforementioned functions are implemented as software functional modules and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0104] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0105] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A modular data center dry-type transformer and UPS collaborative system, characterized in that, include: S1, Distribution Topology Modeling Module, is used to obtain the wiring relationship between the dry-type transformer uninterruptible power supply and the load branch, and generate a distribution topology diagram; S2, Transformer State Modeling Module, is used to collect current, voltage, winding temperature, harmonics and ambient temperature, and calculate thermal time-domain margin fingerprint based on the preset thermal model and short-time overload curve; S3, Uninterruptible Power Supply Operating Condition Acquisition Module, is used to acquire the output power, power factor, harmonics, operating mode, and redundancy capacity of the uninterruptible power supply to form an operating condition vector. S4, Scene Event Recognition Module, is used to identify maintenance and fault switching, capacity expansion access power switching scenarios based on operation and maintenance requests and protection actions, and generate load change vectors and switching sequences. S5, Capacity Envelope Construction Module, is used to calculate the power-time capacity envelope based on thermal time-domain margin fingerprint, base load and operating condition vector, and apply the switching sequence to the distribution topology diagram to pre-simulate the load of each transformer and determine whether the capacity envelope boundary is exceeded. S6, the collaborative execution correction module, is used to generate a collaborative operation plan based on feasible switching sequences, issue load migration instructions and switch control instructions, update the hot time domain margin fingerprint during execution, adjust subsequent operation steps and record operation evidence data when the safety margin is lower than the preset lower limit.

2. The modular data center dry-type transformer and UPS collaborative system according to claim 1, characterized in that, S1 includes: Extract equipment names, equipment identifiers, voltage levels, and circuit names from electrical construction drawings, cable lists, and switchgear nameplates to form timestamped intermediate records; Based on the results of a power supply test, the intermediate records are checked for continuity and disconnection, and the circuits that fail the continuity and disconnection check are marked as pending confirmation. After removing equipment records marked as retired, a configuration record is generated containing upstream equipment identifier, downstream equipment identifier, connection type, voltage level, and rated current value. The configuration record is then written to non-volatile memory and assigned a topology configuration version number to form a power distribution configuration evidence chain.

3. A modular data center dry-type transformer and UPS collaborative system according to claim 1, characterized in that, S2 include: Under unified time synchronization, the transformer condition modeling module collects three-phase current values, bus voltage values, winding hot spot temperatures, harmonic distortion rates, and ambient temperatures according to a preset observation window. Abnormal measurement points are identified based on physical limits. Missing measurement records and measurement records with time deviations are weighted down and then aligned to a unified time axis.

4. A modular data center dry-type transformer and UPS collaborative system according to claim 3, characterized in that: The transformer condition modeling module converts the aligned three-phase current value and bus voltage value into apparent power and active power in each rhythm cycle, and combines the harmonic distortion rate and ambient temperature to form an equivalent heat load. The additional apparent power upper limit of the preset time span set is calculated based on the thermal model parameters and short-time overload curve parameters, and the corresponding thermal time-domain margin fingerprint is generated.

5. A modular data center dry-type transformer and UPS collaborative system according to claim 1, characterized in that, S3 include: The uninterruptible power supply (UPS) status acquisition module receives status messages containing device identifiers, time stamps, power and operating status under unified time synchronization conditions. Qualified messages are selected based on the acquisition rhythm and allowable deviation. Messages that exceed the numerical range are marked as abnormal messages and are not included in the generation of working condition vectors during the rhythm cycle. After deduplicating duplicate messages according to the idempotency flag within the same rhythm cycle, the uninterruptible power supply (UPS) condition vector is generated by summarizing the data according to the device identifier and written into the condition record area for use by the capacity envelope construction module and the scene event recognition module.

6. A modular data center dry-type transformer and UPS collaborative system according to claim 1, characterized in that, S4 includes: Under unified time synchronization, the scene event recognition module obtains operation records and protection records with equipment identification and time tags from the duty officer's operation interface, the operation and maintenance management system and protection devices. Within the scene observation window, records are matched according to the operation type, equipment category, and sequence relationship in the operating procedure to identify scene events that change the power supply structure and load structure; Based on the power distribution topology diagram and the uninterruptible power supply condition vector, a load change vector containing load branch identifiers and apparent power increase / decrease amounts, as well as a candidate switching sequence with sequence identifiers, are generated.

7. A modular data center dry-type transformer and UPS collaborative system according to claim 1, characterized in that, S5 include: The capacity envelope construction module reads the thermal time-domain margin fingerprint from the transformer condition modeling module, reads the uninterruptible power supply condition vector from the uninterruptible power supply condition acquisition module, and calculates the basic load level of each dry-type transformer in combination with the distribution topology diagram. A power time capacity envelope is generated over a preset time span. The time span identifier, additional apparent power limit, base load level, thermal time domain margin fingerprint version information, and capacity envelope model version information are associated with the transformer identifier and stored in the capacity envelope record area of ​​the collaborative control device.

8. A modular data center dry-type transformer and UPS collaborative system according to claim 7, characterized in that: The capacity envelope construction module, based on the power-time capacity envelope, calculates the expected apparent power of each dry-type transformer at each time point within the observation window for the candidate switching sequence given by the scene event recognition module. It is compared with the maximum allowable apparent power limit at the corresponding time point in the power time capacity envelope. When the expected apparent power at a time point exceeds the limit, the candidate switching sequence is marked as infeasible in the planning record area. Candidate switching sequences are marked as feasible when the expected power does not exceed the upper limit at all time points, and a comprehensive score is formed based on the heat margin consumption balance and the number of switching operations, and written into the planning record area.

9. A modular data center dry-type transformer and UPS collaborative system according to claim 1, characterized in that, S6 include: The collaborative execution correction module generates a collaborative operation plan in the collaborative control device based on the feasible switching sequence and comprehensive score given by the capacity envelope construction module; Write the load migration command and switch control command, along with the plan identifier and idempotency key, into the execution queue; Under unified time synchronization, control messages are sent and status feedback messages are received. Calculate the safety margin by reading the thermal time-domain margin fingerprint, the uninterruptible power supply condition vector, and the actual load value; When the safety margin is lower than the preset lower limit, the unexecuted instructions are adjusted according to the preset correction strategy and the running evidence data is written to the evidence chain record area.

Images