A method and system for multi-source collaborative mode switching and hierarchical takeover of an automatic generation control system

Abstract

The application discloses a kind of automatic generation control system multi-power source collaborative mode switching hierarchical takeover method and system, and main station will be the capacity of different types of power supply, climbing ability and system sensitivity converted into unified risk input, combined with the deviation of actual output and takeover reference target Non-dimensional takeover risk is calculated, and power supply is divided into direct takeover, soft takeover and delayed takeover grade;Again according to frequency, key section power flow, single batch capacity and parallel number constraint screening current batch, and hierarchical issue takeover instruction;During takeover process, according to actual output, execution receipt and system measurement, recalculate risk, adjust grade and batch.The application realizes the quantization grading and safe batch takeover of multiple types of power supply recovery remote AGC control.

Description

Technical Field

[0001] This invention relates to the field of power system dispatching technology, and in particular to a hierarchical takeover method and system for switching multi-source cooperative modes in an automatic generation control system. Background Technology

[0002] Automatic generation control systems (AGCs) are used to automatically generate active power output targets for each controlled power source based on information such as frequency deviation, tie-line power deviation, and load forecasting during power system operation. These targets are then distributed to each AGC substation and substation control system for execution, thereby achieving system frequency regulation and optimized active power allocation. Under normal operating conditions, each controlled power source remains in remote AGC control mode. The AGC master station continuously calculates and distributes targets, and each controlled power source adjusts gradually according to these targets, keeping system frequency and power flow within safe limits.

[0003] In actual operation and maintenance, due to reasons such as planned maintenance, fault handling, protection actions, field testing, inverter self-protection, and energy storage status management, some controlled power sources need to be removed from remote automatic generation control and switched to local control mode. After maintenance or fault handling is completed, operators usually hope to reintegrate these power sources into remote automatic generation control in a short period of time to restore the overall regulation and backup capabilities of the system.

[0004] In existing control methods, automatic generation control master stations often employ a simple strategy: at a certain moment, all controlled power sources that have switched back to remote mode are included in the automatic generation control target optimization process. This approach fails to uniformly assess and hierarchically manage the magnitude of deviations, ramp-up capabilities, and impacts on system frequency and power flow at critical sections for each power source. If multiple controlled power sources simultaneously exhibit significant power deviations and are concentrated in automatic generation control within a short period, the targets allocated to these power sources by the automatic generation control master station will be significantly adjusted within a short timeframe, resulting in substantial changes in the output of each power source within the same time window.

[0005] This "centralized takeover" may cause the following problems: the system frequency may shift significantly due to rapid changes in the active power balance in a short period of time; the power flow of key transmission sections may change significantly in a short period of time, causing the power flow of the section to approach or exceed the safety limit; individual power sources may trigger protection actions due to exceeding their allowable ramping capacity or exceeding operating constraints, resulting in the disconnection of the unit or station from the grid.

[0006] Currently, most solutions for AGC systems in mode switching scenarios focus on single-unit constraint handling, coordination between different frequency modulation links, smooth mode switching at the equipment level, or substation-side control integration. However, when multiple heterogeneous power supplies simultaneously restore remote AGC control, there is still a lack of a dedicated decision-making framework for how the master station can uniformly assess takeover risks, how to arrange the batch access sequence under system-level constraints, and how to continuously monitor and dynamically adjust during the takeover process.

[0007] Especially in scenarios where a high proportion of new energy and energy storage participate in regulation, the capacity, ramp-up capability, operational constraints, and impact on frequency and power flow of different power sources vary significantly. If the master station takes over centrally based solely on a single deviation or simple timing rules, it is easy for large system disturbances to accumulate in a short period of time, and it is also difficult to simultaneously meet the needs of rapid recovery of low-risk power sources and delayed processing of high-risk power sources.

[0008] Chinese patent document CN118674282A discloses a multi-dimensional comprehensive assessment system for power dispatch safety risks. This system assesses and warns of dispatch safety risks based on dimensions such as equipment maintenance convenience, backup power emergency capabilities, and shielded cable protection status. The evaluation objects and outputs of this scheme are used for equipment-level safety risk identification. However, it does not provide a mechanism for generating power-level risk values ​​based on power deviation, ramp-up time ratio, and system sensitivity when multiple heterogeneous power sources resume remote AGC control, and for arranging phased takeover accordingly.

[0009] Chinese patent document CN121395314A discloses a smart energy microgrid stabilization system and method for distributed power source collaborative control. It determines virtual inertia and damping parameters through multi-objective optimization and reinforcement learning, and switches the microgrid execution mode based on current saturation risk. This scheme focuses on the coordination of virtual inertia and damping of distributed power sources within the microgrid, but does not provide a decision mechanism for the recovery sequence and phased takeover of multiple power sources under the constraints of frequency and critical sections, oriented towards the grid AGC master station.

[0010] Chinese patent document CN119886802A discloses a method and system for calculating the coordinated risk assessment of a primary and distribution network considering uncertainties. It assesses the power difference, power flow constraints, and uncertainties between the primary network dispatching command and the distribution network response, and outputs dispatching optimization suggestions. However, this scheme does not provide the three-level states (direct takeover, soft takeover, and delayed takeover) when a controlled power source reverts from local mode to remote AGC mode, nor does it address the batch execution and dynamic reassessment links driven by system constraints.

[0011] To address the shortcomings of the existing technologies, providing a hierarchical management method and system for multi-power source coordinated mode switching in an automatic power generation control system is a worthwhile research topic. Summary of the Invention

[0012] This invention aims to solve the problems of difficulty in applying the same risk caliber to the type-specific operating boundaries, the inability of risk levels to directly represent system safety batches, and the failure of execution feedback to update subsequent takeover when restoring remote AGC control for multiple types of controlled power sources. It provides a hierarchical takeover method and system with parameter conversion, unified risk, three-level takeover, safety batches, and dynamic reclassification as the main chain. Risk calculation and instruction generation share the same takeover reference target. The specific formation method of the takeover reference target is only an optional implementation method and does not constitute the protection center of this case.

[0013] The objective of this invention is achieved through the following technical solution: A hierarchical takeover method for multi-power source cooperative mode switching in an automatic power generation control system includes the following steps: Step 1: Obtain the status data, capacity data, and system safety data of each controlled power source. The status data shall include at least the actual active power output and control mode. The capacity data shall include at least the adjustable power, allowable ramp rate, reference takeover duration, and type-specific operating constraints. The system safety data shall include at least the frequency, critical section power flow, and the sensitivity of each controlled power source to the frequency and critical section power flow. Step 2: Convert the type-specific operating constraints of each controlled power source into equivalent rated adjustable power and equivalent allowable ramp rate with a unified caliber; normalize the power deviation of the actual active power output relative to the control reference target used in this round of control evaluation according to the equivalent rated adjustable power; normalize the deviation elimination time obtained according to the equivalent allowable ramp rate according to the reference control duration; normalize the sensitivity of each controlled power source to frequency and critical section power flow according to the corresponding safety margin; and obtain the unified control risk value of each controlled power source based on the obtained dimensionless components. Step 3: Based on the unified takeover risk value, classify each controlled power source into direct takeover level, soft takeover level, or delayed takeover level, and generate the current takeover batch from the controlled power sources of the direct takeover level and soft takeover level under the constraints of frequency, power flow at critical sections, single batch takeover capacity, and number of parallel takeovers. Step 4: Perform direct takeover only on the directly managed power supply entering the current takeover batch, perform rate-limited takeover on the soft managed power supply entering the current takeover batch, and do not send new takeover commands to the managed power supply with a delayed takeover level. Step 5: Receive the actual active power output, control mode, execution feedback, frequency, and key section measurements of each controlled power source, and recalculate the unified takeover risk value, adjust the takeover level, and update subsequent takeover batches based on the updated reliable data.

[0014] The takeover risk value can be calculated using the following formula: , In the formula Dimensionless The unit is MW; The units are MW, MW / min, and min, respectively; Formula (3) is a simplified notation that omits equivalent parameters. , and Determined and frozen as a constant at the beginning of the same takeover evaluation cycle; and All first As a reference power cancellation sensitivity coefficient The dimensions are then normalized and limited to [0,1] according to the frequency change rate limit and the available margin of the key section, respectively. Each weight is dimensionless, positive, and sums to 1.

[0015] and The tuning is based on historical takeover samples with labels of actual results. Direct takeover samples refer to events in which the frequency change rate, frequency deviation, key section power flow, and target tracking error all meet the limits after direct takeover. Soft takeover samples refer to events in which direct takeover does not meet the limits but speed-limiting ramp takeover does. Delayed takeover samples refer to events in which the speed-limiting ramp still does not meet the limits or is not completed within the maximum takeover time window.

[0016] The samples were divided into a tuning set and an independent backtesting set based on time. Under the given conditions, the threshold that minimizes the three-level weighted misclassification rate is selected; the statistical threshold is used only when the backtest accuracy and its lower confidence interval reach the threshold, otherwise the stability threshold obtained from transient simulation is used.

[0017] Physical security verification of candidate batches at each assessment starting point Re-establish the baseline: and order When not yet initiated, the planned takeover command trajectory will be used; when a soft takeover has been initiated and a rolling reassessment is conducted, the subsequent trajectory will be used. Trajectory; power generation surplus is positive and power generation deficit is negative, and each reassessment uses the current credible frequency as the new initial value for integration.

[0018] The starting point for candidate batch evaluation is denoted as . ,in The continuous time period is calculated from the start point of the assessment, and the unit is... , for The evaluation time window length is calculated and is greater than 0; the measured frequency deviation at the evaluation starting point is used as the evaluation time window length. As an initial value, Internal Press Predict the frequency response; A positive power generation surplus and a negative power generation deficit are defined as power generation surpluses. The compensation amount has the same sign and is subtracted from the equation; the predicted rate of change of frequency, frequency deviation and power flow change at key sections shall not exceed the corresponding limits.

[0019] The takeover plan includes power identification, level, start-up time, takeover target, ramp coefficient increment, maximum takeover window, and abnormal rollback conditions; direct takeover uses the current AGC target after the takeover reference target and the current AGC target enter a synchronization dead zone; soft takeover is based on... Generate instructions The speed limit increases from 0 to 1, and Delay takeover and do not send new targets.

[0020] Type-specific operating constraints are converted into equivalent parameters consistent with the risk formula.

[0021] For thermal power units, the current direction feedback from the unit control system can be taken as the smaller value between the upper limit of the ramp and the nominal ramp rate; for hydropower units, the equivalent allowable ramp rate is tightened according to the water level safety margin.

[0022] The energy storage system reads the available power and ramp-up limit of the battery management system (BMS) in the corresponding direction based on the state of charge (SOC) safety margin and the planned command direction, and jointly corrects the equivalent capacity and equivalent allowable ramp-up rate; when crossing zero power, the charging and discharging directions are checked in segments; wind farms and photovoltaic power plants determine the equivalent capacity based on the real-time power generation, power rationing command and available equipment capacity, and when the available capacity is 0, it is directly postponed without calculating the risk division.

[0023] During the takeover process, the risk value is periodically recalculated; when Dynamic reclassification is triggered when the risk transition threshold is reached, any physical safety constraint is not met, or the actual output or mode data exceeds the allowable delay or the quality flag is invalid.

[0024] Upon triggering, freeze commands that have not yet been initiated, reduce the ramp factor increment of initiated soft takeovers, and recalculate risk, level, and batch based on the most recent trusted data; if the data is untrusted, the power supply will maintain the most recent trusted command or enter the safe target by rolling back the ramp.

[0025] Upon completion of the takeover, the control mode, ramp progress, target tracking, and system safety status are verified; all target power supplies are in remote AGC mode; the ramp coefficient of all soft-takeover power supplies reaches 1, and The difference between the actual output of all target power sources and the current AGC target enters the completion dead zone, and the frequency change rate, frequency deviation, and power flow at critical sections meet the limits.

[0026] The takeover is considered complete when the above conditions are met consecutively within the minimum observation window of the maximum values ​​of the model confirmation time, the actual target tracking stabilization time, the takeover reference target convergence time, the soft takeover slope tail stabilization time, the frequency recovery time, and the cross-sectional recovery time.

[0027] This invention also provides a multi-power source coordinated mode switching hierarchical takeover system for an automatic power generation control system, including an AGC master station and multiple controlled power sources communicatively connected to the AGC master station; the multiple controlled power sources are selected from at least two types of thermal power, hydropower, wind power, photovoltaic power, and energy storage; the AGC master station is equipped with the following six unique functional modules: The data acquisition and takeover reference target module is used to collect the current AGC optimization target, takeover reference target, actual active power output, control mode, type-specific operating constraints, power-by-power reference takeover duration, frequency and critical section sensitivity parameters, risk threshold, system safety parameters, execution receipts and data quality status, and output raw data and configuration parameters to subsequent modules.

[0028] The parameter conversion and sensitivity normalization module is used to generate power deviation. Equivalent rated adjustable power Equivalent permissible climbing rate Per-power reference takeover time and system sensitivity terms Capacity normalization bias Time normalization bias The unified output is generated and generated by this module.

[0029] The risk assessment and classification module includes a risk assessment unit and a classification unit. The risk assessment unit receives the parameter conversion results and normalization deviation, calculates the unified takeover risk value, and outputs the risk value to the classification unit and the physical constraints and batch planning module, respectively. The classification unit outputs the takeover level, soft takeover level, or delayed takeover level, and outputs the takeover level and system sensitivity item to the physical constraints and batch planning module.

[0030] The Physical Constraints and Batch Planning module receives unified takeover risk values, takeover levels, system sensitivity items, and system safety parameters; verifies frequency change rate, frequency deviation, power flow at critical sections, maximum single-batch takeover capacity, and maximum number of parallel takeovers; and outputs the current takeover batch that passes all constraints.

[0031] The takeover instruction generation module is used to receive the current takeover batch, takeover level, current AGC optimization target, and takeover reference target. It issues direct takeover instructions only to power supplies that are simultaneously at the direct takeover level and have entered the current takeover batch, and issues speed-limiting soft takeover instructions with the takeover reference target as the instruction target to soft takeover power supplies within the batch. The controlled power supply executes the corresponding instructions and returns to the execution status.

[0032] The monitoring confirmation and dynamic reclassification module is used to receive actual active power output, control mode, execution receipt, frequency measurement, key section measurement and data quality status, and feed back the updated reliable data to the parameter conversion and unified risk chain, and feed back the constraint status to the batch planning chain. When there is a risk transition, data failure or constraint triggering, the risk is recalculated, the batch is reclassified and re-screened, and the takeover completion confirmation is output when the criteria are met.

[0033] The above six modules transmit information in the following order: data acquisition and takeover reference target, parameter conversion and sensitivity normalization, risk assessment and level classification, physical constraints and batch planning, takeover instruction generation, monitoring confirmation and dynamic reclassification. The monitoring confirmation and dynamic reclassification module returns updated data and constraint status to form a closed loop.

[0034] One of the beneficial effects is that the unique operating constraints of different power supply types are transformed into equivalent capacity and equivalent ramp-up capability through parameter conversion. After normalization, the frequency and critical section sensitivity are included in the same risk caliber as the reference take-over time and power deviation of each power supply, so that risk input can be traced back to the specific operating boundary.

[0035] The second beneficial effect is that the three-level risk classification only forms candidate takeover methods. Only power supplies that pass the constraints of frequency, critical section, single batch capacity and parallel quantity will enter the current batch, avoiding the substitution of risk level for physical safety verification.

[0036] The third beneficial effect is that risk calculation and instruction generation share the same takeover reference target; direct takeover is simultaneously limited by direct level and security batch, while the change in adjacent instructions of soft takeover is limited by equivalent ramp capability, thereby keeping the risk assessment object consistent with the actual execution object.

[0037] The fourth beneficial effect is that the execution receipts, actual output, control mode, frequency and key section measurement return parameters, risks and batch chains can be promptly paused, downgraded and rearranged when risk transitions, data breaches or safety boundary triggers occur.

[0038] The fifth beneficial effect is that the frequency response is initialized from the measured frequency deviation at the evaluation starting point, and the continuous time in seconds, system inertia, primary frequency modulation response and load damping are explicitly adopted, so that the frequency change rate and frequency deviation verification have consistent time and power positive direction caliber.

[0039] Therefore, the present invention focuses on parameter conversion, unified risk, three-level takeover, safety batching, and dynamic reclassification as its main components; the specific formation method of the takeover reference target is only an optional implementation method and does not change or replace the invention center of multi-power source risk classification and physical safety screening.

[0040] The system of the present invention is applicable to at least two types of scenarios in which thermal power, hydropower, wind power, photovoltaic power and energy storage simultaneously recover remote AGC mode from local mode; the same method can also be used to evaluate each controlled power source in different combinations of types. Attached Figure Description

[0041] Figure 1 This is a schematic diagram illustrating the functional relationships between the various types of controlled power supplies, system safety input, parameter conversion and sensitivity normalization, unified risk classification, physical constraints and batch planning, takeover command generation, and monitoring confirmation and dynamic reclassification of the present invention. Figure 2 This diagram illustrates the data acquisition and takeover reference target module, parameter conversion and sensitivity normalization module, risk assessment and level classification module, physical constraint and batch planning module, takeover instruction generation module, monitoring confirmation and dynamic reclassification module, external safety parameters, and input / output signal flow between at least two types of controlled power sources, including thermal power, hydropower, wind power, photovoltaic power, and energy storage. Detailed Implementation

[0042] Embodiments of the present invention are described in detail below, examples of which are illustrated in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain the present invention, and should not be construed as limiting the present invention. Example 1

[0043] like Figures 1 to 2 As shown, this embodiment uses the same takeover reference target to connect risk calculation and takeover instructions, and completes physical constraint verification and dynamic reclassification on the main station side.

[0044] Step 1: Collect the actual active power output of each controlled power source Freeze target when exiting remote AGC Current AGC optimization goals Control mode, rated capacity, permissible ramp rate in the current direction, reference take-off time, type-specific operating constraints, and frequency and critical section safety parameters.

[0045] Define the system has A controlled power supply, numbered as follows: For each controlled power supply Collect the following information: control mode state variables For the first The controlled power supply in the first The control mode status for each sampling period is defined as follows: a value of 1 indicates that the power supply is in remote AGC control mode, and a value of 0 indicates that the power supply is in local control mode. (1) In the formula For power supply number, This refers to the sampling period number. The sampling period is used in the calculation of target migration, ramp-up, and takeover risks. Reference takeover time All takeover time windows are uniformly measured in minutes; continuous time in the frequency state equation. Evaluation time window and Unified with count, In Hz, In MW, In MW / Hz. If the same discrete step enters the frequency integration, then use... Complete min to Explicit conversion, without or Substitute the risk items directly into the minimum value.

[0046] and The values ​​are, in order, the actual active power output, the last reliable target frozen when exiting the remote AGC, the current AGC optimization target, and the takeover reference target, all in MW. The unit is MW. The unit is MW / min. The unit is min; type-specific constraints are used to obtain the same unit. and .

[0047] Step 2: Obtain the takeover reference target by migrating the frozen target to the current AGC target at a limited speed, and calculate the takeover risk based on the difference between the actual output and the takeover reference target.

[0048] The takeover reference targets and takeover deviations are defined as follows: (2) In the formula This marks the start of the target migration cycle for this round. This represents the synchronization difference between the current AGC target and the previous cycle's takeover reference target. This is the maximum allowed synchronous power step size for this cycle; . The unit is MW / min; to simultaneously constrain the migration of the take-off reference target and the slope advancement, a dimensionless budget coefficient is introduced. and demand This recursion ensures that both increasing and decreasing output are guaranteed. ; Update from the previous cycle Continue the recursion.

[0049] The AGC master station calculates the takeover risk value based on the takeover deviation, equivalent capacity, equivalent allowable ramp rate, reference takeover time, and normalized system sensitivity: (3) In the formula and All are dimensionless. The unit is MW; The units are MW, MW / min, and min, respectively; each weight satisfies and Formula (3) is not calculated if any denominator is not greater than 0, and the corresponding power supply is directly delayed.

[0050] The risk formula has the dimension of MW / MW for the first term, MW / [(MW / min)·min] for the second term, and the third term is dimensionless. Therefore, the three items and All are dimensionless; the same method is used within the same evaluation period. And the same set of equivalent parameter calibers.

[0051] definition . ,in The unit is (Hz / s) / MW. The unit is Hz / s; ,in The power flow transfer distribution coefficient is a signed coefficient that can be positive or negative, and its unit is MW / MW. The unit is MW; Reference power Implicit for eliminating unity power sensitivity dimension.

[0052] For thermal power units, The smaller of the nominal allowable ramp rate and the current directional ramp limit fed back by the unit's control system is taken; for hydroelectric units, the equivalent allowable ramp rate is further corrected based on the water level safety margin. (3a) In the formula and These are the current water level, the minimum safe operating water level, and the warning water level, respectively, all using the same water level unit, and must meet the following requirements. To ensure that the denominator is positive; The current available ramp rate for the unit control system, in MW / min; clip will Limited to [0,1].

[0053] (3b) when Tighten the equivalent permissible ramp rate at the same time; Or, if the water level data is unreliable, it will not be calculated. As the risk item in the denominator, the hydropower unit was directly classified as a delayed takeover.

[0054] For energy storage systems, the master station is configured with a SOC safety range and early warning boundary, and meets the following requirements: , The safety margin of energy storage SOC is defined as: (3c) (3d) The four SOC boundaries in the formula and Both are dimensionless, and the strict order ensures that both denominators of formula (3c) are positive; The available ramp limit for the BMS in the planned charge / discharge direction, in MW / min. When Alternatively, if the available BMS value is not greater than 0, the process can be directly postponed. For wind farms and photovoltaic power plants, ,in The equivalent capacity is: (3e) when The time is directly delayed, and the capacity normalization term is not calculated; hour .

[0055] Therefore, the takeover risk value in step two can be written in a unified constraint correction form: Water level, SOC, charging / discharging direction, and available power of new energy sources are only corrected using equations (3a) to (3e). and .

[0056] All types of power supplies use only formula (3). As the sole risk value, no second set of risk criteria will be established.

[0057] In the formula and These are the equivalent rated capacity and equivalent allowable ramp rate after type-specific operating constraints correction, respectively; when no correction is needed for a certain dimension, the corresponding equivalent parameter is equal to the nominal parameter. Formula (3) omits the equivalent parameter for simplified notation. ; and They are determined and frozen as constants at the beginning of the same takeover evaluation cycle, and all weighted terms are dimensionless.

[0058] Step 3: Assign risk value Risk threshold of direct takeover Soft takeover risk threshold In comparison, Under the given conditions, three candidate levels are obtained: direct, soft, or delayed. The candidate level must also be combined with type-specific operational constraints and the batch physical safety verification in step four to determine the actual takeover method.

[0059] The automatic generation control master station determines the takeover risk value of each controlled power source based on... The comparison results with preset thresholds classify the controlled power supply into different management levels: (4) In the formula For takeover level variables: These represent direct takeover, soft takeover, and deferred takeover, respectively. The threshold is satisfied. Furthermore, the results are determined by historical samples or transient simulations with labels of actual results; if the statistical threshold and stability threshold for the same power source yield different results, the higher risk level is adopted.

[0060] Level 3 takeover only provides candidate takeover methods and cannot replace the physical security check in step four. Direct takeover level refers to... The power supply must be sufficient, but it must still meet the requirements at the same time. The current AGC target can only be issued after the difference between the actual output and the current AGC target enters the direct takeover dead zone and all batch constraints are met. The unit is MW. The soft-pipe level uses the same method as the risk calculation. As a ramp command target; the delayed takeover level does not send new targets and awaits reassessment.

[0061] Step 4: Generate a candidate set by sorting by takeover level, risk value, and system sensitivity, and only include power supplies that pass the verification of frequency change rate, frequency deviation, critical section power flow, single batch capacity, and parallel quantity in the current batch.

[0062] For candidate batches At the starting point of each assessment definition The unit is MW, and it is ordered that Candidate batches not yet initiated. The trajectory of the direct or soft takeover command to be issued shall be adopted; when the soft takeover rolling reassessment has been initiated, the subsequent data generated by formulas (9) to (11) shall be adopted. Trajectory. Used for batch planning and rolling verification during the execution period, not related to risk formulas. Mixed use; each rolling reassessment re-establishes a baseline based on current actual output and reliable frequency.

[0063] (5) In the formula For the current candidate batch, by And For The evaluation time window for the design; For time window The auxiliary single-batch capacity boundary, configured by online security analysis, is in MW. Formula (5) limits... The sum of the absolute values ​​of the changes in the candidate plans cannot replace the rate of change of frequency (RoCoF) and frequency deviation integral verification in formula (8).

[0064] The constraint on the number of concurrent takeovers is: (6) In the formula The number of candidates when starting the batch. It is a positive integer and takes the minimum value among the master station's computing power, communication concurrency capability, and physical security verification allowable values.

[0065] The power flow constraints at the critical section are: (7) In the formula For power supply For key sections The signed power flow transfer distribution coefficient can be positive, negative, or zero, and its unit is MW / MW; This represents the signed planned power change. The available margin for bidirectional cross-sections is taken as... And deducting online security reserves, requiring Formula (7) adopts Conduct conservative verification, do not Limited to [0,1].

[0066] The frequency response state equation for the candidate batch is: (8) In the formula, The continuous time period is calculated from the starting point of the candidate batch evaluation, and the unit is... ; For The length of the evaluation time window is greater than 0. The unit is ; For the rated frequency, And to evaluate the deviation of the measured frequency at the starting point. As initial values ​​for the state equations; In MW, The net active power imbalance is represented by a signed value, with a positive power generation surplus and a negative power generation deficit. In addition to the compensation Subtract the equations with the same sign from each other. In MW / Hz. If the input is generated by a control cycle measured in min, then first press... Converted to Integrate again; only if The verification is passed when both the internal frequency change rate and frequency deviation do not exceed the corresponding limits.

[0067] When multiple system safety constraints approach their boundaries simultaneously, the AGC master station handles them protectively according to the following priority: frequency safety constraints have the highest priority because system frequency stability is the primary prerequisite for safe grid operation; power flow constraints at critical sections are secondary, protecting critical transmission channels from overload; total takeover capacity constraints and parallel quantity constraints are tertiary, as their constraint boundaries have more adjustable space. The specific handling principle is as follows: when a high-priority constraint triggers an alarm, the master station immediately suspends the initiation of new takeovers related to that constraint and can upgrade or postpone planned but not yet initiated takeovers; when a low-priority constraint triggers an alarm, the master station mitigates the situation by adjusting the takeover sequence and batch intervals without affecting the approved takeover activities for high-priority constraints. When a constraint is released, the master station resumes the takeover process in descending order of severity, first confirming that frequency and critical section risks have exited the alarm range, and then gradually relaxing capacity and parallel quantity restrictions.

[0068] The alarm threshold configuration for constraint conflict priority handling is as follows: Frequency-type constraints simultaneously monitor both frequency change rate and frequency deviation; if either indicator reaches the warning boundary, it will be handled with the highest priority. Table 1 Constraint Conflict Priority Alarm Threshold

[0069]

[0070] When the frequency change rate or frequency deviation reaches the alarm boundary, the master station suspends all newly added takeovers; when the power flow of a critical section reaches the alarm boundary, the power supply with high sensitivity to that section is delayed; when the capacity or parallel quantity reaches the boundary, the batch is split or the batch interval is extended.

[0071] Step 5: Implement tiered takeover. Direct takeover is only applicable to cases where... and Power supplies that do not exceed the direct takeover dead zone; soft takeover power supplies generate start-up time, target, maximum time window, and ramp factor; delayed takeover power supplies do not send new targets and wait for re-evaluation.

[0072] The tiered takeover decision-making follows the principle of "low risk first, then high risk," handling power supplies that are directly taken over, soft takeover, and delayed takeover in sequence, while adhering to the aforementioned constraint priority rules.

[0073] Directly control the power supply: In and Entering the synchronization dead zone and the actual output is the same as Once the direct takeover dead zone is entered, the power supply will be included in the current AGC target allocation.

[0074] Takeover Start Time Power supply i is now officially included in the sampling period of this batch; When the takeover method is direct takeover, the AGC master station issues the current AGC optimization target and continues to be monitored by frequency and key section constraints.

[0075] Soft-pipe power supply: Generates a speed limiting command from the actual active power output at startup to the piped reference target. (9) In the formula and The units are all in MW; Dimensionless and The initial value is 0. Risk calculation and soft takeover instructions use the same... Target.

[0076] slope coefficient Gradually increase according to the following pattern: (10) In the formula ,definition and satisfy and The main station first obtains the result according to formula (2) each cycle. Then determine the slope increment according to formula (11) and update according to formula (10). Finally, calculate according to formula (9). By identity The target migration term and the slope advancement term are respectively affected by and Due to budget constraints, the combination of the two satisfies... Even with limited precision, the final amplitude limit is still applied.

[0077] maximum increment of slope coefficient in a single step The configuration method is as follows: (11) In the formula The unit is MW / min. In min, The unit is MW. This is a dimensionless safety margin coefficient. This is the dimensionless change factor for budgeting. It is used in formulas (2) and (11). In the cycle Before issuing the command, a conservative upper limit is formed by taking the smaller value of the current trust capability, the prediction capability for the next cycle, and the preset security lower limit, and then... Internal freezing allows both items to use the same single-step budget. If a lower trust capability value is received before issuing the command, the upper limit is lowered before calculating the command. The takeover reference target migration satisfies... Formula (11) will leave the remaining Allocated to ramp advancement. The denominator uses... The lower limit will only tighten the ramp increment, not relax the instruction change constraints; when the target difference is zero, or a zero crossing occurs, or The adjacent elements after synthesis are still considered. The change is subject to a final limit.

[0078] when At this time, it only indicates the end of the soft-connector ramp stage; the power supply i must also meet the current AGC target tracking, mode confirmation, frequency and cross-section safety conditions and continuously reach the minimum observation window before it can switch to steady-state AGC coordinated control.

[0079] Delayed takeover of power supply: No new takeover targets are sent; the takeover reference target, actual output, type-specific constraints, and system safety status are continuously monitored.

[0080] Main site continuous calculation and And verify the timestamps and quality indicators of the feedback data; When the risk is reduced and all safety constraints are restored, the classification is re-evaluated, and the new batch is only entered after the risk is reclassified as direct or soft takeover level. If data is lost or execution receipt times out, maintain the most recently trusted instruction or follow the preset backslope to enter the safe target and issue an alarm.

[0081] Step Six: Closed-loop monitoring of risk transitions, target tracking, frequency change rate, frequency deviation, key section flow, and execution receipts; pause, downgrade, and reorder when trigger conditions are met.

[0082] During the takeover process, the automatic generation control master station dynamically monitors the evolution of power deviations of each controlled power source and system operation indicators, and adaptively adjusts the takeover strategy based on the monitoring results.

[0083] The target tracking and monitoring quantity is defined as: (12) In the formula This represents the tracking error between the actual output and the current AGC target. If the dead zone is not reached within the maximum takeover window... If the power supply fails to converge for several consecutive cycles, it will be downgraded and a new plan will be generated.

[0084] System frequency change rate and frequency deviation monitoring: (13) In the formula The unit is Hz / s. The unit is Hz; and These are the alarm thresholds for the same unit. Protective adjustments are triggered when any measurement reaches or exceeds the corresponding alarm threshold; the system is in an untriggered state only when both measurements are below the alarm threshold.

[0085] Key section power flow monitoring: (14) In the formula and All are measured in MW. For signed cross-sectional tidal currents measured in the specified positive direction, and The alarm is triggered when either side reaches the alarm boundary; the lower limit check must not be omitted when the current is below or in the opposite direction.

[0086] When any of the following abnormal situations are detected, the AGC master station will adaptively adjust the takeover strategy: The deviation of the power supply in a certain pipeline did not converge as expected within the preset time window; The system's measured frequency change rate or frequency deviation reaches or exceeds the corresponding alarm threshold; The power flow at the critical section is approaching the safety boundary.

[0087] The adjustment measures include: The management level of the power supply with abnormal deviation convergence will be upgraded from direct management to soft management, or from soft management to delayed management. Adjusting some power supplies that have not yet been taken over from their current risk level to a higher risk level; Slow down the ramp rate of the soft-switch power supply, i.e., reduce ; Suspend the takeover of the new power supply until system indicators return to a safe range.

[0088] The takeover must be completed simultaneously to meet the following requirements: mode confirmation, soft takeover ramp completion, current AGC target tracking, frequency change rate and frequency deviation safety, and power flow safety at critical sections.

[0089] The following four conditions must be met simultaneously to determine whether takeover is complete: Condition 1: All target power control modes are restored. (15) In the formula This is the target power source set for this takeover process. For power supply Control mode state variables ( This indicates that the system is in remote AGC control mode.

[0090] Condition 2: The current AGC target tracking error of all target power supplies enters the completion dead zone, and the ramp of all soft-switch power supplies is completed. (16) In the formula For the target power source set, For soft-pipe power supply collection; To actually contribute to completing the dead zone, To take over the completion dead zone of the reference target relative to the current AGC target, both are measured in MW. This condition also excludes only... ,but Not yet converged Or misjudgment when the actual output has not yet tracked the target.

[0091] Condition 3: The system frequency change rate and frequency deviation are both within the allowable range. (17) In the formula and These are the predicted frequency change rate and predicted frequency deviation given by formula (8) in the current evaluation window, respectively. and These are actual measurements taken within the same window; and This is an allowable upper limit used for both prediction and actual measurement. If either the prediction or the actual measurement fails, the takeover cannot be considered complete.

[0092] Condition 4: The power flow at the critical section is within a safe range. (18) In the formula and All are measured in MW and use the same signed power flow direction, requiring The upper and lower limits are checked while the judgment criteria are completed.

[0093] To avoid misjudgment caused by boundary fluctuations, the mode, takeover reference target convergence, actual target tracking, frequency and bidirectional power flow conditions corresponding to formulas (15) to (18) should be continuously met within the minimum observation window.

[0094] (19) In the formula and All are expressed in min and are all greater than 0; for Time required for consecutive establishment To cover The minimum positive integer number of sampling periods.

[0095] (20) Only if the above conditions are continuous The AGC master station determines that the batch is complete only when all conditions are met within a complete sampling period; if any condition fails, the observation count is reset to zero and the count restarts from the next complete period.

[0096] and Labels are determined based on the actual takeover results.

[0097] The events in the configuration center are categorized into three types: direct takeover safety, soft takeover safety only, and takeover that should be delayed. Under the given conditions, select the threshold that minimizes the three-level weighted misclassification rate.

[0098] Calculate the classification accuracy and its confidence interval on the independent backtest set; if the threshold is not reached, use the transient simulation stability threshold; if both thresholds exist, use the higher risk level.

[0099] The tiered takeover strategy includes target synchronization, risk prioritization, physical security verification, and phased execution.

[0100] Direct takeover is only executed when the takeover reference target has been migrated to the current AGC target and the difference between the actual output and the current target enters the dead zone.

[0101] The soft connector generates a command with an increasing ramp coefficient and limited adjacent changes from the actual output at startup to the connector reference target, and sends back an execution receipt.

[0102] Delay takeover without sending new targets, and wait for type constraints and system security status to improve before reclassifying.

[0103] Type-specific operational constraints are incorporated into risk calculations through defined equivalent parameters.

[0104] Thermal power plants use the current available ramp rate limit, while hydropower plants adjust the equivalent allowable ramp rate according to the water level margin.

[0105] Energy storage is adjusted for equivalent capacity and ramp rate based on SOC, direction, and available power, while new energy is adjusted for equivalent capacity based on real-time generating power, generation restriction instructions, and available equipment capacity.

[0106] Power supplies that exceed the type-specific safety zone are directly classified as delayed takeovers.

[0107] Risk values ​​are periodically recalculated during the takeover process.

[0108] The triggering conditions are risk transition, failure of security constraints, data exceeding the allowable delay, or invalid quality flag.

[0109] Upon triggering, it freezes unstarted instructions, lowers the ramp of started soft takeovers, and reclassifies and rearranges based on the most recent trusted data.

[0110] The priority of constraint processing is frequency change rate and frequency deviation, power flow at critical sections, single batch capacity, and number of parallel operations.

[0111] When frequency constraints are triggered, all newly added takeovers are paused; when critical section constraints are triggered, the high-sensitivity power supply is delayed.

[0112] When capacity or parallel quantity constraints are triggered, batches are split or batch intervals are extended.

[0113] Upon completion of the takeover, the control mode, target tracking, ramp progress, frequency, and power flow are simultaneously verified.

[0114] All target power supplies are in remote AGC mode, the target error has entered the completion dead zone, and the soft-connector ramp is complete.

[0115] The observation is considered complete only after all conditions are met consecutively within the minimum observation window.

[0116] The hierarchical takeover system includes an AGC master station and multiple controlled power sources that are communicatively connected to the AGC master station. The multiple controlled power sources are selected from at least two types of thermal power, hydropower, wind power, photovoltaic power, and energy storage. The AGC master station is equipped with only six functional modules: data acquisition and takeover reference target module, parameter conversion and sensitivity normalization module, risk assessment and level classification module, physical constraint and batch planning module, takeover instruction generation module, and monitoring confirmation and dynamic reclassification module.

[0117] The data acquisition and takeover reference target module outputs the takeover reference target, actual active power output, control mode, type-specific operating constraints, power-by-power reference takeover duration, sensitivity parameters, risk thresholds, system safety parameters, execution receipts, and data quality status.

[0118] Output of parameter conversion and sensitivity normalization module and Furthermore, the capacity normalization bias and time normalization bias are generated by this module and then output to the risk assessment and classification module.

[0119] The risk assessment and classification module outputs a unified takeover risk value, direct takeover / soft takeover / delayed takeover level, and system sensitivity item to the physical constraints and batch planning module; the physical constraints and batch planning module outputs the current takeover batch that passes the constraints of frequency, cross section, single batch capacity, and parallel quantity.

[0120] The takeover command generation module sends direct takeover commands only to power supplies that are simultaneously at the direct takeover level and have entered the current batch, and sends speed-limiting soft takeover commands to power supplies within the batch that are under soft takeover. The controlled power supply executes the corresponding command and returns the execution status. The monitoring confirmation and dynamic reclassification module feeds back the updated reliable data and constraint status to the parameter conversion, risk assessment and batch planning chain, respectively, to recalculate the risk, reclassify, re-screen the batch and output completion confirmation.

[0121] To make the objectives, technical solutions, and advantages of the present invention clearer, the present invention will be described in further detail below with reference to the accompanying drawings and specific embodiments.

[0122] The following is a summary table of key parameter configurations. Table 2 Initial candidate values ​​for risk assessment weights (must be adjusted using labeled samples)

[0123]

[0124] The values ​​in Tables 2 and 3 are only initial candidates for optimization and cannot be used as operational set values. The actual weights must be tuned using labeled samples and validated on an independent backtesting set; they should be readjusted when operating conditions change.

[0125] Table 3 Initial Candidate Values ​​for Risk Weights of Various Power Source Types

[0126] Note: Candidate values ​​in the table are used to initiate the weighted search; ultimately... It is determined by three features of the same dimensionality, a tuning set labeled with real results, and independent backtesting results, and is not manually adjusted without verification due to changes in power source type or penetration rate.

[0127] Table 4. Threshold Setting Criteria for Admission Control Classification

[0128] and Tuning example: Collect takeover events by time and label the actual results. Those that pass the RoCoF, frequency deviation, critical section power flow, and target error after direct takeover are labeled as direct; those that fail direct takeover but pass the speed-limiting ramp takeover are labeled as soft takeover; those that still fail the speed-limiting ramp takeover or are not completed within the maximum time window are labeled as delayed takeover. Earlier events are used as the tuning set, and later events are used as an independent backtesting set. The system searches for the threshold that minimizes the three-level weighted misclassification rate under certain conditions and reports the backtest accuracy and confidence interval; if the threshold is not reached, the transient simulation stability threshold is used.

[0129] When the inertia of a weak power grid or system decreases, the threshold is not mechanically reduced by a fixed proportion, but rather updated. First, assess the frequency regulation response and load damping, then re-execute the frequency safety simulation within the evaluation time window and verify it. .

[0130] Table 5. Methods for Determining System Security Constraint Parameters

[0131] The frequency constraints of the candidate batch are checked by integral of formula (8). If the primary frequency regulation has not yet responded and the load damping term is zero at the initial moment, then the RoCoF constraint gives... Substitute ,have to , Each Hz is excluded. Then it must be done with... The calculation Internal substitution and And verify ; in min You must first multiply by 60 and then convert. , cannot Alternatively, the frequency deviation can be directly substituted into the power formula.

[0132] Take the minimum allowable value from the real-time computing power, communication channel concurrency capability, and physical security verification results of the AGC master station; its function is to limit the number of simultaneous takeovers, and it does not replace frequency and cross-section verification.

[0133] Table 6. Calculation and Verification Standards for Slope Parameters of Flexible Connectors

[0134] Table 7. Initial candidate weights and thresholds for different scenarios (must be verified according to the workflow below before deployment).

[0135] Table 8. Method for Determining Constraint Conflict Early Warning Boundaries

[0136] Parameter tuning and dynamic update workflow To ensure the feasibility and consistency of tiered take-off under different operating conditions, parameter tuning can be performed according to the following procedure: (1) Initial setting of weight coefficients: Statistics are performed within a typical operating window. First, scale the data according to the dispersion of each dimension, and then select the weights through the tuning set with real labels to avoid simply amplifying the high variance dimension based on the current numerical value.

[0137] (2) Threshold dynamic review: New event labels are added according to a preset cycle and backtesting is performed continuously; when When the primary frequency response, load damping, or critical section margin changes significantly, the stability simulation is re-executed and the threshold is updated.

[0138] (3) Slope parameters and connection window: calculated according to formula (11). and the synthesized Verify the changes between adjacent instructions. A complete recalculation example is provided. The total budget for a single period is The target migration budget is The budget for ramp advancement is Therefore In the worst-case scenario, the incremental changes of the two syntheses shall not exceed [a certain value]. Consistent with the total budget. Maximum takeover window. The time limit is no less than the sum of the target migration time, the soft-connection ramp time, and the target tracking stabilization time; if the task is not completed within the time window, it will be downgraded and rearranged, and will not be forced to complete by relaxing the ramp constraint.

[0139] Exception / Degradation / Recovery Logic when When any frequency or section constraint reaches the alarm boundary, or when the data timestamp exceeds the allowable delay, the quality flag is invalid, or the mode and power measurements are missing, the master station enters the degradation strategy.

[0140] (1) Freeze takeover commands that have not yet been initiated; keep the most recently trusted command for data loss power sources or enter the safe target by pressing the back ramp; (2) Adjust the ramp increment for the soft-connector power supply. ,in ; (3) The level of high-sensitivity power supply is upgraded or delayed, and the evaluation is carried out again after the constraints are restored.

[0141] The recovery condition is: continuous Within a complete sampling period, the frequency change rate, frequency deviation, bidirectional cross-sectional power flow, data quality, and current AGC target tracking error all meet the threshold requirements. and All values ​​are in min and are greater than 0. During recovery, gradually increase the value first. And continue to review adjacent The change in quantity will be recorded, and subsequent batches will resume takeover; the count will be reset to zero if any condition fails.

[0142] Example 2

[0143] This embodiment focuses on a regional power grid containing thermal power units, wind farms, and electrochemical energy storage power stations, illustrating the complete implementation process of the method of the present invention.

[0144] The scenario describes a dispatch area containing multiple thermal power units, wind farms, and electrochemical energy storage power stations, all participating in AGC (Automatic Guided Control) active power regulation. Due to planned maintenance, equipment failure handling, and energy storage status management, some thermal power units have exited AGC and switched to local control mode; some wind farms have exited AGC control due to inverter self-protection; and some energy storage power stations have exited AGC control due to state of charge management. Once these power sources have completed maintenance or returned to normal operation, they need to be reintegrated into the AGC system for unified dispatch.

[0145] Step 1: Data collection from the main site The system includes the mode, rated capacity, available ramp limit in the current direction, type-specific constraints, and system safety parameters, and verifies the timestamp and quality flag.

[0146] Different types of power sources use currently available capacity rather than fixed nominal capacity: thermal power... The smaller value between the nominal ramp rate and the current direction fed back by the unit's control system can be used as the ramp limit; for wind power... Determined by real-time available power, power rationing commands, and available equipment capacity; energy storage and The parameters are determined by the SOC safety margin, charging / discharging direction, and available power. These effective parameters are updated with the operating boundary and enter the same risk and command speed limit chain.

[0147] Sensitivity factor Calculated according to the aforementioned sensitivity definition: frequency sensitivity and critical section sensitivity are normalized according to their respective safety margins and limited to [0,1], and then the larger of the two values ​​is taken to avoid directly participating in the classification with undefined empirical quantities.

[0148] Step 2: Deviation Calculation. The AGC master station calculates the power deviation of each controlled power source according to formula (2). In hybrid power scenarios, power deviations in thermal power units are typically caused by load adjustments or local control operations during maintenance; power deviations in wind farms are caused by wind speed variations and differences in local control strategies; and power deviations in energy storage power stations are caused by state-of-charge management operations.

[0149] Step 3: Main site generates origin Speed ​​limit migration to of , and then Substitute into formula (3) to calculate the risk value; It remains unchanged within this takeover evaluation window.

[0150] In the mixed power supply scenario, the second term of formula (3) is the ratio of deviation elimination time to reference takeover time. This term increases when the ramp rate is low or the reference takeover time is short, and decreases when the ramp rate is high or the allowable takeover time is long, thus reflecting the time margin of different power supplies in a dimensionless manner.

[0151] Furthermore, the AGC master station converts the operating boundaries of different types of power sources into equivalent parameters: thermal power uses the current direction available ramp limit fed back by the unit control system; hydropower corrects ramp according to water level margin; energy storage corrects capacity and ramp according to SOC, charging and discharging direction and available power; wind power and photovoltaic power correct capacity according to real-time available power, power rationing instructions and available equipment capacity.

[0152] Step 4: Takeover Level Classification. The AGC main site is classified according to formula (4); and The system is generated from a tuning set labeled with real results and verified by an independent backtesting set. If the threshold is not reached, a transient simulation stability threshold is used.

[0153] Step 5: System security constraints. The main station takes the net active power imbalance of the candidate batch as input and applies it according to formula (8) in... Internally verify RoCoF and frequency difference, and simultaneously verify critical section power flow, auxiliary batch capacity and parallel quantity.

[0154] For different types of power sources, the main station synchronously reads the current available ramp limit for thermal power, the water level margin for hydropower, the SOC and directional power boundary of energy storage, and the real-time available power and power restriction information of new energy sources; when any type-specific constraint exceeds the limit, even if the basic risk value does not exceed the threshold, the power source will be adjusted to postpone takeover.

[0155] Step Six: Tiered Takeover Decision The AGC master station generates a tiered takeover plan based on the principle of "lower risk first, then higher risk": First batch: Only power sources that meet the direct takeover conditions and pass the verification of formula (8), key sections and parallel constraints are selected; no fixed priority order is preset according to power source type.

[0156] Second batch: Transfer the power supply to the flexible connector according to formula (9) Generate speed limit commands and constrain the ramp coefficient and the change in adjacent commands according to formulas (10) and (11); Continue migrating at the speed limit according to formula (2) to .

[0157] Third batch: The risk level of delayed power takeover has been reduced to [a certain value]. Following this, its takeover level will be reassessed, and it will be incorporated into the takeover plan according to the corresponding strategy.

[0158] Step 7: Monitoring and Completion Judgment. The main station monitors the target error, RoCoF, frequency difference, and cross-sectional power flow according to formulas (12) to (14), and continuously satisfies the corresponding conditions in formulas (15) to (18). The takeover was subsequently completed; Take the maximum value of the mode confirmation, target tracking, soft-connector tail segment, frequency recovery, and cross-section recovery time.

[0159] Example of a verifiable scenario record: Sampling period With frequency security assessment time window Configured separately according to the control cycle and the effective time period of the online model; system inertia, primary frequency regulation response, and All scheduling safety data are taken at the same moment. The table below records the inputs, instructions, predictions, and acknowledgments for each cycle, without pre-filling frequency improvement results that have not been verified by the original curves.

[0160] initial state :the following Both refer to the actual effort put in and the reference targets at the time of takeover. The difference. Thermal power G1 takes Wind power W1, photovoltaic P1 and energy storage S1 The risk calculation for each power source follows formula (3), and a frequency safety check is performed separately using formula (8) before entering the batch.

[0161] Takeover plan generation: First press , The candidates are classified and then rigorously screened based on the net active power imbalance of the candidate batch, the RoCoF and frequency difference results of formula (8), the power flow of the key section, and the number of parallel processes. It is only used as an auxiliary batch capacity boundary and does not replace physical model verification.

[0162] Table 9A Takeover Process Record (Batch, Level, and Instructions)

[0163] Note: Enter the batch and power supply at each time point; recalculate the control level according to formula (3) and formula (4); fill in the key parameters column. and adjacent changes.

[0164]

[0165] Table 9B Handover Process Record (Safety Monitoring and Status) Note: Fill in the frequency prediction / actual measurement, total control power and status judgment at each time point; the safety status must be recalculated according to formula (12) to (20) in combination with the curve / log.

[0166]

[0167] The takeover process analysis was reviewed using event logs: each batch was recorded. The system includes mode feedback, data quality, RoCoF, frequency deviation, and cross-sectional power flow. If any prediction or measured value exceeds the limit, new instructions are frozen and reclassified. The completion time, maximum frequency deviation, and target error of each stage are statistically derived from the original curves and are not replaced by preset values.

[0168] The comparison with the centralized takeover baseline must use the same power grid model, initial output, disturbance, primary frequency regulation parameters and safety limits, and report the original curves, event logs, sample size and confidence intervals; differences can only be reported truthfully when the hierarchical scheme simultaneously meets the safety constraints and completion criteria, without pre-claiming fixed frequency improvement values ​​or completely avoiding disturbances.

[0169] Example 3

[0170] This embodiment focuses on multi-unit thermal power plants to illustrate the application of the method of the present invention in a multi-unit scenario with a single power source.

[0171] Scenario Description: A thermal power plant comprises multiple thermal power units of varying capacities, all participating in AGC (Automatic Guided Control) active power regulation. Due to planned maintenance and equipment testing, several units successively exited AGC control and switched to local control mode. After the maintenance is completed, operators need to reintegrate all units into the AGC system.

[0172] Step 1: Status Acquisition - The AGC master station collects the operating status information of each unit. Rated capacity of thermal power units. Take the rated output from the unit's nameplate; allowable ramp rate. Based on the unit type and current operating conditions, the ramp rate parameters for large-capacity and small-capacity units should be configured separately; sensitivity factor The determination is based on the unit's location in the power grid topology and its impact on critical sections.

[0173] Step Two: Deviation Calculation and Risk Assessment. Calculate the power deviation of each unit according to formula (2), and calculate the deviation according to formula (3). and Calculate the risks of takeover. (Second item) This represents the dimensionless ratio of the deviation elimination time to the reference take-off time; within the same evaluation period. And the equivalent parameter caliber should be kept consistent. For large-capacity units, the first factor may be smaller, but if their normalized system sensitivity is consistent... The risk is relatively high, and the third item will still increase the overall risk.

[0174] Weighting coefficient Configuration can be tailored to the operating characteristics of thermal power plants: when deviation elimination time is the primary concern, the configuration can be increased. The weight can be increased when system power flow security is the primary concern. The weight.

[0175] Step 3: The units are divided into three control levels according to formula (4). Threshold and The settings should be based on the operational safety standards of the dispatching area where the power plant is located. Units with small deviations are classified as directly connected, units with medium deviations are classified as softly connected, and units with large deviations are classified as delayed connected.

[0176] Step 4: Safety Constraints and Takeover Scheme. After determining the system safety constraint parameters, the AGC master station generates a tiered takeover scheme. For thermal power plant scenarios, since all units are thermal power type, the ramp rate is limited, and the ramp coefficient increment of the soft takeover scheme is... When calculating according to formula (11), the regulating inertia of the thermal power unit should be fully considered, and an appropriate soft pipe margin coefficient should be configured. To ensure that the actual adjustment rate does not exceed the unit's allowable ramp rate.

[0177] The tiered takeover scheme is implemented in order of risk value from low to high: first takeover of units with small deviations and low risk, then gradually bring units with larger deviations into AGC control through soft takeover, and finally deal with the units with the largest deviations that are taken over later.

[0178] Step 5: Monitoring and Completion Judgment. The main station monitors target tracking, frequency, bidirectional cross-sectional power flow, and continuous observation window according to formulas (12) to (20); target tracking of thermal power units adopts completion dead zone. The undefined convergence ratio field is no longer used. All units meet the completion conditions and system indicators are continuously satisfied. After that, the takeover was deemed complete.

[0179] Numerical Example: A thermal power plant has 4 generating units, with weighting coefficients set as follows: threshold The results of the risk assessment and classification for each unit's takeover are as follows: In this example, all units are considered... .

[0180] Table 10 Results of Risk Assessment and Classification for Thermal Power Units During Operation

[0181] Formula (3) is recalculated item by item: G1=0.30×15 / 600+0.45×15 / (3×1)+0.25×0.12=2.2875≈2.288; G2=0.30×8 / 300+0.45×8 / (4×1)+0.25×0.10=0.933; G3=0.30×3 / 300+0.45×3 / (4×1)+0.25×0.08=0.3605≈0.361; G4=0.30×1 / 150+0.45×1 / (5×1)+0.25×0.05=0.1045≈0.105. MW, MW / min and min are canceled out according to formula (3), and the result is dimensionless. The takeover order still needs to be checked using formulas (5) to (8) and type constraints.

[0182] Example 4 This embodiment focuses on a regional power grid that includes wind farms and photovoltaic power plants, illustrating the application of the method of the present invention in a power supply scenario dominated by new energy sources.

[0183] Scenario Description: A regional power grid includes multiple wind farms and multiple photovoltaic power stations, all participating in AGC (Active Power Control) regulation. Due to inverter self-protection activation, abnormal grid voltage, and communication failures, some wind farms and photovoltaic power stations have exited AGC control. Once the fault is resolved, these renewable energy sources need to be reintegrated into the AGC system.

[0184] Step 1: Status Acquisition - The AGC master station collects operational status information from each wind farm and photovoltaic power station. This includes the rated capacity of the new energy power source. Take the total installed capacity of the station; allowable ramp rate. Based on the response capability of the power control system of the power plant, the ramp rate of new energy power sources is usually higher than that of conventional thermal power units; sensitivity factor The determination is based on the electrical location of the station in the power grid and its impact on key sections and frequencies.

[0185] For new energy power sources, the actual active power output is... Due to the influence of current resource conditions (wind speed, sunlight), significant changes may occur during the period of exiting AGC control, resulting in power deviation. The direction and magnitude of the output are uncertain. Therefore, the fluctuation characteristics of new energy power output need to be considered when calculating deviations and assessing risks.

[0186] Step 2: Deviation Calculation and Risk Assessment. Calculate the power deviation of each power source according to formula (2). The deviation characteristics of new energy power sources are: the deviation of wind farms is closely related to the wind speed changes during the AGC period, and may have large positive or negative deviations; the deviation of photovoltaic power stations is related to the changes in light intensity, and the deviation is larger when the daytime light intensity changes.

[0187] The takeover risk value is calculated according to formula (3). In the new energy scenario, due to the ramp rate... The risk of takeover is usually high, and the contribution of the second term in formula (3) is relatively small. Therefore, the risk of takeover is mainly determined by the first term (capacity normalization bias) and the third term (system sensitivity). The ramp-up dimension is the dimensionless time percentage.

[0188] The weighting coefficients should reflect the operating characteristics of the renewable energy power grid: when the system inertia is low, the impact of renewable energy power generation on system frequency is more sensitive, and the weighting coefficients should be appropriately increased. The weighting should be appropriately increased when the capacity of new energy power stations is large and the deviation ratio is high. The weight.

[0189] Step 3: Classification and Takeover Scheme. Takeover levels are classified according to formula (4), and takeover schemes are determined based on system safety constraints. In new energy scenarios, due to the fast response speed of new energy power sources, the proportion of power sources directly taken over may be relatively high. However, if the system inertia in the regional power grid is low (high new energy penetration rate), system safety constraints may be more stringent, and the maximum total takeover capacity may be limited. and maximum number of concurrent takeovers It may be necessary to set it to a smaller value to prevent power disturbances caused by the takeover from causing excessive frequency impacts on low-inertia systems.

[0190] For soft-pipe power supplies, the ramp transition process in formulas (9) and (10) can utilize the fast response capability of new energy power supplies to set a larger ramp increment. This shortens the transition time for soft take-off.

[0191] Step 4: Monitoring and Completion Judgment. In addition to verifying target tracking, frequency change rate, frequency deviation, and bidirectional cross-sectional power flow according to formulas (12) to (20) in the new energy scenario, real-time generateable power and system inertia are also updated. Target tracking uses a completion dead zone. All conditions must be met consecutively. The undefined convergence ratio field will no longer be used.

[0192] Numerical Example: A regional power grid includes 3 wind farms and 2 photovoltaic power stations. The weighting coefficients are taken as follows: threshold The risk assessment results for each power source takeover are as follows: In this example, all power sources are considered... .

[0193] Table 11 Risk Assessment Results for Wind Power and Solar Power Supply Connection

[0194] Formula (3) is recalculated item by item: .

[0195] Corrected according to real-time available power, the result is dimensionless; after sorting, it still needs to be processed by formulas (5) to (8).

[0196] Example 5 This embodiment focuses on multi-cluster electrochemical energy storage groups to illustrate the application of the method of the present invention in power supply scenarios where energy storage is the primary function.

[0197] Scenario Description: A regional power grid is configured with an energy storage group consisting of multiple electrochemical energy storage clusters, each participating independently in AGC (Automatic Generation Control) active power regulation. Due to battery management system protection actions, state-of-charge exceeding limits, and balancing maintenance, some energy storage clusters have exited AGC control. After maintenance is completed, these energy storage clusters need to be reintegrated into the AGC system.

[0198] Step 1: Status Acquisition. The main station acquires the rated charge / discharge power of each energy storage cluster, the available power in the planned direction given by the Battery Management System (BMS), the available ramp limit in the corresponding direction, the safe range of state of charge (SOC), charge / discharge efficiency, temperature, and fault status. Based on this, calculations are performed. and If the planned instruction crosses zero power, the charging and discharging sections will be checked separately and the more stringent capability will be adopted.

[0199] The energy storage cluster's output range covers both charging and discharging directions, with power deviation. It may correspond to crossing zero power; its adjustability is limited by the state of charge (SOC), so risk assessment and command verification are both read from the BMS boundary in the planned direction.

[0200] The equivalent permissible ramp rate of the energy storage cluster, adjusted for current directional capability, is as follows: (twenty one) in: For energy storage clusters The equivalent permissible ramp rate in the planned charge / discharge direction, in MW / min; For energy storage clusters The nominal allowable climbing rate, in MW / min; The available ramp rate for the corresponding direction is given by the BMS based on SOC, temperature, individual cell voltage, planned direction, and protection status, in MW / min.

[0201] The BMS outputs the available ramp limit for the corresponding direction based on SOC, temperature, cell voltage, planned charge / discharge direction, and protection status. and available power When crossing zero power, the parameters in both directions are read in segments and a stricter constraint is applied. When either boundary is triggered, the corresponding available value is tightened or set to zero.

[0202] Risk calculation premise:

[0203] The safe upper and lower limits of SOC, the available charge and discharge power, and the allowable rate of SOC change are all read from the current valid settings of the battery management system; the fixed 10% or 90% boundary shall not be used under different battery systems, life states, and temperature conditions.

[0204] The above boundaries are entered into the equivalent capacity and equivalent ramp rate through the available value of BMS; when the range is exceeded or the available value is zero, the master station classifies the energy storage cluster as delayed takeover and performs a safe backoff.

[0205] The equivalent rated capacity of the energy storage cluster is corrected for the available power in the current direction as follows: (twenty two) In the formula The available power of the BMS under the current SOC, temperature, and planned charge / discharge direction, in MW; The maximum available climbing speed in the corresponding direction is expressed in MW / min.

[0206] The takeover risk value for energy storage at its current operational boundary will still use a unified risk caliber: (twenty three) Formula (23) is the equivalent parameter expansion of the unified formula (3) in the energy storage scenario, without applying the SOC penalty to the sensitivity term again; SOC, direction and power boundaries are all obtained through and They are all falling under the same risk caliber. The main station also predicts the evolution of SOC within the takeover window using the following formula: (twenty four) In the formula Indicates discharge, Indicates charging. and All are non-negative power and the unit is MW; These are the charging efficiency and discharging efficiency provided by the BMS, respectively. and In h, In MWh, SOC is dimensionless. The discharge term is divided by... , charging item multiplied by The unit is MWh / MWh. If the predicted SOC exceeds the current safe range of the BMS, or the available power in the direction is insufficient for execution... If so, takeover will be suspended, a rollback will be implemented, and the status will be adjusted to a postponement level.

[0207] Formula (23) Substitute into formula (4) to obtain the candidate level, and continue to execute formula (8), cross section, single batch capacity and parallel quantity verification.

[0208] Step 2: Deviation Calculation and Risk Assessment. Calculate the power deviation of each energy storage cluster according to formula (2). The deviation of the energy storage cluster is usually caused by the charge and discharge status management operation during the AGC exit period. The deviation direction may be positive (actual discharge power is greater than the target) or negative (actual charging power is greater than the target or actual discharge power is less than the target).

[0209] The energy storage cluster calculates the takeover risk according to formula (23); when SOC, temperature or planned charge / discharge direction causes or When tightening, the capacity or time term increases. Only equivalent parameters... Formulas (23) and (24) are calculated only when the efficiency in both directions meets their positive range; when the available value or efficiency is zero, the SOC is out of bounds, or the direction data is unreliable, the division is not calculated and the takeover is directly postponed.

[0210] Step 3: Classification and Takeover Scheme. The takeover level is classified according to formula (4). In the scenario of multi-cluster energy storage groups, most energy storage clusters are usually classified as direct takeover or soft takeover level because of their fast ramp rate and moderate deviation ratio. Only when the SOC of the energy storage cluster is close to the limit, resulting in a significant reduction in the effective ramp rate, or when the deviation ratio is large, can it be classified as delayed takeover level.

[0211] The configuration of system safety constraints must consider the unique characteristics of energy storage clusters: simultaneous takeover of multiple clusters can cause rapid and significant power changes; even if the deviation of each cluster is small, the total effect can still cause substantial disturbances to the system frequency. Therefore, constraints on the number of parallel takeovers are necessary. This is especially important in energy storage scenarios, and should be set reasonably according to the total scale of the energy storage group and the system inertia level.

[0212] The tiered grid connection scheme utilizes the rapid response characteristics of energy storage, allowing direct grid connection of energy storage clusters to be completed in a short time. The ramp transition time of soft-grid energy storage clusters is calculated using formula (11). Due to the high ramp rate of energy storage, with the appropriate margin coefficient... Even after that, the transition time for soft take-off was still relatively short.

[0213] Step Four: Monitoring and Completion Judgment In the takeover monitoring of multi-cluster energy storage groups, in addition to standard deviation convergence monitoring and system indicator monitoring, special attention should be paid to the SOC change trend of each energy storage cluster. If the SOC of a certain energy storage cluster rapidly approaches its upper or lower limit during the takeover process, the AGC master station should promptly adjust the takeover strategy for that cluster or limit its adjustment range to prevent SOC exceeding the limit and triggering protection.

[0214] In addition to meeting the standard conditions of formulas (15) to (18), the completion of the takeover should also verify that the SOC of each energy storage cluster is in the normal operating range.

[0215] The present invention also provides a multi-power collaborative mode switching hierarchical takeover device for an AGC system deployed at the AGC master station, which consists of a data acquisition and takeover reference target module, a parameter conversion and sensitivity normalization module, a risk assessment and level classification module, a physical constraint and batch planning module, a takeover instruction generation module, and a monitoring confirmation and dynamic reclassification module.

[0216] The data acquisition and takeover reference target module collects takeover reference targets, actual output, mode, type constraints, power-by-power reference takeover duration, sensitivity parameters, system safety parameters, and receipts; the parameter conversion and sensitivity normalization module generates five risk inputs, and the normalization deviation is only output by this module; the risk assessment and level classification module outputs the risk value and three levels in sequence; the physical constraints and batch planning module outputs the current batch that has passed the constraints; the takeover command generation module issues direct or rate-limited soft takeover commands to the power sources within the batch; the controlled power source executes the corresponding command and returns the execution status; the monitoring confirmation and dynamic reclassification module returns reliable feedback to parameters, risks, and batch chains, and outputs completion confirmation.

[0217] System signal flow such as Figure 2 As shown: At least two types of controlled power sources, including thermal power, hydropower, wind power, photovoltaic power, and energy storage, send actual output, mode, type constraints, and receipts to the data acquisition and takeover reference target module and the monitoring confirmation and dynamic reclassification module; the six modules transmit information in the order of data acquisition and takeover reference target, parameter conversion and sensitivity normalization, risk assessment and level classification, physical constraints and batch planning, takeover instruction generation, and monitoring confirmation and dynamic reclassification, forming a closed loop.

[0218] Comparative Example 1

[0219] Comparative Test of Verifiable Centralized and Hierarchical Control To verify the effectiveness of the distinguishing features, the same power grid model, the same initial output, the same load disturbance, and the same primary frequency regulation parameters were used to execute centralized takeover baseline and the hierarchical takeover of the present invention, respectively; the original curves and event logs were recorded for each test.

[0220] The baseline scheme incorporates all target power sources into the current AGC target; the scheme of this invention enables target migration, risk classification, physical verification of formula (8) and dynamic reclassification.

[0221] The testing direction is uniformly set as the controlled power supply reverting from local mode to remote AGC mode; it must not be replaced by the reverse scenario of remote to local conversion.

[0222] Table 12 Comparison Test Record of Centralized Management and Hierarchical Management

[0223] This comparison employs a verifiable testing protocol, eliminating the need for pre-filling absolute improvement percentages that cannot be verified by raw curves, event logs, and sample statistics. The evaluation focuses on the combined constraints of target mutation amount, RoCoF, frequency deviation, cross-sectional power flow, target tracking error, and completion time.

[0224] Table 13 Example of a batch plan for multiple power supply types (qualitative structure)

[0225] Comparative Example 2 Table 14A: Characteristic Inputs and Equivalent Parameters for Each Power Supply Type

[0226] To demonstrate the refined and differentiated management of different power supply types in this invention, the following table compares the differences in scheduling parameters for each power supply type on the AGC master station side with the optimization strategy of this invention:

[0227] Table 14B Main Station Processing and Takeover Plan Fields

[0228] Table 14C Completion Conditions and Risk Level Description

[0229] The table above only provides the inputs, equivalent parameters, and completion conditions required for each type to enter the same risk and takeover chain, without presupposing success rates, improvement ratios, or fixed batch capacities that have not been validated by samples or physical models.

[0230] This invention uses the AGC master station as the decision-making core, migrates the frozen target to the current AGC target with a speed limit, and calculates the risk using the same takeover reference target; then, it generates batches through three-level classification and frequency, cross-section, capacity and parallel constraints with evaluation time windows, executes them in direct or soft takeover mode, and dynamically reclassifies them using actual output, mode, receipt and safety measurement.

[0231] This invention focuses on protecting the master station closed loop between the target synchronization risk chain, physical security batch verification, and execution feedback reclassification, without limiting the specific implementation of the substation or field station underlying controller.

[0232] The main site is responsible for generating , Grade, batch and Substations or stations are responsible for execution and feedback. Pattern, type-specific constraints, and execution receipts.

[0233] When a risk transition, data breach, or any physical constraint is triggered, the master station freezes new commands, lowers the soft takeover ramp, or delays the relevant power supply, and rearranges the data based on the most recent trusted data.

[0234] Takeover is only completed when the mode, slope, current target tracking, rate of change of frequency, frequency deviation, and key section power flow are continuously satisfied within the minimum observation window.

[0235] The above is only used to illustrate the technical solution of the present invention and not to limit it. Any other modifications or equivalent substitutions made by those skilled in the art to the technical solution of the present invention, as long as they do not depart from the spirit and scope of the technical solution of the present invention, should be covered within the scope of the claims of the present invention.

Claims

1. A hierarchical takeover method for switching multi-power source cooperative modes in an automatic power generation control system, characterized in that, This is executed by the automatic generation control master station when multiple different types of controlled power sources revert from local control mode to remote automatic generation control mode, including: Step 1: Obtain the status data, capacity data, and system safety data of each controlled power source. The status data shall include at least the actual active power output and control mode. The capacity data shall include at least the adjustable power, allowable ramp rate, reference takeover duration, and type-specific operating constraints. The system safety data shall include at least the frequency, critical section power flow, and the sensitivity of each controlled power source to the frequency and critical section power flow. Step 2: Convert the type-specific operating constraints of each controlled power source into equivalent rated adjustable power and equivalent allowable ramp rate with a unified caliber; normalize the power deviation of the actual active power output relative to the control reference target used in this round of control evaluation according to the equivalent rated adjustable power; normalize the deviation elimination time obtained according to the equivalent allowable ramp rate according to the reference control duration; normalize the sensitivity of each controlled power source to frequency and critical section power flow according to the corresponding safety margin; and obtain the unified control risk value of each controlled power source based on the obtained dimensionless components. Step 3: Based on the unified takeover risk value, classify each controlled power source into direct takeover level, soft takeover level, or delayed takeover level, and generate the current takeover batch from the controlled power sources of the direct takeover level and soft takeover level under the constraints of frequency, power flow at critical sections, single batch takeover capacity, and number of parallel takeovers. Step 4: Perform direct takeover only on the directly managed power supply entering the current takeover batch, perform rate-limited takeover on the soft managed power supply entering the current takeover batch, and do not send new takeover commands to the managed power supply with a delayed takeover level. Step 5: Receive the actual active power output, control mode, execution feedback, frequency, and key section measurements of each controlled power source, and recalculate the unified takeover risk value, adjust the takeover level, and update subsequent takeover batches based on the updated reliable data.

2. The method according to claim 1, characterized in that, The multiple controlled power sources of different types are selected from at least two of the following: thermal power units, hydropower units, wind farms, photovoltaic power plants, and energy storage systems. Specifically, for thermal power units, the equivalent permissible ramp rate is determined based on the available ramp limit in the current direction; for hydropower units, the equivalent permissible ramp rate is corrected based on the water level safety margin; for energy storage systems, the equivalent rated adjustable power and equivalent permissible ramp rate are determined based on the state-of-charge safety margin, charging / discharging direction, available power, and available ramp limit; for wind farms and photovoltaic power plants, the equivalent rated adjustable power is determined based on real-time available power, power rationing instructions, and available equipment capacity. When type-specific operating constraints exceed the permissible range, or when any denominator parameter used in the risk calculation is not greater than zero, the corresponding controlled power source is classified as a delayed takeover level.

3. The method according to claim 1, characterized in that, The risk value of the unified takeover is calculated according to the following formula: ， in, Number the controlled power supply. The sampling period number; For the first The controlled power supply in the first The unified takeover risk value for each sampling period is a dimensionless quantity. To contribute practically and effectively The takeover reference targets used in this round of takeover evaluation are as follows: and The unit for all values ​​is megawatts; The equivalent rated adjustable power is expressed in megawatts and is greater than zero. The equivalent permissible ramp rate in the current direction, expressed in megawatts per minute and greater than zero; For reference, the takeover duration is in minutes and is greater than zero, and remains constant within the same takeover evaluation period; The maximum value among the normalized and clipped values ​​of frequency sensitivity and critical section sensitivity is a dimensionless quantity that satisfies the following conditions: ; , and For a dimensionless weight greater than zero, and .

4. The method according to claim 1, characterized in that, Set direct takeover risk threshold and soft takeover risk threshold ,in, and All are dimensionless quantities and satisfy The risk value of unified takeover is no greater than The time is divided into direct takeover level, greater than and not greater than The time is classified into soft pipe level, greater than The risk levels are divided into delayed takeover levels. The direct takeover risk threshold and the soft takeover risk threshold are set by historical samples with takeover result labels and backtested by independent samples. When the backtesting result does not meet the preset conditions, the stability threshold obtained by transient simulation is used. When the classification results of the two thresholds are inconsistent, the higher risk level is used.

5. The method according to claim 1, characterized in that, The current takeover batch is determined in order of candidate order based on takeover level, unified takeover risk value and system sensitivity. For each candidate batch, the frequency change rate and frequency deviation within the time window are evaluated based on the net active power imbalance caused by the candidate batch, system equivalent inertia, primary frequency regulation response and load damping prediction. At the same time, the power flow changes of key sections, single batch takeover capacity and number of parallel takeovers are checked. Only candidate batches that satisfy all the above constraints are determined as the current takeover batch.

6. The method according to claim 1, characterized in that, The tiered execution includes: when a directly controlled power source enters the current takeover batch, and the deviation between its takeover reference target and the current automatic generation control target, as well as the deviation between its actual active power output and the current automatic generation control target, respectively enter the corresponding dead zone, the current automatic generation control target is issued; when a soft takeover controlled power source enters the current takeover batch, a soft takeover command is generated using the same takeover reference target used for the risk calculation of the controlled power source as the command target, and the command change in adjacent control cycles is limited to the range determined according to the equivalent allowable ramp rate of the controlled power source; and the controlled power sources with delayed takeover levels are monitored without sending new takeover targets.

7. The method according to claim 1, characterized in that, When the unified takeover risk value jumps beyond the preset range, any system security constraint is not met, the status data or execution receipt of the controlled power supply exceeds the allowable delay, or the data quality flag is invalid, new takeovers are suspended, the progress of soft takeovers that have already been initiated is slowed down, and the unified takeover risk value is recalculated based on the most recent trusted data, the takeover level is reclassified, and subsequent takeover batches are regenerated; for controlled power supplies with untrusted data, the most recent trusted instruction is maintained or the system enters the security target according to the preset rollback process.

8. The method according to claim 1, characterized in that, The takeover completion is determined only when all target controlled power sources are in remote automatic generation control mode, the soft takeover process has been completed, the actual active power output tracking error of the current automatic generation control target meets the completion dead zone, and the frequency change rate, frequency deviation, and power flow of key sections all meet the corresponding limits; the takeover is determined to be completed when all conditions are met continuously within the minimum observation window.

9. A hierarchical management system for switching multi-power source cooperative modes in an automatic power generation control system, characterized in that, It includes an automatic power generation control master station and multiple controlled power sources that are communicatively connected to the automatic power generation control master station. The multiple controlled power sources are selected from at least two types of thermal power units, hydropower units, wind farms, photovoltaic power stations, and energy storage systems. The automatic power generation control master station includes: The data acquisition and takeover reference target module is used to acquire status data, capability data, takeover reference targets, execution receipts, and system safety data of each controlled power source. The parameter conversion and sensitivity normalization module is used to convert type-specific operating constraints into equivalent rated adjustable power and equivalent allowable ramp rate with a unified caliber, and to form a dimensionless risk component. The risk assessment and classification module is used to calculate the unified takeover risk value based on the dimensionless risk components and output the direct takeover level, soft takeover level, or delayed takeover level. The physical constraints and batch planning module is used to generate the current takeover batch based on the takeover level and unified takeover risk value under constraints of frequency, critical section power flow, single batch takeover capacity and parallel takeover quantity; The takeover command generation module is used to output direct takeover commands only to the directly takeover level controlled power supplies in the current takeover batch, output speed-limiting takeover commands to the soft takeover level controlled power supplies in the current takeover batch, and not output new takeover commands to the controlled power supplies with delayed takeover levels. The monitoring and confirmation and dynamic reclassification module is used to receive the actual active power output, control mode and execution feedback of each controlled power source, as well as frequency and key section measurements, and feed back the updated reliable data to the parameter conversion and sensitivity normalization module, risk assessment and level classification module and physical constraint and batch planning module to update the unified takeover risk value, takeover level and subsequent takeover batch. The multiple controlled power sources are used to execute the direct takeover command or the speed-limiting takeover command, and to provide feedback on the actual active power output, control mode, and execution receipt.

10. The system according to claim 9, characterized in that, The parameter conversion and sensitivity normalization modules respectively generate the ratio of the actual active power output to the equivalent rated adjustable power, the ratio of the power deviation elimination time to the reference takeover duration, and the normalized system sensitivity. The risk assessment and level classification module weights the ratio of the power deviation to the equivalent rated adjustable power, the ratio of the power deviation elimination time to the reference takeover duration, and the normalized system sensitivity to obtain a unified takeover risk value. The physical constraints and batch planning module only includes directly takeover or soft takeover controlled power sources that are constrained by frequency change rate, frequency deviation, critical section power flow, single batch takeover capacity, and parallel takeover quantity into the current takeover batch. When the monitoring confirmation and dynamic reclassification module experiences a risk transition, data breach, or any of the aforementioned constraints, it triggers a pause in adding new takeovers and updates to the unified takeover risk value, takeover level, and subsequent takeover batches.

Images