A hierarchical cooperative optimization scheduling method for multi-source heterogeneous regional power grid

Abstract

The application provides a layered cooperative optimization scheduling method for multi-source heterogeneous regional power grids, and relates to the technical field of power dispatching. The method determines the equivalent inertia time constant of the multi-source heterogeneous regional power grid based on the real-time physical operation state; determines the resource response rate coefficient based on the maximum active power reserve quantity; calculates the frequency deviation and the frequency change rate; if the frequency deviation or the frequency change rate exceeds the corresponding threshold value, the emergency frequency support is triggered; under the emergency frequency support, when the system frequency crosses the minimum point and the frequency change rate is less than the first threshold value, a multi-stage cooperative optimization model is constructed; the alternating direction multiplier method is introduced to solve the layered consensus of the multi-stage cooperative optimization model, and the active power output planning adjustment value of each local control node is obtained to control the multi-source heterogeneous regional power grid. The method can avoid the secondary drop of the system frequency and the dimension disaster, can help to improve the response speed, can provide good data privacy protection, and can improve the safety guarantee.

Description

Technical Field

[0001] This invention relates to the field of power dispatching technology, and in particular to a hierarchical collaborative optimization dispatching method for multi-source heterogeneous regional power grids. Background Technology

[0002] With the deepening of energy transition, the power grid has widely integrated new energy sources such as wind and solar power, along with battery storage, conventional thermal power units, and a large number of flexible loads with regulation potential, forming a multi-source heterogeneous structure. However, the replacement of traditional synchronous generator units by a large number of new energy devices based on power electronic interfaces has led to a continuous decline in the overall mechanical rotational inertia of the power grid, significantly weakening its frequency disturbance immunity. When faced with power imbalance disturbances such as drastic fluctuations in internal new energy output, sudden increases in large-capacity loads, or external faults, the rate of frequency change (RoCoF) of the power grid increases significantly, and the frequency drops extremely rapidly. While the large-scale integration of new energy sources helps improve the low-carbon level of regional power grids, it also poses a severe challenge to the frequency stability of the power grid. If the massive heterogeneous resources cannot be effectively coordinated and rapidly dispatched, it is easy to trigger low-frequency load shedding protection actions of underlying equipment, and even cause power grid collapse.

[0003] The dynamic frequency change of a regional power grid after a power imbalance disturbance is a continuous physical process involving an initial sharp frequency drop (or rise), reaching the lowest (or highest) frequency point, and then gradually recovering to a steady state. Existing frequency optimization scheduling and control research often focuses on a single frequency regulation resource or a single response stage, lacking a systematic decoupled modeling of the entire frequency response process. Especially in the frequency recovery stage, if the physical limits and exit sequences of various heterogeneous resources are not fully considered, it can easily lead to rotor speed instability of wind turbines or the state of charge (SOC) of energy storage devices exceeding limits, thus triggering a "secondary drop" in system frequency. Therefore, constructing a full-process frequency response model covering frequency support and frequency recovery is a necessary means to ensure the long-term physical security of large power grids.

[0004] Faced with massive, dispersed, and heterogeneous resources with vastly different physical response characteristics within a regional power grid, traditional energy management generally employs centralized optimization scheduling methods. This requires uploading the real-time physical parameters and status of all devices to a central control center for centralized solution. However, as the power grid expands and the number of participating entities increases, the centralized architecture not only faces the "curse of dimensionality" problem—characterized by excessively high model dimensionality, large communication latency, and low solution efficiency—but also struggles to meet the data privacy protection requirements among multiple stakeholders. Furthermore, existing scheduling strategies often neglect the comprehensive adjustment costs of each device and the hard constraints of physical self-recovery during resource recovery.

[0005] In existing technologies, local frequency control methods based on single resources or single stages mainly rely on a single flexible resource within the regional power grid (such as battery energy storage systems or flywheel energy storage). By configuring a local controller, they monitor the grid connection point frequency or rate of change of frequency (RoCoF) in real time. When a system frequency exceeding the limit is detected, active power is rapidly output to suppress the frequency drop according to preset droop control or virtual inertia control logic. These methods have a fast response speed in the initial stage of a frequency drop (frequency support phase), providing frequency support for a very short time. However, these methods typically only focus on the single stage of frequency support and lack the coordinated utilization of multi-source heterogeneous resources within the grid, such as wind turbines, photovoltaics, and flexible loads. When the grid's system frequency stabilizes and enters the recovery phase, the lack of global planning and the disorderly withdrawal or drastic reduction in output of resources such as energy storage can easily trigger a "secondary frequency drop."

[0006] Furthermore, the regional power grid multi-source collaborative optimization scheduling method based on a centralized architecture adopts an architecture of "full data upload and centralized unified solution". Wind, solar, energy storage, thermal, and load resources within the power grid all upload their real-time operating parameters (such as current output, power limit, equipment parameters, cost coefficients, etc.) to the centralized Energy Management System (EMS) via a communication network. The EMS constructs a high-dimensional objective function and constraints encompassing all resource nodes, and solves it centrally using mixed-integer linear programming or heuristic algorithms, subsequently issuing instructions to each device. While this type of method can theoretically obtain a globally optimal solution, in practical applications, on the one hand, due to long communication delays and computation times, it is difficult to meet the millisecond / second-level ultra-fast response requirements of the frequency support stage; on the other hand, facing the massive heterogeneous resources within the regional power grid, centralized solution suffers from a severe "curse of dimensionality," and the mandatory requirement for each independent stakeholder to upload private physical parameters cannot guarantee data privacy.

[0007] To overcome the shortcomings of centralized scheduling, some existing studies have applied distributed consensus algorithms, such as consensus algorithms and the Alternating Direction Method of Multipliers (ADMM), to regional power grid energy management. These methods divide the grid by nodes, with each node independently solving its own output plan locally. Adjacent nodes only interact with boundary multipliers (such as price signals or power deviations), achieving supply-demand balance and minimizing operating costs through multiple iterations. This approach effectively alleviates computational pressure and protects local privacy. However, current distributed scheduling focuses primarily on economic optimization under steady-state conditions, failing to deeply integrate dynamic physical constraints. When dealing with the specific dynamic transient process of frequency recovery, conventional distributed methods fail to effectively consider physical requirements such as the smooth recovery of wind turbine speed and the self-recovery of the state of charge (SOC) of energy storage. This can lead to scheduling results that violate the physical limits of equipment during actual execution, failing to provide safety guarantees for full-process frequency control. Summary of the Invention

[0008] To address the aforementioned technical problems in existing technologies, this invention aims to provide a hierarchical collaborative optimization scheduling method for multi-source heterogeneous regional power grids. By considering physical limits and exit timing, it aims to avoid the "secondary drop" in system frequency caused by rotor speed instability and load overload in wind turbines, thus avoiding the "curse of dimensionality" caused by the increasing number of various heterogeneous resources in traditional centralized architectures. It also avoids the risk of a "secondary drop" in system frequency caused by local frequency control methods based on a single resource or single stage when the system frequency stabilizes and enters the recovery phase.

[0009] Specifically, the technical solution is as follows:

[0010] A hierarchical collaborative optimization scheduling method for multi-source heterogeneous regional power grids includes:

[0011] Based on the real-time physical operating status of various multi-source heterogeneous resources, the equivalent inertial time constant of the multi-source heterogeneous regional power grid is determined.

[0012] Based on the maximum active power reserve of various multi-source heterogeneous resources, the corresponding resource response rate coefficient is determined.

[0013] Based on the system frequency of the multi-source heterogeneous regional power grid, the current frequency deviation and frequency change rate are calculated. If the frequency deviation or frequency change rate exceeds the corresponding threshold, emergency frequency support is triggered, and the multi-source heterogeneous regional power grid enters the frequency support stage: based on the equivalent inertial time constant and resource response rate coefficient, local active power control commands are generated to control the corresponding multi-source heterogeneous resources.

[0014] When the system frequency crosses the minimum point and the rate of frequency change is less than the first threshold, the multi-source heterogeneous regional power grid enters the frequency recovery phase; a multi-stage collaborative optimization model is constructed with the goal of minimizing the total comprehensive regulation cost of the frequency recovery phase.

[0015] The alternating direction multiplier method is introduced to solve the multi-stage collaborative optimization model in a hierarchical consensus manner: the active power output plan values ​​of various multi-source heterogeneous resources at the local control node are solved independently to obtain the active power output plan adjustment values, and the multi-source heterogeneous regional power grid is controlled according to each active power output plan adjustment value.

[0016] Preferably, based on the real-time physical operating status of various heterogeneous resources, the equivalent inertial time constant of the heterogeneous regional power grid is determined, including:

[0017] Determine the mechanical inertia time constant and rated apparent power of various multi-source heterogeneous resources;

[0018] Based on the mechanical inertia time constant and rated apparent power of each type of multi-source heterogeneous resource, determine Equivalent inertial time constant of multi-source heterogeneous regional power grid The formula is as follows:

[0019] ;

[0020] In the formula, This refers to the type number of the multi-source heterogeneous resources. The number of types of heterogeneous resources. For the first The mechanical inertia time constant of multi-source heterogeneous resources For the first The rated apparent power of multi-source heterogeneous resources, This serves as the baseline capacity for a multi-source heterogeneous regional power grid.

[0021] Preferably, the resource response rate coefficient is determined based on the maximum active power reserve, including:

[0022] according to Time of the first Maximum active power reserve that can be instantly mobilized by multi-source heterogeneous resources ,calculate Time of the first Resource response rate coefficient of multi-source heterogeneous resources The formula is as follows:

[0023] ;

[0024] In the formula, For the first The inherent delay in the transmission of control commands from multi-source heterogeneous resources to the output response of physical equipment.

[0025] Preferably, the formula for calculating the rate of change of frequency is as follows:

[0026] ;

[0027] In the formula, for Rate of change of frequency at time, for The system frequency at any given time. The system frequency at the previous sampling time. The sampling period.

[0028] Preferably, if the frequency deviation or rate of change exceeds a corresponding threshold, emergency frequency support is triggered, including:

[0029] Update the quick response trigger status bit The formula is as follows:

[0030] ;

[0031] In the formula, This is the maximum permissible limit for frequency deviation. This is the maximum permissible limit for the rate of change of frequency;

[0032] like This triggers emergency frequency support, including:

[0033] calculate Total active power compensation deficit required by multi-source heterogeneous regional power grids at all times The formula is as follows:

[0034] ;

[0035] In the formula, The comprehensive droop control proportional gain coefficient for multi-source heterogeneous regional power grids. for Frequency deviation of multi-source heterogeneous regional power grids at any time For the comprehensive inertia differential control gain coefficient of a multi-source heterogeneous regional power grid, for The equivalent inertial time constant of a multi-source heterogeneous regional power grid at any given time. for Rate of change of frequency at any given moment;

[0036] like If so, then increased output is required; if Then you need to reduce your output;

[0037] calculate Time of the first Dynamic priority weights for multi-source heterogeneous resources in local power allocation The formula is as follows:

[0038] ;

[0039] ;

[0040] In the formula, The number of types of heterogeneous resources. for Time of the first Resource response rate coefficient of multi-source heterogeneous resources;

[0041] By dynamic priority weight Distribute total active power to compensate for deficit ,generate Time of the first Initial active power regulation command for multi-source heterogeneous resources The formula is as follows:

[0042] ;

[0043] Introduction Time of the first Maximum active power reserve that can be instantly mobilized by multi-source heterogeneous resources Limiting initial active power adjustment command Generate local active power control commands using the following formula:

[0044] ;

[0045] In the formula, for Time of the first Local active power control commands for multi-source heterogeneous resources It is a symbolic function;

[0046] Through local active power control commands Real-time control Multi-source heterogeneous resources.

[0047] Preferably, the energy management system constructs a multi-stage collaborative optimization model with the objective function of minimizing the total comprehensive adjustment cost during the frequency recovery phase, including:

[0048] Based on emergency frequency support, the multi-source heterogeneous regional power grid crosses the lowest system frequency point and enters the frequency recovery phase.

[0049] Total comprehensive adjustment cost during the frequency recovery phase The minimum objective function is defined by the following formula:

[0050] ;

[0051] ;

[0052] ;

[0053] ;

[0054] ;

[0055] ;

[0056] ;

[0057] In the formula, , and For the first The comprehensive adjustment cost coefficient of multi-source heterogeneous resources. for Time of the first The planned adjustment value of active power output for multi-source heterogeneous resources. The frequency exceeding the limit penalty weight coefficient, for Frequency deviation of multi-source heterogeneous regional power grids at any time The permissible steady-state dead zone limit for multi-source heterogeneous regional power grids. The number of types of heterogeneous resources. for The total active power compensation deficit required by the multi-source heterogeneous regional power grid at all times. for Active power loss values ​​in multi-source heterogeneous regional power grids at any given time. For the first Minimum active power output of multi-source heterogeneous resources. For the first The maximum active power output of multi-source heterogeneous resources. for The rotational speed of the rotor in the wind turbine unit at all times. for The rotational speed of the rotor in the wind turbine unit at all times. This refers to the maximum permissible rate of change of rotor speed in a wind turbine generator set. This represents the lower limit of the safe rotational speed of the rotor in a wind turbine generator set. This refers to the upper limit of the safe rotational speed of the rotor in a wind turbine generator set. This is the lower limit of the state of charge of energy storage. for The state of charge that stores energy at all times. This represents the upper limit of the state of charge of energy storage. The end time of the frequency recovery phase State of charge of energy storage This represents the lower limit of the target range for the self-recovery of the energy storage state of charge. This represents the upper limit of the target range for the self-recovery of the energy storage state of charge.

[0058] Preferably, the alternating direction multiplier method is introduced to perform hierarchical consensus solution for the multi-stage collaborative optimization model: the active power output plan values ​​of various multi-source heterogeneous resources at the local control node are solved independently to obtain the active power output plan adjustment values; the multi-source heterogeneous regional power grid is controlled according to each active power output plan adjustment value, including:

[0059] Introducing Lagrange multipliers As a boundary consensus variable for coordinating various multi-source heterogeneous resources;

[0060] Minimize the objective function During the iteration process, the local control node receives the boundary consensus variables from the previous iteration. The active power output plan adjustment value is updated using the following formula:

[0061] ;

[0062] In the formula, For the first Round iteration Time of the first The planned adjustment value of active power output for multi-source heterogeneous resources. For the first A set of local physical security constraints for multi-source heterogeneous resources;

[0063] Based on the planned adjustment value of active power output The boundary consensus variable is updated using the following formula:

[0064] ;

[0065] ;

[0066] In the formula, For the first Round iteration Boundary consensus variables at time points, This is the iteration step size factor;

[0067] Updated boundary consensus variables The data is then distributed to each local control node to begin the next iteration.

[0068] After the iteration terminates, a timing control instruction matrix for issuance is generated. The formula is as follows:

[0069] ;

[0070] In the formula, to for Time until the end of the frequency recovery phase Control commands at corresponding moments during the process It is the transpose symbol;

[0071] Control commands are executed in sequence. To control commands Control each local control node;

[0072] The formula for the iteration termination condition is as follows:

[0073] ;

[0074] In the formula, This is the limit for the convergence accuracy of the original residual in power balance. This is the convergence accuracy limit for the dual residuals of the boundary consensus variables.

[0075] Furthermore, it also includes:

[0076] like ,and If maintained continuously, the local control node will reset the fast response trigger status bit. .

[0077] Furthermore, it also includes:

[0078] If each local control node does not receive an updated boundary consensus variable within m consecutive sampling periods If it fails, it will automatically exit the hierarchical consensus solution process and maintain its output according to the active power output plan adjustment value of the previous valid sampling period; among which, .

[0079] Preferably, the multi-source heterogeneous resources include thermal power units, wind power units, photovoltaic power units, energy storage and flexible loads, and the real-time physical operating status includes active power, reactive power, rotor speed and energy storage state of charge.

[0080] Compared to existing technologies, the technical solution provided by this invention is based on real-time physical operating status. By considering the physical limits and exit timing of various heterogeneous resources, it performs local control to enable the heterogeneous regional power grid to enter the frequency recovery phase. After that, it constructs a corresponding multi-stage collaborative optimization model and controls the heterogeneous regional power grid according to the solution results to achieve hierarchical collaborative optimization scheduling. This can avoid the "secondary drop" of system frequency caused by rotor speed instability and load overload in wind turbines. It can also avoid the "curse of dimensionality" caused by the increase in the number of heterogeneous resources in traditional centralized architectures. Furthermore, it can avoid the risk of "secondary drop" of system frequency caused by local frequency control methods based on a single resource or a single stage when the system frequency stabilizes and enters the recovery phase. Attached Figure Description

[0081] Figure 1 This is a flowchart illustrating the hierarchical collaborative optimization scheduling method in this invention. Detailed Implementation

[0082] The technical solution provided by the present invention will be further described in detail below with reference to the accompanying drawings.

[0083] like Figure 1 As shown, the hierarchical collaborative optimization scheduling method provided by this invention is oriented towards multi-source heterogeneous regional power grids and mainly includes 8 steps.

[0084] Step 1: Data Acquisition of Operational Status of Multi-Source Heterogeneous Resources. Based on the collaborative control architecture, the local node-side control node (local control node) collects the real-time physical operational status of multi-source heterogeneous resources within the regional power grid. These resources include thermal power units, wind power units, photovoltaic power units, energy storage, and flexible loads. The real-time physical operational status includes at least active power, reactive power, rotor speed, and the state of charge of the energy storage.

[0085] Step 2: Quantitative Modeling of Frequency Support Capability. Based on the equivalent inertia and power reserve mechanism, a frequency response characteristic model covering multi-source heterogeneous resources is constructed. The real-time physical operating status collected in Step 1 is input, and the model calculates and outputs the frequency support capability indicators and response delay times of various resources at different time scales. The frequency support capability indicators include at least the equivalent inertial time constant, which can be determined based on the inherent mechanical inertial time constant or equivalent virtual inertial time constant of various multi-source heterogeneous resources; the response delay time corresponds to and determines the resource response rate, which can be determined by the inherent delay coefficient of the control command issued to the physical equipment's output response.

[0086] Step 3: Frequency Dynamic Monitoring and Support Triggering. The local node-side control node monitors the dynamic changes of the system frequency in real time, calculates the current frequency deviation and frequency change rate; when the frequency deviation or frequency change rate exceeds its corresponding threshold, it is identified as a power imbalance disturbance, triggering the first stage of emergency frequency support.

[0087] Step 4: Local node-side rapid differentiated power compensation. After triggering emergency frequency support, the local node-side control node generates a local active power control command based on the frequency support capability indicators and response delay time obtained in Step 2, and controls the corresponding multi-source heterogeneous resources to perform rapid power compensation to suppress system frequency drops.

[0088] Step 5: Response Phase Judgment and Smooth Transfer of Control Rights Across Domains. When the system frequency crosses the lowest point (judged based on the sign flip of the frequency change rate or the recovery amplitude of the system frequency from a local minimum), and the frequency change rate is less than a preset first threshold, the multi-source heterogeneous regional power grid is determined to have entered the frequency recovery phase. The local node-side control node transfers frequency control rights to the dispatch-end energy management system, which divides the multi-source heterogeneous regional power grid into multiple optimization levels according to resource category and physical region.

[0089] Step 6: Construction of a multi-stage economic collaborative optimization model for dispatch-side energy management. The dispatch-side energy management system takes the minimum comprehensive regulation cost of the regional power grid system as the objective function, and uses the smooth recovery constraint of wind turbine speed, the self-recovery constraint of energy storage state of charge, and power balance as constraints to construct a multi-stage collaborative optimization model.

[0090] Step 7: Iterative solution of the hierarchical distributed consensus algorithm. The alternating direction multiplier method is introduced to solve the multi-stage collaborative optimization model through hierarchical consensus. Each optimization level solves the recovery strategy independently locally, and the boundary multiplier information of the interaction power and price between levels is iterated until convergence, so as to obtain the optimal collaborative recovery strategy of each level.

[0091] Step 8: Issuance of control commands and orderly exit execution during the recovery period. Based on the iterative convergence results of the multi-stage collaborative optimization model in Step 7, the optimal power allocation commands and exit timing commands for each participating entity in the multi-source heterogeneous regional power grid are generated; the commands are issued to the corresponding underlying multi-source heterogeneous resources to perform power regulation and orderly exit until the system frequency recovers to the rated range.

[0092] It should be noted that the above steps are set up for the convenience of understanding the solution and do not completely follow a chronological order.

[0093] 1. Unified spatiotemporal perception and parameter initialization of the operating status of multi-source heterogeneous resources.

[0094] Based on the collaborative control architecture, local control nodes are deployed on the local node side of the regional power grid (i.e., the local side of each underlying heterogeneous resource) to coordinate with the energy management system at the dispatching end, thereby decoupling data perception and decision-making.

[0095] The system collects the real-time physical operating status of multi-source heterogeneous resources. Specifically, let the set of multi-source heterogeneous resources to be monitored be denoted as . .in, This represents the number and types of heterogeneous resources from multiple sources. The sampling period for the local control node is defined as follows: In any At that moment, the Multi-source heterogeneous resources ( The original physical operating state set of ) is denoted as The formula is as follows:

[0096] ;

[0097] In the formula, for Time of the first Active power of multi-source heterogeneous resources for Time of the first Reactive power of multi-source heterogeneous resources for The rotational speed of the rotor in a generator set or wind turbine set at any given time. for The state of charge that stores energy at all times. for Voltage amplitude at the grid connection point of multi-source heterogeneous resources at any given time This is the transpose symbol.

[0098] To eliminate computational overflow caused by differences in the physical magnitudes and dimensions of different devices within a multi-source heterogeneous regional power grid, the local control node normalizes the original state data and maps it to a standardized coordinate system, forming a resource spatiotemporal perception vector for the multi-source heterogeneous regional power grid, as shown in the following formula:

[0099] ;

[0100] In the formula, In multi-source heterogeneous regional power grids Time of the first Resource spatiotemporal awareness vectors for multi-source heterogeneous resources. For the first The rated apparent power of multi-source heterogeneous resources, For the first The rated rotor speed of multi-source heterogeneous resources, For the first The rated voltage of the grid connection point for multi-source heterogeneous resources. Local control nodes monitor the status in real time at a set sampling period, constructing a resource spatiotemporal awareness vector. This serves as the foundational input parameter for subsequent frequency-wide collaborative optimization scheduling.

[0101] 2. Modeling for assessing the frequency support capability of multi-source heterogeneous resources.

[0102] Based on the real-time physical operating status of the collected multi-source heterogeneous resources, the equivalent inertial time constant of the multi-source heterogeneous regional power grid is determined. This is based on resource spatiotemporal sensing vectors. The local control node calculates the current time of the multi-source heterogeneous regional power grid. The equivalent inertial time constant at time ( ) The formula is as follows:

[0103] ;

[0104] In the formula, This refers to the type number of the multi-source heterogeneous resources. The number of types of heterogeneous resources. For the first Mechanical inertia time constant (or equivalent virtual inertia time constant, in seconds) of multi-source heterogeneous resources. This serves as the baseline capacity for a multi-source heterogeneous regional power grid.

[0105] Secondly, the resource response rate coefficient is determined based on the maximum active power reserve. This is done by obtaining data based on the real-time operating status. Time of the first Maximum active power reserve that can be instantly mobilized from multi-source heterogeneous resources To quantitatively assess the dynamic response priority of multi-source heterogeneous resources in the initial stage of frequency drop, a definition is defined. Time of the first Resource response rate coefficient of multi-source heterogeneous resources The formula is as follows:

[0106] ;

[0107] In the formula, For the first The inherent delay in the transmission of control commands from heterogeneous resources to the output response of physical equipment. The calculated response rate coefficient constitutes an evaluation index of the frequency support capability of heterogeneous resources in the regional power grid, serving as an input condition for the generation of local active power control commands at the local node side.

[0108] 3. Real-time monitoring and rapid response criteria for local node-side frequency security constraints.

[0109] The system frequency of the multi-source heterogeneous regional power grid is dynamically monitored, and the current frequency deviation and frequency change rate are calculated. An independent frequency monitoring and logic comparison module is configured in the local control node to calculate in real time the frequency deviation between the current grid connection point frequency (the system frequency of the multi-source heterogeneous regional power grid) and the rated frequency. Meanwhile, the local control node uses a differential algorithm to calculate the frequency change rate of the multi-source heterogeneous regional power grid in real time, as shown in the following formula:

[0110] ;

[0111] In the formula, for Rate of change of frequency at time, for The system frequency at any given time. The system frequency at the previous sampling time.

[0112] Preset thresholds, including the maximum allowable limit for frequency deviation. With the maximum permissible limit of the rate of change of frequency The logic comparison module evaluates the following logic expression in each control cycle to update the fast response trigger status bit, as shown in the formula below:

[0113] ;

[0114] In the formula, for The fast response trigger status bit is set at any given time; a value of 1 indicates trigger support, and a value of 0 indicates normal operation. If a power imbalance disturbance causes the fast response trigger status bit to be affected... When the frequency deviation or frequency change rate exceeds the corresponding threshold, the multi-source heterogeneous regional power grid directly activates the emergency response logic at the local node side, triggers emergency frequency support, blocks the signal from the regular dispatch terminal, and avoids the first wave of degradation caused by communication delay.

[0115] 4. Local differentiated power compensation execution mechanism.

[0116] Based on the equivalent inertia time constant and resource response rate coefficient, local active power control commands are generated to control the corresponding multi-source heterogeneous resources. When the fast response trigger state bit is determined... Upon activation, a localized differentiated power compensation execution mechanism is immediately initiated. This is first combined with the equivalent inertial time constant. And real-time frequency indicators, calculate the current time ( (Time) Total active power compensation deficit required by multi-source heterogeneous regional power grid The formula is as follows:

[0117] ;

[0118] In the formula, The comprehensive droop control proportional gain coefficient for multi-source heterogeneous regional power grids. for Frequency deviation of multi-source heterogeneous regional power grids at any time This represents the comprehensive inertia differential control gain coefficient for a multi-source heterogeneous regional power grid. If... If so, then increased output is required; if If so, then you need to reduce your output.

[0119] Based on resource response rate coefficient ,calculate Time of the first Dynamic priority weights for multi-source heterogeneous resources in local power allocation The formula is as follows:

[0120] ;

[0121] ;

[0122] The total active power will compensate for the shortfall. By dynamic priority weight Allocation, generation Time of the first Initial active power regulation command for multi-source heterogeneous resources The formula is as follows:

[0123] ;

[0124] Considering the physical output boundaries of each underlying device, the following is introduced: Time of the first Maximum active power reserve that can be instantly mobilized by multi-source heterogeneous resources Limiting initial active power adjustment command Generate local active power control commands using the following formula:

[0125] ;

[0126] In the formula, for Time of the first Local active power control commands for multi-source heterogeneous resources This is the sign function, used to determine the direction of power adjustment. Local active power control command. The data is directly sent to the local inverter or speed controller, prioritizing the use of energy storage and flexible loads with high response rates to achieve rapid local compensation.

[0127] 5. Construction of a multi-stage economic collaborative optimization model for energy management at the dispatching end.

[0128] Under emergency frequency support, when the system frequency crosses the lowest point and the rate of frequency change is less than the first threshold, a multi-stage collaborative optimization model is constructed. When the system crosses the lowest frequency point and enters a stable period (when the system determines that the absolute value of the rate of frequency change is less than the preset first threshold), the control of frequency recovery is transferred to the energy management system at the dispatching end.

[0129] The energy management system's total comprehensive adjustment cost during the frequency recovery phase The objective function is to minimize the heterogeneous resource adjustment cost and the frequency deviation penalty cost. Simultaneously, the system power balance and the dynamic self-recovery physical hard constraints of multi-source heterogeneous resources are introduced, forming the following optimization solution model, as shown in the formula:

[0130] ;

[0131] ;

[0132] ;

[0133] ;

[0134] ;

[0135] ;

[0136] ;

[0137] In the formula, , and For the first The comprehensive adjustment cost coefficient of multi-source heterogeneous resources. for Time of the first The planned adjustment value of active power output for multi-source heterogeneous resources. The frequency exceeding the limit penalty weight coefficient, for Frequency deviation of multi-source heterogeneous regional power grids at any time The permissible steady-state dead zone limit for multi-source heterogeneous regional power grids. The number of types of heterogeneous resources. for The total active power compensation deficit required by the multi-source heterogeneous regional power grid at all times. for Active power loss values ​​in multi-source heterogeneous regional power grids at any given time. For the first Minimum active power output of multi-source heterogeneous resources. For the first The maximum active power output of multi-source heterogeneous resources. for The rotational speed of the rotor in the wind turbine unit at all times. for The rotational speed of the rotor in the wind turbine unit at all times. This refers to the maximum permissible rate of change of rotor speed in a wind turbine generator set. This represents the lower limit of the safe rotational speed of the rotor in a wind turbine generator set. This refers to the upper limit of the safe rotational speed of the rotor in a wind turbine generator set. This is the lower limit of the state of charge of energy storage. for The state of charge that stores energy at all times. This represents the upper limit of the state of charge of energy storage. The end time of the frequency recovery phase State of charge of energy storage This represents the lower limit of the target range for the self-recovery of the energy storage state of charge. This represents the upper limit of the target range for the self-recovery of the energy storage state of charge.

[0138] 6. Collaborative solution logic based on hierarchical distributed consensus.

[0139] This invention introduces the Alternating Direction Method of Multipliers (ADMM) to perform hierarchical consensus solving of a multi-stage collaborative optimization model, obtaining the optimal collaborative recovery strategy at each level. For the aforementioned high-dimensional model, this invention does not employ centralized direct solving, but instead introduces the Alternating Direction Method of Multipliers (ADMM) framework for hierarchical consensus solving. Lagrange multipliers are introduced as boundary consensus variables to coordinate various resources. This is done by minimizing the objective function... The In the nth iteration, the 1st For multi-source heterogeneous resources, local optimization subproblems are solved independently on the local node side. Multi-source heterogeneous resources receive consensus variables from the previous iteration The active power output plan adjustment value is updated using the following formula:

[0140] ;

[0141] In the formula, For the first Round iteration Time of the first The planned adjustment value of active power output for multi-source heterogeneous resources. For the first A set of local physical security constraints for multi-source heterogeneous resources.

[0142] After each local control node completes its local solution, it only uploads the updated active power output plan value to the scheduling coordination layer, without uploading the local cost coefficient and constraint status. The scheduling coordination layer uses a subgradient descent method based on the active power output plan adjustment value. The boundary consensus variable is updated using the following formula:

[0143] ;

[0144] ;

[0145] In the formula, For the first Round iteration Boundary consensus variables at time points, This is the iteration step size factor.

[0146] The scheduling coordination layer will update the boundary consensus variables. The results are distributed to each local control node, initiating the next iteration. The iteration termination condition is set so that both the original residual and the dual residual satisfy the preset convergence accuracy limit, as shown in the following formula:

[0147] ;

[0148] In the formula, This is the limit for the convergence accuracy of the original residual in power balance. This serves as the convergence accuracy limit for the dual residuals of the boundary consensus variables. When the above conditions are met, the iteration terminates, yielding the final active power output planned adjustment values ​​for each local control node, thus forming the optimal collaborative recovery strategy.

[0149] Preferably, as a supplementary method, if each local control node does not receive an updated boundary consensus variable within m consecutive sampling periods... If the communication is interrupted, the hierarchical consensus solving process will automatically exit. Each local control node will maintain its output according to the active power output plan of the previous valid sampling period, triggering the underlying deadlock prevention and safety rollback mechanism. The effective sampling period is the period during which the updated boundary consensus variable is received. The active power output plan adjustment value was updated. The sampling period.

[0150] 7. System optimal control command issuance and closed-loop execution throughout the entire process.

[0151] The multi-source heterogeneous regional power grid is controlled according to the optimal collaborative recovery strategy. Once the hierarchical consensus solution process meets the convergence condition, the dispatching-end energy management system extracts the planned active power output values ​​of each resource and generates a timing control command matrix for issuance, as shown in the following formula:

[0152] ;

[0153] In the formula, to for Time until the end of the frequency recovery phase Control commands at corresponding moments during the process.

[0154] Control commands are executed in sequence. To control commands The dispatching-end energy management system controls each local control node. It sends a timing control command matrix to the corresponding local control nodes via the communication downlink. Upon receiving the commands, the local control nodes parse them into underlying control reference values ​​recognizable by the physical devices and input them to the local actuators of each heterogeneous resource. The local actuators employ closed-loop follow-up control, driving the underlying physical devices to smoothly adjust and orderly deactivate active power according to the specified trajectory.

[0155] During this process, when physical time advances to the end of the scheduling cycle, and the system frequency deviation monitored in real time by the local node continuously remains within the system steady-state dead zone limit, the physical execution of this entire frequency response task is determined to be complete. At this time, the local control node sets the fast response trigger status bit. To perform a reset, use the following formula:

[0156] ;

[0157] After the reset, the power allocation control of the system is returned to the regular steady-state scheduling module, and the local control nodes resume their background monitoring state for daily power fluctuations. At this point, the multi-source heterogeneous regional power grid has completed the closed-loop frequency response control throughout the entire process.

[0158] This invention establishes a quantitative model of frequency response characteristics covering multi-source heterogeneous resources based on a hierarchical collaborative control architecture. During the frequency support phase, local nodes perceive security constraints in real time and drive underlying resources to implement differentiated and rapid responses to suppress the first frequency drop. During the frequency recovery phase, a multi-stage collaborative optimization model is constructed and integrated to adjust costs and physical self-recovery hard constraints of wind turbines, energy storage, etc. Combined with hierarchical consensus solving to iteratively interact boundary information between levels, the optimal power allocation and orderly exit sequence for ensuring stable frequency recovery of the system are finally solved, effectively improving the frequency security and stability level of multi-source heterogeneous regional power grids in the face of power disturbances throughout the entire process.

[0159] The entire frequency control process offers enhanced security. Addressing the issue of existing single-stage control systems prone to frequency "secondary drops" during the recovery period due to disordered resource withdrawal, this invention forcibly embeds underlying physical constraints, such as the smooth recovery of rotor speed in wind turbines and the self-recovery of the state of charge of energy storage, into the multi-stage optimization at the scheduling end. This mechanism effectively breaks down the barriers between transient physical control and steady-state economic scheduling, fundamentally eliminating the risk of secondary frequency drops.

[0160] The frequency support response is significantly faster in the initial stage of disturbances. Addressing the long communication delays of traditional centralized scheduling, this invention relies on a hierarchical collaborative architecture to achieve closed-loop perception of limit exceedances and directly generate commands at the local node level. Simultaneously, it proposes a "response rate coefficient" to precisely and preferentially allocate compensation quotas to high-response-rate resources such as energy storage and flexible loads. This successfully avoids communication delays with the scheduling end, minimizing the initial frequency drop in the system.

[0161] The model solves more efficiently and effectively protects the privacy of stakeholders. Addressing the "curse of dimensionality" and privacy leaks inherent in centralized architectures, this invention employs a layered consensus solution. Each layer independently solves optimization subproblems locally, with local nodes only iteratively interacting with the scheduler on boundary multipliers (such as power deviation and consistent price signals). This mechanism significantly reduces the computational dimensionality and communication bandwidth pressure of centralized systems and effectively isolates private parameters from multiple stakeholders.

[0162] The unified characteristics and economic synergy of multi-source heterogeneous resources are superior. Addressing the challenge of coordinating and optimizing heterogeneous equipment, this invention uses equivalent inertia and power reserve mechanisms to quantify and evaluate various heterogeneous resources in a unified manner. Simultaneously, a multi-stage optimization model considering comprehensive resource regulation costs and frequency deviation penalties is established at the dispatching end. Under the premise of ensuring stable frequency recovery, the optimal power allocation and orderly exit timing are globally coordinated, significantly reducing the overall operating costs of the regional power grid.

[0163] In summary, compared with existing technologies, the technical solution provided by this invention is based on real-time physical operating status. By considering the physical limits and exit timing of various heterogeneous resources, it performs local control to enable the multi-source heterogeneous regional power grid to enter the frequency recovery phase. After that, it constructs a corresponding multi-stage collaborative optimization model and controls the multi-source heterogeneous regional power grid according to the solution results to achieve hierarchical collaborative optimization scheduling. This can avoid the "secondary drop" of system frequency caused by rotor speed instability and load overload in wind turbines. It can also avoid the "curse of dimensionality" caused by the increase in the number of various heterogeneous resources in traditional centralized architectures, such as communication delay, low solution efficiency, and poor data privacy. Furthermore, it can avoid the risk of "secondary drop" of system frequency caused by local frequency control methods based on a single resource or a single stage when the system frequency stabilizes and enters the recovery phase.

Claims

1. A hierarchical collaborative optimization scheduling method for multi-source heterogeneous regional power grids, characterized in that, include: Based on the real-time physical operating status of various multi-source heterogeneous resources, the equivalent inertial time constant of the multi-source heterogeneous regional power grid is determined. Based on the maximum active power reserve of various multi-source heterogeneous resources, the corresponding resource response rate coefficient is determined. Based on the system frequency of the multi-source heterogeneous regional power grid, the current frequency deviation and frequency change rate are calculated. If the frequency deviation or frequency change rate exceeds the corresponding threshold, emergency frequency support is triggered, and the multi-source heterogeneous regional power grid enters the frequency support stage: based on the equivalent inertial time constant and resource response rate coefficient, local active power control commands are generated to control the corresponding multi-source heterogeneous resources. When the system frequency crosses the lowest point and the frequency change rate is less than the first threshold, the multi-source heterogeneous regional power grid enters the frequency recovery stage; a multi-stage collaborative optimization model is constructed with the goal of minimizing the total comprehensive regulation cost of the frequency recovery stage. The alternating direction multiplier method is introduced to solve the multi-stage collaborative optimization model in a hierarchical consensus manner: the active power output plan values ​​of various multi-source heterogeneous resources at the local control node are solved independently to obtain the active power output plan adjustment values; the multi-source heterogeneous regional power grid is controlled according to each active power output plan adjustment value.

2. The hierarchical collaborative optimization scheduling method for multi-source heterogeneous regional power grids according to claim 1, characterized in that, The determination of the equivalent inertial time constant of the multi-source heterogeneous regional power grid based on the real-time physical operating status of various multi-source heterogeneous resources includes: Determine the mechanical inertia time constant and rated apparent power of various multi-source heterogeneous resources; Based on the mechanical inertia time constant and rated apparent power of each type of multi-source heterogeneous resource, determine Equivalent inertial time constant of multi-source heterogeneous regional power grid The formula is as follows: ； In the formula, This refers to the type number of the multi-source heterogeneous resources. The number of types of heterogeneous resources. For the first The mechanical inertia time constant of multi-source heterogeneous resources For the first The rated apparent power of multi-source heterogeneous resources, This serves as the baseline capacity for a multi-source heterogeneous regional power grid.

3. The hierarchical collaborative optimization scheduling method for multi-source heterogeneous regional power grids according to claim 1, characterized in that, The determination of the corresponding resource response rate coefficient based on the maximum active power reserve of various multi-source heterogeneous resources includes: according to Time of the first Maximum active power reserve that can be instantly mobilized by multi-source heterogeneous resources ,calculate Time of the first Resource response rate coefficient of multi-source heterogeneous resources The formula is as follows: ； In the formula, For the first The inherent delay in the transmission of control commands from multi-source heterogeneous resources to the output response of physical equipment.

4. The hierarchical collaborative optimization scheduling method for multi-source heterogeneous regional power grids according to claim 1, characterized in that, The formula for calculating the rate of change of frequency is as follows: ； In the formula, for Rate of change of frequency at time, for The system frequency at any given time. The system frequency at the previous sampling time. The sampling period.

5. A hierarchical collaborative optimization scheduling method for multi-source heterogeneous regional power grids according to claim 1, characterized in that, If the frequency deviation or frequency change rate exceeds the corresponding threshold, emergency frequency support is triggered, including: Update the quick response trigger status bit The formula is as follows: ； In the formula, This is the maximum permissible limit for frequency deviation. This is the maximum permissible limit for the rate of change of frequency; like Then calculate Total active power compensation deficit required by multi-source heterogeneous regional power grids at all times The formula is as follows: ； In the formula, The comprehensive droop control proportional gain coefficient for multi-source heterogeneous regional power grids. for Frequency deviation of multi-source heterogeneous regional power grids at any time For the comprehensive inertia differential control gain coefficient of a multi-source heterogeneous regional power grid, for The equivalent inertial time constant of a multi-source heterogeneous regional power grid at any given time. for Rate of change of frequency at any given moment; like If so, then increased output is required; if Then you need to reduce your output; calculate Time of the first Dynamic priority weights for multi-source heterogeneous resources in local power allocation The formula is as follows: ； ； In the formula, The number of types of heterogeneous resources. for Time of the first Resource response rate coefficient of multi-source heterogeneous resources; By dynamic priority weight Distribute total active power to compensate for deficit ,generate Time of the first Initial active power regulation command for multi-source heterogeneous resources The formula is as follows: ； Introduction Time of the first Maximum active power reserve that can be instantly mobilized by multi-source heterogeneous resources Limiting initial active power adjustment command Generate local active power control commands using the following formula: ； In the formula, for Time of the first Local active power control commands for multi-source heterogeneous resources It is a symbolic function; Through local active power control commands Real-time control Multi-source heterogeneous resources.

6. The hierarchical collaborative optimization scheduling method for multi-source heterogeneous regional power grids according to claim 1, characterized in that, The multi-stage collaborative optimization model, with the objective function of minimizing the total integrated adjustment cost during the frequency recovery phase, is constructed, including: Based on emergency frequency support, the multi-source heterogeneous regional power grid crosses the lowest system frequency point and enters the frequency recovery phase. Total comprehensive adjustment cost during the frequency recovery phase The minimum objective function is defined by the following formula: ； ； ； ； ； ； ； In the formula, , and For the first The comprehensive adjustment cost coefficient of multi-source heterogeneous resources. for Time of the first The planned adjustment value of active power output for multi-source heterogeneous resources. The frequency exceeding the limit penalty weight coefficient, for Frequency deviation of multi-source heterogeneous regional power grids at any time The permissible steady-state dead zone limit for multi-source heterogeneous regional power grids. The number of types of heterogeneous resources. for The total active power compensation deficit required by the multi-source heterogeneous regional power grid at all times. for Active power loss values ​​in multi-source heterogeneous regional power grids at any given time. For the first Minimum active power output of multi-source heterogeneous resources. For the first The maximum active power output of multi-source heterogeneous resources. for The rotational speed of the rotor in the wind turbine unit at all times. for The rotational speed of the rotor in the wind turbine unit at all times. This refers to the maximum permissible rate of change of rotor speed in a wind turbine generator set. This represents the lower limit of the safe rotational speed of the rotor in a wind turbine generator set. This refers to the upper limit of the safe rotational speed of the rotor in a wind turbine generator set. This is the lower limit of the state of charge of energy storage. for The state of charge that stores energy at all times. This represents the upper limit of the state of charge of energy storage. The end time of the frequency recovery phase State of charge of energy storage This represents the lower limit of the target range for the self-recovery of the energy storage state of charge. This represents the upper limit of the target range for the self-recovery of the energy storage state of charge.

7. A hierarchical collaborative optimization scheduling method for multi-source heterogeneous regional power grids according to claim 6, characterized in that, The method of alternating direction multipliers is introduced to solve the multi-stage collaborative optimization model in a hierarchical consensus manner: the active power output plan values ​​of various multi-source heterogeneous resources at the local control node are solved independently to obtain the active power output plan adjustment values. Controlling the multi-source heterogeneous regional power grid according to the planned adjustment values ​​of each active power output includes: Introducing Lagrange multipliers As a boundary consensus variable for coordinating various multi-source heterogeneous resources; Minimize the objective function During the iteration process, the local control node receives the boundary consensus variables from the previous iteration. The active power output plan adjustment value is updated using the following formula: ； In the formula, For the first Round iteration Time of the first The planned adjustment value of active power output for multi-source heterogeneous resources. For the first A set of local physical security constraints for multi-source heterogeneous resources; Based on the planned adjustment value of active power output The boundary consensus variable is updated using the following formula: ； ； In the formula, For the first Round iteration Boundary consensus variables at time points, This is the iteration step size factor; Updated boundary consensus variables The data is then distributed to each local control node to begin the next iteration. After the iteration terminates, a timing control instruction matrix for issuance is generated. The formula is as follows: ； In the formula, to for Time until the end of the frequency recovery phase Control commands at corresponding moments during the process It is the transpose symbol; Control commands are executed in sequence. To control commands Control each local control node; The formula for the iteration termination condition is as follows: ； In the formula, This is the limit for the convergence accuracy of the original residual in power balance. This is the convergence accuracy limit for the dual residuals of the boundary consensus variables.

8. A hierarchical collaborative optimization scheduling method for multi-source heterogeneous regional power grids according to claim 7, characterized in that, Also includes: like ,and If maintained continuously, the local control node will reset the fast response trigger status bit. .

9. A hierarchical collaborative optimization scheduling method for multi-source heterogeneous regional power grids according to claim 7, characterized in that, If each local control node does not receive an updated boundary consensus variable within m consecutive sampling periods If it fails, it will automatically exit the hierarchical consensus solution process and maintain its output according to the active power output plan adjustment value of the previous valid sampling period; among which, .

10. A hierarchical collaborative optimization scheduling method for multi-source heterogeneous regional power grids according to any one of claims 1 to 9, characterized in that, The multi-source heterogeneous resources include thermal power units, wind power units, photovoltaic power units, energy storage and flexible loads, and the real-time physical operating status includes active power, reactive power, rotor speed and energy storage state of charge.

Images