Intelligent control method for multi-energy coordinated scheduling

Abstract

The invention discloses an intelligent control method for multi-energy coordinated scheduling, belongs to the technical field of power system control, and aims to solve the technical problems that in an existing virtual synchronous generator control method, steady-state economy and transient stability are difficult to consider at the same time, and control resources cannot be accurately distributed according to disturbance positions. The method comprises the following steps: executing distributed economic dispatching in a steady state and setting a VSG as an economic parameter; the source node broadcasts and announces the disturbance; each agent calculates an electrical distance according to the power grid impedance matrix and is dynamically divided into a core group, a cooperative group or a backup group; differentiated VSG parameters are generated according to the groups so as to accurately suppress disturbance; and finally, smoothly returning to a steady state through cooperative recovery control. Unification of economical efficiency and stability is achieved, and the method has the advantages of being rapid in response, optimizing resources, avoiding secondary impact and the like.

Description

Technical Field

[0001] This invention relates to the field of power system control technology, specifically an intelligent control method for multi-energy coordinated dispatch. Background Technology

[0002] With the large-scale grid connection of distributed energy sources such as photovoltaic, wind power, and energy storage systems, the dynamic characteristics of the power system are undergoing profound changes. These distributed energy sources are mainly connected to the main grid through power electronic inverters, lacking the mechanical rotational inertia and inherent damping characteristics of traditional synchronous generator sets. This reduction in system inertia leads to a significant decrease in the frequency stability margin of the grid when subjected to power disturbances, an increase in the rate of frequency change, and an increased risk of system frequency collapse.

[0003] To address this challenge, Virtual Synchronous Generator (VSG) technology has been proposed. VSG technology, through specific control algorithms, enables power electronic inverters to simulate the electromagnetic transient characteristics of synchronous generators, actively providing virtual inertia and virtual damping to the grid, thereby effectively participating in system frequency regulation and enhancing the system's frequency support capability.

[0004] However, existing VSG control strategies still face challenges in practical applications. On the one hand, there is an inherent contradiction in the selection of VSG control parameters (such as virtual inertia and virtual damping): if large parameter values ​​are set to ensure the stability of the system under severe disturbances, unnecessary energy loss and potential reduction in dynamic response speed will occur during steady-state operation, sacrificing the system's operational economy; if small parameter values ​​are set to pursue economy, sufficient transient support cannot be provided when disturbances occur.

[0005] On the other hand, when local disturbances occur in the power grid, existing control methods typically employ a uniform response strategy for all distributed energy units (VSGs), meaning all VSGs perform the same parameter adjustments or control logic. This globally consistent response ignores the actual location of the disturbance and the electrical proximity of each energy unit to the disturbance source within the grid topology. The result may be that the unit closest to the disturbance source and most in need of strong support responds insufficiently, while units far away and almost unaffected undergo unnecessary power regulation. This not only wastes valuable control resources but may even induce unnecessary power oscillations in the system.

[0006] Furthermore, the backoff mechanism of control parameters also has flaws when the system successfully suppresses disturbances and needs to recover from emergency control to normal economic operation. If the VSG parameters suddenly jump back from the emergency support value to the economic mode value, this parameter abrupt change will act as a new disturbance, causing a secondary impact on the system, triggering oscillations in frequency and power, which is not conducive to the smooth recovery of the system.

[0007] In summary, existing virtual synchronous generator (VSG) control strategies have the following problems in practical applications:

[0008] There is a contradiction in the selection of control parameters: large parameter values ​​sacrifice economy, while small parameter values ​​cannot provide sufficient transient support;

[0009] The global unified response strategy ignores the relationship between the location of the disturbance and the electrical distance, resulting in a waste of control resources and even causing oscillations;

[0010] The parameter rollback mechanism is imperfect and can easily cause secondary impacts. Summary of the Invention

[0011] To address the technical problems of existing virtual synchronous generator control methods, which struggle to balance steady-state economy and transient stability and cannot accurately allocate control resources based on disturbance location, this invention provides an intelligent control method for multi-energy coordinated scheduling.

[0012] The present invention discloses an intelligent control method for multi-energy coordinated scheduling, applicable to a power system comprising multiple distributed energy units, each energy unit being configured with a local controller and a distributed intelligent agent. The method includes the following steps:

[0013] S101. Steady-state operation steps: In steady-state economic operation mode, the distributed intelligent agent network determines the optimal active power reference value of each energy unit through distributed optimization calculation, and sets the virtual synchronous generator VSG control parameters of each local controller to economic mode parameters.

[0014] S102, Disturbance Triggering Step: When the grid frequency change rate is detected to exceed the preset threshold, the distributed intelligent agent corresponding to the energy unit that detected the disturbance acts as the disturbance source node and broadcasts a disturbance notice containing its node identifier.

[0015] S103, Group self-organization steps: Each distributed intelligent agent that receives the disturbance notification calculates the electrical distance between itself and the disturbance source node according to the pre-stored system grid impedance matrix, and dynamically assigns itself to the core response group, collaborative support group or backup standby group based on the comparison results of the electrical distance with at least two preset thresholds.

[0016] S104. Parameter reconfiguration step: Each distributed intelligent agent generates the corresponding VSG control parameters according to its response group and sends them to the local controller so that the energy units of different groups can perform differentiated disturbance suppression control.

[0017] S105. Collaborative recovery step: After the disturbance is suppressed, all distributed intelligent agents collaboratively execute recovery control to eliminate the steady-state frequency deviation of the system and smoothly restore the VSG control parameters to the economic mode parameters, ultimately enabling the system to return to the steady-state economic operation mode.

[0018] Preferably, the electrical distance is calculated by calculating the reciprocal of the absolute value of the mutual impedance elements in the power grid impedance matrix corresponding to its own identifier and the identifier of the disturbance source node.

[0019] Preferably, in the group self-organization step, the at least two preset thresholds include a core response threshold and a collaborative support threshold, and satisfy the following: Core Response Threshold Collaborative support threshold;

[0020] The allocation rules are as follows:

[0021] If the electrical distance The core response threshold is then assigned to the core response group;

[0022] If the core response threshold The electrical distance If the collaborative support threshold is reached, it will be assigned to the collaborative support group;

[0023] If the electrical distance If the collaborative support threshold is reached, it will be assigned to the backup standby group.

[0024] Preferably, in the parameter reconstruction step:

[0025] For the distributed intelligent agents assigned to the core response group, their VSG control parameters are set to the preset maximum support mode parameters;

[0026] For distributed intelligent agents assigned to the backup standby group, their VSG control parameters are maintained at the economic mode parameters.

[0027] For the distributed intelligent agent assigned to the collaborative support group, its VSG control parameters are generated by interpolation calculation, and the interpolation calculation makes the generated parameter values ​​between the economic mode parameter and the maximum support mode parameter, and negatively correlated with the electrical distance.

[0028] Preferably, the VSG control parameters of the collaborative support group are generated using the following interpolation formula:

[0029]

[0030] In the formula, For the new VSG control parameters, including virtual inertia or virtual damping coefficient; These are the corresponding values ​​under the economic model parameters; This refers to the corresponding value under the maximum support mode parameter;

[0031] These are the weighting coefficients, and

[0032]

[0033] For electrical distance, To support the threshold for collaboration, The core response threshold.

[0034] Preferably, the elimination of the system steady-state frequency deviation in the collaborative recovery step is achieved through distributed secondary frequency recovery control, which includes:

[0035] Each distributed intelligent agent integrates the local frequency deviation to generate a local secondary adjustment signal;

[0036] The consensus algorithm is used to converge the local secondary adjustment signals of all agents to a unified global adjustment signal.

[0037] The global adjustment signal is superimposed on the optimal active power reference value of each energy unit.

[0038] Preferably, the smooth recovery of the VSG control parameters in the collaborative recovery step is achieved through the following process:

[0039] Each distributed intelligent agent dynamically calculates the real-time VSG parameters based on a return control function that decays over time;

[0040] The calculation method for the real-time VSG parameters is as follows:

[0041]

[0042] In the formula, Here are the real-time VSG parameters at time t. The return control function has a value that decays from 1 to 0 over time.

[0043] Preferably, the distributed optimization calculation in the steady-state operation step is a consensus-based distributed economic scheduling algorithm. Its goal is to minimize the total operating cost of the system under the conditions of satisfying the system power balance and the output constraints of each unit, and to make the marginal cost of all distributed intelligent agents converge to a unified value, thereby calculating the optimal active power reference value of each unit in reverse.

[0044] The beneficial effects of this invention are:

[0045] 1. This method uses the system grid impedance matrix to calculate the electrical distance between each unit and the disturbance source, and dynamically divides them into core response group, coordinated support group and backup standby group. By configuring different VSG control parameters for different groups (for example, configuring the maximum support mode parameter for the core response group and maintaining the economic mode parameter for the backup standby group), the inertial support and damping response are precisely concentrated in the area most affected by the disturbance, avoiding ineffective response of the whole network and improving the efficiency of frequency suppression.

[0046] 2. This invention enables the system to operate in the optimal economic state through distributed economic scheduling calculation. When a disturbance occurs, only the units that are electrically close are activated to perform high-intensity support, while the units that are electrically far away maintain their economic mode parameters without incurring unnecessary control costs. This strategy minimizes the global economy sacrificed to cope with local disturbances.

[0047] 3. After parameter reconstruction, this invention performs distributed secondary frequency recovery and smooth return of VSG parameters in parallel. The former eliminates the steady-state frequency deviation of the system by negotiating a unified adjustment signal through a consensus algorithm, ensuring that the frequency is accurately restored to the rated value. The latter realizes the smooth transition from emergency parameters to economic mode parameters through a return control function that decays over time, effectively avoiding secondary oscillations caused by parameter mutations. Attached Figure Description

[0048] Figure 1 This is a schematic diagram of the method flow of the present invention;

[0049] Figure 2 This is a schematic diagram of the steady-state economic operation mode of the present invention;

[0050] Figure 3 This is a schematic diagram of the disturbance event detection and triggering mechanism of the present invention;

[0051] Figure 4 This is a schematic diagram of the collaborative recovery and mode return phase of the present invention. Detailed Implementation

[0052] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0053] It should be noted that, unless otherwise specified, the embodiments and features described in the present invention can be combined with each other.

[0054] The present invention will be further described below with reference to the accompanying drawings and specific embodiments, but this is not intended to limit the scope of the invention.

[0055] Specific Implementation Method 1: The following is combined with... Figures 1 to 4 This embodiment describes a smart control method for multi-energy coordinated dispatch, applied to a power system comprising multiple distributed energy units. Each energy unit is equipped with a local controller and a distributed smart agent. The method includes the following steps:

[0056] S101. Steady-state operation steps: In steady-state economic operation mode, the distributed intelligent agent network determines the optimal active power reference value of each energy unit through distributed optimization calculation, and sets the virtual synchronous generator (VSG) control parameters of each local controller to economic mode parameters; specifically, the distributed intelligent agent solves an optimization problem with the objective of minimizing the total operating cost of all energy units:

[0057] This optimization satisfies the system power balance constraint. (in The calculation is performed under the conditions of total system load and upper and lower limits of output of each unit. The calculated optimal active power is used as a reference value. The data is then sent to the local controller. At this point, the VSG control parameters, i.e., the virtual inertia, in the local controller... and virtual damping coefficient The economic model parameters are set to the preset parameters. .

[0058] S102, Disturbance Triggering Step: When the grid frequency change rate is detected to exceed the preset threshold, the distributed intelligent agent corresponding to the energy unit that detected the disturbance acts as the disturbance source node and broadcasts a disturbance notice containing its node identifier.

[0059] Each local controller in the system monitors the rate of change of the grid frequency at its connection point in real time. When any local controller detects that the absolute value of this value exceeds a preset frequency change rate threshold... When this occurs, a disturbance event is determined to have taken place. The triggering condition for this event can be expressed as:

[0060] ;

[0061] Where t represents time. After the event is triggered, the distributed intelligent agent corresponding to the local controller (denoted as the disturbance source node)... Immediately generate a disturbance notification containing the disturbance source node. It is a unique identifier and is broadcast to all other distributed intelligent agents through the communication network.

[0062] S103, Group self-organization steps: Each distributed intelligent agent that receives the disturbance notification calculates the electrical distance between itself and the disturbance source node according to the pre-stored system grid impedance matrix, and dynamically assigns itself to the core response group, collaborative support group or backup standby group based on the comparison results of the electrical distance with at least two preset thresholds.

[0063] Upon receiving a disturbance notification, all distributed intelligent agents in the network immediately initiate group self-partitioning computation. Each distributed intelligent agent (denoted as node j) calculates its partitioning based on the pre-stored system grid impedance matrix. Calculate the relationship between itself and the disturbance source node. electrical distance between The calculation formula is as follows:

[0064] ;

[0065] in, For the j-th and j-th impedance matrices The elements of the column. Then, node j, based on this electrical distance and two preset thresholds, and It categorizes itself into one of the three response groups: the core response group. Collaborative Support Group or reserve standby group .

[0066] absolute value of mutual impedance The larger the value, the stronger the electrical coupling between the two nodes. In this embodiment of the invention, the reciprocal of the absolute value of the mutual impedance is defined as the electrical distance. Therefore, a smaller electrical distance value corresponds to a stronger electrical coupling, indicating a greater susceptibility to disturbances.

[0067] S104. Parameter reconfiguration step: Each distributed intelligent agent generates the corresponding VSG control parameters according to its response group and sends them to the local controller so that the energy units of different groups can perform differentiated disturbance suppression control.

[0068] After group division is completed, each distributed intelligent agent generates a new set of VSG parameters for its corresponding local controller based on its group. And instruct it to apply immediately. The parameter reconstruction rules are as follows: if node j is in the core response group Its parameters are set to the preset maximum support value. ;

[0069] If node j belongs to the collaborative support group Its parameters are calculated using a weighted average based on electrical distance; if node j belongs to the backup standby group Its parameters remain the same as those of the economic model. constant.

[0070] S105. Collaborative recovery step: After the disturbance is suppressed, all distributed intelligent agents collaboratively execute recovery control to eliminate the steady-state frequency deviation of the system and smoothly restore the VSG control parameters to the economic mode parameters, ultimately enabling the system to return to the steady-state economic operation mode.

[0071] The control comprises two parallel processes: first, eliminating the steady-state frequency deviation of the system through distributed integral control to generate a secondary power adjustment command; second, controlling the VSG parameters of each unit. The reconstructed parameter values ​​are smoothly returned to the economic model parameter values ​​according to a preset time decay function. After the system frequency, voltage, and other state variables have stabilized within the preset normal range for a period of time, the system returns to a steady state, and the method flow returns to step S101.

[0072] The system architecture is logically divided into three layers: the physical layer, the control layer, and the information decision layer.

[0073] Physical layer: Multi-energy power system

[0074] The physical layer is the controlled object of the method of this invention, and it constitutes a multi-energy power system. This system includes:

[0075] Multiple distributed energy units, such as photovoltaic power generation units, wind power generation units, energy storage units (such as battery energy storage systems), and other controllable power generation equipment, are connected to the power grid via power electronic inverters. The power electronic inverter is the final executor of control commands in the method of this invention, regulating the power output of the energy units by controlling the on / off switching of its internal power electronic switches (such as IGBTs).

[0076] Multiple electrical loads, including adjustable and non-adjustable loads.

[0077] The power transmission and distribution network includes equipment such as power lines, transformers, and switches. These devices constitute the physical connection between all energy units and loads within the system, and its network topology and line parameters are fixed.

[0078] Control layer: Local controller

[0079] At each controllable distributed energy unit, a local controller is configured. The local controller is a physical hardware device responsible for performing low-level, real-time control tasks. A specific embodiment of the local controller includes:

[0080] Data Acquisition Unit: This unit consists of measuring sensors such as voltage transformers and current transformers, along with their signal conditioning circuits. Its function is to acquire the three-phase voltage and current signals at the grid connection point of the energy unit in real time and at high frequency.

[0081] State Calculation Unit: This unit receives signals from the data acquisition unit and calculates (e.g., via a phase-locked loop, PLL) the real-time electrical state quantities of the grid connection point. These state quantities include at least the grid frequency f and the rate of frequency change. Output active power and output reactive power .

[0082] Control Algorithm Execution Unit: This unit is the core of the local controller and is typically composed of a digital signal processor (DSP) or a microcontroller (MCU). It implements the control algorithm based on a virtual synchronous generator (VSG). This unit receives real-time state data from the state calculation unit and active power reference values ​​from the information decision layer. The virtual inertia J and virtual damping coefficient D are calculated and discretized iteratively according to the mathematical model of VSG (as in the formulas in the previous chapters) to generate the voltage and frequency reference signals required to control the inverter.

[0083] Parameter storage and management unit: This unit is a non-volatile memory used to store multiple sets of VSG control parameters, including at least one set of economic mode parameters. and a set of maximum support mode parameters This unit switches between different parameter groups according to the instructions of the information decision layer and provides the currently effective parameters to the control algorithm execution unit.

[0084] Pulse Width Modulation (PWM) Signal Generation Unit: This unit receives voltage and frequency reference signals from the control algorithm execution unit and generates PWM control signals for directly driving the power switching transistors inside the power electronic inverter.

[0085] Communication Interface Unit: This unit provides a physical communication interface (such as an Ethernet or CAN bus interface) to enable bidirectional data communication between the local controller and the distributed intelligent agent of the information decision layer.

[0086] Information Decision Layer: Distributed Intelligent Agent Network

[0087] The information decision-making layer consists of distributed intelligent agents, each corresponding to a local controller. Each distributed intelligent agent is a logical functional entity that can run as a software program on the processor inside the local controller or on an independent computing device that is communicatively connected to the local controller. All distributed intelligent agents form a peer-to-peer information interaction and decision-making system through a communication network. A specific implementation of a distributed intelligent agent includes the following functional units:

[0088] Communication Unit: Responsible for handling information transmission and reception with other distributed intelligent agents. This includes broadcasting disturbance notices, receiving disturbance notices, and exchanging necessary data (such as marginal cost, state information, etc.) during specific algorithm execution (such as economic scheduling or coordinated recovery).

[0089] Economic Dispatch Calculation Unit: This unit stores the cost function of its corresponding energy unit. During steady-state operation, it interacts with the communication units of other agents in the network to execute a consensus-based distributed optimization algorithm, calculating the optimal active power reference value that minimizes the total operating cost of the system. It then sends the data to the control algorithm execution unit of the local controller.

[0090] Topology Information Storage and Calculation Unit: This unit pre-stores the network topology information of the entire power system, specifically represented by the system's impedance matrix. When a disturbance notification is received, the unit is activated based on the identifier of the disturbance source node. And its own identifier j, extract the corresponding element from the impedance matrix. And calculate the electrical distance .

[0091] Group Decision Unit: This unit receives electrical distance values ​​from the topology information storage and calculation unit. Internally, it has embedded logical rules for group partitioning, which involves comparing the input electrical distance values ​​with preset thresholds. and Compare the responses and output the response group to which the agent belongs. or ).

[0092] Control Parameter Generation Unit: This unit generates new VSG parameters that the local controller needs to execute based on the group affiliation output by the group decision unit. For those belonging to and The agent, this unit directly outputs the pre-stored data. or For those belonging to The agent unit executes a preset weighted calculation formula to generate specific parameters that fall between the maximum value and the economic value. The generated parameters are then sent to the parameter storage and management unit of the local controller.

[0093] Recovery Strategy Execution Unit: This unit is activated after parameter reconstruction is completed and the system enters the recovery phase. It is responsible for performing distributed secondary frequency recovery calculations and generating dynamic decay signals to control the smooth return of VSG parameters from the reconstructed values ​​to the economic values, ensuring a stable transition of the system back to normal operation.

[0094] See attached document Figure 2 This mode is the normal operating state of the method of the present invention when the power system is not disturbed.

[0095] Step 1: Implementation of Steady-State Economic Operation Mode. In steady-state economic operation mode, the primary goal of the system is to adjust the active power output of each distributed energy unit to meet the total system load demand while minimizing the total operating cost of all units. This process is collaboratively completed by the distributed intelligent agent network of the information decision-making layer.

[0096] Operating Objectives and Constraints: The control problem in this mode is constructed as a constrained optimization problem. Its objective function aims to minimize the total operating cost of all N controllable energy units within the system.

[0097] ;

[0098] Where i is the index of the energy unit, and N is the total number of units. Let be the output active power of the i-th unit. Let be the cost function of the i-th unit, which is a quadratic function of the output power: ,in It is a predetermined cost coefficient, stored in the i-th distributed intelligent agent.

[0099] The optimization process must meet the following constraints:

[0100] Power balance constraint: The total active power output of all energy units in the system must be equal to the current total load of the system.

[0101] ;

[0102] in, This represents the total active power load of the system.

[0103] Unit output constraint: The output power of each energy unit must be within its physically permissible upper and lower limits.

[0104] ;

[0105] in, and These are the minimum and maximum output active power of the i-th unit, respectively.

[0106] Distributed optimization process based on consensus algorithm:

[0107] To avoid using a centralized controller, this embodiment of the invention employs a distributed method based on a consensus algorithm to solve the aforementioned optimization problem. The core of this method is to enable all distributed intelligent agents to reach a consensus on a unified system incremental cost.

[0108] The process is implemented as follows:

[0109] First, before the iteration begins ( Each distributed intelligent agent i calculates its initial marginal cost based on its current output power. Marginal cost is defined as the first derivative of the cost function with respect to output power:

[0110] ;

[0111] in, This represents a quantity related to the current output state or processing node i. This is right The derivative of the cost function represents the cost function. With output The rate of change Used to represent the derivative of a quadratic function. It is a coefficient related to node i. It is a constant term, representing the offset or intercept.

[0112] Next, the system enters the iterative calculation process. In each discrete iteration step... Each distributed intelligent agent i performs the following operations:

[0113] Information broadcast: It will be broadcast in the... Marginal cost value obtained by step calculation It sends data to all its predefined neighboring distributed intelligent agents via a communication network.

[0114] Information reception and update: Receive information from its neighbor set (This set includes the marginal cost values ​​of agent i itself and all its directly communicating neighboring agents.) Then, agent i updates its own marginal cost using a weighted average based on all received values, obtaining... :

[0115] ;

[0116] in, is the state or output value of node i at time k, and j is the index of the neighbor agent. It is the set of neighbors of node i. It represents the connection weight between node i and its neighbor node j. Is node j at time... The status or output value.

[0117] The above iterative process continues until the marginal cost of all distributed intelligent agents in the network converges to a common value. The convergence criterion is that in two consecutive iterations, the change in marginal cost of each agent i is less than a preset minimum threshold. That is, when Once the convergence condition is met, each distributed intelligent agent i will determine the marginal cost of reaching a final consensus. In reverse, the corresponding optimal local active power reference value is calculated. :

[0118] ;

[0119] Calculated It is sent to its corresponding local controller as the active power output command of the energy unit under steady state.

[0120] VSG parameter configuration in economic mode:

[0121] While performing distributed economic scheduling, each distributed intelligent agent sends an instruction to its corresponding local controller to configure the control parameters of its internal VSG.

[0122] In steady-state economic operation mode, the local controller is instructed to select the economic mode parameter set pre-stored in its parameter storage and management unit. Its characteristic is its virtual inertia. and virtual damping coefficient All values ​​are set to relatively small values. The purpose of this setting is to reduce the autonomous response strength of the energy unit to small, normal frequency fluctuations in the power grid, thereby ensuring that the unit's actual power output accurately follows the optimal power reference value calculated by economic dispatch. This ensures that the overall economic efficiency of the system is given priority.

[0123] See attached document Figure 3 This mechanism serves as the entry point for the system to switch from a steady-state economic operation mode to an emergency response mode.

[0124] Step 2: Detection and Triggering Mechanism of Disturbance Events

[0125] Real-time monitoring of critical electrical quantities:

[0126] Each local controller in the system continuously samples the grid voltage waveform at its connection point at high frequency through its internal data acquisition unit.

[0127] The local controller's state calculation unit incorporates a phase-locked loop (PLL) module. This PLL module uses the acquired voltage signal as input to track the phase angle of the grid voltage in real time. By differentiating this phase angle, the accurate real-time angular frequency of the power grid can be obtained. This leads to the calculation of the power grid frequency. .

[0128] Rate of Change of Frequency (RoCoF) It is through the calculated power grid frequency It is obtained by performing time-difference operations. Within a discrete control cycle, its calculation method is as follows:

[0129] ;

[0130] in, This is the calculated frequency value for the current control cycle. It is the frequency calculation value of the previous control cycle. This refers to the duration of the control cycle. This is to avoid measurement noise or non-critical fluctuations affecting the calculated value. A spike in the value triggers a false trigger, prompting the state calculation unit to further calculate the value. The original value is filtered by a low-pass filter to obtain a smooth frequency change rate monitoring value that reflects the true trend of the disturbance.

[0131] The specific definition of the event triggering conditions:

[0132] In each control cycle, the local controller's control algorithm execution unit compares the absolute value of the filtered frequency change rate monitoring value with a preset frequency change rate threshold. Compare them.

[0133] The threshold This is a key setting parameter, and its value is set based on the following: this value is greater than the value caused by normal load fluctuations under normal operating conditions. The maximum value, but less than that caused by severe faults (such as generator disconnection, large-capacity load switching, or line faults). Initial minimum value.

[0134] The local controller determines that a disturbance event has occurred if and only if the following conditions are met:

[0135] ;

[0136] in, It is the absolute value of the filtered rate of change monitoring value obtained by the i-th local controller at time t.

[0137] Once the condition is met, the local controller immediately switches its internal status flag from steady state to disturbance and generates an internal trigger signal.

[0138] Generation and broadcasting of disturbance notices:

[0139] Once the local controller generates an internal trigger signal, this signal is immediately transmitted to its corresponding distributed intelligent agent. The distributed intelligent agent that receives this trigger signal (this agent is then identified as the source node of the disturbance)... Immediately generate a standard-format disturbance notification message data packet in its communication unit. The data structure of this packet must contain at least the following two key fields:

[0140] Source node identifier (source_id): Records the unique address or number of the disturbance source node in the network. This field is a necessary input for subsequent topology-aware calculations.

[0141] Event timestamp: Records the precise time when the event was triggered. This field can be used for subsequent event analysis or time-series determination of multiple events.

[0142] After generating this data packet, the distributed intelligent agent The system immediately broadcasts the data packet to the communication network via its communication unit. All other distributed intelligent agents in the network will receive this disturbance notification, ensuring that the entire system can learn about the disturbance and its initial location in the shortest possible time.

[0143] This process is triggered when the distributed intelligent agent receives a disturbance notification.

[0144] Step 3: Dynamic Partitioning of Response Groups Based on Topology Awareness

[0145] The principle and implementation of electrical distance calculation:

[0146] In this embodiment of the invention, electrical distance is used to quantify the correlation strength between any two nodes in a power grid in terms of their electrophysical characteristics. Specifically, it characterizes the extent to which a disturbance in one node (the source node) affects another node (the responding node).

[0147] Calculation principle:

[0148] The calculation of electrical distance is based on the nodal impedance matrix of the power system. The node impedance matrix is ​​the node admittance matrix. inverse matrix ( ), each of its elements All of these have clear physical meaning. According to the basic superposition principle of power networks, when a current is injected into node i... At that time, the voltage change caused at node j The following relationship must be satisfied:

[0149] ;

[0150] in, It is the element in the j-th row and i-th column of the node impedance matrix, and is called the mutual impedance between node j and node i.

[0151] From this relationship, we can see that mutual impedance The magnitude of directly reflects the degree to which the current change at node i affects the voltage stability at node j. Its absolute value The larger the value, the stronger the electrical coupling between the two nodes, and the greater the impact of the disturbance propagating from node i to node j.

[0152] Therefore, in this embodiment of the invention, the absolute value of the mutual impedance is defined as the difference between node j and the disturbance source node. electrical distance between :

[0153] ;

[0154] The electrical distance defined here is a scalar value that can be directly used for subsequent numerical comparisons and grouping.

[0155] Implementation method:

[0156] The calculation of electrical distance is divided into two stages: offline preparation and online calculation, to ensure a rapid response when disturbances occur.

[0157] Offline preparation phase:

[0158] This phase is executed once before the system is put into operation or after the system topology has changed.

[0159] Construct the node admittance matrix ( First, based on the network topology data of the power system, including the series impedance and parallel admittance of all transmission lines and transformers, as well as the ground admittance of generators and loads, the nodal admittance matrix of the system is constructed. The construction of this matrix follows standard power system analysis methods.

[0160] Calculate the node impedance matrix ( Subsequently, the constructed node admittance matrix was... Perform matrix inversion to obtain the nodal impedance matrix. This calculation is typically performed by a back-end computing platform or a systems engineer.

[0161] Matrix storage: The calculated complete node impedance matrix The topology information is distributed and stored within the topology storage and computing unit of each distributed smart agent. This means that each distributed smart agent has a complete topology view of the entire power grid.

[0162] Online computation phase:

[0163] This phase is triggered in real time after any distributed intelligent agent receives a disturbance notification.

[0164] Information Extraction: Distributed intelligent agent j extracts the identifier of the disturbance source node from the received disturbance notification data packet. .

[0165] Matrix query: Agent j accesses the node impedance matrix stored locally. .

[0166] Numerical acquisition: Agent j uses its own identifier j as the row index, and extracts the identifier of the perturbation source node. As a column index, directly from Search for and retrieve the corresponding complex form of the mutual impedance element in the matrix. .

[0167] Calculate the absolute value: the complex number obtained by agent j. Perform a modulo operation, that is, calculate its absolute value, to obtain the final scalar electrical distance value. .

[0168] By placing the computationally complex matrix inversion process offline, while the online stage only requires a single fast matrix lookup and modulo operation, this invention ensures that the electrical distance calculation can be completed in a very short time, meeting the response speed requirements of emergency control. The calculated electrical distance value is then transmitted to the group decision unit for the next step of group division.

[0169] After calculating the electrical distance, the process enters the response group division phase.

[0170] Response group division rules and logic:

[0171] Within the group decision-making unit of each distributed intelligent agent j, two scalar thresholds for group partitioning are pre-stored: the core response threshold. and collaborative support threshold These two thresholds were derived from offline analysis and simulation tuning of the power system topology and dynamic characteristics, and satisfy... .

[0172] When the topology information storage and computing unit of the distributed intelligent agent j completes the electrical distance After calculation, the value is immediately transmitted to the group decision-making unit. The group decision-making unit, based on the following set of mutually exclusive conditional judgments, classifies agent j into one of three predefined response groups:

[0173] Condition 1: Core Response Group ( The division of )

[0174] The group decision unit will input the electrical distance With core response threshold Compare them. If the following inequalities are satisfied:

[0175] ;

[0176] Then, the distributed intelligent agent j is determined to belong to the core response group. The energy units identified by this condition are the set of units that are electrically closest to the disturbance source.

[0177] Condition 2: Collaborative Support Group ( (division)

[0178] If condition one is not met, the group decision unit will continue to determine the input electrical distance. Does it fall between the two thresholds? If the following inequality is satisfied:

[0179] ;

[0180] If so, the distributed intelligent agent j is determined to belong to the collaborative support group. Energy units identified under this condition are electrically located at a moderate distance from the disturbance source. They are less affected by disturbances than the core response group, and their primary task is to collaborate with the core response group to suppress disturbances and prevent them from further propagating and expanding within the power grid.

[0181] Condition 3: Backup standby group ( (division)

[0182] If neither condition one nor condition two is met, i.e., the input electrical distance Greater than the collaborative support threshold ;

[0183] ;

[0184] Then, the distributed intelligent agent j is determined to belong to the backup standby group. The energy unit identified by this condition is electrically furthest from the disturbance source.

[0185] Disturbances have significantly attenuated by the time they reach these units, minimizing their impact on their operational status. They are grouped into this category to prevent over-response across the entire network and to preserve their regulation capabilities as backup resources for any subsequent necessary adjustments or recovery processes.

[0186] After completing the above logical judgment, the group decision-making unit will output a definite group identifier. or This identifier will serve as the direct basis for parameter generation in the next step, controlling the parameter generation unit. The entire partitioning process is based entirely on local computation, requiring no additional communication with other agents, thus ensuring rapid decision-making.

[0187] The dynamic self-organizing property of group partitioning:

[0188] The response group partitioning process in this embodiment of the invention has the characteristics of dynamic adaptability and self-organization.

[0189] The dynamic adaptability of this process is reflected in the fact that the final composition of the response group is not a pre-defined static set, but is entirely determined by the actual location of the disturbance event. For example, when a disturbance event is detected at node A of the system, all distributed intelligent agents within the system will use node A as a reference to calculate their respective electrical distances, thus forming the first type of grouping result. In this result, nodes B, C, etc., which are closest to node A in electrical distance, are assigned to the core response group. If another independent disturbance event is detected at node D of the system, all distributed intelligent agents will use node D as a new reference to recalculate the electrical distances, forming a second type of grouping result that is completely different from the previous one. In this result, nodes E, F, etc., which are closest to node D in electrical distance, will be assigned to the core response group. Therefore, this method can dynamically and adaptively generate a response structure that best matches the propagation path of each disturbance based on its specific spatial location.

[0190] The self-organizing nature of this process is reflected in the fact that the global, ordered group structure is spontaneously formed by each distributed intelligent agent through the execution of unified local rules, without any central coordination or instruction allocation. The specific formation process is as follows: the disturbance source node broadcasts only a trigger signal containing its own location identifier to the network; upon receiving this single message, each distributed intelligent agent in the network independently and in parallel performs electrical distance calculations and group partitioning logic judgments; each agent determines its own group affiliation solely based on its own calculation results, without needing to negotiate with other agents or wait for centralized instructions.

[0191] This distributed decision-making mechanism enables a globally consistent and spatially rational response group structure to emerge from a large number of parallel, independent local decision-making actions. This feature avoids the communication bottlenecks and single points of failure risks present in centralized control architectures, and significantly reduces the time delay between disturbance detection and the formation of an effective response structure.

[0192] After the response group is divided, each distributed intelligent agent immediately performs this step to generate and distribute differentiated control parameters to its corresponding local controller.

[0193] Step 4: Reconstructing the control parameters for group differentiation

[0194] The core of this step lies in configuring different virtual synchronous generator (VSG)-based control parameters—namely, virtual inertia J and virtual damping coefficient D—for energy units with different response levels, based on the group affiliation determined in the previous stage. This results in a spatially precisely distributed, hierarchical dynamic response. This process is executed by the control parameter generation unit of the distributed intelligent agent. This unit receives the group identifier from the group decision-making unit and executes the corresponding parameter generation logic based on that identifier.

[0195] Core response group parameter reconstruction strategy:

[0196] If the identifier output by the group decision unit of distributed intelligent agent j is the core response group ( If the control parameter generation unit is triggered to execute the maximum support response strategy, then the unit retrieves a set of preset maximum support mode parameters from its internal memory. .in, It is the maximum effective virtual inertia value that can be safely provided based on the physical hardware of the j-th energy unit (such as the inverter overload capacity and the instantaneous power throughput capacity of the energy storage unit). It is the maximum effective virtual damping coefficient value that can be provided.

[0197] The newly generated control parameters are set as follows:

[0198] ;

[0199] in, This represents the new control parameters associated with node j. This parameter configuration enables the energy units in the core response group to provide the strongest inertial support to resist abrupt frequency changes and to provide maximum damping to rapidly decay frequency oscillations, acting directly on the root location of the disturbance.

[0200] Parameter reconfiguration strategy of the collaborative support group:

[0201] If the identifier output by the group decision unit of distributed intelligent agent j is the collaborative support group ( If this occurs, the control parameter generation unit is triggered to execute the weighted support response strategy. The parameter values ​​generated by this strategy are between those of the economic model parameters. and maximum support mode parameters An interpolation value between the two, the specific value of which is related to the electrical distance of the unit. Related. The calculation formula is as follows:

[0202] ;

[0203] ;

[0204] in, and This refers to the aforementioned grouping threshold. This calculation method ensures that the response strength of energy units within a collaborative support group is inversely proportional to their electrical distance from the disturbance source. The closer the electrical distance (i.e., ... The closer The closer the parameter value is to the maximum support mode parameter, the greater the electrical distance (i.e., The closer The closer the parameter value is to the economic model parameter, the better.

[0205] Parameter maintenance strategy for backup standby groups:

[0206] If the identifier output by the group decision unit of distributed intelligent agent j is the backup standby group ( If the control parameter generation unit is triggered, the parameter maintenance strategy will be executed.

[0207] The unit retrieves economic model parameters from its internal memory. The newly generated control parameters are set as follows:

[0208] ;

[0209] This operation essentially keeps the current parameters unchanged, ensuring that the units with the greatest electrical distance do not participate in the inertial and damped response to disturbances, thus preserving their full adjustment margin for subsequent system recovery.

[0210] The instantaneous execution mechanism generated by the parameters:

[0211] The new parameter set was determined in the control parameter generation unit. Then, the parameter set is encapsulated in a control command and sent to its corresponding local controller through the communication unit of the distributed intelligent agent.

[0212] Upon receiving the instruction, the local controller's internal parameter storage and management unit immediately uses the received new parameter values. The parameter registers or memory addresses currently being used by the VSG control algorithm are overwritten. Starting from the next control cycle, the local controller's control algorithm execution unit will automatically use this updated set of parameters when iteratively calculating the VSG dynamic equations. This seamless parameter switching process ensures that changes to the response strategy take effect in the shortest possible physical time (one communication delay plus one control cycle), achieving instantaneous and differentiated responses to disturbances.

[0213] See attached document Figure 4 After the system has initially suppressed the disturbance through group-differentiated parameter reconstruction, it enters this step to fully restore the system state to normal and prepare to return to the economic operation mode.

[0214] Step 5: Cooperative Recovery and Mode Return This step consists of two parallel control processes: distributed secondary frequency recovery control and smooth return based on virtual synchronous generator (VSG) parameters.

[0215] The goal of distributed secondary frequency recovery control is to eliminate the steady-state frequency deviation that may result from disturbances and accurately restore the system frequency to its rated value (e.g., 50Hz). This process is jointly executed by all distributed intelligent agents.

[0216] In the recovery strategy execution unit of each distributed intelligent agent i, an integrator is set up. This integrator continuously calculates the locally measured grid angular frequency. With the rated angular frequency The deviation between the two is calculated, and the deviation is integrated to generate a local secondary adjustment signal. :

[0217] ;

[0218] in, This is the starting point for restoring control. It is the preset integral gain coefficient. It represents the actual frequency, speed, or value calculated at time t according to a certain rule.

[0219] To ensure consistency in the regulatory actions across the entire system, all distributed intelligent agents execute a consensus algorithm similar to that used in steady-state economic scheduling to evaluate the secondary regulation signals they have calculated locally. Iterative negotiation is conducted until all agents in the network agree on a unified secondary adjustment signal. A consensus was reached.

[0220] Once consensus is reached, each distributed intelligent agent i updates the active power reference value of its corresponding local controller, which is the unified secondary adjustment signal. Superimposed on the original power reference value Above:

[0221] ;

[0222] Updated active power reference value The command is sent to the local controller for execution. This process continues, adjusting the active power output of all units until the system frequency deviation is completely eliminated.

[0223] Smooth return mechanism for VSG parameters:

[0224] In parallel with the secondary frequency recovery, to avoid secondary disturbances caused by the VSG parameters abruptly returning from the reconstructed emergency value to the economic value, this embodiment of the invention adopts a smooth return mechanism.

[0225] In the recovery strategy execution unit of each distributed intelligent agent j, a return control function that decays over time is defined. :

[0226] ;

[0227] Where t is the current time, and e represents the base of the natural logarithm, approximately equal to , To restore control at the start time, This is a preset recovery time constant, which determines the rate at which parameters are returned.

[0228] At any time t during the recovery process The distributed intelligent agent j calculates the VSG parameters that its corresponding local controller should currently use based on this function. The calculation formula is as follows:

[0229] ;

[0230] ;

[0231] in, These are the economic model parameters for this unit. This represents the new control parameters associated with node j.

[0232] The distributed intelligent agent j performs the above calculations in each control cycle and returns the calculated real-time parameter values. The data is then sent to the local controller. This mechanism ensures a continuous and smooth transition of the VSG's virtual inertia and damping coefficient from emergency values ​​to economic values.

[0233] Recovery completion determination and mode switching:

[0234] The distributed intelligent agent continuously monitors the system status to determine whether the collaborative recovery process has been completed. The criteria for this determination include:

[0235] The system frequency is within a preset duration. Within this range, it is always maintained within a very small tolerance range around the rated value. Within.

[0236] Return control function The value has decayed to a preset threshold close to zero. The following indicates that the VSG parameters have been largely restored to the economic model parameters.

[0237] When all the above conditions are met, the distributed intelligent agent determines that the collaborative recovery process has ended. At this point, the agent stops executing the recovery strategy and switches its internal working state back to steady-state economic operation. The entire process returns to step one, and the system restarts the distributed economic scheduling based on the consensus algorithm.

[0238] While the invention has been described herein with reference to specific embodiments, it should be understood that these embodiments are merely examples of the principles and applications of the invention. Therefore, it should be understood that many modifications can be made to the exemplary embodiments, and other arrangements can be designed without departing from the spirit and scope of the invention as defined by the appended claims. It should be understood that different dependent claims and features described herein can be combined in ways different from those described in the original claims. It is also understood that features described in conjunction with individual embodiments can be used in other described embodiments.

Claims

1. A method for intelligent control of multi-energy coordinated scheduling, applied to a power system comprising a plurality of distributed energy units, characterized in that, Each of the energy units is configured with a local controller and a distributed intelligent agent, and the method comprises the following steps: S101, a steady-state operation step: in a steady-state economic operation mode, optimal active power reference values of the energy units are determined by the distributed intelligent agent network through distributed optimization calculation, and virtual synchronous generator (VSG) control parameters of each local controller are set as economic mode parameters; S102, a disturbance triggering step: when a grid frequency change rate exceeds a preset threshold, a distributed intelligent agent corresponding to an energy unit detecting the disturbance broadcasts a disturbance announcement containing a node identifier of the disturbance source node; S103, a group self-organizing step: each distributed intelligent agent receiving the disturbance announcement calculates an electrical distance between itself and the disturbance source node according to a pre-stored system grid impedance matrix, and dynamically classifies itself into a core response group, a collaborative support group or a standby group according to a comparison result of the electrical distance with at least two preset threshold values; S104, a parameter reconstruction step: each distributed intelligent agent generates corresponding VSG control parameters and issues them to the local controller according to the response group to which it belongs, so that the energy units of different groups perform differentiated disturbance suppression control; S105, a collaborative recovery step: after the disturbance is suppressed, all distributed intelligent agents collaboratively perform recovery control to eliminate system steady-state frequency deviation and smoothly restore the VSG control parameters to the economic mode parameters, so that the system returns to the steady-state economic operation mode. The electrical distance is calculated by calculating the reciprocal of the absolute value of the mutual impedance element corresponding to the identifier of the disturbance source node and the identifier of the disturbance source node in the grid impedance matrix. The classification rule is: In the parameter reconstruction step: for the distributed intelligent agent classified into the core response group, the VSG control parameter is set as a preset maximum support mode parameter; 2.The intelligent control method of multi-energy coordinated scheduling according to claim 1, wherein, for the distributed intelligent agent classified into the standby group, the VSG control parameter is maintained as the economic mode parameter; 3.The intelligent control method of multi-energy coordinated scheduling according to claim 2, wherein, In the self-organizing step of the group, the at least two preset thresholds include a core response threshold and a cooperative support threshold, and satisfy: the core response threshold the cooperative support threshold. for the distributed intelligent agent classified into the collaborative support group, the VSG control parameter is generated by interpolation calculation, and the interpolation calculation makes the generated parameter value between the economic mode parameter and the maximum support mode parameter, and negatively related to the electrical distance. If the electrical distance If the core response threshold, then it is classified into the core response group; if the core response threshold the electrical distance if the synergistic support threshold, then assigned to the synergistic support group; If the electrical distance Cooperative support threshold, then relegated to the standby group.

4. The intelligent control method for multi-energy coordinated scheduling according to claim 3, characterized in that, The VSG control parameter of the collaborative support group is generated by the following interpolation formula: In the collaborative recovery step, the elimination of system steady-state frequency deviation is achieved by distributed secondary frequency recovery control, which comprises: each distributed intelligent agent integrates the local frequency deviation to generate a local secondary regulation signal; the local secondary regulation signals of all agents are converged to a unified global regulation signal through a consensus algorithm; 5. The intelligent control method for multi-energy coordinated scheduling according to claim 4, characterized in that, the global regulation signal is superimposed on the optimal active power reference value of each energy unit. wherein is a new VSG control parameter, including a virtual inertia or virtual damping coefficient; is the corresponding value under the economic mode parameter; is the corresponding value under the maximum support mode parameter; are weight coefficients, and is an electrical distance, is a cooperative support threshold, is a core response threshold. 6.The intelligent control method of multi-energy coordinated scheduling according to claim 1, wherein, In the collaborative recovery step, the smooth recovery of the VSG control parameter is achieved by the following process: each distributed intelligent agent dynamically calculates a real-time VSG parameter according to a return control function decaying over time; the calculation method of the real-time VSG parameter is: ​ 7. The intelligent control method for multi-energy coordinated scheduling according to claim 5, characterized in that, ​ ​ ​ wherein is the real-time VSG parameter at time t, is the return control function, which value decays from 1 to 0 over time. 8.The intelligent control method of multi-energy coordinated scheduling according to claim 1, wherein, The distributed optimization calculation in the steady-state operation step is a consistent-based distributed economic dispatch algorithm, which aims to minimize the total operation cost of the system under the conditions of meeting system power balance and unit output constraints, and make the marginal cost of all distributed intelligent agents converge to a unified value, thereby inversely calculating the optimal active power reference value of each unit.

Images