Distributed control method for off-grid operation of micro-grid based on consistency collaboration

Abstract

The invention discloses a micro-grid off-grid operation distributed control method based on consistency cooperation, belongs to the field of new energy micro-grid control, and aims to solve the problems of high single-point fault risk and single control target of existing centralized control. The method is operated in a three-layer architecture formed by a distribution network master station control layer, a micro-grid coordination control layer and an equipment distributed coordination control layer, a distributed communication network is constructed through edge agents, and weight coefficients and security constraints are configured; the edge agent receives a master station total power target, collects and interacts equipment data, calculates and limits amplitude to obtain a final power instruction based on a consistency updating law fusing global power difference distribution, energy storage SOC balance and system loss minimization, and then decomposes the final power instruction into equipment-level instructions to be issued and executed. According to the method, a single-point fault is eliminated, multi-target collaborative optimization is realized, control is simplified, expansibility is high, and power balance and stable and efficient operation when the micro-grid is off-grid are effectively guaranteed.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of new energy micro-grid control, and particularly relates to a micro-grid off-grid operation distributed control method based on consistency collaboration. BACKGROUND

[0002] When the micro-grid is in off-grid operation, the randomness and volatility of the internal new energy power generation (such as photovoltaic) and load (especially electric vehicle charging piles) pose a serious challenge to the real-time power balance and long-term stable operation of the system.

[0003] At present, in the micro-grid coordination control layer of the micro-grid, the existing control method mainly has the following defects:

[0004] 1. Single point failure risk exists in the control architecture, and the reliability is low. The existing method mainly adopts an architecture based on master-slave control or centralized energy management. This kind of method seriously depends on a central controller to collect global information, perform calculation and distribute instructions to all subordinate units. This architecture has obvious "single point failure" bottleneck, and once the central controller fails, the coordination control function of the whole system will be paralyzed.

[0005] 2. The control target is single, and the inherent conflict of heterogeneous resources cannot be coordinated at the instruction generation level. The energy storage system (aiming to maintain the state of charge SOC) and the charging pile (aiming to provide charging service) in the micro-grid are completely different adjustable resources, and their inherent operation targets are naturally conflicting. The existing technology usually only considers a single power balance target when generating control instructions, and ignores the dynamic balance of the SOC of the energy storage, resulting in excessive charging and discharging of part of the energy storage units, damaging the service life and even causing system failure. At the same time, the existing instruction generation mechanism also generally lacks consideration of the overall operation efficiency of the system, resulting in high energy consumption of the system and low efficiency. SUMMARY

[0006] The application aims to provide a micro-grid off-grid operation distributed control method based on consistency collaboration to overcome the defects of single point failure risk and single control target in the control architecture of the prior art.

[0007] To achieve the above-mentioned purpose, the application provides the following technical scheme:

[0008] A micro-grid off-grid operation distributed control method based on consistency collaboration, which runs in a three-layer micro-grid system architecture including a distribution network master station control layer, a micro-grid coordination control layer and a device distributed collaborative control layer, wherein the distribution network master station control layer is provided with a master station management platform, the micro-grid coordination control layer is provided with a plurality of edge agents, and the device distributed collaborative control layer is provided with photovoltaic, energy storage and charging piles.

[0009] The method comprises the following steps:

[0010] S1: System initialization and parameter setting, configuring neighbor agent set for each edge agent to form a distributed communication network, setting system weight coefficient, setting adjustable device total power upper and lower limits, and reading device initial state; the master station management platform determines the total power target according to the device initial state;

[0011] S2: Each edge agent receives the total power target issued by the master station management platform, collects jurisdictional device data and calculates local adjustable device total power, local energy storage state of charge and local equivalent efficiency function, the local equivalent efficiency function is used to determine the loss minimization term;

[0012] S3: Each edge agent sends its jurisdictional device total power and energy storage state of charge data to neighbor agents, and receives corresponding data sent by neighbor agents;

[0013] S4: Based on the comprehensive distributed consistency update law, the power instruction update amount is calculated combined with the system weight coefficient, the global power difference allocation term, the SOC balancing term and the loss minimization term are fused, and the final power instruction is obtained after amplitude limiting processing by the total power upper and lower limits;

[0014] S5: The final power instruction is decomposed into total power instructions of the energy storage system and the charging pile, and then decomposed into device-level instructions and executed through the collector.

[0015] Further, the system weight coefficient includes α, β, γ, k, α, β, γ, k are all greater than 0, α is the power distribution weight coefficient, β is the SOC balancing weight coefficient, γ is the power convergence coefficient, and k is the loss optimization weight coefficient; α, β, γ, k can be uniformly set in the whole system range, or adaptively fine-tuned according to the device properties of the edge agent.

[0016] Further, the calculation formula of the total power target is as follows:

[0017] , is the total power target, is the system real-time total load power, is the total power of all photovoltaic real-time output of the system.

[0018] Further, the calculation formula of the global power difference allocation term is as follows:

[0019] ;

[0020] wherein is the global power distribution term of the i-th edge agent, P[i] is the total power of the i-th edge agent, P[j] is the total power of the neighbor edge agent j of the i-th edge agent, The sum of real-time total power of all adjustable devices, including energy storage and charging piles, governed by the mth edge agent; N[i] is the neighbor agent set of the ith edge agent, and n is the total number of edge agents.

[0021] Further, the calculation formula of the SOC balancing term is as follows:

[0022] ;

[0023] The average state of charge of the neighbor edge agent j of the ith edge agent, The average state of charge of the ith edge agent, The SOC balancing term of the ith edge agent; N[i] is the neighbor agent set of the ith edge agent.

[0024] Further, the calculation formula of the loss minimization term is as follows:

[0025] ; the loss optimization weight coefficient k is a positive weight coefficient. The loss minimization term of the ith edge agent; is the first-order derivative of the local equivalent efficiency function of the ith edge agent.

[0026] Further, the edge agent determines the total power instruction P_ess_cmd of the energy storage system governed by the edge agent and the total power P_ev_cmd of the charging pile system governed by the edge agent by solving the optimization objective function, which aims to minimize the power distribution error and maintain the energy storage SOC at an ideal level, with constraint conditions including the total energy storage power limit, the total charging pile power limit, and the average energy storage SOC limit, wherein the optimization objective function is:

[0027] ;

[0028] wherein and are two decision variables, is the total power value to be issued by the edge agent to the energy storage system (the total power of the energy storage system, discharging is positive, and charging is negative), is the total power value to be issued by the edge agent to the charging pile (the total power of the charging pile, charging is positive). is the average state of charge (measured value) of the ith edge agent. is the SOC target value of the energy storage system of the ith edge agent, E[i] is the final power instruction of the i-th edge agent, E[i] is the total capacity of the energy storage under the jurisdiction of the i-th edge agent, T is the control period, and lambda is the SOC balancing weight factor.

[0029] Further, if there is only energy storage without charging piles under the edge agent, P_ev is always 0 when solving; if there is only charging piles without energy storage, P_ess is always 0 when solving; when decomposing the total power instruction of the energy storage system and the total power instruction of the charging pile, an average allocation strategy or an allocation strategy combined with priority is adopted.

[0030] Further, the method further comprises a fault response mechanism: if the communication between the edge agent and the neighbor agent is interrupted, the neighbor agent is removed from the neighbor agent set; if the key equipment under the edge agent fails, the local power distribution is recalculated and the adjacent agent is notified.

[0031] Compared with the prior art, the beneficial effects of the present application are:

[0032] 1. The method provided by the present application can eliminate single point failure: through a completely distributed control architecture, the centralized or master-slave architecture depending on the central controller is replaced, and the reliability and robustness of the system are fundamentally improved. Each edge agent has an equal status in the consistency algorithm and makes autonomous decisions based on local and neighbor information. Failure of any edge agent or interruption of any communication link only means local reconstruction of the communication topology of the system, and does not cause the collapse of the entire collaborative control system. The remaining normal edge agents can continue to run the consistency algorithm autonomously through the remaining communication paths to maintain the power balance and optimization of the system. The new unit only needs to access the communication network to automatically participate in collaboration without modifying the upper control logic.

[0033] 2. Multi-objective collaborative optimization is realized (core invention point): at the microgrid coordination control layer, the problem of inherent conflict between the energy storage system (power target) and the charging pile (power and service target) due to the single target of the prior art is solved. The present application builds a distributed collaborative framework, so that each edge agent can simultaneously consider and optimize the three key targets of "system power balance", "energy storage SOC dynamic balance" and "system operation loss minimization" when making autonomous decisions, thereby guiding the entire system to the optimal operating state of safety, economy and efficiency.

[0034] 3. The present application has simplified control and strong scalability: intelligent collaboration at the device layer eliminates the need for complex operating condition decision logic at the upper microgrid coordination control layer. System design, debugging and maintenance are simplified. The communication topology is flexible and easy to expand the system, especially suitable for microgrid projects built in stages. BRIEF DESCRIPTION OF DRAWINGS

[0035] Figure 1A schematic diagram of the distributed collaborative control method for microgrid sources, loads and storage based on consensus algorithm provided in an embodiment of the present invention;

[0036] Figure 2 This is a diagram of a three-layer microgrid system architecture in an embodiment of the present invention;

[0037] Figure 3 This is a diagram of a three-layer microgrid system architecture in another embodiment of the present invention. Detailed Implementation

[0038] 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.

[0039] The following is in conjunction with the appendix Figures 1-3 The specific embodiments of the present invention will be described in detail below.

[0040] Example: A distributed control method for off-grid operation of microgrids based on consensus and coordination, which is implemented in a microgrid system architecture.

[0041] Pre-built microgrid system architecture such as Figure 2 As shown, the system comprises a three-tier architecture: a distribution network master station control layer, a microgrid coordination control layer, and a device distributed collaborative control layer. Specifically: The distribution network master station control layer houses a master station management platform responsible for global monitoring, data aggregation, long-term strategy formulation, and cross-regional coordination. The microgrid coordination control layer includes edge agents, which act as local intelligent units responsible for collecting lower-level data, executing short-term control strategies, coordinating local resources, and reporting to the upper-level platform. The device distributed collaborative control layer includes data collectors, meters, energy storage, photovoltaic systems, and charging piles. Data collectors, acting as data acquisition and execution units, directly connect to physical devices (such as energy storage, photovoltaic systems, and charging piles), which are the final objects of control execution. Additionally, physical devices can connect to meters for measurement purposes, such as photovoltaic output meters to measure the actual output of photovoltaic systems and load feeder meters to measure the power of the load.

[0042] In this system, devices (meters, energy storage, photovoltaics, and charging piles) are connected through data collectors and edge agents. The data collectors and edge agents can communicate via wired or wireless communication protocols, including Modbus, Ethernet, and 104. Energy storage units connected to different data collectors can communicate indirectly.

[0043] The edge agents can exchange data through a local area network or a dedicated protocol (such as MQTT, DDS, etc.), simulating the "neighbor communication" required by the consensus algorithm.

[0044] The device distributed cooperative control layer of the system includes energy storage, photovoltaic and charging piles, but there can be at least one of energy storage, photovoltaic and charging piles under each collector, that is, the device group under the collector can be heterogeneous.

[0045] In the edge agent of the microgrid coordination control layer, a distributed control strategy based on the consensus algorithm is used to replace the original centralized energy storage control strategy. Each edge agent calculates the power instruction and issues the calculation result to the corresponding collector and device for execution. The edge agents exchange the state information of the energy storage units they manage through the communication network.

[0046] As shown in Figure 1 , the microgrid off-grid operation distributed control method based on consensus cooperation specifically includes the following steps:

[0047] Step S1: system initialization and parameter setting.

[0048] Each edge agent is configured with a communication module, and each edge agent in the microgrid is configured with a neighbor agent set N[i] to form a distributed communication network (such as a ring, star or network topology). Each edge agent only needs to exchange data with the neighbor edge agent nodes in its communication topology.

[0049] Set α, β, γ, k, all greater than 0, where α is the power distribution weight coefficient, β is the SOC balancing weight coefficient, γ is the power convergence coefficient, and k is the loss optimization weight coefficient. The power convergence coefficient γ controls the global dynamic response speed of the system tracking the total power target ΔP net . The power distribution weight coefficient α determines the relative influence of the global power distribution item in the total update amount. The SOC balancing weight coefficient β controls the speed of the SOC balancing process. The loss optimization weight coefficient k determines the importance of the loss optimization target.

[0050] α, β, γ, k are preferably uniformly set in the entire system to ensure cooperative consistency. For example, a uniform power distribution weight coefficient α can ensure that all agents respond to power deviations in unison, which is the basis for cooperation, and a uniform β can ensure that all energy storages compete fairly in the balancing process, avoiding some units from changing power too quickly due to excessive weight, while other units are slow to react.

[0051] α, β, γ, k can also be adaptively fine-tuned according to the properties of the devices under the jurisdiction of the specific edge agent to cope with system heterogeneity. For example, the values of the weight coefficients (α, β, γ, k) can be flexibly adjusted according to actual needs to emphasize different optimization objectives in different scenarios (such as more emphasis on balance, more emphasis on energy storage life, or more emphasis on economy). This design enables the method to adapt to microgrid systems with different configurations and different operation strategies, and has strong universality.

[0052] The security constraint setting includes setting the upper and lower limits of the total power P min [i] of each edge agent's jurisdictional adjustable device. max [i], and reading the initial state of the device. Optionally, each edge agent reads the initial state of all devices under its jurisdiction and uploads it to the master management platform, so that the master management platform determines the total power target based on the initial state of the device.

[0053] Step S2: Each edge agent receives the total power target issued by the master management platform.

[0054] The master management platform calculates the system net power difference, i.e. the total power target, according to the initial state of the device, and broadcasts the total power target to all edge agents. The calculation formula of the total power target is as follows:

[0055] , is the total power target, is the real-time total load power of the system, including the power of the charging pile, and one of the method objectives is to optimally allocate the limited power to the charging pile and other loads; is the real-time total power of all photovoltaic systems, does not participate in the control cycle update. The total power target item is to ensure that the total power of all energy storage units can compensate for the net power shortage of the microgrid.

[0056] and It can be measured by a power meter, and data aggregation and communication are performed by a collector, which collects data from the power meter, as well as state data from energy storage, photovoltaic inverters and charging pile controllers, and then packages and uploads them to the edge agent, which in turn uploads them to the master management platform.

[0057] Each edge agent i periodically collects data from the devices under its jurisdiction, including: the real-time total power P_current_ess of all energy storage units under each edge agent i, and the real-time total power P_current_ev of all charging piles under each edge agent i.

[0058] P_current_ess= ; P_current_ev= .

[0059] where P ess is the real-time power of each energy storage unit (positive for discharging, negative for charging), P ev is the real-time power of each charging pile (positive for charging).

[0060] Each edge agent i computes the local total power of adjustable devices:

[0061] , is the sum of real-time total power of all adjustable devices (energy storage, charging pile) under the jurisdiction of the i-th edge agent.

[0062] Each edge agent i computes the local energy storage priority-based allocation strategy "allocation strategy combined with charging urgency and user level, allocates state of charge : , is the state of charge value of a single energy storage under the edge agent i.

[0063] Each edge agent i computes the local equivalent efficiency function , represents the comprehensive efficiency of the i-th edge agent under the total power of local adjustable devices (which can be obtained by looking up the preset efficiency-power curve); and calculates , is the first derivative (slope) of this efficiency function .

[0064] Step S3: Each edge agent i sends to all neighbor agents j in its neighbor agent set , and receives sent by all neighbor agents j .

[0065] Step S4: Based on the comprehensive distributed consensus update law, the power instruction update amount is calculated in combination with the system weight coefficient.

[0066] The core of the present application is to design a comprehensive distributed consensus update law, which for the first time fuses the following three optimization objectives through weighted summation.

[0067] A comprehensive optimization objective is designed, and is converted into a distributed consensus update law. Each energy storage node periodically performs the following power instruction update calculation:

[0068] 1. Calculate the global power difference allocation item to drive the total power of the system to approach the global target.

[0069] The power deviation is dynamically averaged by a consensus algorithm, and the expression is as follows:

[0070] ;

[0071] Wherein is the global power allocation item of the i th edge agent, P[i] is the total power of the i th edge agent (real-time total power), P[j] is the total power of the neighbor edge agent j of the i th edge agent (real-time total power), is the sum of the real-time total power of all adjustable devices (energy storage, charging pile) under the jurisdiction of the m th edge agent.

[0072] The global total power deviation ( ) is evenly distributed to the n edge agents in the system, and then the power convergence coefficient γ is used to adjust the influence strength of this average deviation on the update amount.

[0073] The term ensures that the total output of all edge agents matches the system net power demand. The subscripts i and j are the numbers of edge agents. N[i] is the neighbor agent set of the i th edge agent, that is, the set of all other energy storage units directly connected to the i th edge agent in the communication network.

[0074] is the summation operator, which means summing up the corresponding values of all neighbor edge agents j in the set N[i], and n is the total number of edge agents participating in collaborative control.

[0075] γ is the power convergence coefficient, which is a configurable parameter greater than zero. This coefficient affects the speed and stability of the system in tracking the total power demand. Its specific value can be adjusted and fine-tuned on site according to the communication performance of the actual microgrid, system inertia, and desired dynamic response characteristics.

[0076] 2. SOC balancing term. This term is to drive the SOC of all energy storage units to be consistent and avoid excessive charging and discharging of some units. It is achieved by comparing the local SOC with the average value of neighbor SOCs.

[0077] The calculation formula is as follows: ;

[0078] is the average state of charge of the neighbor edge agent j of the i th edge agent, is the average state of charge of the i th edge agent, is the SOC balancing term of the i th edge agent. This term drives the SOCs of all edge agents to be consistent and avoids excessive charging and discharging of individual units.

[0079] If there is no energy storage under a certain edge agent, its If the value is null or invalid, the edge agent will still broadcast its total adjustable power P[i] (from the charging pile) to other neighboring edge agents; it will not broadcast... The value, or its SOC status marked as invalid in the communication protocol.

[0080] The algorithm should be configured so that when there is no stored energy, it does not participate in the iteration of the SOC equalization term, i.e. .

[0081] 3. Loss Minimization Term: This term aims to guide the system towards minimizing total losses. This is achieved by considering marginal losses, i.e., the derivative of efficiency with respect to power (…). This is used to adjust power distribution. High-efficiency (low-loss) units should handle more power.

[0082] k is a positive loss optimization weight coefficient. Let be the loss minimization term for the i-th edge agent. This is the first derivative (slope) of the efficiency function. When the slope is positive, it means that increasing the total power P[i] by a little will improve the overall efficiency. A negative slope indicates the algorithm will tend to increase the agent's power; a negative slope means that increasing power will actually reduce efficiency. If the value is positive, the algorithm will tend to reduce the power of that edge agent. The end result is that the power allocation of all agents is adjusted to near the peak of their efficiency function (i.e., the region where the derivative is close to zero), thereby minimizing the total system loss. The loss optimization weight coefficient k is determined by... It indirectly affects the comprehensive update law, ultimately influencing the power command. The value of the loss optimization weight coefficient k can be tuned according to the actual system's requirements for operational economy.

[0083] Comprehensive Update Law: The above three terms are weighted and combined to obtain the update amount of the power instruction of the i-th edge agent. :

[0084] ;

[0085] Finally, the power command P[i] of the i-th edge agent in the previous control cycle and the update amount The sum is determined as the power command for the next time step after the update:

[0086] ;

[0087] in This is the power command for the i-th edge agent during the current control cycle.

[0088] Constraints: In calculation Then, the i-th edge agent needs to adjust the total power of the devices under its jurisdiction according to the upper and lower limits P max [i] and P min [i] to determine the final power instruction of the i-th edge agent in the current control period , ensuring that the instruction is within the safe operating range of the device, and the expression is:

[0089] .

[0090] Step S5, instruction decomposition and issuance.

[0091] The i-th edge agent obtains the total power instruction P_ess_cmd of the energy storage system under its jurisdiction and the total power instruction P_ev_cmd of all charging piles under its jurisdiction by solving the following local optimization objective function according to the final power instruction of the i-th edge agent in the current control period , and then decomposes the total power instruction P_ess_cmd and the total power instruction P_ev_cmd to the subordinate devices to generate specific device-level instructions.

[0092] The i-th edge agent issues specific instructions to each energy storage converter and charging pile controller through the local collector. The i-th edge agent needs to perform intelligent allocation among the devices under its jurisdiction, i.e., after obtaining P_ess_cmd and P_ev_cmd, the edge agent will immediately execute a local allocation algorithm.

[0093] For the energy storage cluster, an average allocation strategy or a priority strategy can be used to allocate P_ess_cmd to each energy storage unit to generate specific instructions P_cmd_1, P_cmd_2,..., P_cmd_N1, N1 being the number of energy storage units under the edge agent.

[0094] For the charging pile cluster, an average allocation strategy or a priority strategy can be used to allocate P_ev_cmd to each charging pile to generate specific instructions P_ev_cmd_1, P_ev_cmd_2,..., P_ev_cmd_N2, N2 being the number of charging piles under the edge agent.

[0095] The method further includes step S6, continuously detecting the execution situation and system state changes, waiting for the next control period to come, returning to step S2, and receiving the latest total power target calculated by the master station management platform.

[0096] Optionally, if the communication between a certain edge agent and a certain neighbor agent is interrupted, it is temporarily removed from the neighbor agent set N[i].

[0097] Optionally, if a certain critical device fails under an edge agent, recalculate the local power allocation and notify the neighboring agents.

[0098] In some embodiments, the i-th edge agent calculates the i-th edge agent power instruction P_ess_cmd[i] according to the final current control period Get the total power instruction P_ess_cmd of the energy storage system under the jurisdiction of the edge agent, and the total power instruction P_ev_cmd of all charging piles under the jurisdiction of the edge agent, including determining the following optimization objective function:

[0099]

[0100] Wherein and Two decision variables, is the total power value (positive for discharging and negative for charging) that the edge agent will issue to the energy storage system, is the total power value (positive for charging) that the edge agent will issue to the charging pile. is the average state of charge (measured value) of the i-th edge agent. is the SOC target value of the energy storage system of the i-th edge agent (such as 50%~70%, etc., to ensure the regulation ability of the energy storage system). It can be a dynamic value, for example: when photovoltaic power is large during the day, It can be set higher (such as 70%) to prepare for the evening load peak. At night, It can be set lower (such as 50%) to reserve space for photovoltaic charging in the morning of the next day; E[i] is the total capacity of the energy storage under the jurisdiction of the i-th edge agent (unit: kWh, kilowatt-hour), T is the control period (unit: hour, h). λ is the SOC balance weight factor, used to balance the importance of power allocation error and SOC target deviation.

[0101] The first goal is to minimize the power allocation error, and the second goal is to maintain the energy storage SOC at an ideal level.

[0102] Constraint: P_ess_min<=P_ess<=P_ess_max; (total energy storage power limit)

[0103] P_ev_min<=P_ev<=P_ev_max; (total charging pile power limit)

[0104] SOC_min[i] <= SOC[i] <= SOC_max[i] (average SOC limit of energy storage).

[0105] ​P_ess_min represents the minimum total energy storage power of the edge agent, and P_ess_max represents the maximum total energy storage power of the edge agent; P_ev_min represents the minimum total charging pile power of the edge agent, and P_ev_max represents the maximum total charging pile power of the edge agent; SOC_min[i] represents the minimum average SOC of the edge agent's energy storage, and SOC_max[i] represents the maximum average SOC of the edge agent's energy storage. The power capacity range and load safety range of each energy storage unit, as well as the rated power safety range of each charging pile, are predetermined hardware specifications. These specifications can be used to determine the energy storage capacity range and load safety range of each energy storage unit.

[0106] The method calculates the agent-level final power command. The system safely and rationally decomposes the commands into energy storage system commands (P_ess_cmd) and charging pile commands (P_ev_cmd), which are then ultimately sent to each physical device. This step not only ensures that the coordinated power commands are sent to specific devices, but also adds safety constraints such as power upper and lower limits and SOC limits at the device level, improving system safety and ensuring the safe and stable operation of the entire system.

[0107] If an edge agent only has energy storage and no charging stations, then P_ev will always be 0 when solving the problem; conversely, if there are only charging stations and no energy storage, then P_ess will always be 0.

[0108] Each edge agent can pre-collect the power capacity range, load safety range, and rated power safety range of each of its subordinate energy storage units.

[0109] Based on the constraints, the optimal P_ess and P_ev are obtained by solving the above optimization objective function, which are P_ess_cmd and P_ev_cmd.

[0110] Continuously monitor the execution status and system status, and wait for the next control cycle to return to step S2.

[0111] In another embodiment, such as Figure 3 As shown, this embodiment demonstrates a three-layer control architecture comprising three edge agents (A1, A2, A3) connected via a ring communication topology (i.e., each agent communicates directly with the other two agents). The specific devices managed by each agent are as follows:

[0112] Edge Agent A1: 1 photovoltaic power generation unit and 1 energy storage system.

[0113] Edge Agent A2: 1 energy storage system and 4 charging piles.

[0114] Edge Proxy A3: 2 charging stations.

[0115] Communication topology: A1's neighbors are A2 and A3, A2's neighbors are A1 and A3, and A3's neighbors are A1 and A2.

[0116] The main station management platform is responsible for calculating the global power target ΔP. net .

[0117] The first iteration calculation (using A2 as an example below for a brief illustration):

[0118] After agent A2 obtains data from its neighbors A1 and A3, it begins calculations:

[0119] Global power differential allocation item:

[0120] ;

[0121] SOC equilibrium term:

[0122] There is no energy storage under A3, so there is no SOC[3].

[0123] Loss minimization term:

[0124] , 0.1 is acceptable.

[0125] .

[0126] The power command for the next moment is: ;

[0127] Finalized Then, the instructions are broken down and issued for execution.

[0128] The contents not described in detail in this description are existing technologies known to those skilled in the art. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A distributed control method for off-grid operation of microgrids based on consensus and coordination, characterized in that, The system operates on a three-layer microgrid architecture that includes a distribution network master station control layer, a microgrid coordination control layer, and a device distributed collaborative control layer. The distribution network master station control layer is equipped with a master station management platform, the microgrid coordination control layer is equipped with multiple edge agents, and the device distributed collaborative control layer is equipped with photovoltaic, energy storage, and charging piles. The method includes the following steps: S1: System initialization and parameter settings, configure a set of neighbor agents for each edge agent to form a distributed communication network, set the system weight coefficient, set the upper and lower limits of the total power of adjustable devices, and read the initial state of the devices. The main station management platform determines the total power target based on the initial state of the devices. S2: Each edge agent receives the total power target issued by the main station management platform, collects data of the managed devices and calculates the total power of the local adjustable devices, the local energy storage state of charge and the local equivalent efficiency function. The local equivalent efficiency function is used to determine the loss minimization term. S3: Each edge agent sends the total power and energy storage state of charge data of the devices under its jurisdiction to the neighbor agents, and receives the corresponding data sent by the neighbor agents; S4: Combining the system weight coefficients, the power command update amount is calculated based on the comprehensive distributed consistency update law. The global power difference allocation term, SOC equalization term and loss minimization term are integrated, and the final power command is obtained by limiting the total power upper and lower limits. S5: Decompose the final power command into the total power command of the energy storage system and the charging pile, then decompose it into device-level commands and send them out for execution through the data collector.

2. The distributed control method for off-grid operation of microgrids based on consensus and coordination as described in claim 1, characterized in that, The system weighting coefficients include α, β, γ, and k, where α, β, γ, and k are all greater than 0. α is the power allocation weighting coefficient, β is the SOC equalization weighting coefficient, γ is the power convergence coefficient, and k is the loss optimization weighting coefficient.

3. The distributed control method for off-grid operation of microgrids based on consensus and coordination according to claim 1, characterized in that, The formula for calculating the total power target is as follows: , For the total power target, This represents the system's real-time total load power. This represents the total real-time output power of all photovoltaic systems in the system.

4. The distributed control method for off-grid operation of microgrids based on consensus and coordination according to claim 2, characterized in that, The formula for calculating the global power difference allocation term is as follows: ； in Let P[i] be the global power allocation term for the i-th edge agent, where P[i] is the total power of the i-th edge agent, and P[j] is the total power of the i-th edge agent's neighboring edge agent j. For the total power target, N[i] is the sum of the real-time total power of all adjustable devices under the jurisdiction of the m-th edge agent, including energy storage and charging piles; N[i] is the set of neighbor agents of the i-th edge agent, and n is the total number of edge agents.

5. The distributed control method for off-grid operation of microgrids based on consensus and coordination according to claim 1, characterized in that, The formula for calculating the SOC equilibrium term is as follows: ； Let j be the average charge state of the neighboring edge agent j of the i-th edge agent. Let i be the average state of charge of the i-th edge agent. Let N[i] be the SOC equilibrium term of the i-th edge agent; N[i] is the set of neighboring agents of the i-th edge agent.

6. The distributed control method for off-grid operation of microgrids based on consensus and coordination according to claim 2, characterized in that, The formula for calculating the loss minimization term is as follows: ； Let i be the loss minimization term for the i-th edge agent. It is the local equivalent efficiency function of the i-th edge agent. The first derivative.

7. The distributed control method for off-grid operation of microgrids based on consensus and coordination according to claim 1, characterized in that, The edge agent determines the total power command P_ess_cmd of the energy storage system and the total power P_ev_cmd of the charging pile system under its jurisdiction by solving an optimization objective function. The optimization objective function aims to minimize power allocation error and maintain the energy storage SOC at an ideal level. The constraints include a total energy storage power limit, a total charging pile power limit, and a total energy storage SOC limit. The optimization objective function is as follows: ； in and There are two decision variables. This is the total power value that the edge agent will send to the energy storage system. This is the total power value that the edge agent will send to the charging station. It is the average state of charge of the i-th edge agent. It is the SOC target value of the i-th edge agent energy storage system. Let E[i] be the final power command of the i-th edge agent, E[i] be the total energy storage capacity under the jurisdiction of the i-th edge agent, T be the control period, and λ be the SOC balance weight factor.

8. The distributed control method for off-grid operation of microgrids based on consensus and coordination according to claim 7, characterized in that, If there is only energy storage and no charging piles under the edge agent, P_ev will always be 0 when solving; if there are only charging piles and no energy storage, P_ess will always be 0 when solving; when decomposing the total power command of the energy storage system and the total power command of the charging piles, an average allocation strategy or an allocation strategy combined with priority is adopted.

9. The distributed control method for off-grid operation of microgrids based on consensus and coordination according to claim 1, characterized in that, The method also includes a fault response mechanism: if communication between the edge agent and the neighbor agent is interrupted, the neighbor agent is removed from the set of neighbor agents; if a critical device under the edge agent fails, the local power allocation is recalculated and the neighbor agents are notified.

Images