Cooperative control method for alternating current and direct current hybrid micro-grid

Abstract

The invention relates to the technical field of micro-grids, in particular to an alternating current and direct current hybrid micro-grid cooperative control method. According to the technical scheme, the alternating current and direct current hybrid micro-grid cooperative control method comprises the following steps that S1, a multi-stage micro-grid framework is constructed and comprises a main micro-grid, a direct current sub-micro-grid and an alternating current sub-micro-grid, and the main micro-grid and the direct current sub-micro-grid are connected through a bidirectional interconnection converter; s2, deploying a hierarchical Multi-Agent control architecture, wherein the hierarchical Multi-Agent control architecture comprises a main micro-grid agent, a direct current sub-micro-grid agent and an alternating current sub-micro-grid agent; and S3, executing three-level cooperative control, receiving a main network scheduling instruction by the main micro-grid agent, decomposing the main network scheduling instruction into sub-micro-grid control targets, coordinating power distribution among sub-micro-grids based on a global optimization algorithm, and maintaining voltage and frequency stability by the DC sub-micro-grid agent and the AC sub-micro-grid agent through a local control algorithm. According to the method, the power grid consumption efficiency and the operation stability in a high-proportion new energy access scene are remarkably improved.

Description

Technical Field

[0001] The present invention relates to the technical field of microgrids, and particularly to a coordinated control method for an AC-DC hybrid microgrid. Background Art

[0002] Currently, with the large-scale access of renewable energy, the microgrid control system is gradually shifting from the centralized management of single energy to the distributed architecture of multi-energy complementarity. In the prior art, the AC-DC hybrid microgrid mainly realizes energy scheduling through a centralized controller or decentralized local control. For example, some solutions adopt a centralized control architecture dominated by an AC microgrid, relying on a central controller to uniformly allocate wind-solar-storage resources, but it is difficult to meet the plug-and-play requirements of distributed energy; other solutions are based on decentralized control, achieving power balance through local droop control, but lack the ability of global optimization and are prone to power conflicts between multiple microgrids.

[0003] With the advancement of the construction of the new power system, the high proportion of renewable energy access has led to severe challenges in the consumption of traditional regional power grids. In the prior art, the AC-DC hybrid microgrid mostly adopts a centralized control architecture or a completely decentralized local control strategy: centralized control relies on a central controller for unified scheduling, requires the construction of a complex global model, is difficult to adapt to the random volatility of distributed energy such as wind, solar, water, and storage, and has a lag in response speed; decentralized control can quickly respond to local disturbances, but lacks the coordinated optimization between multi-level microgrids and is prone to power distribution conflicts and voltage over-limit problems. In addition, existing solutions are mostly limited to a single type of AC / DC networking mode, are difficult to be compatible with the interconnection requirements of multi-level and multi-form microgrids, and do not deeply couple the power market mechanism with grid regulation, resulting in a lack of economic incentives for new energy consumption and low utilization rate of flexible resources. How to construct a microgrid coordinated control system with global optimization ability, dynamic response efficiency, and market adaptability has become the core problem that needs to be urgently solved in the current technology.

[0004] Chinese Patent CN120016559A discloses a power coordination control method and device for a hybrid microgrid group, including: establishing a topology structure of the hybrid microgrid group; each AC subnet collects the bus frequency of the corresponding AC bus, and each DC subnet collects the bus voltage of the corresponding DC bus; when the bus frequency of the AC subnet or the bus voltage fluctuation of the DC subnet is within the first preset range, the AC subnet and the DC subnet adopt adaptive droop control to achieve power coordination; when the bus frequency of the AC subnet or the bus voltage fluctuation of the DC subnet is within the second preset range, the AC subnet and the DC subnet achieve power coordination through the exchange of power; when the bus frequency of the AC subnet or the bus voltage fluctuation of the DC subnet exceeds the second preset range, control the power output of the energy storage subnet to achieve power coordination. This method can improve the robustness of the system, but it does not achieve the coordinated optimization of consumption efficiency, dynamic stability and economy in the scenario of high proportion of new energy access, and there are certain limitations in use. Summary of the Invention

[0005] The present invention proposes a coordinated control method for an AC-DC hybrid microgrid, which solves the problem that in the existing AC-DC hybrid microgrid system, due to the rigid control architecture being out of touch with the market mechanism, it is difficult to achieve the coordinated optimization of consumption efficiency, dynamic stability and economy in the scenario of high proportion of new energy access.

[0006] To solve the above technical problems, the technical solution adopted by the present invention is:

[0007] A coordinated control method for an AC-DC hybrid microgrid includes the following steps:

[0008] Step S1: Construct a multi-level microgrid architecture, including a main microgrid, a DC sub-microgrid and an AC sub-microgrid, and the main microgrid is connected to the DC sub-microgrid through a bidirectional interconnection converter;

[0009] Step S2: Deploy a hierarchical Multi-Agent control architecture, including a main microgrid agent, a DC sub-microgrid agent and an AC sub-microgrid agent;

[0010] Step S3: Execute three-level coordinated control. The main microgrid agent receives the main network dispatching instruction and decomposes it into sub-microgrid control objectives, coordinates the power distribution among the sub-microgrids based on the global optimization algorithm, and at the same time, the DC sub-microgrid agent and the AC sub-microgrid agent maintain the voltage and frequency stability through the local control algorithm and perform dynamic interaction with the main microgrid agent through the real-time feedback mechanism.

[0011] Further, the control method of the bidirectional interconnection converter in step S1 includes:

[0012] When the main microgrid supplies power to the DC sub - microgrid, a constant - voltage control mode is adopted, and the DC bus voltage is maintained within a preset threshold range by dynamically adjusting the converter duty cycle;

[0013] When the surplus power of the DC sub - microgrid is fed back to the main microgrid, it switches to a constant - power control mode, and the output power is matched in real - time based on the power demand curve issued by the main microgrid agent;

[0014] The trigger conditions for control - mode switching include: the DC bus voltage exceeding the limit, the emergency - level identification of the main - grid dispatching instruction, and the combined determination of the state of charge (SOC) of the energy storage unit in the sub - microgrid.

[0015] Furthermore, the execution of the global optimization algorithm in step S3 includes:

[0016] Construct a multi - objective optimization model. The optimization objectives include minimizing the power transmission loss between the main microgrid and the sub - microgrid, maximizing the local wind - solar - hydro - storage energy consumption rate, and restricting the AC bus voltage deviation and frequency fluctuation range;

[0017] Use an improved particle - swarm optimization algorithm to solve the optimization model, where the particle - swarm inertia weight is dynamically adjusted according to the number of iterations and the convergence speed;

[0018] When the local control strategy of the sub - microgrid conflicts with the main - grid objective, a coordination strategy is generated based on a dynamic - game model, which specifically includes: establishing a game revenue function for the main - microgrid agent and the sub - microgrid agent, where the revenue weights are associated with grid stability, economy, and carbon - emission indicators; determining the optimal power - distribution scheme through Nash - equilibrium solution, and decomposing the result into the power instruction and constraint conditions of the sub - microgrid.

[0019] Furthermore, the local control algorithm of the DC sub - microgrid in step S3 includes:

[0020] Based on the DC - bus voltage droop characteristic, the energy - storage charge - discharge power is adjusted in real - time. The droop coefficient is dynamically adjusted according to the SOC of the energy - storage unit, and the adjustment rule is: when the SOC is lower than 40%, the droop coefficient is reduced to give priority to charging; when the SOC is higher than 80%, the droop coefficient is increased to give priority to discharging;

[0021] When abnormal voltage fluctuations are detected, the super - capacitor energy - storage unit is started for transient compensation, which specifically includes: when the voltage fluctuation exceeds the preset threshold, triggering the calculation of the compensation power based on fuzzy logic; dynamically adjusting the response speed of the super - capacitor according to the output - power prediction error of the wind - solar power generation unit.

[0022] Furthermore, the local control algorithm of the AC sub - microgrid in step S3 includes:

[0023] Harmonics are suppressed through the coordinated control of the active filtering device and the static var compensator, which specifically includes: using a harmonic current detection algorithm to extract the load harmonic components in real time; generating a compensation current command through a proportional-resonant controller to drive the active filtering device to inject reverse harmonic current;

[0024] Frequency-active power droop control is used to maintain frequency stability. When the frequency deviation exceeds the threshold, secondary frequency modulation of the main microgrid agent is triggered, which specifically includes: the main microgrid agent adjusts the active power output of the energy storage unit according to the frequency deviation level; synchronously sending a frequency modulation command to adjacent sub-microgrid agents for coordinated regulation.

[0025] Furthermore, the coordinated control strategy of the energy storage unit in step S3 includes:

[0026] The energy storage unit on the main microgrid side responds to long-term power fluctuations, and its charge-discharge strategy is linked to the dynamic electricity price, which specifically includes: preferentially charging during low electricity price periods and preferentially discharging during peak periods; predicting the electricity price fluctuation trend based on the long short-term memory network (LSTM) and dynamically optimizing the charge-discharge time window;

[0027] The energy storage unit on the DC sub-microgrid side dynamically adjusts its priority based on the local new energy output change rate, which specifically includes: calculating the minute-level change rate of the wind and solar power output. When the change rate exceeds 10% / min, the response priority of the supercapacitor is increased; the charge-discharge threshold is differentially adjusted according to the rising / falling direction of the change rate.

[0028] Furthermore, the implementation of the dynamic interaction mechanism in step S3 includes:

[0029] The main microgrid agent periodically sends global policies to the sub-microgrid agents, which specifically includes: receiving the main grid dispatch command every 5 minutes, generating a global optimization result through an improved particle swarm optimization algorithm; encoding the power distribution command and constraints into the FIPA-ACL message format and sending them down;

[0030] The sub-microgrid agent adjusts local parameters in real time and feedbacks abnormal events through a low-latency communication channel, which specifically includes: updating local control parameters every 30 seconds, and the parameter adjustment is based on the weighted fusion of local sensor data and main microgrid instructions; when voltage or frequency over-limit is detected, immediate feedback is triggered, and the feedback information includes the over-limit level, timestamp, and recommended regulation measures.

[0031] Furthermore, the deployed Multi-Agent architecture in step S2 also includes a cross-market agent module for performing the following operations:

[0032] Point-to-point power trading between sub-microgrids, specifically including: generating trading orders based on blockchain smart contracts, with the order priority arranged in ascending order of electricity price and descending order of carbon emissions; matching buyer and seller agents through the Dutch auction mechanism, and the transaction price is dynamically generated by historical transaction data and real-time supply-demand ratio;

[0033] Submitting a flexibility service bid to the main grid, specifically including: when the local new energy consumption rate is lower than the preset threshold, generating a bidding strategy based on the Markov decision process; the bidding capacity is related to the available power of the energy storage unit and the regulation demand of the main microgrid.

[0034] Furthermore, the Multi-Agent communication and fault tolerance mechanism in step S2 includes:

[0035] The main microgrid agent subscribes to the real-time data of the sub-microgrid, and the communication protocol adopts the OPC UA architecture. The data packet encapsulation format includes: real-time operation data (voltage, frequency, power) is transmitted in JSON format; control instructions are transmitted in Protobuf binary format to ensure low latency;

[0036] Adopting a distributed auction mechanism to allocate main grid dispatching resources, specifically including: the sub-microgrid agent submits resource requirements and quotes to the auction pool; the main microgrid agent determines the optimal resource allocation plan according to the linear programming model;

[0037] Fault tolerance processing for faulty agent nodes, specifically including: monitoring the status of agent nodes through a heartbeat detection mechanism; when a node failure is detected, the redundant agent takes over the task and restarts the faulty process, and the historical status data is restored from the distributed database.

[0038] Furthermore, the deployment of the control algorithm in step S3 includes:

[0039] The local control algorithm realizes real-time calculation through a hardware acceleration module, specifically including: the droop control algorithm is executed by the FPGA, and the calculation period ≤ 1ms; the harmonic suppression algorithm is parallelly calculated by the GPU, and the processing delay of a single harmonic component < 10μs;

[0040] The global optimization algorithm is deployed on the edge computing node, specifically including: the improved particle swarm algorithm runs in a containerized environment, and resource isolation ensures task stability; the task scheduling period is configured as 10ms, and the priority is dynamically adjusted according to the grid emergency state.

[0041] The positive effects of the present invention are as follows: Through the multi-level architecture design of the main microgrid, DC sub-microgrid and AC sub-microgrid, the flexible expansion ability of AC-DC hybrid networking is realized, and the plug-and-play of diversified energy sources such as wind, light, water and storage is supported; Based on the hierarchical control architecture of Multi-Agent, the global optimization is combined with local autonomous control, which not only meets the main grid dispatching requirements, but also improves the rapid response ability of the sub-microgrid to fluctuating power sources; The three-level collaborative control mechanism effectively balances the contradiction between centralized dispatching and decentralized execution, and significantly improves the grid consumption efficiency and operation stability in the scenario of high proportion of new energy access. Specific embodiments

[0042] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative work shall fall within the scope of protection of the present invention.

[0043] A collaborative control method for an AC-DC hybrid microgrid includes the following steps:

[0044] Step S1: Construct a multi-level microgrid architecture, including a main microgrid, a DC sub-microgrid and an AC sub-microgrid, and connect the main microgrid and the DC sub-microgrid through a bidirectional interconnection converter;

[0045] Step S2: Deploy a hierarchical Multi-Agent control architecture, including a main microgrid agent, a DC sub-microgrid agent and an AC sub-microgrid agent;

[0046] Step S3: Execute three-level collaborative control. The main microgrid agent receives the main grid dispatching instruction and decomposes it into sub-microgrid control objectives, coordinates the power distribution among the sub-microgrids based on the global optimization algorithm, and at the same time, the DC sub-microgrid agent and the AC sub-microgrid agent maintain the voltage and frequency stability through the local control algorithm, and dynamically interact with the main microgrid agent through the real-time feedback mechanism.

[0047] This method first constructs a multi-level architecture including a main microgrid, a DC sub-microgrid and an AC sub-microgrid. The main microgrid integrates an energy storage unit, a photovoltaic power generation unit, an active power filter device and an AC load through an AC bus (AC BUS); The DC sub-microgrid connects an energy storage unit, a wind-solar power generation unit and a DC load through a DC bus (DC BUS), and the two realize AC-DC power interaction through a bidirectional interconnection converter. The AC sub-microgrid simulates a small microgrid system and is equipped with a distributed energy monitoring unit.

[0048] In terms of the control architecture, a hierarchical Multi-Agent system is deployed: the main microgrid agent is responsible for receiving the main grid dispatching instructions and generating global strategies, while the DC sub-microgrid agent and the AC sub-microgrid agent respectively execute local control. The main microgrid agent optimizes the power distribution among sub-microgrids by parsing the main grid instructions (such as power distribution targets and carbon emission constraints) using an improved particle swarm optimization algorithm, and at the same time coordinates the conflicts between the sub-microgrid control strategies and the main grid objectives. The sub-microgrid agents maintain the grid stability through local algorithms (such as voltage droop control and harmonic suppression), and upload the operating status to the main microgrid agent through a real-time feedback channel, forming a three-level closed-loop control of "main grid dispatching - global optimization - local execution".

[0049] Through the multi-level architecture design of the main microgrid, DC sub-microgrid and AC sub-microgrid, the flexible expansion ability of AC-DC hybrid networking is realized, supporting the plug-and-play of diverse energy sources such as wind, light, water and energy storage; based on the hierarchical control architecture of Multi-Agent, the global optimization is combined with local autonomous control, which not only meets the main grid dispatching requirements, but also improves the fast response ability of sub-microgrids to volatile power sources; the three-level collaborative control mechanism effectively balances the contradiction between centralized dispatching and decentralized execution, significantly improving the grid consumption efficiency and operating stability in scenarios with high proportion of new energy access.

[0050] The control method of the bidirectional interconnection converter in step S1 includes:

[0051] When the main microgrid supplies power to the DC sub-microgrid, a constant voltage control mode is adopted, and the DC bus voltage is maintained within a preset threshold range by dynamically adjusting the converter duty ratio;

[0052] When the surplus power of the DC sub-microgrid is fed back to the main microgrid, it switches to a constant power control mode, and the output power is matched in real time based on the power demand curve issued by the main microgrid agent;

[0053] The triggering conditions for the control mode switch include: the DC bus voltage exceeding the limit, the emergency level identification of the main grid dispatching instruction, and the combined determination of the state of charge (SOC) of the energy storage unit in the sub-microgrid.

[0054] Specifically, the control of the bidirectional interconnection converter dynamically switches the mode based on the supply and demand status of the main microgrid and the DC sub-microgrid:

[0055] Constant voltage control mode: When the main microgrid supplies power to the DC sub-microgrid, the converter maintains the stability of the DC bus voltage by adjusting the IGBT switching frequency, and the voltage reference value is set by the main microgrid agent according to the global strategy;

[0056] Constant power control mode: When the remaining power of the DC sub-microgrid is fed back to the main microgrid, the converter switches to the power tracking mode, and the output power is dynamically adjusted according to the demand curve issued by the main microgrid agent. The power command is updated in real time through the communication protocol.

[0057] Mode switching mechanism: Based on the comprehensive judgment of the DC bus voltage deviation, the emergency flag of the main grid dispatching instruction (such as frequency modulation demand), and the energy storage SOC state, the main microgrid agent triggers the switching instruction.

[0058] The dynamic mode switching mechanism of the bidirectional interconnection converter realizes the flexible two-way power mutual assistance between the AC and DC microgrids, avoiding power blockage or voltage instability caused by a single control mode; the switching logic based on multi-factor criteria (voltage, dispatching instruction, energy storage state) enhances the adaptability of the system to sudden supply and demand changes and reduces the transmission loss during the energy feedback process; it provides a standardized power interaction interface for the AC-DC hybrid microgrid, reducing the integration and operation and maintenance difficulties of complex energy systems.

[0059] The execution of the global optimization algorithm in step S3 includes:

[0060] Construct a multi-objective optimization model. The optimization objectives include minimizing the power transmission loss between the main microgrid and the sub-microgrid, maximizing the local wind-solar-hydro-storage energy consumption rate, and restricting the AC bus voltage deviation and frequency fluctuation range.

[0061] Use an improved particle swarm algorithm to solve the optimization model, where the particle swarm inertia weight is dynamically adjusted according to the iteration times and the convergence speed.

[0062] When the local control strategy of the sub-microgrid conflicts with the main grid objective, a coordination strategy is generated based on the dynamic game model, which specifically includes: establishing the game revenue functions of the main microgrid agent and the sub-microgrid agent, and associating the revenue weights with the grid stability, economy, and carbon emission indicators; determining the optimal power distribution plan through Nash equilibrium solution and decomposing the results into the power command and constraint conditions of the sub-microgrid.

[0063] Among them, the execution process of the global optimization algorithm is divided into three stages:

[0064] Multi-objective modeling: Establish an optimization model including power transmission loss, new energy consumption rate, voltage / frequency stability, and dynamically allocate the weight of the objective function according to the grid operation state.

[0065] Improved particle swarm solution: Introduce an adaptive inertia weight into the standard particle swarm algorithm, and the weight value decreases linearly with the increase of the iteration times to balance the global search and local convergence capabilities.

[0066] Conflict coordination: When the local strategy of the sub-microgrid (such as the energy storage charge and discharge plan) conflicts with the main grid objectives, a game revenue matrix of the main microgrid agent and the sub-microgrid agent is constructed. The revenue indicators cover economy, stability, and environmental protection. The optimal solution that takes into account the interests of multiple parties is generated through Nash equilibrium solution, and the result is decomposed into power instructions that can be executed by the sub-microgrid.

[0067] The multi-objective optimization model takes into account economy (reducing losses), environmental protection (improving new energy consumption), and stability (voltage / frequency constraints), breaking through the limitations of traditional single-objective optimization; the improved particle swarm optimization algorithm balances the global search speed and local convergence accuracy through dynamic weight adjustment, improving the optimization efficiency in complex scenarios; the conflict coordination mechanism based on game theory solves the interest conflicts between the main grid and the sub-microgrid, and realizes a win-win power distribution scheme for multiple parties.

[0068] The local control algorithm of the DC sub-microgrid in step S3 includes:

[0069] Based on the droop characteristics of the DC bus voltage, the energy storage charge and discharge power is adjusted in real time. The droop coefficient is dynamically adjusted according to the SOC of the energy storage unit. The adjustment rule is: when the SOC is lower than 40%, the droop coefficient is reduced to give priority to charging; when the SOC is higher than 80%, the droop coefficient is increased to give priority to discharging;

[0070] When abnormal voltage fluctuations are detected, the supercapacitor energy storage unit is started for transient compensation, specifically including: when the voltage fluctuation exceeds the preset threshold, the compensation power calculation based on fuzzy logic is triggered; the response speed of the supercapacitor is dynamically adjusted according to the output prediction error of the wind-solar power generation unit.

[0071] Among them, the local control of the DC sub-microgrid includes:

[0072] Voltage droop control: Dynamically adjust the energy storage charge and discharge power according to the real-time value of the DC bus voltage. The droop coefficient is linked with the energy storage SOC - when the SOC is lower, the droop coefficient is reduced to give priority to charging; when the SOC is higher, the coefficient is increased to give priority to discharging;

[0073] Transient compensation mechanism: When the voltage fluctuation exceeds the threshold, the supercapacitor energy storage unit is started for rapid compensation. The compensation power is calculated by a fuzzy logic controller, and the input variables include voltage deviation, wind-solar power generation prediction error, and load change rate;

[0074] Wind-solar fluctuation suppression: Based on the sliding window algorithm, calculate the minute-level wind-solar output change rate. When the change rate exceeds the set threshold, the response priority of the supercapacitor is increased and the charge and discharge rate is adjusted.

[0075] The SOC adaptive droop control strategy avoids overcharging / overdischarging of energy storage units, extends the equipment life and improves energy utilization efficiency; the fuzzy logic-driven transient compensation mechanism realizes precise suppression of second-level disturbances of wind and light fluctuations, and ensures the power supply quality of the DC microgrid; the hierarchical response strategy (energy storage regulation + supercapacitor compensation) effectively deals with power fluctuations on different time scales and reduces the main grid frequency modulation pressure.

[0076] The local control algorithm of the AC sub-microgrid in step S3 includes:

[0077] Harmonics are suppressed through the coordinated control of the active filtering device and the static var compensator, specifically including: using the harmonic current detection algorithm to extract the load harmonic components in real time; generating the compensation current command through the proportional-resonant controller to drive the active filtering device to inject reverse harmonic current;

[0078] The frequency-active power droop control is adopted to maintain frequency stability. When the frequency deviation exceeds the threshold, the secondary frequency modulation of the main microgrid agent is triggered, specifically including: the main microgrid agent adjusts the active power output of the energy storage unit according to the frequency deviation level; synchronously sending the frequency modulation command to the adjacent sub-microgrid agents for coordinated regulation.

[0079] Among them, the AC sub-microgrid control is divided into two parts: harmonic suppression and frequency regulation:

[0080] Harmonic suppression: The active filtering device is used to detect the load harmonic current in real time, the proportional-resonant controller is used to generate the reverse compensation current, and at the same time, the static var compensator (SVG) is coordinated to dynamically adjust the reactive power, and the total harmonic distortion rate (THD) is controlled within the allowable range;

[0081] Frequency regulation: The frequency-active power droop control is adopted. When the frequency deviation is detected, the active power output of the local energy storage unit is preferentially adjusted; if the deviation continues to expand, a frequency modulation request is sent to the main microgrid agent, and the main microgrid agent coordinates the energy storage units of adjacent sub-microgrids to participate in frequency modulation together to form regional frequency coordinated support.

[0082] The harmonic and reactive power coordinated control strategy solves the power quality problems caused by nonlinear loads and reduces the risk of harmonic damage to sensitive equipment; the main-sub microgrid linked frequency regulation mechanism expands the scale of the frequency modulation resource pool and enhances the system's resistance to load mutations; through the division of labor between local fast frequency modulation and global coordinated support, the dependence on traditional frequency modulation power plants is reduced and the grid operation cost is lowered.

[0083] The coordinated control strategy of the energy storage unit in step S3 includes:

[0084] The energy storage unit on the main microgrid side responds to long-term power fluctuations, and its charging and discharging strategy is linked to dynamic electricity prices, specifically including: preferentially charging during low electricity price periods and preferentially discharging during peak periods; predicting the electricity price fluctuation trend based on the long short-term memory network (LSTM) and dynamically optimizing the charging and discharging time window;

[0085] The energy storage unit on the DC sub-microgrid side dynamically adjusts the priority based on the local new energy output change rate, specifically including: calculating the minute-level change rate of the wind and solar power generation output, and when the change rate exceeds 10% / min, enhancing the response priority of the supercapacitor; differentially adjusting the charging and discharging thresholds according to the rising / falling direction of the change rate.

[0086] Among them, the energy storage collaborative control strategy includes:

[0087] Energy storage on the main microgrid side: Based on the time-of-use electricity price signal and the LSTM neural network prediction model, it charges during low electricity price periods and discharges during peak periods, and the charging and discharging plan is updated every 15 minutes; at the same time, the SOC balancing algorithm is adopted, and when the SOC difference of each energy storage unit exceeds 20%, the balancing control module is triggered to reallocate the charging and discharging tasks;

[0088] Energy storage on the DC sub-microgrid side: Real-time calculates the change rate of the wind and solar power generation output. When the change rate exceeds the set threshold, the supercapacitor is started for second-level compensation, and the charging and discharging trigger thresholds are differentially set according to the change direction (such as a sudden increase in wind speed leading to a sudden increase in power).

[0089] Linking electricity price prediction with charging and discharging strategies to maximize the economic benefits of the energy storage system and shorten the investment return period; dynamically adjusting the priority based on the change rate of new energy output to optimize the division of labor efficiency of the energy storage unit and avoid resource idleness or overload; designing differential charging and discharging thresholds to improve the response accuracy and reliability of the energy storage in the face of wind and solar uncertainties.

[0090] The implementation of the dynamic interaction mechanism in step S3 includes:

[0091] The main microgrid agent periodically issues global policies to the sub-microgrid agents, specifically including: receiving the main grid dispatch instructions every 5 minutes, generating global optimization results through the improved particle swarm optimization algorithm; encoding the power distribution instructions and constraint conditions into the FIPA-ACL message format and issuing them;

[0092] The sub-microgrid agent adjusts local parameters in real time and feedbacks abnormal events through a low-latency communication channel, specifically including: updating local control parameters every 30 seconds, and the parameter adjustment is based on the weighted fusion of local sensor data and main microgrid instructions; when detecting voltage or frequency over-limit, triggering immediate feedback, and the feedback information includes the over-limit level, timestamp, and recommended control measures.

[0093] Among them, the implementation process of the dynamic interaction mechanism:

[0094] Policy distribution: The main microgrid agent receives the main grid dispatching instructions every 5 minutes, generates a global policy through an improved particle swarm optimization algorithm, encodes the power distribution instructions into the FIPA-ACL standard message format, and distributes them to the sub-microgrid agents via the OPC UA communication protocol;

[0095] Local execution and feedback: The sub-microgrid agent collects local voltage, frequency, and power data every 30 seconds, adjusts the control parameters after weighted fusion with the main microgrid instructions; when voltage / frequency over-limit is detected, immediately package the over-limit level, timestamp, and recommended measures (such as increasing the energy storage power) to form a feedback message, and upload it to the main microgrid agent through a low-latency channel;

[0096] Conflict coordination: After receiving the feedback, the main microgrid agent calls the dynamic game model to recalculate the global policy, update the instructions, and distribute them.

[0097] Standardized communication protocols (FIPA-ACL, OPC UA) ensure the efficient parsing and reliable transmission of cross-layer control instructions; the local parameter fusion mechanism (sensor data + main grid instructions) enhances the accuracy and environmental adaptability of sub-microgrid decision-making; the instant feedback of abnormal events and the dynamic policy update mechanism reduce the risk of cascading failures caused by voltage / frequency over-limit.

[0098] The Multi-Agent architecture deployed in step S2 also includes a cross-market agent module for performing the following operations:

[0099] Point-to-point power trading between sub-microgrids, specifically including: generating trading orders based on blockchain smart contracts, and arranging the order priorities in ascending order of electricity price and descending order of carbon emissions; matching buyer and seller agents through the Dutch auction mechanism, and the transaction price is dynamically generated by historical transaction data and real-time supply-demand ratio;

[0100] Submitting flexibility service bids to the main grid, specifically including: when the local new energy consumption rate is lower than the preset threshold, generating a bidding strategy based on the Markov decision process; the bidding capacity is related to the available power of the energy storage unit and the regulation requirements of the main microgrid.

[0101] The operation process of the cross-market coordination mechanism is as follows:

[0102] Point-to-point trading: The sub-microgrid agent publishes electricity supply and demand information through blockchain smart contracts, and the contract terms include the electricity price cap, carbon emission constraints, and delivery time. The transaction matching adopts the Dutch auction mechanism: the buyer agent submits an offer, and the seller agent matches in ascending order of price, and the transaction price is dynamically generated according to historical transaction data and real-time supply-demand ratio;

[0103] Flexibility service bidding: When the local new energy consumption rate is lower than the set threshold, the cross-market agent module starts the Markov decision process, generates a bidding strategy based on historical electricity prices, weather forecasts, and available energy storage capacity, and the bidding capacity is associated with the regulation demand priority of the main microgrid.

[0104] The combination of blockchain smart contracts and the Dutch auction mechanism constructs a transparent and trustworthy distributed power trading market to promote the local consumption of new energy; the bidding strategy driven by Markov decision-making improves the success rate and economic benefits of flexibility service bidding; embedding the market mechanism into the microgrid control system provides economic incentives for grid regulation and accelerates the market-oriented transformation of the energy system.

[0105] The Multi-Agent communication and fault tolerance mechanism in step S2 includes:

[0106] The main microgrid agent subscribes to the real-time data of the sub-microgrid. The communication protocol adopts the OPC UA architecture. The data packet encapsulation format includes: real-time operation data (voltage, frequency, power) is transmitted in JSON format; control instructions are transmitted in Protobuf binary format to ensure low latency;

[0107] Adopt a distributed bidding mechanism to allocate main grid dispatching resources, specifically including: the sub-microgrid agent submits resource requirements and bids to the bidding pool; the main microgrid agent determines the optimal resource allocation plan according to the linear programming model;

[0108] Fault tolerance processing for faulty agent nodes, specifically including: monitoring the status of agent nodes through a heartbeat detection mechanism; when a node failure is detected, the redundant agent takes over the task and restarts the faulty process, and the historical status data is restored from the distributed database.

[0109] Among them, the Multi-Agent communication and fault tolerance mechanism:

[0110] Data communication: The main microgrid agent subscribes to the real-time operation data of the sub-microgrid. The data message encapsulates key parameters such as voltage, frequency, and power in JSON format, and the control instruction is compressed and transmitted in Protobuf binary format to ensure that the end-to-end delay is less than 50ms;

[0111] Resource bidding: The sub-microgrid agent submits resource requirements (such as frequency modulation capacity, standby power) and bids to the bidding pool. The main microgrid agent calculates the optimal allocation plan based on the linear programming model, and the winning bid results are publicized through the blockchain ledger;

[0112] Fault tolerance: Monitor the status of agent nodes through a heartbeat detection mechanism. When a node failure is detected, the redundant agent automatically takes over the control task and restores the latest status data before the failure from the distributed database to ensure control continuity.

[0113] The heterogeneous data encapsulation format (JSON + Protobuf) balances readability and transmission efficiency, meeting the dual requirements of real-time control and historical analysis; the distributed auction mechanism optimizes the fair allocation of main network resources, avoiding resource monopolies caused by a single decision-making node; the redundant proxy and heartbeat detection mechanism significantly improve system availability, ensuring continuous and reliable operation under extreme conditions.

[0114] The deployment of the control algorithm in step S3 includes:

[0115] The local control algorithm is implemented by the hardware acceleration module for real-time calculation, specifically including: the droop control algorithm is executed by the FPGA, and the calculation period ≤ 1ms; the harmonic suppression algorithm is parallel computed by the GPU, and the processing delay of a single harmonic component < 10μs;

[0116] The global optimization algorithm is deployed on the edge computing node, specifically including: the improved particle swarm algorithm runs in a containerized environment, and resource isolation ensures task stability; the task scheduling period is configured as 10ms, and the priority is dynamically adjusted according to the grid emergency status.

[0117] Among them, the control algorithm deployment plan:

[0118] Local control algorithm: Algorithms with high real-time requirements such as droop control and harmonic suppression are deployed on the FPGA hardware acceleration module, and the parallel computing architecture is used to achieve microsecond-level response;

[0119] Global optimization algorithm: Computationally intensive algorithms such as the improved particle swarm algorithm and dynamic game model run on containerized edge computing nodes, and resource scheduling is performed through Kubernetes. The task period is dynamically adjusted according to the grid emergency status (e.g., shortened to 1 second during a fault);

[0120] Communication interface: Supports Modbus TCP and IEC 61850 protocols. The 5G NR wireless communication module is used for remote instruction transmission to ensure the real-time performance and reliability of the wide-area control network.

[0121] The FPGA / GPU hardware acceleration achieves microsecond-level response for key algorithms, breaking through the real-time bottleneck of traditional software computing; containerized deployment and dynamic resource scheduling improve the utilization rate of edge computing nodes and reduce the hardware investment cost; multi-protocol communication interfaces support seamless compatibility with existing grid equipment, accelerating the large-scale implementation of technical solutions.

[0122] The above-described embodiments are described in detail and specifically, expressing the preferred embodiments of the present invention. They are only used to illustrate the technical ideas and features of the present invention. The purpose is to enable those skilled in the art to understand the content of the present invention and implement it accordingly. However, it is not limited to the present invention alone. The patent scope of the present invention cannot be limited only by this embodiment. That is, any equivalent changes or modifications made in accordance with the spirit disclosed by the present invention, for researchers or technicians in the field, within the structure of the present invention, local improvements within the system and changes and transformations between subsystems are still within the patent scope of the present invention.

Claims

1. A coordinated control method for AC-DC hybrid microgrids, characterized in that, Including the following steps: Step S1: Construct a multi-level microgrid architecture, including a main microgrid, a DC sub-microgrid, and an AC sub-microgrid. The main microgrid is connected to the DC sub-microgrid through a bidirectional interconnection converter; Step S2: Deploy a hierarchical Multi-Agent control architecture, including a main microgrid agent, a DC sub-microgrid agent, and an AC sub-microgrid agent; Step S3: Execute three-level coordinated control. The main microgrid agent receives the main grid dispatch instruction and decomposes it into sub-microgrid control objectives, coordinates the power distribution among sub-microgrids based on a global optimization algorithm. At the same time, the DC sub-microgrid agent and the AC sub-microgrid agent maintain voltage and frequency stability through local control algorithms and dynamically interact with the main microgrid agent through a real-time feedback mechanism.

2. The collaborative control method for an AC-DC hybrid microgrid according to claim 1, wherein The control method of the bidirectional interconnection converter in Step S1 includes: When the main microgrid supplies power to the DC sub-microgrid, adopt a constant voltage control mode, and maintain the DC bus voltage within a preset threshold range by dynamically adjusting the converter duty cycle; When the surplus power of the DC sub-microgrid is fed back to the main microgrid, switch to a constant power control mode, and match the output power in real time based on the power demand curve issued by the main microgrid agent; The triggering conditions for control mode switching include: the DC bus voltage exceeding the limit, the emergency level identifier of the main grid dispatch instruction, and the combined determination of the state of charge (SOC) of the energy storage unit in the sub-microgrid.

3. A coordinated control method for an AC-DC hybrid microgrid according to claim 1, characterized in that The execution of the global optimization algorithm in Step S3 includes: Construct a multi-objective optimization model. The optimization objectives include minimizing the power transmission loss between the main microgrid and the sub-microgrid, maximizing the local wind-solar-hydro energy consumption rate, and constraining the AC bus voltage deviation and frequency fluctuation range; Adopt an improved particle swarm optimization algorithm to solve the optimization model, where the particle swarm inertia weight is dynamically adjusted according to the iteration number and convergence speed; When the local control strategy of the sub-microgrid conflicts with the main grid objective, generate a coordination strategy based on a dynamic game model, specifically including: establishing a game revenue function for the main microgrid agent and the sub-microgrid agent, and the revenue weights are associated with grid stability, economy, and carbon emission indicators; determining the optimal power distribution plan through Nash equilibrium solution, and decomposing the result into the power instruction and constraint conditions of the sub-microgrid.

4. A coordinated control method for an AC / DC hybrid microgrid according to claim 1, characterized in that, The local control algorithm of the DC sub-microgrid in Step S3 includes: Based on the DC bus voltage droop characteristic, adjust the energy storage charge and discharge power in real time. The droop coefficient is dynamically adjusted according to the SOC of the energy storage unit. The adjustment rule is: when the SOC is lower than 40%, reduce the droop coefficient to give priority to charging; when the SOC is higher than 80%, increase the droop coefficient to give priority to discharging; When detecting abnormal voltage fluctuations, start the supercapacitor energy storage unit for transient compensation, specifically including: when the voltage fluctuation exceeds the preset threshold, trigger the compensation power calculation based on fuzzy logic; dynamically adjust the response speed of the supercapacitor according to the output power prediction error of the wind-solar power generation unit.

5. A coordinated control method for an AC-DC hybrid microgrid according to claim 1, characterized in that The local control algorithm of the AC sub-microgrid in Step S3 includes: Harmonics are suppressed through coordinated control of active power filter devices and static VAR compensators. This includes: using a harmonic current detection algorithm to extract load harmonic components in real time; generating compensation current commands through a proportional resonant controller to drive the active power filter device to inject reverse harmonic current; Frequency-active power droop control is used to maintain frequency stability. When the frequency deviation exceeds the threshold, the secondary frequency regulation of the main microgrid agent is triggered. Specifically, the main microgrid agent adjusts the active output of the energy storage unit according to the frequency deviation level; and simultaneously issues frequency regulation instructions to adjacent sub-microgrid agents for coordinated regulation.

6. The collaborative control method of an AC-DC hybrid microgrid according to claim 1, wherein The coordinated control strategy of the energy storage unit in step S3 includes: The energy storage unit on the main microgrid responds to long-term power fluctuations, and its charging and discharging strategies are linked to dynamic electricity prices. Specifically, this includes: prioritizing charging during low-price periods and discharging during peak periods; using a long short-term memory (LSTM) network to predict electricity price fluctuation trends and dynamically optimize charging and discharging time windows; The energy storage unit on the DC sub-microgrid side dynamically adjusts its priority based on the change rate of local renewable energy output. Specifically, it calculates the minute-by-minute change rate of wind and solar power generation output, and when the change rate exceeds 10% / min, increases the response priority of the supercapacitor; and differentially adjusts the charge and discharge thresholds according to the direction of the change rate, whether it rises or falls.

7. A coordinated control method for an AC-DC hybrid microgrid according to claim 1, characterized in that The implementation of the dynamic interaction mechanism in step S3 includes: The master microgrid agent periodically sends global strategies to the sub-microgrid agents. Specifically, it receives dispatch instructions from the master grid every 5 minutes, generates global optimization results using an improved particle swarm optimization algorithm, and encodes power allocation instructions and constraints into FIPA-ACL message format for transmission. The sub-microgrid agent adjusts local parameters in real time and reports abnormal events through a low-latency communication channel. Specifically, local control parameters are updated every 30 seconds, and parameter adjustments are based on a weighted fusion of local sensor data and main microgrid instructions. When voltage or frequency exceeds the limit, immediate feedback is triggered, and the feedback information includes the limit level, timestamp, and recommended control measures.

8. A coordinated control method for an AC-DC hybrid microgrid according to claim 1, characterized in that, The Multi-Agent architecture deployed in step S2 also includes a cross-market agent module for performing the following operations: Peer-to-peer electricity trading between microgrids involves: generating transaction orders based on blockchain smart contracts, with order priorities sorted by ascending electricity price and descending carbon emissions; matching buyers and sellers through a Dutch auction mechanism, with transaction prices dynamically generated based on historical transaction data and real-time supply-demand ratios; Submitting a flexibility service bid to the main grid, specifically including: generating a bidding strategy based on a Markov decision process when the local renewable energy consumption rate is lower than a preset threshold; associating the bidding capacity with the available power of the energy storage unit and the regulation requirements of the main microgrid.

9. A coordinated control method for an AC-DC hybrid microgrid according to claim 1, characterized in that, The Multi-Agent communication and fault tolerance mechanism in step S2 includes: The master microgrid agent subscribes to the real-time data of the sub-microgrid. The communication protocol adopts the OPC UA architecture. The data packet encapsulation format includes: real-time operation data including voltage, frequency, and power are transmitted in JSON format; control instructions are transmitted in Protobuf binary format to ensure low latency; The main network scheduling resources are allocated using a distributed auction mechanism, specifically including: the sub - microgrid agent submits resource requirements and bids to the auction pool; the main - microgrid agent determines the optimal resource allocation plan according to the linear programming model; Fault - tolerant processing of the fault - agent node, specifically including: monitoring the status of the agent node through the heartbeat detection mechanism; when a node failure is detected, the redundant agent takes over the task and restarts the failed process, and the historical status data is restored from the distributed database.

10. A coordinated control method for an AC-DC hybrid microgrid according to claim 1, characterized in that The deployment of the control algorithm in step S3 includes: The local control algorithm is implemented by the hardware acceleration module for real - time calculation, specifically including: the droop control algorithm is executed by the FPGA, and the calculation period ≤ 1ms; the harmonic suppression algorithm is calculated in parallel by the GPU, and the processing delay of a single harmonic component < 10μs; The global optimization algorithm is deployed on the edge - computing node, specifically including: the improved particle swarm algorithm runs in a containerized environment, and resource isolation ensures task stability; the task scheduling period is configured as 10ms, and the priority is dynamically adjusted according to the emergency state of the power grid.