A micro-grid autonomous optimization control method based on a multi-agent system

Abstract

The application discloses a kind of microgrid autonomous optimization control method based on multi-agent system, and the application relates to microgrid optimization control technical field, solve in the extreme working condition of communication interruption or island startup, only rely on local measurable signal, realize the technical problem of autonomous optimization control of frequency steady-state deviation, power is according to actual regulation ability fair distribution, and with adaptive disturbance response capability, the application introduces integral gain on the basis of traditional droop item to eliminate frequency steady-state deviation, simultaneously design the dynamic base power that can be event triggered to update, so that system can maintain frequency stability, and can also dynamically adjust output reference according to unit actual capacity, break through the limitation of traditional droop fixed proportion distribution, based on local SOC, rated capacity and safety boundary, calculate normalized available energy margin, set initial using proportional distribution method, ensure that high SOC unit bears more initial output, avoid transient SOC imbalance from source, prolong the overall service life of battery.

Description

Technical Field

[0001] This invention relates to the field of microgrid optimization control technology, specifically to a microgrid autonomous optimization control method based on a multi-agent system. Background Technology

[0002] As the penetration rate of renewable energy continues to increase, microgrids, as a key carrier for enhancing energy resilience and flexibility, face severe challenges in their operation and control strategies.

[0003] According to patent publication number CN121055464A, a unified distributed dynamic optimization control method and device for an autonomous microgrid is disclosed. The method includes: embedding a frequency regulator at a virtual synchronous converter, estimating the average frequency of the autonomous microgrid through the frequency regulator, comparing the average frequency with the rated frequency to generate an error term, and inputting the error term into an integrator to obtain a first power correction term; embedding a benefit optimizer at the virtual synchronous converter, generating a neighborhood error through the benefit optimizer, and inputting the neighborhood error into an integrator to obtain a second power correction term; adding the first power correction term and the second power correction term to a reference power term to obtain a power setting term, and using the power setting term as the input of the virtual synchronous converter.

[0004] Furthermore, this paper combines existing autonomous optimization control methods, including communication-free droop control methods and multi-agent system-based optimization control methods. Firstly, regarding the communication-free droop control method, this method achieves autonomous power allocation among multiple power sources through frequency-active power droop characteristics, and is simple in structure and plug-and-play. However, it has an inherent steady-state frequency deviation, which cannot meet the frequency accuracy requirements for islanded operation; at the same time, power allocation relies solely on a preset droop coefficient, without considering the state of charge (SOC) of each energy storage unit, which can easily lead to over-discharge of low SOC units and idleness of high SOC units, accelerating battery aging and disrupting system lifespan balance. Secondly, regarding the optimization control method for multi-agent systems, this method achieves economic scheduling, SOC balancing, and dynamic frequency adjustment through distributed agent cooperative optimization, theoretically approaching the global optimum. However, it heavily relies on a reliable communication network. Once communication is interrupted or delayed, the system will lose its coordination ability and may even degenerate into an uncontrolled state. Although some solutions propose communication failure degradation mechanisms, the degraded local control still uses the traditional droop strategy, which cannot eliminate frequency deviation or guarantee the fairness of power allocation. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a microgrid autonomous optimization control method based on a multi-agent system. This method solves the problem of achieving frequency steady-state control without deviation, fair power allocation according to actual adjustment capabilities, and adaptive disturbance response capability under extreme operating conditions such as communication interruption or islanded startup, relying solely on locally measurable signals.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a microgrid autonomous optimization control method based on a multi-agent system, comprising the following steps: In an autonomous control mode, data is collected from multiple agents of the microgrid to construct a self-healing droop reference power calculation model. The local active power reference value of the i-th unit is defined as P. ref，i = -m(f-f0)+ Where f0 is the rated frequency, m is the droop factor, and k a >0 represents the integral gain. For dynamic base power; The dynamic base power is determined by acquiring the local output active power in real time. If Pi>0, it is determined to be in a discharging state; otherwise, it is determined to be in a charging state. Based on the current state of charge (SOC) i Rated capacity C rated，i Permissible discharge lower limit SOC min，i Allowable maximum SOC max，i and maximum discharge power P max，i Calculate the normalized available energy margin; The initial dynamic base power is set using a proportional allocation method, so that the initial output of each energy storage unit is proportional to its available energy margin. When the preset event triggering conditions are met, the dynamic base power is updated, and the updated value is substituted into the self-recovering droop model to generate the final active power command, so as to achieve autonomous control with no frequency steady-state error and fair power distribution according to the regulation capability.

[0007] As a further aspect of the present invention, the normalization can be calculated using the energy margin as follows: If it is in a discharge state, then ; If it is in a charging state, then ; Initial dynamic base power Where N is the total number of energy storage units participating in regulation, and j is the number of distributed energy storage units. This represents the total power demand of the system.

[0008] As a further aspect of the present invention, the update of the dynamic base power is based on the local adjustment capability weight: If charging is in progress, weight It is proportional to the current remaining dischargeable amount, then SOC i The current state of charge (SOC) min The minimum SOC threshold for permissible discharge, SOC max The maximum SOC threshold that allows charging; If it is in a discharge state, weight It is proportional to the current remaining rechargeable capacity, then .

[0009] As a further aspect of the present invention, the preset event triggering conditions include: Frequency events: consecutive T f Within 2 seconds, |f-f0| > frequency deviation threshold ; State events: | |> ; Mode switching event: Grid connection switch status S grid Changes occurred, S grid =0 indicates island mode, S grid =1 indicates grid connection mode.

[0010] As a further aspect of the present invention, the frequency deviation threshold is dynamically set based on the system's equivalent inertia: System equivalent observation H eq According to the formula, through a short-time power step... Perform calculations, where This represents the sudden change in active power, and df / dt represents the rate of change of frequency. Then, according to the calculation formula... Calculate the frequency deviation threshold ,in As a reference threshold, To adjust the index, and ∈[0.3, 0.7].

[0011] As a further aspect of the present invention, the SOC change threshold Calculation based on available capacity after minimum effective regulation energy and temperature correction: Obtain the rated capacity C of the energy storage unit rated The minimum effective adjustment step size is preset according to the battery type. Then according to the formula Calculate the SOC change threshold C available (T) represents the actual usable capacity at the current temperature T, and C available (T) = , This is the temperature correction factor.

[0012] As a further aspect of the present invention, the update rule for the dynamic base power is as follows: ,in The global adjustment step size is dynamically calculated based on the disturbance intensity and the weight of the local adjustment capability.

[0013] As a further aspect of the present invention, the global adjustment step size is dynamically calculated based on the disturbance intensity and the local adjustment capability weight as follows: According to the formula Calculate the severity of the current disturbance, where f is the local measurement frequency, f0 is the rated frequency, and df / dt is the rate of frequency change. and These are weighting coefficients, based on the severity of the obtained perturbation. Normalize it. ,in The preset maximum disturbance intensity; Obtain local adjustment capability weights By fusing the disturbance intensity with the local capability, the global adjustment step size of the i-th unit is obtained. ,in Used as the baseline step size.

[0014] As a further aspect of the present invention, the global adjustment step size is processed by a first-order low-pass filter: ,in These are the filter coefficients. ∈[0.1, 0.3].

[0015] This invention provides a microgrid autonomous optimization control method based on a multi-agent system. Compared with existing technologies, it has the following advantages: This invention employs a dual-mode hierarchical collaborative control architecture. When communication is normal, multi-agent collaborative optimization is enabled to achieve economic scheduling and demand-side response. When communication is interrupted, it automatically and seamlessly degrades to pure local autonomous control, with each unit operating solely based on local sensing signals. This completely eliminates dependence on the communication network and significantly improves the system's robustness under extreme conditions.

[0016] This invention introduces integral gain to eliminate steady-state frequency deviation based on the traditional droop term. At the same time, it designs a dynamic base power that can be updated by event triggering, so that the system can maintain frequency stability and dynamically adjust the output benchmark according to the actual capacity of the cells. This breaks through the limitations of the traditional fixed ratio allocation of droop. Secondly, based on the local SOC, rated capacity and safety boundary, it calculates the normalized available energy margin and uses a proportional allocation method to set the initial value, ensuring that high SOC cells can bear more initial output, avoiding transient SOC imbalance from the source and extending the overall battery life.

[0017] The frequency event threshold of this invention varies with the equivalent inertia of the system. Smaller inertia systems are more sensitive, while larger inertia systems are more robust. The SOC change threshold is based on the minimum effective adjustment energy and incorporates temperature compensation to ensure that each adjustment has physical meaning and adapts to environmental changes. The adjustment step size is jointly determined by the local frequency deviation, the frequency change rate, and the adjustment capability weight, and is smoothed by a first-order low-pass filter. It has a fast response to large disturbances and stable convergence to small disturbances, completely solving the oscillation or sluggishness problem caused by fixed step size. Attached Figure Description

[0018] Figure 1 This is a flowchart of the microgrid autonomous optimization control method of the present invention. Detailed Implementation

[0019] 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.

[0020] Please see Figure 1 This application provides a microgrid autonomous optimization control method based on a multi-agent system, which specifically includes the following steps: Step 1: Construct a hierarchical collaborative control architecture. When communication is normal, the multi-agent collaborative optimization mode is activated. When communication is interrupted, it automatically degrades to pure local autonomous control. Each unit operates solely based on local sensing information, requiring no communication. Under normal communication conditions, the microgrid's multi-agent system is collected, specifically including distributed power generation agents, energy storage system agents, load agents, microgrid central coordination agents, and market / price agents. Among them, the distributed power generation agent is responsible for local power generation forecasting, output regulation, and fault response; the energy storage system agent manages charging and discharging strategies, maintains state of charge, and participates in frequency regulation and peak shaving; the load agent can adjust the load to achieve demand-side response or load transfer; the microgrid central coordination agent is responsible for global target setting, mode switching judgment, and issuing safety boundary constraints; and the market / price agent, in the scenario of participating in the electricity market, is responsible for price forecasting and bidding strategy formulation. Then, local sensing units are deployed in each distributed power controller to collect the following signals in real time: local bus frequency f, output active power P, energy storage unit state of charge (SOC), and grid connection switch status (S). grid ∈{0, 1}, the specific S grid =0 indicates that the microgrid is in islanded mode, S grid =1 indicates that the microgrid is in grid-connected mode.

[0021] Step 2: Based on the obtained parameters, design a self-recovering droop reference power calculation model and define the local active power reference value P. ref，i For: P ref，i = -m(f-f0)+ Where f0 is the rated frequency, m is the droop factor, and k a >0 represents the integral gain. This is the dynamic base power, and the dynamic base power The specific determination method is as follows: For the i-th distributed energy storage unit, the local controller of each distributed energy storage unit acquires its output active power Pi in real time through the power measurement module, and defines the positive direction of power: when Pi>0, it indicates that the energy storage unit is outputting electrical energy to the microgrid bus, which is determined to be a discharging state; when Pi<0, it indicates that the energy storage unit is absorbing electrical energy from the microgrid bus, which is determined to be a charging state. The initial dynamic power is calculated based on different states, and the method is as follows: Obtain the energy storage state parameters of the local distributed energy storage unit i, including the current state of charge (SOC). i Rated capacity C rated，i Permissible discharge lower limit SOC min，i Allowable maximum SOC max，i and maximum discharge power P max，i If it is in a discharge state, then according to the formula Calculate the normalized available energy margin If it is in a charging state, then according to the formula Calculate the normalized available energy margin Next, a proportional distribution method is used to ensure that the initial output of each unit is proportional to its available energy margin. The specific calculation formula is as follows: Where N is the total number of energy storage units participating in regulation, and j is the number of distributed energy storage units. The total power demand of the system can be calculated by using the initial frequency deviation and droop factor: m j This represents the droop coefficient of energy storage unit j; Step 3: Next, calculate the local weights based on their type. If the energy storage system is in a charging state, the weighting It is proportional to the current remaining dischargeable amount, then SOC i The current state of charge (SOC) min The minimum SOC threshold for permissible discharge, SOC max The maximum SOC threshold that allows charging, if the energy storage system is in a discharging state, the weights are... It is proportional to the current remaining rechargeable capacity, then ; Next, set the event triggering conditions, including frequency events: consecutive T. f Within 2 seconds, |f-f0|> Among them, regarding frequency events The threshold is not fixed; the specific calculation method is as follows: during the microgrid initialization phase or periodically online identification of the system's equivalent inertia H. eq And the system's equivalent inertia H eq According to the formula, through a short-time power step... Perform calculations, where The active power variation is represented by df / dt, which represents the rate of frequency change, based on the obtained system equivalent inertia H. eq Dynamically set frequency deviation threshold According to the calculation formula calculate ,in As a reference threshold, if =0.2Hz, then the corresponding H eq =2s, To adjust the index, and ∈[0.3, 0.7]; State events: | |> ,in The calculation method is to obtain the rated capacity C of the energy storage unit. rated The minimum effective adjustment step size is preset according to the battery type. Then according to the formula Calculate the SOC change threshold and introduce temperature compensation. Compensation will be provided, and the specific compensation methods are as follows: Based on the microgrid control requirements and battery characteristics, a physically meaningful energy step size is set. Furthermore, this value is calibrated at a reference temperature of 25°C. The ambient temperature T is collected in real time by a temperature sensor installed near the energy storage unit, and then a mapping relationship is established using the temperature-usable capacity curve provided by the battery manufacturer. C available (T) represents the actual usable capacity at the current temperature T. If a precise curve is unavailable, a piecewise linear approximation can be used. ; Next, calculate the current temperature. The corresponding SOC percentage is calculated using the following formula: and impose boundary restrictions ; Mode switching event: The grid-connected / islanded status changes; Based on the above events, the dynamic base power P is triggered only when any one of the conditions is met. base，i renew.

[0022] Step 4: When the triggering event occurs, execute the update rules: , The initial dynamic power is denoted as , where For adjustment amount, and ,in To adjust the step size globally, and >0, This indicates that output is increased when the frequency is low and decreased when the frequency is high, with the step size adjusted globally. It can easily lead to oscillations or slow response, affecting the global adjustment step size. Adaptive processing is performed, and the specific processing method is as follows: Based on locally measurable frequency dynamic information, the severity of the current disturbance is calculated. Where f is the local measurement frequency, f0 is the rated frequency, and df / dt is the rate of frequency change. and These are weighting coefficients, based on the severity of the obtained perturbation. Normalize it. ,in The preset maximum disturbance intensity; Next, obtain the local adjustment capability weights. By fusing the perturbation intensity with local capabilities, the exclusive step size of the i-th unit is obtained. ,in The reference step size is used, and a smoothing filter is introduced to prevent sudden changes in the step size. Specifically, a first-order low-pass filter is used. Perform adaptive step size Calculation, where These are the filter coefficients. ∈[0.1, 0.3], then the resulting adaptive step size Substitution get Next, the updated dynamic base power will be... Substituting the values ​​back into the computational model, we get P. ref，i = -m(f-f0)+ Local active power P ref，i .

[0023] Some of the data in the above formulas are numerical calculations with dimensions removed, and the contents not described in detail in this specification are all prior art known to those skilled in the art.

[0024] The above embodiments are only used to illustrate the technical methods of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical methods of the present invention without departing from the spirit and scope of the technical methods of the present invention.

Claims

1. A microgrid autonomous optimization control method based on a multi-agent system, characterized in that, The method includes the following steps: In an autonomous control mode, data is collected from multiple agents of the microgrid to construct a self-healing droop reference power calculation model. The local active power reference value of the i-th unit is defined as P. ref，i = -m(f-f0)+ Where f0 is the rated frequency, m is the droop factor, and k a >0 represents the integral gain. For dynamic base power; The dynamic base power is determined by acquiring the local output active power in real time. If Pi>0, it is determined to be in a discharging state; otherwise, it is determined to be in a charging state. Based on the current state of charge (SOC) i Rated capacity C rated，i Permissible discharge lower limit SOC min，i Allowable maximum SOC max，i and maximum discharge power P max，i Calculate the normalized available energy margin; The initial dynamic base power is set using a proportional allocation method, so that the initial output of each energy storage unit is proportional to its available energy margin. When the preset event triggering conditions are met, the dynamic base power is updated, and the updated value is substituted into the self-recovering droop model to generate the final active power command, so as to achieve autonomous control with no frequency steady-state error and fair power distribution according to the regulation capability.

2. The microgrid autonomous optimization control method based on a multi-agent system according to claim 1, characterized in that, The normalization can be calculated using the energy margin as follows: If it is in a discharge state, then ; If it is in a charging state, then ; Initial dynamic base power Where N is the total number of energy storage units participating in regulation, and j is the number of distributed energy storage units. This represents the total power demand of the system.

3. The microgrid autonomous optimization control method based on a multi-agent system according to claim 1, characterized in that, The update of the dynamic base power is based on local regulation capability weights: If charging is in progress, weight It is proportional to the current remaining dischargeable amount, then SOC i The current state of charge (SOC) min The minimum SOC threshold for permissible discharge, SOC max The maximum SOC threshold that allows charging; If it is in a discharge state, weight It is proportional to the current remaining rechargeable capacity, then .

4. The microgrid autonomous optimization control method based on a multi-agent system according to claim 1, characterized in that, The preset event triggering conditions include: Frequency events: consecutive T f Within 2 seconds, |f-f0| > frequency deviation threshold ; State events: | |> ; Mode switching event: Grid connection switch status S grid Changes occurred, S grid =0 indicates island mode, S grid =1 indicates grid connection mode.

5. The microgrid autonomous optimization control method based on a multi-agent system according to claim 4, characterized in that, The frequency deviation threshold is dynamically set based on the system's equivalent inertia. System equivalent observation H eq According to the formula, through a short-time power step... Perform calculations, where This represents the sudden change in active power, and df / dt represents the rate of change of frequency. Then, according to the calculation formula... Calculate the frequency deviation threshold ,in As a reference threshold, To adjust the index, and ∈[0.3, 0.7].

6. The microgrid autonomous optimization control method based on a multi-agent system according to claim 4, characterized in that, The SOC change threshold Calculation based on available capacity after minimum effective regulation energy and temperature correction: Obtain the rated capacity C of the energy storage unit rated The minimum effective adjustment step size is preset according to the battery type. Then according to the formula Calculate the SOC change threshold C available (T) represents the actual usable capacity at the current temperature T, and C available (T) = , This is the temperature correction factor.

7. The microgrid autonomous optimization control method based on a multi-agent system according to claim 1, characterized in that, The update rule for the dynamic base power is as follows: ,in The global adjustment step size is dynamically calculated based on the disturbance intensity and the weight of the local adjustment capability.

8. The microgrid autonomous optimization control method based on a multi-agent system according to claim 7, characterized in that, The global adjustment step size is dynamically calculated based on the disturbance intensity and the weight of the local adjustment capability as follows: According to the formula Calculate the severity of the current disturbance, where f is the local measurement frequency, f0 is the rated frequency, and df / dt is the rate of frequency change. and These are weighting coefficients, based on the severity of the obtained perturbation. Normalize it. ,in The preset maximum disturbance intensity; Obtain local adjustment capability weights By fusing the disturbance intensity with the local capability, the global adjustment step size of the i-th unit is obtained. ,in Used as the baseline step size.

9. The microgrid autonomous optimization control method based on a multi-agent system according to claim 8, characterized in that, The global adjustment step size is processed by a first-order low-pass filter: ,in These are the filter coefficients. ∈[0.1, 0.3].

Images