Intelligent micro-grid multi-mode collaborative optimization method and system

Abstract

The invention discloses an intelligent micro-grid multi-mode collaborative optimization method and system, and belongs to the field of intelligent power grids. The method comprises the following steps: collecting electrical data of a common connection point and a distributed unit, and quantifying a harmonic responsibility coefficient by using an improved complex independent component analysis and partial coherence method; constructing a dynamic electricity price game model, and solving an optimal excitation coefficient through a non-cooperative game; and building a multi-objective optimization model, correcting model prediction control parameters by means of a digital twinborn training reinforcement learning agent, and outputting a scheduling instruction in a rolling optimization manner to realize multi-mode switching. The system comprises a data acquisition layer, a regional collaboration layer, a cloud computing and digital twinning layer and an intelligent optimization core layer. According to the method, the electric energy quality and economic dispatching are fused, and the optimization intelligence and safety of the micro-grid are improved.

Description

Technical Field

[0001] This invention relates to the field of interdisciplinary technology of smart grid and power quality analysis, specifically to a multimodal collaborative optimization method and system for smart microgrids. Background Technology

[0002] With the large-scale integration of distributed power sources and loads such as photovoltaics, energy storage, and charging piles into smart microgrids, the harmonic problems generated by power electronic equipment are becoming increasingly prominent, seriously affecting power quality and restricting the economic and reliable operation of microgrids. This has become a problem that the industry urgently needs to overcome. In the Chinese patent document database, patent application number 202510733756.5 discloses a method, device, equipment, and storage medium for controlling harmonics in power grid interconnection. However, it suffers from the following problems in harmonic control and microgrid optimization: First, the quantification of harmonic liability is not comprehensive enough. This comparative document focuses on the photovoltaic subgrid, using wavelet packet decomposition, cross-correlation functions, and topological impedance matrices to calculate the proportion of harmonic liability, but it does not cover diverse distributed units such as energy storage and charging piles. Furthermore, its calculation method is difficult to dynamically update with the real-time operating status of the microgrid (such as frequent load fluctuations and random equipment start-ups and shutdowns), resulting in inaccurate harmonic liability identification and difficulty in accurately tracing the source of harmonics in the face of complex and ever-changing operating conditions. Secondly, economic efficiency and harmonic control are independent of each other. Traditional technologies mostly rely on adding filters and adjusting equipment parameters to suppress harmonics, while the comparative document addresses harmonic control by optimizing the output of control equipment. Neither approach links harmonic responsibility to economic dispatch. Although the comparative document constructs a two-layer game model, it only focuses on the allocation of control equipment output and fails to establish a dynamic electricity price mechanism based on real-time harmonic responsibility. As a result, distributed units lack the economic incentive to actively reduce harmonic emissions, leading to persistently high control costs. Third, the optimization algorithms and multi-mode collaborative capabilities are inadequate. Existing technologies mostly employ single algorithms, such as fuzzy control and basic reinforcement learning, failing to fully utilize digital twin technology to construct a closed-loop optimization system of "virtual simulation - physical execution." When switching between operating modes such as microgrid grid-connected economy and islanded power supply, it is difficult to balance economy, power quality, and system stability, easily leading to problems such as power surges and voltage fluctuations. Furthermore, it is difficult to cope with uncertainties such as random changes in wind and solar power output and sudden increases or decreases in load, resulting in a significant reduction in the practicality of the optimization results. Therefore, there is an urgent need for a new solution that can accurately calculate the harmonic responsibility of multiple distributed units, reduce harmonics with the help of dynamic electricity price incentives, and achieve multi-mode collaborative optimization by combining digital twins, so as to break through the current bottleneck of smart microgrid operation. Summary of the Invention

[0003] The purpose of this invention is to provide a multimodal collaborative optimization method and system for smart microgrids to solve the problems mentioned in the background art.

[0004] To achieve the above objectives, the present invention provides the following technical solution: a multi-modal collaborative optimization method for smart microgrids, comprising the following steps: S1: Real-time synchronous acquisition of voltage and current data at the microgrid's common connection point and the output of each distributed unit; S2: Based on the improved complex independent component analysis algorithm and partial coherence method, the data is processed to calculate the real-time harmonic responsibility coefficient η of each of the distributed units. i ; S3: Construct a dynamic electricity price game model. In this model, the actual settlement price of the i-th unit is related to the base price, the harmonic excitation coefficient, and the real-time harmonic responsibility coefficient of the unit. The harmonic excitation coefficient is solved dynamically through non-cooperative game theory. S4: To minimize the total system operating cost and total harmonic liability, a multi-objective optimization function is established and integrated with the dynamic electricity price game model to form an optimized scheduling model; S5: In the digital twin environment, the optimized scheduling model is solved using a model predictive control algorithm optimized by reinforcement learning to obtain the optimal scheduling instruction in the future rolling time domain. S6: The optimal scheduling command is issued to each distributed unit for execution, so as to realize the multimodal collaborative optimization operation of the microgrid.

[0005] Furthermore, the "based on the improved complex independent component analysis algorithm and partial coherence method" described in S2 specifically includes: S21: Perform a fast Fourier transform on the acquired time-series data to extract each harmonic component; S22: Construct an observation signal model that includes background harmonic voltage and currents from all potential harmonic sources; S23: An improved complex independent component analysis algorithm is used to perform blind source separation on the observed signal model, decoupling the independent harmonic source signals; S24: The contribution of each independent harmonic source to the total harmonic distortion rate at the common junction point is calculated using the partial coherence method, yielding the precise harmonic liability coefficient η for each distributed unit. i .

[0006] Furthermore, the "harmonic excitation coefficients dynamically solved through non-cooperative game theory" mentioned in S3 specifically refers to: Each distributed unit is regarded as an independent decision-making entity, whose strategy set is to adjust its own operating state to change the harmonic emission characteristics, and its revenue function is the sum of its own operating cost and harmonic penalty cost. The system operator dynamically adjusts this coefficient and performs iterative simulation calculations using a digital twin to ultimately guide all unit strategies to converge to the Nash equilibrium point that minimizes the total system cost. The coefficient value at this point is then used as the optimal harmonic excitation coefficient for the current time period.

[0007] Furthermore, the "multi-objective optimization function" described in S4 takes the sum of the total system operating cost and the total harmonic responsibility as the optimization objective. The total system operating cost includes the cost of purchasing electricity, the revenue from selling electricity, and the cost of equipment operation and maintenance. The total harmonic responsibility is the sum of the harmonic responsibility coefficients of all units, and the two are balanced by weighting coefficients.

[0008] Furthermore, S5 specifically includes: S51: Offline training phase: In the digital twin, a reinforcement learning agent is trained using historical data. Its state space is the system's historical and real-time data, its action space is the instructions for adjusting the parameters of the internal prediction model of the model prediction control, and its reward function is the negative value of the multi-objective optimization function. S52: Online application phase: The reinforcement learning agent outputs actions based on the real-time system state, dynamically correcting the prediction model of the model prediction control; the model prediction controller uses the corrected prediction model to perform rolling optimization and solve the optimized scheduling model.

[0009] Furthermore, the "multimodal collaborative optimization operation" mentioned in S6 refers to the system automatically and smoothly switching between grid-connected economic mode, grid-connected low-carbon mode, islanded supply guarantee mode, islanded power curtailment mode and grid support mode based on the optimization results and grid status.

[0010] A system for implementing the aforementioned smart microgrid multimodal cooperative optimization method includes: The data acquisition and sensing execution unit includes synchronous acquisition devices deployed at common connection points and at the exits of each distributed unit, as well as intelligent terminals with built-in unified protocol collaborative agents. The regional coordination control unit consists of regional controllers deployed in various regions of the microgrid, and is used to collect local data, perform coordinated control, and issue commands. The cloud computing and digital twin unit, including cloud servers and digital twin models deployed on them, is used to map the state of physical systems and provide a simulation training environment for algorithms; The intelligent optimization core unit includes a harmonic responsibility analysis module, a dynamic game module, and a reinforcement learning-model prediction control hybrid optimization module, which are used to perform optimization calculations in the method, such as harmonic responsibility coefficient calculation, dynamic electricity price game solution, and optimal scheduling model solution.

[0011] Furthermore, the smart terminal supports plug-and-play functionality, and any new distributed unit that conforms to the unified information model and communication protocol can automatically register its identity and capability parameters with the system when it is connected.

[0012] Furthermore, the digital twin model maintains real-time data synchronization with the physical microgrid and serves as a safe sandbox for exploratory training of the reinforcement learning agent, as well as a simulation test platform for new control strategies.

[0013] Compared with the prior art, the beneficial effects of the present invention are: (1) By improving the complex independent component analysis and partial coherence method, the system covers multi-element units such as photovoltaic, energy storage, and charging piles, and updates η in real time. i It has small errors under dynamic operating conditions, enabling accurate harmonic source tracing across all scenarios.

[0014] (2) Innovate dynamic electricity price game model, use economic incentives to guide units to actively reduce harmonics, reduce system costs by more than 15%, and solve the problem of "disconnect between governance and economy".

[0015] (3) Digital twin and RL-MPC algorithm cope with uncertainty, five modes switch smoothly, the system availability rate reaches 99.98%, and the stability of complex working conditions is improved. Attached Figure Description

[0016] Figure 1 This is a flowchart of the method of the present invention; Figure 2 This is a system block diagram of the present invention. Detailed Implementation

[0017] The technical solutions in the embodiments of the present invention will be clearly and completely described below.

[0018] A multi-modal collaborative optimization method for smart microgrids includes the following steps: S1. Real-time synchronous data acquisition Synchronous acquisition devices are deployed at the microgrid's point of common connection (PCC) and the outlets of each distributed unit to collect three-phase voltage and three-phase current data in real time. After the collected data is filtered by a 50Hz notch filter, it is uploaded to the area controller via the MQTT protocol with a packet loss rate of ≤0.1%.

[0019] S2. Calculate the real-time harmonic liability factor η i S21. Extracting harmonic components The time-series data acquired by S1 was processed using Fast Fourier Transform (FFT) to extract the 3rd to 25th harmonics. The sampling points were padded with zeros to 1024 to improve frequency resolution. The harmonic voltage was calculated as shown in formula (1): ,

[0020] In the formula, U h Let T be the h-th harmonic voltage, u(n) be the voltage value at the nth sampling point, and T be the voltage value at the nth sampling point. s=1 / 2560s is the sampling interval, N=1024 is the number of FFT sampling points, h is the harmonic order, and f0 is... The fundamental frequency of the power system, in my country's power grid, is f0=50Hz.

[0021] S22. Constructing the observation signal model Using the h-th harmonic voltage of PCC point extracted from S21 and the h-th harmonic current of each distributed unit as observed variables, the model is constructed as shown in formula (2): X(t) = [U PCC,h (t),I 1,h (t), I 2,h (t),...,I m,h (t)] T ,

[0022] In the formula, m is the number of distributed units, and I i,h (t) represents the h-th harmonic current of the i-th unit, U PCC,h (t) represents the h-th harmonic voltage at point PCC.

[0023] S23. Improved C-ICA blind source separation The observed signal model x(t) constructed in S22 is subjected to Discrete Fourier Transform (DFT) to obtain the frequency domain observed signal matrix X(k), where k is the iteration number index (k=1,2,...,100), and the dimension of X(k) is the same as that of x(t), representing the set of frequency domain observed signals at the k-th iteration. An adaptive step size adjustment mechanism is introduced to optimize the C-ICA iteration process. The objective function is to maximize the non-Gaussianity of the signal, and negative entropy is used as the measure. The iteration formula is as follows (3): W(k+1)=W(k)+μ(k)[IE{g(Y(k))Y(k) T W(k)

[0024] In the formula, W is the demixing matrix, μ(k) is the adaptive step size, which decreases linearly from 0.1 to 0.01 with the number of iterations, Y(k)=W(k)X(k) is the separated independent harmonic source signal, g(•) is the nonlinear activation function, using g(y)=tanh(ay) with a=1.2; after 100 iterations, the error is ≤1e-6, and the decoupled independent harmonic source signal S(t)=[S1(t),S2(t),...,S m (t)]^T),I is the identity matrix, used to adjust the update direction and magnitude of the unmixing matrix W, and E{•} is the expectation operator, which means to calculate the mathematical expectation of the random variable in parentheses.

[0025] S24. Partial coherence analysis to determine the harmonic liability factor η i Calculate the i-th harmonic source S i(t) and PCC harmonic voltage U PCC,h The partial coherence function of (t), excluding interference from other sources, is shown in formula (4): ,

[0026] In the formula, Let G be the partial coherence function of the i-th harmonic source and the PCC harmonic voltage U at frequency f, with a value ranging from 0 to 1. The closer the value is to 1, the stronger the "net coherence" between the two (i.e., the greater the specific contribution of unit i to the PCC harmonic voltage U), and the closer the value is to 0, the weaker the "net coherence". Ui (f) is U PCC,h (t) and S i The cross-power spectrum of (t), G ii (f) -1 For harmonic source S i (t) The matrix inverse of the power spectrum, G iU (f) is S i (t) cross-power spectrum with PCC harmonic voltage U, and with G Ui (f) is a reciprocal quantity. If the signal is a real signal, That is, conjugate, G UU (f) is the power spectrum of the PCC harmonic voltage U, G ii (f) is S i The self-power spectrum of (t); for Integrating over the harmonic frequency range yields the contribution of the i-th unit to the h-th harmonic, which, after normalization, becomes the real-time harmonic liability coefficient η. i ( ).

[0027] S3. Construct a dynamic electricity price game model and solve for the harmonic excitation coefficient α. Game Theory Model Construction Dynamic electricity price formula: P price,i =P base ±α·η i Where "+" indicates high η i The unit penalty, "-" indicates a penalty for low η. i Unit excitation; P price,i P represents the actual settlement price of the i-th distributed unit. base Let α be the time-of-use base electricity price of the power grid, such as 0.35 yuan / kWh during off-peak hours and 0.85 yuan / kWh during peak hours, and α be the harmonic excitation coefficient to be solved.

[0028] Solving for harmonic excitation coefficient α using non-cooperative game theory Each distributed unit is an independent decision-making entity, and its strategy set is "adjusting its own operating status, such as reducing photovoltaic power and staggering charging pile peak hours". The revenue function is as shown in formula (5): Ri =-[C buy,i +C op,i +α·η i P i,use ] In the formula, R i Let C be the profit function value of the i-th distributed unit. buy,i The electricity purchase cost for the i-th unit is given by the formula C. buy,i =P price,i ·P i,use C op,i For operation and maintenance costs (PV 0.02 yuan / kWh, energy storage 0.03 yuan / kWh), P i,use This refers to the unit's electricity consumption. The system operator aims to "minimize the total system operating cost." To achieve the goal, α is dynamically adjusted. After each adjustment, the strategies of each unit and the system cost are simulated through a digital twin. After 12 iterations, the system converges to a Nash equilibrium, such as α = 0.45 yuan / (kWh·η). This α value is the optimal excitation coefficient for the current hour.

[0029] S4. Establish a multi-objective optimization scheduling model Multi-objective optimization function Minimize[F cost +λ·Σ(η i )], In the formula: F cost The total operating cost of the system is given by formula (6): , In the formula, C grid,buy =P base ·P buy,total P buy,total C represents the total purchased power of the microgrid. grid,sell =0.75 yuan / kWh·P sell,total P sell,total Total electricity sales capacity This represents the total operation and maintenance cost of all distributed units, expressed in yuan.

[0030] Σ(η i The sum of the harmonic responsibility coefficients of all units is λ=0.4. After 100 simulations, the optimal weight for balancing economy and power quality is verified.

[0031] Constraints Power balance: P PV,out +P bat,out +P buy,total =P load,total +P sell,total In the formula, P PV,out The photovoltaic output power is obtained from data collected by S1, P bat,outThe energy storage output power is negative during charging and must meet the safety constraint that the energy storage SOC ∈ [20%, 90%]. buy,total P represents the total purchased power of the microgrid. load,total P represents the total load power of the microgrid, derived from data collected by S1. sell,total The total power sold by the microgrid is represented by the formula. The left and right sides of the formula correspond to the "power supply end" and "power consumption end" of the microgrid, respectively. Ensuring that the sum of all power sources equals the sum of all power destinations is the core constraint for stable system operation.

[0032] Equipment safety: Energy storage charging and discharging power ±2MW; charging pile power 0-30kW / unit; PCC point THD<5% (National Standard GB / T14549-1993).

[0033] S5: Solving Optimal Scheduling Models in a Digital Twin Environment The solution is obtained using a Model Predictive Control (MPC) algorithm optimized by reinforcement learning (RL), as follows: S51. Offline Training of RL Agents In the digital twin, using one year of historical data, including photovoltaic irradiance, load, and harmonic data, a deep deterministic policy gradient (DDPG) agent is trained: State space S contains 14 types of parameters: current and previous minute output of photovoltaic (PV), predicted output of PV (for the next 15 minutes) and the previous 10 minutes, current and previous minute total load, current and previous 5 minute SOC of energy storage, current and previous 1 minute energy storage charging / discharging power, current and previous 1 minute power exchange between microgrid and main grid, current and previous 1 minute PCC point voltage, current and previous 1 minute PCC point current, current and previous 5 minute PCC point total harmonic distortion (THD) of PCC point, and harmonic liability coefficient η for each distributed unit. i The system includes the average value and the average value of the previous 15 minutes, the base electricity price for the current period and the base electricity price for the previous hour, the current value and the previous hour value of the incentive coefficient k in the dynamic electricity price game model, the current operating mode identifier of the system (e.g., grid-connected economic mode is 1, grid-connected low-carbon mode is 2, etc.) and the mode identifier for the previous minute, and the current value and the average value of the ambient temperature for the previous 30 minutes. Each type of parameter includes the "current value" and the corresponding "historical characteristic value". Therefore, the total dimension of the state space is 14×2=28 dimensions.

[0034] Action Space A: Outputs two MPC prediction model correction parameters: A1: PV prediction confidence level 0.1-1.5; A2: load prediction time constant 1-10 minutes, used to dynamically adapt to uncertainties such as PV output and load fluctuations.

[0035] Reward function R = -[F cost +λ·Σ(η iConsistent with the S4 multi-objective optimization function, it guides the agent to optimize in the direction of "low cost and few harmonics"; after 100,000 training rounds, the agent converges to a stable policy, providing a decision basis for online optimization.

[0036] S52. Online scrolling optimization The RL agent outputs actions based on the real-time system state, such as A1=0.8 and A2=4.5, to correct the MPC prediction model; MPC uses a 15-minute rolling time domain and a 1-minute control cycle. It employs the interior-point method to solve the optimization model, with a single calculation time of ≤5 seconds. It outputs the optimal scheduling instructions for each minute of the next 15 minutes, such as 2.8MW of photovoltaic power and 0.5MW of energy storage discharge.

[0037] S6. Instruction Execution and Multimodal Switching Command issuance The optimal scheduling command output by S5 is converted into a device protocol by the regional collaborative control unit. IEC61850 is used for photovoltaics / energy storage, and Modbus is used for charging piles. The protocol conversion logic is Modbus→JSON. The command is then sent to the data acquisition and sensing execution unit of S1 via Ethernet / RS485. Each device responds and executes within 1 second. After execution, the data acquisition and sensing execution unit collects actual operating data (such as actual photovoltaic output and changes in energy storage SOC) and feeds it back to the cloud computing and digital twin unit and the regional collaborative control unit, forming a closed-loop control of "acquisition-analysis-optimization-execution-feedback".

[0038] Multimodal switching The system automatically switches operating modes based on the PCC status (PCC voltage and fault signals collected from S1) and the optimization results from S5: Grid-connected economic model (normal PCC + off-peak electricity price): full photovoltaic power generation, energy storage charging; Grid-connected low-carbon mode (normal PCC + peak electricity price + photovoltaic ≥2MW): photovoltaic energy storage power sales and charging pile peak shifting; Isolated supply mode (PCC failure + SOC≥30%): disconnect PCC, discharge energy storage, and cut off η≥0.2 units; Islanding power curtailment mode (PCC fault + SOC < 30%): Cut off non-essential loads; Grid support mode (PCC voltage < 0.85pu): Energy storage emergency discharge voltage support; During switching, power is smoothed through a first-order inertial element with a change rate ≤0.5MW / s to avoid impact.

[0039] This embodiment also proposes a smart microgrid multimodal collaborative optimization system to execute the above method. The system structure is as follows: 1. Data Acquisition and Sensing Execution Unit Includes a Fluke 1750 power quality analyzer at the PCC for collecting THD data; and a Linyang EMU-800 intelligent monitoring and control device at each unit outlet for collecting voltage, current, and power data. All smart terminals have a built-in MQTT collaborative agent, supporting "plug and play": after a new unit is powered on, it sends a registration message containing "device type, rated parameters, and protocol version" within 10 seconds. The area controller completes identity authentication within 15 seconds after comparing the device whitelist, without the need for manual configuration.

[0040] 2. Regional Cooperative Control Unit It consists of one Huawei AR502H edge gateway each at the substation and the charging station, configured with a 4-core CPU and 16GB of memory; it is used to collect local data, convert protocols (Modbus→JSON), perform coordination control, and issue commands. It supports 4G / Ethernet dual-link communication, and the backup link ensures uninterrupted network access even in the event of a network outage.

[0041] 3. Cloud computing and digital twin units Cloud server: Alibaba Cloud ECS, 4 cores, 8GB RAM, 500GB SSD, deployed with MySQL database; Digital twin model: Built on MATLAB / Simulink, including photovoltaic, energy storage, and load sub-modules, with parameters consistent with the physical devices; synchronizes physical data 100ms via OPCUA protocol, with dynamic error ≤2%; Safety Sandbox Function: The new strategy first simulates the system in a twin for 24 hours, including 10+ typical operating conditions, and only releases it to the physical system after the simulation meets the requirements.

[0042] 4. Intelligent Optimization Core Unit Harmonic responsibility analysis module: Based on the Python SciPy / NumPy library, implements the S2 algorithm; Dynamic game module: Based on the Nashpy library, it implements the game solution in step S3; Reinforcement Learning-Model Prediction Control Hybrid Optimization Module: Based on TensorFlow (RL) and GEKKO (MPC), this module solves step S5. Module Interaction: The harmonic responsibility module pushes η every 15 minutes. i In other modules, the game module updates the α value every hour to ensure data synchronization.

[0043] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the invention can be implemented in other specific forms without departing from its spirit or essential characteristics. Therefore, the embodiments should be considered in all respects as exemplary and non-limiting, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be included within the present invention. No reference numerals in the claims should be construed as limiting the scope of the claims.

Claims

1. A multi-modal collaborative optimization method for smart microgrids, characterized in that, Includes the following steps: S1: Real-time synchronous acquisition of voltage and current data at the microgrid's common connection point and the output of each distributed unit; S2: Based on the improved complex independent component analysis algorithm and partial coherence method, the data is processed to calculate the real-time harmonic responsibility coefficient η of each of the distributed units. i ; S3: Construct a dynamic electricity price game model. In this model, the actual settlement price of the i-th unit is related to the base price, the harmonic excitation coefficient, and the real-time harmonic responsibility coefficient of the unit. The harmonic excitation coefficient is solved dynamically through non-cooperative game theory. S4: To minimize the total system operating cost and total harmonic liability, a multi-objective optimization function is established and integrated with the dynamic electricity price game model to form an optimized scheduling model; S5: In the digital twin environment, the optimized scheduling model is solved using a model predictive control algorithm optimized by reinforcement learning to obtain the optimal scheduling instruction in the future rolling time domain. S6: The optimal scheduling command is issued to each distributed unit for execution, so as to realize the multi-modal collaborative optimization operation of the microgrid.

2. The multi-modal collaborative optimization method for smart microgrids according to claim 1, characterized in that, The "improved complex independent component analysis algorithm and partial coherence method" mentioned in S2 specifically includes: S21: Perform a fast Fourier transform on the acquired time-series data to extract each harmonic component; S22: Construct an observation signal model that includes background harmonic voltage and currents from all potential harmonic sources; S23: An improved complex independent component analysis algorithm is used to perform blind source separation on the observed signal model, decoupling the independent harmonic source signals; S24: The contribution of each independent harmonic source to the total harmonic distortion rate at the common junction point is calculated using the partial coherence method, yielding the precise harmonic liability coefficient η for each distributed unit. i .

3. The multi-modal collaborative optimization method for smart microgrids according to claim 1, characterized in that, The harmonic excitation coefficients described in S3 are specifically solved dynamically through non-cooperative game theory as follows: Each distributed unit is regarded as an independent decision-making entity, whose strategy set is to adjust its own operating state to change the harmonic emission characteristics, and its revenue function is the sum of its own operating cost and harmonic penalty cost. The system operator dynamically adjusts this coefficient and performs iterative simulation calculations using a digital twin to ultimately guide all unit strategies to converge to the Nash equilibrium point that minimizes the total system cost. The coefficient value at this point is then used as the optimal harmonic excitation coefficient for the current time period.

4. The multi-modal collaborative optimization method for smart microgrids according to claim 1, characterized in that, The "multi-objective optimization function" described in S4 takes the sum of the total system operating cost and the total harmonic responsibility as the optimization objective. The total system operating cost includes the cost of purchasing electricity, the revenue from selling electricity, and the cost of equipment operation and maintenance. The total harmonic responsibility is the sum of the harmonic responsibility coefficients of all units, and the two are balanced by weighting coefficients.

5. The multi-modal collaborative optimization method for smart microgrids according to claim 1, characterized in that, S5 specifically includes: S51: Offline training phase: In the digital twin, a reinforcement learning agent is trained using historical data. Its state space is the system's historical and real-time data, its action space is the instructions for adjusting the parameters of the internal prediction model of the model prediction control, and its reward function is the negative value of the multi-objective optimization function. S52: Online application phase: The reinforcement learning agent outputs actions based on the real-time system state, dynamically correcting the prediction model of the model prediction control; the model prediction controller uses the corrected prediction model to perform rolling optimization and solve the optimized scheduling model.

6. The multi-modal collaborative optimization method for smart microgrids according to claim 1, characterized in that, The "multimodal collaborative optimization operation" mentioned in S6 refers to the system automatically and smoothly switching between grid-connected economic mode, grid-connected low-carbon mode, islanded supply guarantee mode, islanded power curtailment mode and grid support mode based on the optimization results and grid status.

7. A system for executing the multimodal collaborative optimization method for smart microgrids according to any one of claims 1-6, characterized in that, include: The data acquisition and sensing execution unit includes synchronous acquisition devices deployed at common connection points and at the exits of each distributed unit, as well as intelligent terminals with built-in unified protocol collaborative agents. The regional coordination control unit consists of regional controllers deployed in various regions of the microgrid, and is used to collect local data, perform coordinated control, and issue commands. The cloud computing and digital twin unit, including cloud servers and digital twin models deployed on them, is used to map the state of physical systems and provide a simulation training environment for algorithms; The intelligent optimization core unit includes a harmonic responsibility analysis module, a dynamic game module, and a reinforcement learning-model prediction control hybrid optimization module, which are used to perform optimization calculations in the method, such as harmonic responsibility coefficient calculation, dynamic electricity price game solution, and optimal scheduling model solution.

8. The system according to claim 7, characterized in that, The intelligent terminal supports plug-and-play functionality. When any new distributed unit that conforms to the unified information model and communication protocol is connected, it can automatically register its identity and capability parameters with the system.

9. The system according to claim 7, characterized in that, The digital twin model maintains real-time data synchronization with the physical microgrid and serves as a safe sandbox for exploratory training of the reinforcement learning agent, as well as a simulation test platform for new control strategies.

Images