Multi-microgrid coordinated adaptive switching method and control system considering direct current cost

Abstract

The application discloses a kind of considering the multi-microgrid cooperative adaptive switching method and control system of direct current cost, belong to park direct current microgrid energy management and control technical field.The application first constructs each plot digital twin model, predicts photovoltaic, load and energy storage state;Second, establish the direct current conversion cost function of fusion efficiency loss, carbon emission and equipment aging;Subsequently, with power supply power as consistency variable, using ADMM framework for distributed cooperative scheduling, and based on the ratio of conversion cost and real-time electricity price Dynamic threshold is set, AC / DC start-stop binary decision is generated, which is embedded as a constraint into the optimization model iterative solution, realize scheduling decision and physical execution Causal coordination;Finally, through 15 minutes rolling time domain optimization, issue instruction.The application avoids invalid power supply under high-cost operating condition under the premise of guaranteeing data privacy, significantly reduces the total operating cost of park, reduces the AC / DC equipment start-stop frequency, prolongs the service life of equipment.

Description

Technical Field

[0001] This invention relates to the field of energy management and control technology for DC microgrids in industrial parks, specifically to a multi-microgrid collaborative adaptive switching method and control system that takes into account DC costs. Background Technology

[0002] Coordination challenges under multi-site energy isolation constraints: Each site possesses its own photovoltaic, energy storage, vehicle-to-grid (V2G), and DC loads, but due to property division, energy transfer across sites is not permitted. Existing distributed optimization research (such as ADMM) implicitly assumes that microgrids can directly exchange power, which cannot be directly applied.

[0003] The conversion costs of AC / DC converters are overlooked: converting AC power from the mains to DC power incurs costs such as efficiency loss (5%~10%), equipment aging, and carbon emissions. Existing strategies either unconditionally enable AC / DC or use fixed mode switching, failing to achieve dynamic decision-making based on real-time costs.

[0004] The separation of coordinated scheduling and conversion decisions: The separation between upper-level optimization (how to allocate mains power) and lower-level execution (whether to enable AC / DC) leads to suboptimal or even infeasible decisions.

[0005] Existing technologies neglect the hard constraint of "energy isolation": literature on multi-microgrid coordinated dispatch generally assumes that microgrids can directly exchange power, which is not applicable to parks with fragmented ownership.

[0006] The cost of AC / DC conversion is simplified to a fixed value or ignored: existing AC / DC hybrid systems only treat the commutation loss as a term in the objective function, and never couple the enabling / disabling of AC / DC as a binary decision with the upper-level optimization.

[0007] Scheduling and execution are disconnected: Traditional distributed optimization methods such as ADMM do not include dynamic availability constraints for AC / DC devices, which makes it impossible to execute the theoretically optimal solution in practice due to the high conversion cost. Summary of the Invention

[0008] To address the aforementioned shortcomings in the prior art, this invention provides a multi-microgrid cooperative adaptive switching method that considers DC cost.

[0009] To achieve the above-mentioned objectives, the technical solution adopted by this invention is as follows: A multi-microgrid cooperative adaptive handover method considering DC cost includes the following steps: S1. Establish digital twin prediction models for each microgrid in the park to obtain data on photovoltaic output, load demand, energy storage operation status, and vehicle-to-grid (V2G) dispatchable capacity of each microgrid within a preset time domain in the future. S2. Construct the unit DC conversion cost function for each microgrid to quantify the efficiency loss, carbon emissions, and equipment aging costs of AC-DC converters; S3. Using the grid power consumption of each microgrid as a consistency variable, construct a distributed cooperative scheduling model based on the alternating direction multiplier method (ADMM) to solve for the initial value of grid power consumption of each microgrid. S4. Calculate the adaptive threshold based on the ratio of unit DC conversion cost to real-time electricity price, and generate AC / DC enable / disable decision in combination with local power deficit. Embed this decision as a binary constraint into the feasible domain of the distributed cooperative scheduling model, and iteratively update the optimal value of the mains power drawdown for each microgrid. S5. Using a preset period as a rolling window, repeat steps S1 to S4 to distribute the converged current optimal power output to each plot for execution, thereby realizing the coordinated scheduling and adaptive AC / DC switching of multiple microgrids in the park.

[0010] Furthermore, step S1 specifically includes the following steps: S11. Collect historical photovoltaic output data, historical load data, numerical weather forecast data, energy storage nameplate parameters, and vehicle grid connection records for each microgrid. S12. Using a physical model combined with machine learning residual correction, a photovoltaic power output prediction sequence for each microgrid within a preset time domain is generated based on historical photovoltaic data and numerical weather prediction data. ,in For microgrid indexing, For discrete-time indexing; S13. Using a long short-term memory network model, input time and meteorological characteristics to generate a DC load prediction sequence for each microgrid within a preset time domain. ; S14. Based on the current state of charge of the energy storage and the charging and discharging plan, the boundary of the energy storage state of charge in the future preset time domain of each microgrid is obtained recursively. Based on the current state of charge of the V2G connected vehicles, the expected departure time, and the vehicle owner's discharge willingness coefficient, the V2G dispatchable power range in the future preset time domain of each microgrid is calculated.

[0011] Furthermore, step S2 specifically includes the following steps: S21. Obtain the conversion efficiency of each AC / DC converter. Time to commissioning, total replacement cost Carbon emission cost per unit of electricity loss ,as well as Real-time time-of-use electricity price The remaining lifetime of the AC / DC converter It is obtained by linearly subtracting the time in operation from the design life, and the unit is hours; S22, Constructing the first microgrid The unit DC conversion cost function at time t:

[0012] In the formula, The rated power of the AC / DC converter, The cost per unit of DC-DC conversion.

[0013] Furthermore, step S3 specifically includes the following steps: S31. Set the global optimization objective as minimizing the total operating cost of the park. The objective function is:

[0014] In the formula, To predict the length of the time domain, The total number of microgrids, For the first microgrid The amount of mains power drawn at any given time; for V2G discharge benefit coefficient at time; For the first microgrid V2G discharge power at any given time; This is the load shedding penalty factor; For the first microgrid The load shedding power at any given moment; This is the energy storage lifespan degradation cost coefficient; For the first microgrid The energy storage and discharge power at any given moment; for Real-time time-of-use electricity price at any given moment; S32. Set global constraints, including: Transformer capacity constraints:

[0015] Local power balance constraints:

[0016] Energy storage operation constraints include recursive state of charge constraints, charge and discharge power limit constraints, and upper and lower limits of state of charge constraints; V2G operation constraints include charging and discharging power limits and minimum battery capacity constraints for vehicle owners:

[0017] Among the above constraints, For the first The power that a microgrid draws from the mains electricity via an AC / DC converter, Its upper limit; This refers to the rated capacity of the main transformer in the industrial park. For the first Each microgrid at any time The projected output of photovoltaic power; These represent the charging and discharging power of V2G, respectively. For the first Each microgrid at any time DC load forecast; S33, Power drawn from mains electricity To ensure consistency, an auxiliary variable is introduced. Construct the augmented Lagrange function:

[0018] in, For the first Local cost function of a microgrid For ADMM penalty parameters; For the first microgrid Lagrange multipliers of time, To predict the time domain length, Calculate the norm; S34. Perform ADMM iterations to solve the local subproblems of each microgrid in parallel. :

[0019] In the formula, For the first The first microgrid The power drawn from the mains through the AC / DC converter in the next iteration The secondary constraints generated by the decision to enable AC / DC; S35. Perform projection update on the auxiliary variables to obtain the updated auxiliary variables. The projection operator projects variables onto a plane that satisfies... The feasible set; S36, Update the Lagrange multipliers

[0020] With the original residual As a convergence criterion, among which... , The first and The next iteration is the Lagrange operator. These are the preset convergence thresholds for the original residual and the dual residual, respectively.

[0021] Furthermore, step S4 specifically includes the following steps: S41. Calculate the local power deficit of each microgrid without considering AC / DC power intake:

[0022] In the formula, for Local power deficit at all times and These represent the current available discharge limits for energy storage and the power grid, respectively. For the first Each microgrid at any time DC load forecast value, For the first Each microgrid at any time The projected output of photovoltaic power; S42. Calculate the conversion cost-electricity price ratio:

[0023] in, To prevent small positive numbers from being divided by zero, For the first microgrid The cost per unit DC-DC conversion at any given time. Let t be the real-time time-of-use electricity price of the mains power. calculate Adaptive threshold at time: in, For adaptive threshold, The preset baseline power is set to 1kW or 10% of the local average load. S43, Generate the first AC / DC enable flag for a microgrid : when hour, and forced ; when hour, satisfy ,in, To reserve a margin in power, For the first The maximum power limit for a microgrid to draw power from the mains via an AC / DC converter; S44, will As a feasible region of binary constraint embedded in the ADMM local subproblem, when Time fixed ,when Time By imposing upper limit constraints, the global power extraction task can be transferred to a microgrid with lower conversion costs through the cumulative adjustment of Lagrange multipliers.

[0024] Furthermore, S5 specifically includes the following steps: S51. After the ADMM iteration converges, the central controller extracts the current time step. Optimal power extraction The data is transmitted to the local controllers of each microgrid via the communication network. S52. The local controller adjusts the actual charging and discharging power of the energy storage and V2G based on the received power extraction command, combined with the real-time state of charge of the energy storage and the real-time available capacity of the V2G, and controls the start-stop status of the AC / DC converter. When real-time power fluctuations such as photovoltaic power drop cause the actual power deficit to exceed the preset value, the V2G fast discharge function is called first to supplement the power, and the mains power extraction command is not adjusted immediately. S53. Update the digital twin prediction data of each microgrid with a rolling cycle of 15 minutes, and repeat steps S1 to S4 to achieve dynamic scheduling across the entire time domain.

[0025] Furthermore, in step S3, the boundary information uploaded by each microgrid to the main controller is de-identified data, including: net load forecast range, upper and lower limits of adjustable energy storage capacity, and V2G dispatchable capacity range, without uploading internal topology and detailed load data.

[0026] A multi-microgrid cooperative adaptive switching control system considering DC cost includes: A central controller, deployed on a central server, is used to execute the ADMM coordination algorithm and issue scheduling instructions; Multiple local controllers are deployed in each microgrid plot to execute local optimization sub-problems and control local devices; Multiple AC / DC converters are connected to the DC bus of each microgrid and the AC grid, and are started and stopped by a local controller. In addition to the photovoltaic arrays, energy storage batteries and V2G charging piles deployed in each plot; The system executes steps S1 to S5 every 15 minutes in a rolling manner. The present invention has the following beneficial effects: For the first time, distributed collaborative scheduling of multiple microgrids was achieved under energy isolation constraints: by setting the consistency variable of ADMM to the grid power drawn by each microgrid, rather than the power exchanged between microgrids, it is perfectly adapted to parks with segmented ownership.

[0027] Introducing DC-DC conversion cost function and dynamic threshold: quantifying the losses, carbon emissions, and aging costs of AC / DC converters, and comparing them with real-time electricity prices to generate adaptive enable / disable decisions, avoiding energy waste and premature equipment aging caused by "unconditional power access".

[0028] A causal coordination mechanism is formed between upper and lower layers: AC / DC decision-making serves as a binary constraint for the local subproblem of ADMM, enabling global optimization to automatically avoid microgrids with high conversion costs, thus achieving a balance between overall optimization of the park and local physical constraints.

[0029] Protecting data privacy: Each plot only exchanges anonymized boundary information (net load forecast, energy storage margin), and does not share detailed internal data, in order to meet the management requirements of multiple owners.

[0030] Rolling optimization and real-time execution closed loop: forecasts are updated every 15 minutes to ensure adaptability to weather changes such as typhoons. Attached Figure Description

[0031] Figure 1 This is a schematic diagram of the multi-microgrid cooperative adaptive switching method considering DC cost in this invention. Figure 2 This is a diagram illustrating the architecture of a collaborative control system for a DC microgrid in a park, provided in an embodiment of the present invention.

[0032] Figure 3 This is a convergence curve of the ADMM distributed optimization algorithm in an embodiment of the present invention.

[0033] Figure 4 Adaptive threshold for embodiments of the present invention Follow Change curve.

[0034] Figure 5 This is a heatmap of the AC / DC activation status under different combinations of electricity prices and equipment aging in an embodiment of the present invention. The vertical axis represents the equipment type (new equipment / aging equipment), and the horizontal axis represents the electricity price scenario (off-peak / flat / peak). Dark colors indicate activation, and light colors indicate disabling. Detailed Implementation

[0035] The specific embodiments of the present invention are described below to enable those skilled in the art to understand the present invention. However, it should be understood that the present invention is not limited to the scope of the specific embodiments. For those skilled in the art, various changes are obvious as long as they are within the spirit and scope of the present invention as defined and determined by the appended claims. All inventions utilizing the concept of the present invention are protected.

[0036] A multi-microgrid cooperative adaptive switching method considering DC cost, such as Figure 1 As shown, it includes the following steps: S1. Establish digital twin prediction models for each microgrid in the park to obtain data on photovoltaic output, load demand, energy storage operation status, and vehicle-to-grid (V2G) dispatchable capacity of each microgrid within a preset time domain in the future. Figure 2 This is an architecture diagram of a collaborative control system for a DC microgrid in a park, provided in an embodiment of the present invention. The system includes a central controller (deployed on a central server), N local controllers per plot (one per plot, N=37), N AC / DC converters, and photovoltaic arrays, energy storage batteries, V2G charging piles, and DC loads within each plot. The central controller and the local controllers exchange anonymized boundary information (net load forecast range, upper and lower limits of adjustable energy storage capacity, and V2G dispatchable range) and ADMM iterative variables (Lagrange multipliers) via the MQTT protocol. Auxiliary variables Local controllers at each site communicate with their respective devices via Modbus TCP or CAN bus for real-time data acquisition and command transmission. Solid arrows in the diagram represent energy flow, while dashed arrows represent information flow. Energy flow cannot cross the site boundary line, while information flow is allowed to coordinate across sites.

[0037] In this embodiment, the specific operations for each part, including historical photovoltaic output, historical load, numerical weather prediction (NWP), energy storage nameplate parameters, and V2G vehicle access records (from the park's SCADA or IoT platform), are as follows: Photovoltaic prediction: using physical models Residual correction, output and the standard deviation of prediction error Load forecasting: An LSTM model is used as input, taking time characteristics (hours, days of the week, holidays) and temperature as input, and outputting... Energy storage model: Based on the current SOC and charge / discharge plan, the dynamic equation of SOC is recursively derived.

[0038] V2G model: Each charging station accesses vehicle information (current SOC, estimated departure time). Car owner's willingness to discharge Calculate the schedulable capacity.

[0039] Output: Photovoltaic, load, energy storage SOC boundary, and V2G dispatchable power range for each plot in the next 24 hours (every 15 minutes) → sent to steps 2 and 3.

[0040] S2. Construct the unit DC conversion cost function for each microgrid to quantify the efficiency loss, carbon emissions, and equipment aging costs of AC-DC converters; In this embodiment, the system input is the AC / DC converter parameters ( Service life Carbon emission cost coefficient Real-time electricity price (From the meteorological and market data interface in step 1), the specific processing in this step is as follows: Define the cost of unit DC-DC conversion For example: 1 kWh of AC power is drawn from the mains, but the actual energy input to the DC bus is only... kWh, the remaining losses are converted into heat and carbon emissions, while also shortening the converter's lifespan.

[0041]

[0042] in: Carbon emission cost per unit of electricity loss (RMB / kWh, can be taken as 0.10). Total cost of replacing AC / DC converter (RMB); The remaining lifespan of the converter (in hours) is linearly calculated based on the time already spent in operation. : Rated power of the converter (kW); Output: (RMB / kWh) → Sent to step 4 for dynamic threshold calculation.

[0043] Innovation: Existing literature only incorporates converter loss as a linear term into the objective function. This invention, for the first time, integrates loss, carbon emissions, and lifetime into a single enable / disable threshold criterion. With real-time electricity price The decision to allow power extraction after comparison is a qualitative difference.

[0044] S3. Using the grid power consumption of each microgrid as a consistency variable, construct a distributed cooperative scheduling model based on the alternating direction multiplier method (ADMM) to solve for the initial value of grid power consumption of each microgrid. This embodiment specifically includes the following steps: The global optimization objective is set as minimizing the total operating cost of the park, and the objective function is:

[0045] In the formula, To predict the length of the time domain, The total number of microgrids, For the first microgrid The amount of mains power drawn at any given time; for V2G discharge benefit coefficient at time; For the first microgrid V2G discharge power at any given time; This is the load shedding penalty factor; For the first microgrid The load shedding power at any given moment; This is the energy storage lifespan degradation cost coefficient; For the first microgrid The energy storage and discharge power at any given moment; for Real-time time-of-use electricity price at any given moment; Set global constraints, including: Transformer capacity constraints:

[0046] Local power balance constraints:

[0047] Energy storage operation constraints include recursive state of charge constraints, charge and discharge power limit constraints, and upper and lower limits of state of charge constraints; V2G operation constraints include charging and discharging power limits and minimum battery capacity constraints for vehicle owners:

[0048] Among the above constraints, For the first The power that a microgrid draws from the mains electricity via an AC / DC converter, Its upper limit; This refers to the rated capacity of the main transformer in the industrial park. For the first Each microgrid at any time The projected output of photovoltaic power; These represent the charging and discharging power of V2G, respectively. For the first Each microgrid at any time DC load forecast; Power drawn from mains electricity To ensure consistency, an auxiliary variable is introduced. Construct the augmented Lagrange function:

[0049] in, For the first Local cost function of a microgrid For ADMM penalty parameters; For the first microgrid Lagrange multipliers of time, To predict the time domain length, Calculate the norm; By performing ADMM iterations and solving the local subproblems of each microgrid in parallel, the following results are obtained. :

[0050] In the formula, For the first The first microgrid The power drawn from the mains through the AC / DC converter in the next iteration The two-level constraints generated by the decision to enable AC / DC; The updated auxiliary variables are obtained by projecting them onto the auxiliary variables. The projection operator projects variables onto a plane that satisfies... The feasible set; Update Lagrange multipliers

[0051] With the original residual As a convergence criterion, among which... , The first and The next iteration is the Lagrange operator. These are the preset convergence thresholds for the original residual and the dual residual, respectively.

[0052] S4. Calculate the adaptive threshold based on the ratio of unit DC conversion cost to real-time electricity price, and generate AC / DC enable / disable decision in combination with local power deficit. Embed this decision as a binary constraint into the feasible domain of the distributed cooperative scheduling model, and iteratively update the optimal value of the mains power drawdown for each microgrid. In this embodiment, the system input is: From step S2 From step 3 (Initial value or value from the previous iteration), and local power deficit. (Calculated based on the predicted data from step 1).

[0053] The execution process includes: Calculate local power deficit (if AC / DC is not used)

[0054] in and These represent the current available discharge limits.

[0055] Dynamic threshold calculation Define the conversion cost-electricity price ratio:

[0056] in Use a small positive number (such as 0.001) to prevent division by zero.

[0057] Let the adaptive threshold be:

[0058] in The base power (can be 1 kW or 10% of the local average load).

[0059] Physical meaning: When the conversion cost Greater than or equal to the electricity price hour, When the threshold drops to 0, power extraction is forcibly prohibited; when the conversion cost is much lower than the electricity price, the threshold is larger, and power extraction is allowed.

[0060] In this embodiment, the binary enable decision is specifically as follows: Define AC / DC enable flag

[0061] and, Forced ;when hour,

[0062] Feedback to ADMM feasible domain Will As in step 3.3 Binary constraints: like =0, then in the subproblem Fixed at 0; if 1, then It is positive and subject to an upper limit.

[0063] Causal Coordination: During ADMM Iteration An attempt was made to obtain electricity from a certain plot of land, but the plot was rejected due to high conversion costs. When the value is 0 and the user rejects the application, the Lagrange multiplier λi will increase, prompting the user to transfer the power extraction task to a plot of land with a lower conversion cost in the next iteration.

[0064] Output: Decision and the updated Feasible region → Return to step 3 for the next iteration.

[0065] S5. Using a preset period as a rolling window, repeat steps S1 to S4 to distribute the converged current optimal power output to each plot for execution, thereby realizing the coordinated scheduling and adaptive AC / DC switching of multiple microgrids in the park.

[0066] In this embodiment, the system input is: The optimal result after ADMM convergence (Current moment only) ), the real-time status of local digital twins for each block.

[0067] The execution process includes: The main controller will The message is sent to the local controllers of each microgrid via MQTT.

[0068] Based on this power command, the local controller, combined with the real-time SOC and V2G available capacity, adjusts the actual charging and discharging power of the energy storage and V2G, and controls the start and stop of the AC / DC converter.

[0069] If real-time power fluctuations (such as a sudden drop in photovoltaic power) cause the actual deficit to exceed the preset value, the local controller can temporarily enable V2G fast discharge (direction three function) without immediately changing the settings. .

[0070] Every 15 minutes, update all forecast data and repeat steps 1-5 (rolling time domain).

[0071] The output is: Execution instructions (PWM signals to energy storage converters, V2G charging piles, AC / DC switches) → sent to physical devices.

[0072] To verify the effectiveness of the method proposed in the above steps, this embodiment demonstrates its effectiveness through the following theoretical analysis and simulation experiments.

[0073] Model convergence and stability analysis To prove the convergence of the ADMM distributed optimization framework proposed in this invention, according to standard ADMM theory, when the objective function fi of the local subproblem is a convex function and the constraint set Xi is a non-empty closed convex set, the algorithm converges to the global optimum at a residual sum of squares of O(1 / k). In this invention, the local subproblem of each microgrid is a quadratic programming problem under linear constraints (the objective function contains squared terms), and its convexity is obviously true, thus satisfying the convergence condition.

[0074] Simulation verification: Taking a campus with 37 microgrids and 8 switching stations as an example, the ADMM iteration in step 3 was executed on an Intel i7 processor. After... After the second iteration, both residuals decreased to The following conditions are met, including the convergence threshold. The average solution time is 26.8 seconds, which meets the real-time requirement of 15-minute rolling optimization.

[0075] In comparison, solving the same problem using centralized MILP (using Gurobi) takes over 180 seconds and requires uploading all data from each local data patch, making it impractical in engineering. This invention improves the solution speed by approximately 6.7 times while ensuring privacy.

[0076] Figure 3 This is a convergence curve of the ADMM distributed optimization algorithm in this embodiment of the invention. The horizontal axis represents the number of iterations k, and the vertical axis represents the original residual. With dual residual (Logarithmic coordinates). The simulation was based on typical daily data from 37 microgrids and 8 switching stations in the industrial park, and underwent 50 iterations. As can be seen from the figure, after approximately 25 iterations, both types of residuals decreased to... The following conditions are met, and the preset convergence threshold is satisfied. The average solution time was 26.8 seconds, which meets the real-time requirement of 15-minute rolling optimization.

[0077] Physical effectiveness analysis of conversion cost function and dynamic threshold 1. The physical meaning of each component of the transformation cost Defined in step 2 The three items correspond to: Efficiency loss item: When AC / DC conversion efficiency When the cost drops from 0.95 to 0.85, the cost increases by approximately 2.1 times, reflecting an economic signal that inefficient equipment should be used less frequently.

[0078] Carbon emission penalties: Based on an average grid carbon emission factor of 0.581 kg CO2 / kWh and a carbon price of 60 yuan / ton, For an old converter with an efficiency of 85%, every 1 kWh of AC power drawn results in an additional emission of 0.102 kg of CO2, with a penalty of approximately 0.006 yuan. Although small, this can accumulate to a considerable amount over the long term.

[0079] Equipment aging costs: Assuming an AC / DC replacement cost of 5000 yuan, a design life of 87600 hours (10 years), and a rated power of 50 kW, the initial aging cost is 0.00114 yuan / kWh. After 5 years of operation, the remaining lifespan is halved, and the aging cost rises to 0.00228 yuan / kWh. This cost increases linearly with equipment aging, prompting the system to prioritize the use of younger equipment.

[0080] 2. Dynamic threshold characteristics Figure 4 The ratios of different conversion costs to electricity prices were plotted. Below, adaptive threshold (Pick The curve of change. (Right now )hour, Allowing a maximum power deficit of 1 kW; when hour, At this time, even if there is a power shortage, power extraction is forcibly prohibited; when When the conversion cost is twice the electricity price, the threshold becomes negative, but is actually set to 0, and power is still prohibited.

[0081] This design ensures that AC / DC is only allowed to be activated when the conversion cost is lower than the real-time electricity price, thus eliminating the phenomenon of "loss-making electricity" from an economic perspective. Figure 4 Demonstrates adaptive threshold The pattern of change (taking) The horizontal axis represents the ratio. The vertical axis represents the threshold power. .when At that time, the threshold varies Increase linearly decrease; when When the threshold drops to zero, it indicates that the conversion cost is not lower than the real-time electricity price, and AC / DC power extraction should be forcibly prohibited.

[0082] To verify the optimization effect of this invention on the overall operating cost of the park, based on actual data from 37 microgrids (each plot has a photovoltaic capacity of 100-500 kW, an energy storage capacity of 200-1000 kWh, and 2-10 V2G charging piles), a 24-hour simulation was conducted on a typical summer day (peak photovoltaic hours 12:00-14:00, peak electricity prices 9:00-11:00 and 19:00-21:00). The following four strategies were compared, as shown in Table 1: Table 1 Simulation results for each strategy

[0083] Compared to independent operation (S1), this invention (S4) saves 3,420 yuan (approximately 18.2%) per day in costs. This is mainly due to the use of price signals to guide V2G charging during photovoltaic surpluses (reducing curtailment) and the reduction of AC / DC power consumption during peak electricity prices (utilizing local energy storage and V2G). Since strategy S3 cannot be implemented in this scenario due to the inability of energy to cross regions, cost data is not provided in this embodiment.

[0084] Compared to the ideal centralized system (S2), this invention is approximately 8.6% more expensive. However, S2 requires all privacy data and lacks engineering feasibility. The loss in this invention represents an acceptable privacy / cost tradeoff. The number of AC / DC start-stop cycles is reduced from 143 to 68 (a 52% reduction), effectively extending converter lifespan. The transformer peak load is reduced from 2.36 MW to 1.80 MW (a 23.7% reduction), which can delay investment in distribution expansion, as shown in Table 2 below.

[0085] Table 2. Comparison of overall performance of different scheduling strategies

[0086] Electricity price fluctuation scenarios Three electricity pricing scenarios were set up: off-peak (0.25 yuan / kWh), average (0.65 yuan / kWh), and peak (1.20 yuan / kWh). An aging AC / DC converter (η=0.85, remaining lifespan 2 years) and a new device (η=0.95, remaining lifespan 10 years) were also set up. The strategy of this invention was run for 24 hours, and the actual electricity consumption of each was recorded, as shown in Table 3 below.

[0087] Table 3 Actual power consumption of the present invention's strategy in different scenarios

[0088] The results show that aging equipment is automatically disabled during peak electricity price periods, while new equipment can still be enabled, demonstrating the adaptive capability of this invention to the health status of equipment.

[0089] Extreme weather (typhoon) scenarios A typhoon model is introduced into the digital twin in step 1: the photovoltaic prediction error increases by 1.5 times, and the load decreases by 30% due to personnel evacuation. A TWTI index-triggered early warning is set (this will be detailed later, but this embodiment only demonstrates the robustness of the conversion cost). Simulation results show that even in typhoon scenarios with large prediction errors, the dynamic threshold of this invention can still make correct decisions: when the mains power is available (non-islanded), aging AC / DC converters are still used cautiously; when the mains power may be interrupted (early warning), the system will increase local energy storage / V2G discharge in advance. This method does not rely on precise prediction but on real-time comparison logic, thus exhibiting strong robustness. Figure 5 This study demonstrates whether AC / DC converters are enabled under different electricity price scenarios (off-peak 0.25 yuan / kWh, flat 0.65 yuan / kWh, peak 1.20 yuan / kWh) and two equipment aging states (new equipment: η=0.95, remaining lifespan 10 years; aging equipment: η=0.85, remaining lifespan 2 years). The simulation period was 24 hours, and the actual electricity consumption was recorded. The results show that aging equipment is forcibly disabled during peak electricity price periods because the conversion cost Cconv is higher than the electricity price (R...). i>1), while the new equipment can still be used under the same electricity price. This shows that the dynamic threshold strategy of the present invention can adaptively adjust according to the health status of the equipment, avoid overuse of inefficient equipment, and extend the overall equipment life. Figure 5 In the diagram, the vertical axis represents the equipment type (new equipment / aging equipment), and the horizontal axis represents the electricity price scenario (off-peak / flat / peak). Darker blocks indicate that the AC / DC converter is enabled under this condition, while lighter blocks indicate that the AC / DC converter is disabled under this condition.

[0090] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0091] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0092] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0093] Specific embodiments have been used to illustrate the principles and implementation methods of this invention. The descriptions of the embodiments above are only for the purpose of helping to understand the method and core ideas of this invention. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of this invention. Therefore, the content of this specification should not be construed as a limitation of this invention.

[0094] Those skilled in the art will recognize that the embodiments described herein are intended to help the reader understand the principles of the invention, and should be understood that the scope of protection of the invention is not limited to such specific statements and embodiments. Those skilled in the art can make various other specific modifications and combinations based on the technical teachings disclosed in this invention without departing from the spirit of the invention, and these modifications and combinations are still within the scope of protection of this invention.

Claims

1. A multi-microgrid cooperative adaptive handover method considering DC cost, characterized in that, Includes the following steps: S1. Establish digital twin prediction models for each microgrid in the park to obtain data on photovoltaic output, load demand, energy storage operation status, and vehicle-to-grid (V2G) dispatchable capacity of each microgrid within a preset time domain in the future. S2. Construct the unit DC conversion cost function for each microgrid to quantify the efficiency loss, carbon emissions, and equipment aging costs of AC-DC converters; S3. Using the grid power consumption of each microgrid as a consistency variable, construct a distributed cooperative scheduling model based on the alternating direction multiplier method (ADMM) to solve for the initial value of grid power consumption of each microgrid. S4. Calculate the adaptive threshold based on the ratio of unit DC conversion cost to real-time electricity price, and generate AC / DC enable / disable decision in combination with local power deficit. Embed this decision as a binary constraint into the feasible domain of the distributed cooperative scheduling model, and iteratively update the optimal value of the mains power drawdown for each microgrid. S5. Using a preset period as a rolling window, repeat steps S1 to S4 to distribute the converged current optimal power output to each plot for execution, thereby realizing the coordinated scheduling and adaptive AC / DC switching of multiple microgrids in the park.

2. The multi-microgrid cooperative adaptive switching method considering DC cost according to claim 1, characterized in that, S1 specifically includes the following sub-steps: S11. Collect historical photovoltaic output data, historical load data, numerical weather forecast data, energy storage nameplate parameters, and vehicle grid connection records for each microgrid. S12. Using a physical model combined with machine learning residual correction, a photovoltaic power output prediction sequence for each microgrid within a preset time domain is generated based on historical photovoltaic data and numerical weather prediction data. ,in For microgrid indexing, For discrete-time indexing; S13. Using a long short-term memory network model, input time and meteorological characteristics to generate a DC load prediction sequence for each microgrid within a preset time domain. ; S14. Based on the current state of charge of the energy storage and the charging and discharging plan, the boundary of the energy storage state of charge in the future preset time domain of each microgrid is obtained recursively. Based on the current state of charge of the V2G connected vehicles, the expected departure time, and the vehicle owner's discharge willingness coefficient, the V2G dispatchable power range in the future preset time domain of each microgrid is calculated.

3. The multi-microgrid cooperative adaptive switching method considering DC cost according to claim 1, characterized in that, S2 specifically includes the following steps: S21. Obtain the conversion efficiency of each AC / DC converter. Time to commissioning, total replacement cost Carbon emission cost per unit of electricity loss ,as well as Real-time time-of-use electricity price The remaining lifetime of the AC / DC converter It is obtained by linearly subtracting the time in operation from the design life, and the unit is hours; S22, Constructing the first microgrid The unit DC conversion cost function at time t: In the formula, The rated power of the AC / DC converter, The cost per unit of DC-DC conversion.

4. The multi-microgrid cooperative adaptive switching method considering DC cost according to claim 1, characterized in that, S3 specifically includes the following steps: S31. Set the global optimization objective as minimizing the total operating cost of the park. The objective function is: In the formula, To predict the length of the time domain, The total number of microgrids, For the first microgrid The amount of mains power drawn at any given time; for V2G discharge benefit coefficient at time; For the first microgrid V2G discharge power at any given moment; This is the load shedding penalty factor; For the first microgrid The load shedding power at any given moment; This is the energy storage lifespan degradation cost coefficient; For the first microgrid The energy storage and discharge power at any given moment; for Real-time time-of-use electricity price at any given moment; S32. Set global constraints, including: Transformer capacity constraints: Local power balance constraints: Energy storage operation constraints include recursive state of charge constraints, charge and discharge power limit constraints, and upper and lower limits of state of charge constraints; V2G operation constraints include charging and discharging power limits and minimum battery capacity constraints for vehicle owners: Among the above constraints, For the first The power that a microgrid draws from the mains through an AC / DC converter, Its upper limit; This refers to the rated capacity of the main transformer in the industrial park. For the first A microgrid at any time The projected output of photovoltaic power; These represent the charging and discharging power of V2G, respectively. For the first A microgrid at any time DC load forecast; S33, Power drawn from mains electricity To ensure consistency, an auxiliary variable is introduced. Construct the augmented Lagrange function: in, For the first Local cost function of a microgrid For ADMM penalty parameters; For the first microgrid Lagrange multipliers of time, To predict the time domain length, Calculate the norm; S34. Perform ADMM iterations to solve the local subproblems of each microgrid in parallel. : In the formula, For the first The first microgrid The power drawn from the mains through the AC / DC converter in the next iteration The two-level constraints generated by the decision to enable AC / DC; S35. Perform a projection update on the auxiliary variables to obtain the updated auxiliary variables. The projection operator projects variables onto a plane that satisfies... The feasible set; S36, Update the Lagrange multipliers With the original residual As a convergence criterion, among which... , The first and The next iteration is the Lagrange operator. These are the preset convergence thresholds for the original residual and the dual residual, respectively.

5. The multi-microgrid cooperative adaptive switching method considering DC cost according to claim 1, characterized in that, S4 specifically includes the following steps: S41. Calculate the local power deficit of each microgrid without considering AC / DC power intake: In the formula, for Local power deficit at all times and These represent the current available discharge limits for energy storage and the power grid, respectively. For the first A microgrid at any time DC load forecast value, For the first A microgrid at any time The projected output of photovoltaic power; S42. Calculate the conversion cost-electricity price ratio: in, To prevent small positive numbers from being divided by zero, For the first microgrid The cost per unit DC conversion at any given time. Let t be the real-time time-of-use electricity price of the mains power. calculate Adaptive threshold at time: in, For adaptive threshold, The preset baseline power is set to 1kW or 10% of the local average load. S43, Generate the first AC / DC enable flag for a microgrid : when hour, and forced ; when hour, satisfy ,in, To reserve a margin in power, For the first The maximum power limit for a microgrid to draw power from the mains via an AC / DC converter; S44, will As a feasible region of binary constraint embedded in the ADMM local subproblem, when Time fixed ,when Time By imposing upper limit constraints, the global power extraction task can be transferred to a microgrid with lower conversion costs through the cumulative adjustment of Lagrange multipliers.

6. The multi-microgrid cooperative adaptive switching method considering DC cost according to claim 1, characterized in that, S5 specifically includes the following steps: S51. After the ADMM iteration converges, the central controller extracts the current time step. Optimal power extraction It is transmitted to the local controllers of each microgrid via the communication network; S52. The local controller adjusts the actual charging and discharging power of the energy storage and V2G based on the received power extraction command, combined with the real-time state of charge of the energy storage and the real-time available capacity of the V2G, and controls the start-stop status of the AC / DC converter. When real-time power fluctuations such as photovoltaic power drop cause the actual power deficit to exceed the preset value, the V2G fast discharge function is called first to supplement the power, and the mains power extraction command is not adjusted immediately. S53. Update the digital twin prediction data of each microgrid with a rolling cycle of 15 minutes, and repeat steps S1 to S4 to achieve dynamic scheduling across the entire time domain.

7. The multi-microgrid cooperative adaptive switching method considering DC cost according to claim 1, characterized in that, In S3, the boundary information uploaded by each microgrid to the main controller is de-identified data, including: net load forecast range, upper and lower limits of adjustable energy storage capacity, and V2G dispatchable capacity range. Internal topology and detailed load data are not uploaded.

8. An adaptive switching control system applying the method of any one of claims 1 to 7, characterized in that, include: A central controller, deployed on a central server, is used to execute the ADMM coordination algorithm and issue scheduling instructions; Multiple local controllers are deployed in each microgrid plot to execute local optimization sub-problems and control local devices; Multiple AC / DC converters are connected to the DC bus of each microgrid and the AC grid, and are started and stopped by a local controller. In addition to the photovoltaic arrays, energy storage batteries and V2G charging piles deployed in each plot; The system executes steps S1 to S5 every 15 minutes in a rolling fashion.

Images