New energy low-carbon operation scheduling method and system for island microgrid

Abstract

The invention discloses a new energy low-carbon operation scheduling method and system for an island microgrid, and the method comprises the steps: generating a day-ahead scheduling plan with the minimization of the weighted average carbon emission intensity of a system as a core target through an upper-layer day-ahead optimization model based on load prediction data and renewable energy power prediction data; according to the day-ahead scheduling plan, generating an adjustment instruction of output of each unit through a lower-layer intra-day rolling optimization model; based on the adjustment instruction, calculating the wind and light abandoning power of the renewable energy source, and optimizing the renewable energy source consumption through a cost punishment mechanism; according to the optimized unit output and renewable energy consumption scheme, coordinating a charging and discharging strategy of an energy storage system to obtain an optimized scheduling instruction; and executing the scheduling instruction, and updating the system state through a closed-loop feedback mechanism. According to the embodiment of the invention, power balance and low-carbon operation of the island microgrid under a complex fluctuation working condition can be realized, and the utilization rate of renewable energy sources and the overall environmental protection property of the system are improved.

Description

Technical Field

[0001] This invention belongs to the field of new energy technology, specifically a new energy low-carbon operation scheduling method and system for isolated microgrids. Background Technology

[0002] With the advancement of dual-carbon goals, the penetration rate of new energy sources, represented by wind power and photovoltaics, in isolated microgrids is continuously increasing. However, the inherent intermittency and volatility of renewable energy bring enormous power balance and stable operation pressures to isolated microgrid systems. Existing operation and scheduling methods mostly focus on economic optimization or take power balance as the sole objective, lacking refined control over the carbon emission intensity during operation. Most strategies use optimization models on a single time scale, which are difficult to cope with real-time uncertainties on both the source and load sides, easily leading to deviations between scheduling plans and actual operating conditions, resulting in the abandonment of renewable energy or inefficient operation of conventional units, failing to achieve the goal of low-carbon operation while ensuring power supply reliability. Summary of the Invention

[0003] The purpose of this invention is to provide a new energy low-carbon operation scheduling method and system for islanded microgrids, so as to overcome the shortcomings of the prior art, realize the power balance and low-carbon operation of islanded microgrids under complex fluctuation conditions, and improve the utilization rate of renewable energy and the overall environmental protection of the system.

[0004] One embodiment of this application provides a method for scheduling new energy low-carbon operation in isolated microgrids, the method comprising: Based on load forecast data and renewable energy power forecast data, a day-ahead scheduling plan with the core objective of minimizing the system's weighted average carbon emission intensity is generated through an upper-level day-ahead optimization model. Based on the day-ahead scheduling plan, combined with ultra-short-term forecast data and real-time system status data, adjustment instructions for the output of each unit are generated through the lower-level intraday rolling optimization model. Based on the aforementioned adjustment instructions, the curtailment power of wind and solar power of renewable energy is calculated, and the consumption of renewable energy is optimized through a penalty cost mechanism. Based on the optimized unit output and renewable energy consumption plan, the charging and discharging strategies of the energy storage system are coordinated to achieve power balance and carbon emission control, resulting in optimized dispatch instructions. The system executes optimized scheduling instructions and updates the system status through a closed-loop feedback mechanism to complete the scheduling of new energy and low-carbon operations across multiple time scales.

[0005] Optionally, the generation of a day-ahead scheduling plan based on load forecast data and renewable energy power forecast data, with the core objective of minimizing the system's weighted average carbon emission intensity, through an upper-level day-ahead optimization model, includes: Collect load forecast data, photovoltaic power forecast data, wind power forecast data, and grid electricity price for the next few hours to generate a multi-source forecast dataset; Based on the multi-source prediction dataset, an upper-level optimization objective function is constructed, which includes economic items such as coal-fired unit operating costs, electricity purchase and sale costs, and energy storage degradation costs, as well as low-carbon items such as the system weighted average carbon emission intensity, to generate a comprehensive optimization objective function. Based on the system equipment parameters, set power balance constraints, upper and lower limit constraints of unit output, unit ramp rate constraints, energy storage system power and energy constraints, and power constraints for interaction with the main grid to generate a complete set of constraints. A mixed-integer linear programming solver is used to solve the comprehensive optimization objective function and the complete set of constraints, and outputs the day-ahead scheduling plan, including the output plan of each unit, the energy storage charging and discharging plan, the power purchase and sale plan, and the renewable energy consumption plan.

[0006] Optionally, the step of generating adjustment instructions for the output of each unit based on the day-ahead scheduling plan, combined with ultra-short-term forecast data and real-time system status data, through a lower-level intraday rolling optimization model includes: At the start of the current rolling cycle, real-time system status data is collected, including energy storage charge status, current unit output, and ultra-short-term load forecast data, ultra-short-term photovoltaic power forecast data, and ultra-short-term wind power forecast data, to generate a real-time dataset. Based on real-time datasets and day-ahead scheduling plans, a lower-level optimization objective function is constructed, which includes unit adjustment costs, wind and solar curtailment penalty costs, energy storage degradation costs, and deviation penalty terms from day-ahead plans, thus obtaining a rolling optimization objective function. Based on ultra-short-term forecast data, real-time power balance constraints, renewable energy output constraints, and plan tracking constraints are set. Among them, the plan tracking constraints limit the maximum allowable deviation between the unit output adjustment and the day-ahead plan, and generate a set of real-time constraint conditions. A quadratic programming solver is used to solve the rolling optimization objective function and the real-time constraint set, and outputs adjustment commands for the output of each unit, energy storage, and power purchase and sale.

[0007] Optionally, the step of calculating the curtailment of wind and solar power of renewable energy based on the adjustment instructions, and optimizing the consumption of renewable energy through a penalty cost mechanism, includes: Extract the maximum short-term forecast values ​​of photovoltaic power output and wind power output from the ultra-short-term forecast data to generate ultra-short-term forecast power output values ​​of renewable energy. The actual dispatch output of renewable energy is determined based on the actual dispatch output of photovoltaic power and wind power in the adjustment instructions. The solar curtailment power is calculated as the difference between the solar ultra-short-term predicted output value and the actual dispatched output value. The wind curtailment power is calculated as the difference between the wind power ultra-short-term predicted output value and the actual dispatched output value. The curtailment power values ​​are then generated. The lower-level optimization objective function applies a cost term for wind and solar curtailment penalties, using a cost mechanism to incentivize maximizing renewable energy consumption, and outputs an optimized renewable energy consumption scheme.

[0008] Optionally, the step of coordinating the charging and discharging strategies of the energy storage system based on the optimized unit output and renewable energy consumption scheme to achieve power balance and carbon emission control, and obtaining optimized dispatch instructions, includes: Based on the optimized unit output and renewable energy consumption scheme, the net power demand of the system is calculated, the power balance status is assessed, and the power balance assessment results are generated. Based on the power balance assessment results, the required charging and discharging power and direction of the energy storage system are determined. At the same time, the energy storage state of charge boundary and charging and discharging efficiency are considered to generate the energy storage charging and discharging requirements. By combining the system's weighted average carbon emission intensity target, the energy storage charging and discharging strategy is optimized, prioritizing charging during periods of renewable energy surplus and discharging during periods of high carbon intensity, thus generating a low-carbon coordination strategy. By integrating unit output adjustment, renewable energy consumption schemes, and low-carbon coordination strategies, optimized dispatch instructions are formed to ensure power balance and carbon emission control, and the final dispatch instruction set is output.

[0009] Optionally, the execution of optimized scheduling instructions and the updating of system status through a closed-loop feedback mechanism to complete multi-timescale new energy and low-carbon operation scheduling includes: The optimized scheduling instructions are issued to each execution unit, including coal-fired units, energy storage systems, photovoltaic power plants, wind power plants, and grid interaction units, and the instruction execution status is generated. The system monitors real-time operational data, including actual unit output, actual energy storage charging and discharging power, actual renewable energy output, actual load value, and actual grid interaction power, and generates an actual operational dataset. Based on the actual operation dataset, update the system status parameters, including the energy storage charge status and the unit operation status, and calculate the deviation between the actual values ​​and the planned values ​​to generate the updated system status. The updated system status and deviation information are used as input to the lower-level intraday rolling optimization model to start the next rolling cycle optimization, realize closed-loop feedback control, and complete the multi-timescale new energy low-carbon operation scheduling.

[0010] Another embodiment of this application provides a new energy low-carbon operation scheduling system for isolated microgrids, the system comprising: The first generation module is used to generate a day-ahead scheduling plan with the core objective of minimizing the system's weighted average carbon emission intensity, based on load forecast data and renewable energy power forecast data, through an upper-level day-ahead optimization model. The second generation module is used to generate adjustment instructions for the output of each unit based on the day-ahead scheduling plan, combined with ultra-short-term forecast data and real-time system status data, through the lower-level intraday rolling optimization model. The optimization module is used to calculate the curtailment power of renewable energy based on the adjustment instructions, and optimize the consumption of renewable energy through a penalty cost mechanism; The coordination module is used to coordinate the charging and discharging strategies of the energy storage system based on the optimized unit output and renewable energy consumption plan, so as to achieve power balance and carbon emission control and obtain optimized dispatch instructions. The execution module is used to execute optimized scheduling instructions and update the system status through a closed-loop feedback mechanism to complete the scheduling of new energy and low-carbon operation across multiple time scales.

[0011] Another embodiment of this application provides a storage medium storing a computer program, wherein the computer program is configured to execute the method described in any of the preceding claims when running.

[0012] Another embodiment of this application provides an electronic device including a memory and a processor, wherein the memory stores a computer program and the processor is configured to run the computer program to perform the method described in any of the preceding claims.

[0013] Compared with existing technologies, this invention provides a new energy low-carbon operation scheduling method for isolated microgrids. Based on load forecast data and renewable energy power forecast data, it generates a day-ahead scheduling plan with the core objective of minimizing the system's weighted average carbon emission intensity through an upper-level day-ahead optimization model. According to the day-ahead scheduling plan, it generates adjustment instructions for the output of each unit through a lower-level intraday rolling optimization model. Based on the adjustment instructions, it calculates the curtailment of wind and solar power in renewable energy and optimizes renewable energy consumption through a penalty cost mechanism. According to the optimized unit output and renewable energy consumption scheme, it coordinates the charging and discharging strategies of the energy storage system to obtain optimized scheduling instructions. It executes the optimized scheduling instructions and updates the system status through a closed-loop feedback mechanism, thereby enabling power balance and low-carbon operation of isolated microgrids under complex and fluctuating conditions, improving the utilization rate of renewable energy and the overall environmental friendliness of the system. Attached Figure Description

[0014] Figure 1 A hardware structure block diagram of a computer terminal for a new energy low-carbon operation scheduling method for an islanded microgrid, provided in an embodiment of the present invention; Figure 2A flowchart illustrating a new energy low-carbon operation scheduling method for an islanded microgrid, provided by an embodiment of the present invention; Figure 3 This is a schematic diagram of a new energy low-carbon operation and scheduling system for an isolated microgrid, provided as an embodiment of the present invention. Detailed Implementation

[0015] The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and should not be construed as limiting the present invention.

[0016] The present invention first provides a new energy low-carbon operation scheduling method for islanded microgrids. This method can be applied to electronic devices, such as computer terminals, specifically ordinary computers.

[0017] The following detailed explanation uses a computer terminal as an example. Figure 1 This is a hardware structure block diagram of a computer terminal for a new energy low-carbon operation scheduling method for an islanded microgrid, provided as an embodiment of the present invention. Figure 1 As shown, the computer device includes a processor, memory, and network interface connected via a system bus, wherein the memory may include non-volatile storage media and internal memory.

[0018] Non-volatile storage media can store operating systems and computer programs. These computer programs include program instructions that, when executed, cause the processor to perform any renewable energy and low-carbon operation scheduling method for isolated microgrids.

[0019] The processor provides computing and control capabilities, supporting the operation of the entire computer device.

[0020] The internal memory provides an environment for the execution of computer programs in non-volatile storage media. When the computer program is executed by the processor, it enables the processor to execute any new energy and low-carbon operation scheduling method for isolated microgrids.

[0021] This network interface is used for network communication, such as sending assigned tasks. Those skilled in the art will understand that... Figure 1 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0022] It should be understood that the processor can be a central processing unit, but it can also be other general-purpose processors, digital signal processors, application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. Among these, a general-purpose processor can be a microprocessor or any conventional processor.

[0023] See Figure 2 The present invention provides a method for scheduling new energy low-carbon operation in isolated microgrids, which may include the following steps: S201, based on load forecast data and renewable energy power forecast data, generates a day-ahead scheduling plan with the core objective of minimizing the system's weighted average carbon emission intensity through an upper-level day-ahead optimization model; Specifically, it can collect load forecast data, photovoltaic power forecast data, wind power forecast data, and grid electricity price for the next few hours to generate a multi-source forecast dataset; This step prepares the foundational data for the previous day's optimization. By comprehensively collecting core predictive information that influences scheduling decisions, it ensures the completeness and timeliness of the optimization model's input. The specific implementation is as follows: The forecast duration is set to 24 hours (covering the entire operating day), with a time resolution of 1 hour. All data must be organized in the format of time period-data type-value-precision. The load forecast data comes from the microgrid load management system and is predicted using a nonlinear regression model based on historical load and meteorological factors (such as temperature and humidity). Each time period value is accompanied by an error range of ±0.5MW to reflect the forecast uncertainty.

[0024] The photovoltaic power prediction data is based on the historical output of photovoltaic power plants and solar irradiance prediction calculations. It adopts a prediction method that combines clear sky model and machine learning, with a prediction accuracy of ≥85%. The corresponding irradiance is labeled for each time period, which makes it easy to trace the prediction basis.

[0025] The wind power prediction data is combined with the historical wind speed and direction data of the wind farm, and the wind power curve model is used for prediction, with a prediction accuracy of ≥80%.

[0026] The power grid price forecast data comes from the time-of-use pricing policy released by the main grid dispatch center. It is divided into three periods: peak, flat, and valley. The price data needs to be synchronized with the local power market policy and is updated monthly.

[0027] The grid carbon intensity forecast data is based on the marginal carbon emission factor of the main grid. The carbon intensity value for each period is accompanied by a fluctuation range of ±50kgCO2 / MWh to adapt to changes in the power structure of the main grid.

[0028] The above data is integrated into a multi-source prediction dataset. For example, the dataset for a certain period is as follows: period 12:00-13:00, load 15MW (±0.5MW), photovoltaic 10MW (irradiance 1000W / m²), wind power 5MW (wind speed 6m / s), electricity purchase price 1.2 yuan / kWh, electricity sales price 0.8 yuan / kWh. This ensures that the data for each period is completely correlated and provides a unified input for subsequent optimization modeling.

[0029] Based on the multi-source prediction dataset, an upper-level optimization objective function is constructed, which includes economic items such as coal-fired unit operating costs, electricity purchase and sale costs, and energy storage degradation costs, as well as low-carbon items such as the system weighted average carbon emission intensity, to generate a comprehensive optimization objective function.

[0030] This step integrates the dual objectives of economic growth and low carbon emissions into the core of optimization through mathematical modeling, balancing operating costs and carbon emission control. The specific implementation is as follows: The objective function of the upper-level optimization is a comprehensive optimization that balances economic efficiency and low carbon emissions. It explicitly controls carbon emission levels by introducing a system-weighted average carbon emission intensity index. The optimization model is shown below: in, To optimize the objective function value (numerical) of the upper layer, T is the time period index. This refers to the number of units participating in scheduling within the system. Let be the operating cost coefficient (yuan / MWh) for the i-th coal-fired boiler. Let be the power generation capacity (MW) of the i-th coal-fired boiler during time period t. , The electricity purchase and sale price between the microgrid and the main grid is (RMB / MWh). , Power purchased / sold (MW), The energy storage charging and discharging degradation cost coefficient (yuan / MW). The charging and discharging power (MW) of the energy storage system is represented by the positive value for discharging and the negative value for charging. The duration is expressed in hours (h). This is a carbon weighting coefficient used to balance economic efficiency and low-carbon performance. The system-weighted average carbon emission intensity is defined as follows: in, Let be the carbon emission factor (kgCO2 / MWh) of the i-th coal-fired boiler. The marginal carbon emission factor of the main grid (kgCO2 / MWh).

[0031] Based on the system equipment parameters, set power balance constraints, upper and lower limit constraints of unit output, unit ramp rate constraints, energy storage system power and energy constraints, and power constraints for interaction with the main grid to generate a complete set of constraints. This step ensures that the optimization results meet the requirements of equipment safety and system stability by setting physical and operational constraints, as specifically implemented as follows: Power balance constraint: Ensure that the power generation, energy storage capacity, power purchase and sale capacity within the microgrid are balanced with the load demand in real time. The formula is: in, Forecast of day-ahead solar and wind power loads; This is a day-ahead load demand forecast.

[0032] Unit output upper and lower limit constraints: Limit the output range of each coal-fired unit, the formula is: in For minimum output, To maximize output.

[0033] Unit ramp rate constraint: Limits the rate of change of unit output to avoid equipment damage caused by frequent and large adjustments. The formula is: in, and These represent the maximum climb and descent rates of the unit, respectively, with Δt = 1 hour.

[0034] Energy storage system constraints include two types of constraints: power and energy. Power constraints: ,in Maximum charging power, This represents the maximum discharge power. Energy status update: in, The remaining energy stored in time period t; , This refers to the charge / discharge efficiency.

[0035] Power constraint for interaction with the main grid: Limits the maximum range of power purchase and sale, the formula is as follows: .

[0036] All constraints are organized into a complete set of constraints according to constraint type, formula, parameter example, and verification logic, ensuring that each constraint can be quantified and verified, and providing clear boundaries for optimization solutions.

[0037] A mixed-integer linear programming solver is used to solve the comprehensive optimization objective function and the complete set of constraints, and outputs the day-ahead scheduling plan, including the output plan of each unit, the energy storage charging and discharging plan, the power purchase and sale plan, and the renewable energy consumption plan.

[0038] This step solves the constrained optimization problem using a specialized solver to generate an executable day-ahead scheduling baseline plan, specifically implemented as follows: For mixed-integer linear programming, commercial high-performance solvers (such as CPLEX and Gurobi) are selected. These solvers support large-scale linear constraint problems, have fast convergence speeds (solving a 24-hour scheduling problem takes less than 10 minutes), and can handle integer variables (such as unit start-up and shutdown status, where 0 represents shutdown and 1 represents operation). Before solving, the comprehensive optimization objective function and the complete set of constraints must be converted into a solver-compatible mathematical model format to ensure accurate mapping of variable definitions, objective function coefficients, and constraint boundaries.

[0039] The solution process is divided into three stages: initialization, iterative optimization, and convergence judgment. Initialization: Set the solution parameters, including the upper limit of the number of iterations, the convergence threshold (termination when the objective function value changes <1e-4), the precision of integer variables (1e-6, to ensure that the unit start-up and shutdown status is an integer 0 or 1), and input the multi-source prediction dataset and constraint parameters (such as unit cost coefficient and energy storage capacity). Iterative optimization: The solver uses the branch and bound method to handle integer variables and the simplex method to optimize continuous variables (such as unit output and energy storage power) to gradually reduce the objective function value; Convergence judgment: When the iteration meets the convergence threshold or reaches the upper limit of iteration, the optimal solution is output. If there is an infeasible solution (such as a constraint conflict), the solver will return the conflicting constraint (such as the energy storage constraint not being met), and the constraint parameters need to be adjusted and the solution is solved again.

[0040] The output day-ahead scheduling plan needs to be divided into detailed time periods, with each plan including specific values ​​and execution instructions: Output plan for each unit: List the output values ​​by unit number and time period. For example, coal-fired unit No. 1: 2MW output from 00:00 to 01:00, and indicate the start-up and shutdown status of the unit. Energy storage charging and discharging plan: includes charging and discharging power and state of charge for each time period, for example, energy storage system: 00:00-01:00 charging 4MW ( =5MWh), discharge is positive and charging is negative, and the state of charge must be in the range of 2-10MWh; Electricity purchase and sale plan: Distinguish between the time period and power of electricity purchase and sale. For example, electricity purchase: 5MW of electricity will be purchased from 08:00 to 11:00; electricity sale: 2MW of electricity will be sold from 12:00 to 14:00. The power value shall not exceed the upper limit of electricity purchase and sale. Renewable energy consumption plan: List the planned consumption power (i.e., predicted power, assuming no curtailment) of photovoltaic and wind power. For example, photovoltaic: 5MW to be consumed from 09:00 to 10:00; wind power: 8MW to be consumed from 08:00 to 09:00. If there is curtailment, the curtailed photovoltaic power of 2MW should be noted.

[0041] The scheduling plan needs to be in a standardized tabular format to ensure that the power balance and constraints are met for each time period, providing a clear benchmark for the next level of intraday optimization.

[0042] S202, Based on the day-ahead scheduling plan, combined with ultra-short-term forecast data and real-time system status data, adjustment instructions for the output of each unit are generated through the lower-level intraday rolling optimization model; Specifically, at the start of the current rolling cycle, real-time system status data, including energy storage charge status, current unit output, and ultra-short-term load forecast data, ultra-short-term photovoltaic power forecast data, and ultra-short-term wind power forecast data, can be collected to generate a real-time dataset. This step is the core data input stage for the lower-level intraday rolling optimization. By frequently collecting real-time status and high-precision ultra-short-term forecast data, it provides a precise basis for dynamic adjustments. The specific implementation is as follows: The current rolling cycle follows a 15-minute cycle and a 4-hour prediction time range, with a starting time of 10:00:00 (optimization starts every 15 minutes from 00:00 each day).

[0043] State of Charge (SOC): Obtained through the energy storage management system, reflecting the proportion of remaining energy capacity to the total capacity, expressed as a percentage. This is used to calculate the remaining energy. , (This refers to the total energy storage capacity, such as 10MWh). For example, if the current SOC is 62%, then... =62%×10=6.2MWh, and at the same time record the charging and discharging status (currently discharging, power 0.3MW) to ensure that subsequent adjustments are in line with the current state of energy storage.

[0044] Current Unit Output: Real-time power generation of each coal-fired unit is collected, in MW. For example, Unit 1 (rated 10MW) currently outputs 7.8MW, and Unit 2 (rated 8MW) currently outputs 5.1MW, both in stable operation (no start-stop fluctuations). When recording, it is necessary to indicate whether the unit is in an adjustable state (Unit 1 is adjustable, Unit 2 has a smaller adjustable margin because it is close to its minimum output of 0.8MW). Ultra-short-term forecast data uses a real-time correction model based on machine learning / deep learning to predict load, photovoltaic, and wind power. The forecast time domain is the next 4 hours, with a resolution of 15 minutes. Among them, the ultra-short-term load forecast data: combined with recent load fluctuations and user electricity consumption patterns, the load forecast for 10:00-10:05 is 14.8MW; the ultra-short-term photovoltaic power forecast data: based on real-time solar irradiance and photovoltaic array operating status, the photovoltaic output forecast for 10:00-10:05 is 9.7MW; the ultra-short-term wind power forecast data: based on real-time wind speed and wind power curves, the wind power output forecast for 10:00-10:05 is 4.7MW.

[0045] The above data is integrated into a real-time dataset by combining the rolling cycle start time, real-time status data, and ultra-short-term forecast data formats.

[0046] Based on real-time datasets and day-ahead scheduling plans, a lower-level optimization objective function is constructed, which includes unit adjustment costs, wind and solar curtailment penalty costs, energy storage degradation costs, and deviation penalty terms from day-ahead plans, thus obtaining a rolling optimization objective function. This step ensures that intraday adjustments are both economical and do not deviate from the day-ahead baseline by constructing an objective function that balances cost minimization and plan tracking. The specific implementation is as follows: The lower-level optimization objective function focuses on minimizing the sum of real-time operating costs and planned deviation costs. The formula is designed with reference to the system's dynamic characteristics as follows: in, To optimize the objective function value for the lower level; Optimize the start time for the current scrolling; To optimize the prediction time domain length for rolling; To adjust each equipment unit j in Power during the time period Optimize the planned value for the corresponding upper-level day; It serves as a penalty factor, used to limit deviations from planned output. , For the curtailment of solar and wind power, , To incur the penalties for abandoning solar and wind power, It serves as a penalty factor, used to limit deviations from planned output.

[0047] The cost of curtailing wind and solar power ( Incentives to maximize the absorption of renewable energy The penalty factor (yuan / MW) indicates that the loss from solar power curtailment is even higher. =500 yuan / MW, wind power =400 yuan / MW; (Maximum forecast value for ultra-short-term photovoltaic power output - actual adjusted output) .

[0048] Energy storage degradation cost ( Consistent with the upper layer, =0.5 yuan / MW, The charging and discharging power is adjusted for energy storage (discharging is positive, charging is negative). This represents the absolute value of the power.

[0049] Penalty for deviation from the current plan ( To limit the deviation between unit output and the current day's plan, and to avoid frequent and significant adjustments. The deviation penalty coefficient (yuan / (MW²)) is higher for units with greater inertia, such as Unit 1. =200 yuan / (MW²), Unit 2 =250 yuan / (MW²).

[0050] Δt is the time step within the rolling cycle (5 minutes = 1 / 12 hours). Multiplying the above terms by Δt and summing them gives the objective function value for that time step.

[0051] Based on ultra-short-term forecast data, real-time power balance constraints, renewable energy output constraints, and plan tracking constraints are set. Among them, the plan tracking constraints limit the maximum allowable deviation between the unit output adjustment and the day-ahead plan, and generate a set of real-time constraint conditions.

[0052] This step ensures that the optimization results meet the requirements of equipment safety and system stability by setting rigid constraints. The specific implementation is as follows: Real-time power balance constraints: Based on ultra-short-term forecast data, ensure that the adjusted power sources are balanced with load demand, as shown in the formula. All variables are adjusted or ultra-short-term forecast values ​​(in MW).

[0053] Renewable energy output constraints: Limiting the upper and lower limits of actual renewable energy output to prevent over-generation or waste. in, To contribute to the actual scheduling of the photovoltaic array, The maximum short-term forecast value for photovoltaic power output.

[0054] in, To provide power for the actual dispatching of wind turbine units This represents the ultra-short-term maximum forecast value for solar and wind power output.

[0055] Plan tracking constraints: in, To minimize the maximum allowable power deviation and prevent frequent and significant adjustments to coal-fired power units during the daytime dispatch.

[0056] All constraints are integrated into a real-time constraint set based on constraint type, formula, parameter example, and validation logic.

[0057] A quadratic programming solver is used to solve the rolling optimization objective function and the real-time constraint set, and outputs adjustment commands for the output of each unit, energy storage, and power purchase and sale.

[0058] This step uses an efficient solver to process optimization problems with quadratic terms, generating executable real-time adjustment instructions, as specifically implemented below: The quadratic programming solver is selected based on the underlying objective function. Such solvers excel at handling optimization problems involving quadratic terms (with deviation penalty terms being quadratic functions) and linear constraints (power balance, output limits), exhibiting fast convergence speed (single-cycle solution time < 1 second, meeting the 15-minute rolling cycle requirement). Before solving, the rolling optimization objective function must be transformed into a standard quadratic programming form. Where x is the optimization variable (e.g., , , , H is the quadratic coefficient matrix (composed of the deviation penalty coefficient). The configuration is as follows: For example, the position corresponding to machine number 1 is H(1,1)=2. =400), f is the linear term coefficient vector (composed of adjustment cost, penalty cost, and degradation cost coefficients), and at the same time, the real-time constraint condition set is transformed into the linear constraint form of Ax≤b and Aeqx=beq (such as power balance is the equality constraint Aeqx=beq, and the upper and lower limits of output are the inequality constraints Ax≤b).

[0059] Solution process parameter settings: The convergence threshold is set to 1e-4 (termination occurs when the change in the objective function value between two iterations is <1e-4), the maximum number of iterations is set to 500 (to ensure completion within 1 second), and the variable boundaries are set to the physically feasible range (e.g., ∈[1,10]MW, ∈[-5,5]MW).

[0060] The output adjustment instructions need to be categorized by execution unit, clearly specifying the adjusted values, the deviation from the current day's plan, and the basis for execution: Unit output adjustment instructions: Unit 1 coal-fired power unit shall be adjusted to 7.8MW (currently planned to be 8MW, deviation -0.2MW, output is appropriately reduced due to a slight decrease of 0.5MW in ultra-short-term load); Unit 2 coal-fired power unit shall be adjusted to 5.1MW (currently planned to be 5MW, deviation +0.1MW, output is slightly reduced due to insufficient adjustable margin of Unit 1). The instructions must indicate the operating status of the units (all are in stable operation) to avoid start-up and shutdown operations.

[0061] Energy storage adjustment instruction: The energy storage system is adjusted to discharge 0.5MW (the planned discharge daytime discharge is 0.3MW, with a deviation of +0.2MW, which is used to balance the power reduction caused by wind power curtailment and reduce the demand for electricity purchase), and the current SOC is marked as 62%.

[0062] Power purchase and sales adjustment instructions: The power purchase is adjusted in conjunction with the main grid to 2.1MW (the current planned power purchase is 2MW, with a deviation of +0.1MW, to supplement the remaining power gap and avoid over-reliance on energy storage discharge), while the power sales instruction remains at 0MW (no surplus power available for sale). Instructions must comply with the constraints of a power purchase cap of 10MW and a power sales cap of 8MW. All instructions must be output in a standardized text format.

[0063] S203, based on the adjustment instructions, calculate the wind and solar power curtailment of renewable energy, and optimize the consumption of renewable energy through a penalty cost mechanism; Specifically, the maximum short-term forecast values ​​of photovoltaic power output and wind power output can be extracted from the ultra-short-term forecast data to generate ultra-short-term forecast power output values ​​of renewable energy. This step is fundamental to calculating wind and solar power curtailment. By accurately extracting the maximum potential output of renewable energy, the upper limit benchmark for consumption is determined. The specific implementation is as follows: The ultra-short-term forecast data comes from a real-time corrected forecast model based on machine learning / deep learning. The extraction of the ultra-short-term maximum forecast value of photovoltaic power output needs to combine real-time solar irradiance and photovoltaic array operating status, while the extraction of the ultra-short-term maximum forecast value of wind power output needs to be based on real-time wind speed and wind power curves.

[0064] The generated ultra-short-term renewable energy power output forecasts need to be organized by time period to form a sequence.

[0065] The actual dispatch output of renewable energy is determined based on the actual dispatch output of photovoltaic power and wind power in the adjustment instructions. This step, by associating with previously generated adjustment instructions, clarifies the actual operating output of renewable energy, providing a practical reference for energy curtailment calculations. The specific implementation is as follows: The adjustment command is the output of the lower-level intraday rolling optimization model and is stored in the real-time database of the command execution module. It contains the specific operating parameters of each time period and each device. The renewable energy dispatching zone clearly marks the actual dispatching output of photovoltaic and wind power. This output value is the optimal operating value determined by the model after satisfying the objectives of power balance, equipment constraints and cost minimization. It needs to be directly extracted as the basis for actual dispatching output without secondary calculation.

[0066] The determination of actual photovoltaic (PV) dispatch output needs to match the ultra-short-term forecast period: for example, in the adjustment instruction for the period 10:00-10:05, the actual PV dispatch output is marked as... =9.7MW, which is completely consistent with the maximum predicted value for the same period in the ultra-short term (9.7MW), indicating that the model has achieved full photovoltaic consumption and no curtailment by adjusting other equipment (such as increasing energy storage discharge by 0.5MW and purchasing electricity by 2.1MW). If the actual photovoltaic dispatch output in the adjustment instruction is 9.5MW, it means that due to the limited system acceptance capacity (such as insufficient load and full load of energy storage), 0.2MW of photovoltaic power needs to be curtailed. In this case, 9.5MW should be used as the actual dispatch output.

[0067] Determining the actual dispatch output of wind power requires considering the feasibility of adjustment commands: among adjustment commands for the same time period, the actual dispatch output of wind power is marked as... =4.5MW, lower than the maximum forecast value for the ultra-short term (4.7MW). The adjustment instruction included a note that because the energy storage state of charge was 62% (remaining energy 6.2MWh), the acceptable charging power was 0.2MW. Therefore, the actual wind power output was 4.5MW, and the wind curtailment was 0.2MW.

[0068] Once determined, the actual dispatch output of renewable energy needs to be matched one-to-one with the ultra-short-term forecast output by time period to form a forecast-actual comparison sequence, while also indicating the execution status of the adjustment instructions.

[0069] The solar curtailment power is calculated as the difference between the solar ultra-short-term predicted output value and the actual dispatched output value. The wind curtailment power is calculated as the difference between the wind power ultra-short-term predicted output value and the actual dispatched output value. The curtailment power values ​​are then generated.

[0070] This step, by quantifying the difference between the forecast and the actual output, clarifies the scale of renewable energy curtailment and is a core step in assessing the effectiveness of energy absorption. The specific implementation is as follows: The formula for calculating the curtailment power of photovoltaic power is as follows: ,in The value represents the amount of solar power curtailed. If the calculated result is ≥0, it indicates that curtailment exists; if the result is <0, it means that the actual dispatched output exceeds the maximum predicted value, which is an unreasonable situation and needs to be addressed. Corrected to 0.

[0071] The formula for calculating wind curtailment power is as follows: The parameter meanings are the same as those for photovoltaics, and the result processing logic is the same.

[0072] When generating the curtailment power values, it is necessary to organize them by time period and include an explanation of the cause. For example, 10:00-10:05: 0MW of solar curtailment (fully absorbed), 0.2MW of wind curtailment (insufficient energy storage charging capacity, with only 3.8MWh remaining, unable to accommodate the additional 0.2MW); 10:05-10:10: 0MW of solar curtailment, 0.2MW of wind curtailment (load drops to 29.3MW, system acceptance capacity decreases).

[0073] The lower-level optimization objective function applies a cost term for wind and solar curtailment penalties, using a cost mechanism to incentivize maximizing renewable energy consumption, and outputs an optimized renewable energy consumption scheme.

[0074] This step, by introducing economic penalties, guides the optimization model to prioritize the consumption of renewable energy and reduce energy curtailment. It is a key means to achieve low-carbon goals, and is implemented as follows: The cost of curtailing wind and solar power needs to be incorporated into the lower-level optimization objective function, as shown in the formula. ,in, The unit is the photovoltaic curtailment penalty coefficient (yuan / MW). This is the wind curtailment penalty coefficient (unit: yuan / MW). The coefficient value needs to be determined by combining the development cost and low-carbon value of renewable energy: photovoltaic power has a higher penalty coefficient due to its high construction cost and low carbon footprint (close to 0 kgCO2 / MWh), set at 500 yuan / MW; wind power has a slightly lower construction cost, and the penalty coefficient is set at 400 yuan / MW. This value can be adjusted according to policy guidance (for example, when carbon prices rise, the penalty coefficient can be increased to 600 yuan / MW or 500 yuan / MW to further strengthen the incentive for consumption).

[0075] The optimization process must meet all real-time constraints: for example, when adjusting the energy storage charging power, it is necessary to ensure that the SOC after charging is ≤80% (safe upper limit). The calculated SOC after adjustment is 62% + (0.7×5 / 60) / 10×100%≈62.058%<80%, which meets the energy constraint; when reducing the purchased power, it is necessary to ensure that the purchased power is ≥0MW (lower limit), and 1.9MW>0, which meets the interaction constraint with the main grid.

[0076] The optimized renewable energy consumption plan output should include a comparison before and after optimization, along with specific consumption strategies. For example, before optimization: 9.7MW of solar power consumption (0% curtailment), 4.5MW of wind power consumption (0.2MW curtailment), with a penalty cost of 80 yuan; after optimization: 9.7MW of solar power consumption (0% curtailment), 4.7MW of wind power consumption (0% curtailment), with a penalty cost of 0 yuan. The strategy section should include increasing energy storage charging power by 0.2MW (from 0.5MW to 0.7MW) and reducing purchased power by 0.2MW (from 2.1MW to 1.9MW) to achieve full wind power consumption. The plan should cover the entire rolling cycle (4 hours).

[0077] S204, based on the optimized unit output and renewable energy consumption plan, coordinates the charging and discharging strategy of the energy storage system to achieve power balance and carbon emission control, and obtains the optimized dispatch instructions; Specifically, based on the optimized unit output and renewable energy consumption plan, the net power demand of the system can be calculated, the power balance status can be assessed, and the power balance assessment results can be generated. This step quantifies the difference between power generation and load within the system to identify the power supply-demand gap or surplus, providing a core basis for energy storage coordination. The specific implementation is as follows: The calculation of the system's net power demand requires the integration of three key data types: optimized total unit output, actual total renewable energy consumption, and ultra-short-term load forecast. The formula is defined as: System Net Power Demand = (Optimized Total Unit Output + Actual Total Renewable Energy Consumption) - Ultra-Short-Term Load Forecast. Specifically, the optimized total unit output is the sum of the adjusted outputs of the coal-fired units determined by the intraday rolling optimization at the lower level (e.g., Unit 1: 7.8MW, Unit 2: 5.1MW, total output 12.9MW); the actual total renewable energy consumption is the sum of the actual dispatched outputs of photovoltaic and wind power in the consumption plan (e.g., photovoltaic 9.7MW, wind power 4.7MW, total consumption 14.4MW); and the ultra-short-term load forecast is the load demand within the current rolling cycle (e.g., 29.5MW from 10:00 to 10:05). All three are measured in MW to ensure a consistent calculation dimension.

[0078] The power balance assessment results should include three parts: supply and demand status gap, potential supplementation of surplus, and absorption methods. For example, supply and demand status: insufficient power; gap: 2.2MW; potential supplementation methods: energy storage discharge (current SOC 62%, sufficient discharge capacity) or grid power purchase (current electricity purchase price 0.8 yuan / kWh, carbon intensity 800kgCO2 / MWh). The assessment should prioritize low-carbon methods (e.g., energy storage discharge has no carbon emissions, taking precedence over power purchase), while also considering equipment constraints (e.g., upper limits on energy storage charging and discharging power, and electricity purchase and sale quotas) to ensure the feasibility of subsequent strategies.

[0079] Based on the power balance assessment results, the required charging and discharging power and direction of the energy storage system are determined. At the same time, the energy storage state of charge boundary and charging and discharging efficiency are considered to generate the energy storage charging and discharging requirements.

[0080] This step, by combining energy storage operation constraints, translates power balance requirements into specific energy storage action commands, ensuring safe and efficient energy storage adjustment. The specific implementation is as follows: The direction of energy storage charging and discharging is determined by the sign of the system's net power demand: when demand is negative (power shortage), energy storage needs to discharge to supplement power; when demand is positive (power surplus), energy storage needs to charge to absorb the surplus. The charging and discharging power must simultaneously meet three constraints: power upper limit constraint (maximum charging / discharging power of energy storage, such as 5MW), energy boundary constraint (state of charge (SOC) must be within a safe range of 20%-80%), and efficiency constraint (energy loss occurs during charging and discharging, and charging efficiency must meet certain requirements). =0.9, discharge efficiency =0.9).

[0081] Taking a power output of less than 2.2MW as an example, the specific calculation process is as follows: Determine the direction: Demand is negative, energy storage needs to discharge; Calculate the maximum discharge power: Current SOC of energy storage = 62%, corresponding to the remaining energy. =62%×10MWh=6.2MWh (total energy storage capacity 10MWh), SOC lower limit 20% (corresponding to =2MWh), the energy released is ΔE=6.2-2=4.2MWh. The current rolling period is 5 minutes, and the theoretical maximum discharge power is... =ΔE / (1 / 12)=4.2×12=50.4MW, which is much larger than the demand gap of 2.2MW. Therefore, the upper limit of power is determined by demand. Efficiency impact correction: During discharge, the actual output power of the energy storage needs to be multiplied by the efficiency to be considered an effective supplement to the system. The formula is as follows: If an additional 2.2MW is needed, the energy storage discharge power will be... ; Verifying SOC safety: Remaining energy after discharge The corresponding SOC is 5.997 / 10×100%≈59.97%, which is within the range of 20%-80% and meets the constraints.

[0082] The generated energy storage charging and discharging demand needs to clearly define the direction of action, power value, and SOC change target. For example, charging / discharging direction: discharging; discharging power: 2.44MW; current SOC: 62%; target SOC after discharging: ≈60%; effective supplementary power: 2.2MW. If it is a power surplus scenario (e.g., demand +1.8MW), then the charging power needs to be calculated. The verification showed that the SOC after charging was ≤80%, and the final requirement was a charging power of 2MW, with an SOC of ≈64% after charging.

[0083] By combining the system's weighted average carbon emission intensity target, the energy storage charging and discharging strategy is optimized, prioritizing charging during periods of renewable energy surplus and discharging during periods of high carbon intensity, thus generating a low-carbon coordination strategy. This step, by linking to carbon emission targets, ensures that energy storage operations not only meet power balance requirements but also serve low-carbon operation, specifically as follows: System weighted average carbon emission intensity ( This is the core optimization metric, and its calculation formula is based on the upper-level optimization logic: The carbon emissions of a coal-fired unit = unit output × carbon emission factor (e.g., Unit 1) Unit 2 Carbon emissions from electricity purchase = purchased power output × grid carbon intensity ( (Peak hours are 1000 kg CO2 / MWh, normal hours are 2800, and valley hours are 500). Total power generation = unit output + purchased power + renewable energy consumption (renewable energy carbon emissions are considered to be 0).

[0084] The optimization strategy follows the principle of prioritizing low carbon emissions and is designed for two types of scenarios: Charging during periods of renewable energy surplus: When the output of photovoltaic and wind power is high (e.g., 12:00-12:05 noon, with a forecast of 10MW of photovoltaic power and 5MW of wind power, totaling 15MW of renewable energy), the system has a power surplus (e.g., net demand +3MW). At this time, the carbon intensity of renewable energy is 0, and energy storage charging is given priority. High carbon intensity period discharge: When the proportion of main grid power purchase or coal-fired unit output is high (such as 19:00-19:05 in the evening, peak power purchase carbon intensity of 1000kgCO2 / MWh) and the system power is insufficient (such as a shortage of 2.5MW), energy storage discharge should be used to supplement (0 carbon emissions) instead of power purchase (carbon emissions of 2.5×1000=2500kgCO2).

[0085] The generated low-carbon coordination strategy needs to clearly define the carbon emission effects of energy storage actions during priority action periods. The strategy needs to cover the current rolling cycle and the next 1-2 cycles to ensure the continuity of low-carbon goals.

[0086] By integrating unit output adjustment, renewable energy consumption schemes, and low-carbon coordination strategies, optimized dispatch instructions are formed to ensure power balance and carbon emission control, and the final dispatch instruction set is output.

[0087] This step systematically integrates all optimization results to generate structured instructions that can be directly issued and executed, ensuring coordinated actions among execution units. The specific implementation is as follows: Integration should follow a three-step logic: power balance verification, low-carbon target verification, and instruction structuring. Power balance verification: After integrating unit output, renewable energy consumption, and energy storage operations, the total power generation capacity is verified to ensure it equals the load demand. For example, the integrated result is: Unit 1 7.8MW, Unit 2 5.1MW (total 12.9MW), photovoltaic 9.7MW, wind power 4.7MW (total 14.4MW), energy storage discharge 2.44MW (effectively supplementing 2.2MW), power purchase 0MW, power sale 0MW; total power generation capacity = 12.9 + 14.4 + 2.2 = 29.5MW, which is completely balanced with the ultra-short-term load forecast of 29.5MW, ensuring no power deficit or surplus. Low-carbon target verification: Calculate the weighted average carbon emission intensity of the integrated system to confirm compliance with the target. For example, the carbon emissions of the integrated coal-fired unit = 7.8 × 800 + 5.1 × 850 = 10575 kg CO2, the carbon emissions from electricity purchase = 0, and the total power generation = 29.5 MW (for a 5-minute period, the total energy = 29.5 × 1 / 12 ≈ 2.458 MWh). (Because it contains a large amount of renewable energy), it is lower than the target value of 400 kg CO2 / MWh, which meets the low-carbon requirements; Instruction structuring: Instructions are generated by classifying them according to execution units, clearly defining the unit name, action parameters, target requirements, and constraint boundaries. Coal-fired units: Unit 1 output maintained at 7.8MW, ramp rate ≤ 1MW / 15 minutes, avoid start-up and shutdown; Unit 2 output maintained at 5.1MW, SOC ≥ 0.8MW (minimum output). Renewable energy power plants: Photovoltaic power plants actually absorbed 9.7MW, tracking ultra-short-term forecasts, and over-generation is prohibited; wind power plants actually absorbed 4.7MW, with no wind curtailment, and can be finely adjusted by ±0.2MW if wind speed fluctuates. Energy storage system: Discharge 2.44MW, current SOC 62%, SOC after discharge ≈ 60%, charge and discharge efficiency calculated at 0.9, and it is forbidden to exceed the power generation limit of 5MW; Grid interaction unit: 0MW of electricity purchased, 0MW of electricity sold. If the energy storage fails, backup electricity purchase can be activated, with a single purchase of ≤2MW (to avoid large-scale electricity purchases during periods of high carbon intensity).

[0088] The final dispatch instruction set must be output in a clear text format, for example, the optimized dispatch instruction set for the period 10:00-10:05 on October 2, 2025: 1. Coal-fired units: Unit 1 7.8MW, Unit 2 5.1MW; 2. Renewable energy: Photovoltaic 9.7MW, Wind power 4.7MW; 3. Energy storage: Discharge 2.44MW (SOC 62%→60%); 4. Grid interaction: Purchase 0MW, Sale 0MW; Power balance: 29.5MW=29.5MW; System weighted average carbon emission intensity ≈ 358.47kgCO2 / MWh (compliant). The instruction set must include execution priorities (e.g., energy storage actions take precedence over power purchase, renewable energy consumption takes precedence over unit adjustments) to ensure that execution units act sequentially and avoid conflicts.

[0089] S205 executes optimized scheduling instructions and updates the system status through a closed-loop feedback mechanism to complete the scheduling of new energy and low-carbon operation across multiple time scales.

[0090] Specifically, the optimized scheduling instructions can be sent to each execution unit, including coal-fired units, energy storage systems, photovoltaic power plants, wind power plants, and grid interaction units, and the instruction execution status can be generated. This step is crucial for the execution of scheduling instructions. It transmits the optimization results to each device via a reliable communication link, while simultaneously tracking the execution progress to ensure accurate receipt and initiation of instructions. The specific implementation is as follows: The optimized scheduling instructions are stored in the system's instruction distribution module and sent to each execution unit via industrial Ethernet. The distribution frequency is synchronized with the rolling cycle (once every 15 minutes, corresponding to the instruction update for each optimization cycle). The instruction content for different execution units must be adapted to their control interfaces and operating logic. Coal-fired power units: Instructions include target output value, adjustment rate, and requirements for maintaining operating status; Energy storage systems: Instructions include charging and discharging direction, target power, and state of charge (SOC) control range; Photovoltaic power plants / wind power plants: Instructions include the upper limit of actual power consumption and the allowable range of power fluctuations; Grid interaction units: Instructions include the upper limit of purchased power, the upper limit of sold power, and carbon intensity constraint reminders.

[0091] The command execution status is generated through feedback signals from each execution unit. The feedback cycle is 1 second / time. The status is divided into received (command successfully delivered to the device), executing (the device is adjusting to the target parameters), execution completed (the device has reached the target parameters and is stable), and execution abnormal (the device has not adjusted according to the command, such as output deviation exceeding ±5%).

[0092] The system monitors real-time operational data, including actual unit output, actual energy storage charging and discharging power, actual renewable energy output, actual load value, and actual grid interaction power, and generates an actual operational dataset. This step obtains the actual operating status of the system through high-frequency monitoring, providing raw data for subsequent status updates and deviation calculations. The specific implementation is as follows: The monitoring system employs Supervisory Control and Data Acquisition (SCADA) with a monitoring frequency of 1 second per instance (balancing data accuracy and storage efficiency). The data acquisition scope covers all core operating parameters, with each parameter associated with a timestamp and the device ID of the source device. Data accuracy is as follows: Actual unit output: acquired through power sensors of coal-fired units; Actual energy storage charging and discharging power: acquired through the power monitoring module of the energy storage converter (PCS), with discharge being positive and charging negative; Actual renewable energy output: acquired by inverter output power sensors for photovoltaic power plants and by wind turbine control cabinets for wind power plants; Actual load value: acquired through load monitoring instruments on the microgrid main bus; Actual grid interaction power: acquired through power meters at the connection point with the main grid. The actual operating dataset is organized in the format of rolling period - parameter type - 10-second average - data source - accuracy.

[0093] Based on the actual operation dataset, update the system status parameters, including the energy storage charge status and the unit operation status, and calculate the deviation between the actual values ​​and the planned values ​​to generate the updated system status. This step transforms monitoring data into system status information through data processing, quantifies command execution deviations, and provides feedback for the next round of optimization. The specific implementation is as follows: System status parameter updates should focus on core operational metrics, and the calculation logic and parameter definitions must be tailored to the characteristics of the equipment. Energy storage state of charge (SOC) update: Based on the actual charge and discharge power of the energy storage and the initial SOC (SOC at the end of the previous cycle, such as 62%), it is calculated using the energy balance formula: (during discharge) Positive indicates charging, negative indicates charging; Δt is the calculation period, taken as 10 seconds = 10 / 3600 hours; (Total energy storage capacity, such as 10MWh). Unit operating status update: Based on the deviation between actual output and target output (deviation = actual value - target value), combined with the unit operating characteristics, the status is judged: the absolute value of the deviation ≤ 0.05MW is stable operation, 0.05MW < absolute value of deviation ≤ 0.1MW is slight fluctuation, and the absolute value of deviation > 0.1MW requires attention.

[0094] The deviation calculation between actual and planned values ​​must cover all key parameters. The deviation formula is: Deviation = Actual Average Value - Target Value. Unit output deviation: Unit 1 deviation = 7.795MW - 7.8MW = -0.005MW (negative deviation, slight deviation), Unit 2 deviation = 5.12MW - 5.1MW = 0.02MW (positive deviation, slight deviation). Energy storage charging and discharging power deviation: Energy storage deviation = 2.44MW - 2.44MW = 0MW (no deviation), SOC deviation = 61.99% - 62% = -0.01% (negative deviation, slight deviation). Renewable energy output deviation: Photovoltaic deviation = 9.697MW - 9.7MW = -0.003MW (negative deviation, slight deviation), Wind power deviation = 4.69MW - 4.7MW = -0.01MW (negative deviation, slight deviation). Load-grid interaction deviation: Load deviation = 29.5MW - 29.5MW = 0MW (no deviation), Grid interaction deviation = 0MW - 0MW = 0MW (no deviation).

[0095] The updated system status needs to integrate parameter name, updated value, deviation information, and operating status. For example: 1. Unit 1 (coal-fired power plant): updated output 7.795MW, deviation -0.005MW (minor deviation), stable operation; 2. Energy storage system: updated SOC 61.99%, discharge power 2.44MW, deviation 0MW (no deviation), stable discharge; 3. Photovoltaic power station: updated output 9.697MW, deviation -0.003MW (minor deviation), full absorption; 4. Grid interaction: updated power 0MW, deviation 0MW, no power purchase or sale, providing a clear status benchmark for the next round of optimization.

[0096] The updated system status and deviation information are used as input to the lower-level intraday rolling optimization model to start the next rolling cycle optimization, realize closed-loop feedback control, and complete the multi-timescale new energy low-carbon operation scheduling.

[0097] This step incorporates actual operational deviations into the next round of optimization through closed-loop feedback, dynamically correcting the scheduling strategy to ensure the system operates stably in the low-carbon and economical range in the long term. The specific implementation is as follows: The start time of the next rolling cycle is synchronized with the end of the previous cycle. For example, if the current cycle is 10:00-10:15, the next cycle will automatically start at 10:15. When starting, the lower-level intraday rolling optimization model first reads the updated system status (such as energy storage SOC 61.99% and Unit 1 output 7.795MW) and deviation information (such as small deviations of various parameters) from the system status storage module. At the same time, it re-collects the latest ultra-short-term forecast data (such as load forecast of 29.4MW, photovoltaic forecast of 9.6MW, and wind power forecast of 4.6MW for the period of 10:15-10:20) to form a complete input dataset for a new round of optimization.

[0098] The model incorporates the updated system state and deviation information as feedback correction terms into the optimization process: for small deviations (such as a photovoltaic deviation of -0.003MW), the model only fine-tunes the target parameters (such as correcting the next round of photovoltaic target output from 9.6MW to 9.597MW) to avoid large adjustments that could cause system fluctuations; for larger deviations (if they exist, such as a wind power deviation of -0.2MW due to a sudden drop in wind speed), the model will prioritize adjusting energy storage or power purchase strategies (such as increasing energy storage discharge power by 0.2MW, rather than forcing wind power to meet the target) to balance consumption and system stability.

[0099] The core of closed-loop feedback control is a cyclical mechanism of prediction-optimization-execution-monitoring-update-re-optimization: the upper-level day-ahead optimization model provides a 24-hour baseline plan (the long-scale among multiple time scales), and the lower-level intraday rolling optimization model makes short-scale corrections every 15 minutes using real-time status and deviations. The two are linked through plan tracking constraints (the lower-level optimization must not deviate significantly from the upper-level plan).

[0100] Ultimately, through a closed-loop feedback of multiple rounds of rolling optimization, the system achieves multi-timescale scheduling: long-term (day-ahead) to ensure global low-carbon and economic goals, and short-term (intraday) to cope with real-time fluctuations and prediction errors.

[0101] As can be seen, based on load forecast data and renewable energy power forecast data, a day-ahead dispatch plan with the core objective of minimizing the system's weighted average carbon emission intensity is generated through an upper-level day-ahead optimization model. According to the day-ahead dispatch plan, adjustment instructions for each unit's output are generated through a lower-level intraday rolling optimization model. Based on the adjustment instructions, the curtailment of wind and solar power is calculated, and renewable energy consumption is optimized through a penalty cost mechanism. Based on the optimized unit output and renewable energy consumption scheme, the charging and discharging strategies of the energy storage system are coordinated to obtain optimized dispatch instructions. The optimized dispatch instructions are executed, and the system status is updated through a closed-loop feedback mechanism. This enables islanded microgrids to achieve power balance and low-carbon operation under complex and fluctuating conditions, improving the utilization rate of renewable energy and the overall environmental friendliness of the system.

[0102] Another embodiment of the present invention provides a new energy low-carbon operation and scheduling system for isolated microgrids, see [link to relevant documentation]. Figure 3 The system may include: The first generation module 301 is used to generate a day-ahead scheduling plan with the core objective of minimizing the system's weighted average carbon emission intensity based on load forecast data and renewable energy power forecast data through an upper-level day-ahead optimization model. The second generation module 302 is used to generate adjustment instructions for the output of each unit based on the day-ahead scheduling plan, combined with ultra-short-term forecast data and real-time system status data, through the lower-level intraday rolling optimization model. Optimization module 303 is used to calculate the curtailment power of renewable energy based on the adjustment instructions, and optimize the consumption of renewable energy through a penalty cost mechanism; The coordination module 304 is used to coordinate the charging and discharging strategy of the energy storage system according to the optimized unit output and renewable energy consumption plan, so as to achieve power balance and carbon emission control and obtain optimized dispatch instructions. The execution module 305 is used to execute the optimized scheduling instructions and update the system status through a closed-loop feedback mechanism to complete the scheduling of new energy and low-carbon operation at multiple time scales.

[0103] This invention also provides a storage medium storing a computer program, wherein the computer program is configured to execute the steps in any of the above method embodiments when it is run.

[0104] This invention also provides an electronic device, including a memory and a processor, wherein the memory stores a computer program, and the processor is configured to run the computer program to perform the steps in any of the above method embodiments.

[0105] Specifically, the aforementioned electronic device may further include a transmission device and an input / output device, wherein the transmission device is connected to the aforementioned processor, and the input / output device is connected to the aforementioned processor.

[0106] The above description, based on the embodiments shown in the figures, details the structure, features, and effects of the present invention. The above description is only a preferred embodiment of the present invention, but the present invention is not limited to the scope of implementation shown in the figures. Any changes made in accordance with the concept of the present invention, or equivalent embodiments modified to have equivalent changes, that do not exceed the spirit covered by the specification and figures, should be within the protection scope of the present invention.

Claims

1. A method for scheduling low-carbon operation of new energy sources in isolated microgrids, characterized in that, The method includes: Based on load forecast data and renewable energy power forecast data, a day-ahead scheduling plan with the core objective of minimizing the system's weighted average carbon emission intensity is generated through an upper-level day-ahead optimization model. Based on the day-ahead scheduling plan, combined with ultra-short-term forecast data and real-time system status data, adjustment instructions for the output of each unit are generated through the lower-level intraday rolling optimization model. Based on the aforementioned adjustment instructions, the curtailment power of wind and solar power of renewable energy is calculated, and the consumption of renewable energy is optimized through a penalty cost mechanism. Based on the optimized unit output and renewable energy consumption plan, the charging and discharging strategies of the energy storage system are coordinated to achieve power balance and carbon emission control, resulting in optimized dispatch instructions. The system executes optimized scheduling instructions and updates the system status through a closed-loop feedback mechanism to complete the scheduling of new energy and low-carbon operations across multiple time scales.

2. The method according to claim 1, characterized in that, The day-ahead scheduling plan, based on load forecast data and renewable energy power forecast data, is generated through an upper-level day-ahead optimization model with the core objective of minimizing the system's weighted average carbon emission intensity. This includes: Collect load forecast data, photovoltaic power forecast data, wind power forecast data, and grid electricity price for the next few hours to generate a multi-source forecast dataset; Based on the multi-source prediction dataset, an upper-level optimization objective function is constructed, which includes economic items such as coal-fired unit operating costs, electricity purchase and sale costs, and energy storage degradation costs, as well as low-carbon items such as the system weighted average carbon emission intensity, to generate a comprehensive optimization objective function. Based on the system equipment parameters, set power balance constraints, upper and lower limit constraints of unit output, unit ramp rate constraints, energy storage system power and energy constraints, and power interaction constraints with the main grid to generate a complete set of constraints. A mixed-integer linear programming solver is used to solve the comprehensive optimization objective function and the complete set of constraints, and outputs the day-ahead scheduling plan, including the output plan of each unit, the energy storage charging and discharging plan, the power purchase and sale plan, and the renewable energy consumption plan.

3. The method according to claim 2, characterized in that, The step of generating adjustment instructions for each unit's output based on the day-ahead scheduling plan, combined with ultra-short-term forecast data and real-time system status data, through a lower-level intraday rolling optimization model, includes: At the start of the current rolling cycle, real-time system status data is collected, including energy storage charge status, current unit output, and ultra-short-term load forecast data, ultra-short-term photovoltaic power forecast data, and ultra-short-term wind power forecast data, to generate a real-time dataset. Based on real-time datasets and day-ahead scheduling plans, a lower-level optimization objective function is constructed, which includes unit adjustment costs, wind and solar curtailment penalty costs, energy storage degradation costs, and deviation penalty terms from day-ahead plans, thus obtaining a rolling optimization objective function; Based on ultra-short-term forecast data, real-time power balance constraints, renewable energy output constraints, and plan tracking constraints are set. Among them, the plan tracking constraints limit the maximum allowable deviation between the unit output adjustment and the day-ahead plan, and generate a set of real-time constraint conditions. A quadratic programming solver is used to solve the rolling optimization objective function and the real-time constraint set, and outputs adjustment commands for the output of each unit, energy storage, and power purchase and sale.

4. The method according to claim 3, characterized in that, The calculation of wind and solar curtailment power based on the adjustment instructions, and the optimization of renewable energy consumption through a penalty cost mechanism, includes: Extract the maximum short-term forecast values ​​of photovoltaic power output and wind power output from the ultra-short-term forecast data to generate ultra-short-term forecast power output values ​​of renewable energy. The actual dispatch output of renewable energy is determined based on the actual dispatch output of photovoltaic power and wind power in the adjustment instructions. The solar curtailment power is calculated as the difference between the solar ultra-short-term predicted output value and the actual dispatched output value. The wind curtailment power is calculated as the difference between the wind power ultra-short-term predicted output value and the actual dispatched output value. The curtailment power values ​​are then generated. The lower-level optimization objective function applies a cost term for wind and solar curtailment penalties, using a cost mechanism to incentivize maximizing renewable energy consumption, and outputs an optimized renewable energy consumption scheme.

5. The method according to claim 4, characterized in that, The process involves coordinating the charging and discharging strategies of the energy storage system based on the optimized unit output and renewable energy consumption plan to achieve power balance and carbon emission control, resulting in optimized dispatch instructions, including: Based on the optimized unit output and renewable energy consumption scheme, the net power demand of the system is calculated, the power balance status is assessed, and the power balance assessment results are generated. Based on the power balance assessment results, the required charging and discharging power and direction of the energy storage system are determined. At the same time, the energy storage state of charge boundary and charging and discharging efficiency are considered to generate the energy storage charging and discharging requirements. By combining the system's weighted average carbon emission intensity target, the energy storage charging and discharging strategy is optimized, prioritizing charging during periods of renewable energy surplus and discharging during periods of high carbon intensity, thus generating a low-carbon coordination strategy. By integrating unit output adjustment, renewable energy consumption schemes, and low-carbon coordination strategies, optimized dispatch instructions are formed to ensure power balance and carbon emission control, and the final dispatch instruction set is output.

6. The method according to claim 5, characterized in that, The process of executing optimized scheduling instructions and updating the system state through a closed-loop feedback mechanism to complete the scheduling of new energy and low-carbon operations across multiple time scales includes: The optimized scheduling instructions are issued to each execution unit, including coal-fired units, energy storage systems, photovoltaic power plants, wind power plants, and grid interaction units, and the instruction execution status is generated. The system monitors real-time operational data, including actual unit output, actual energy storage charging and discharging power, actual renewable energy output, actual load value, and actual grid interaction power, and generates an actual operational dataset. Based on the actual operation dataset, update the system status parameters, including the energy storage charge status and the unit operation status, and calculate the deviation between the actual values ​​and the planned values ​​to generate the updated system status. The updated system status and deviation information are used as input to the lower-level intraday rolling optimization model to start the next rolling cycle optimization, realize closed-loop feedback control, and complete the multi-timescale new energy low-carbon operation scheduling.

7. A new energy low-carbon operation and scheduling system for isolated microgrids, characterized in that, The system includes: The first generation module is used to generate a day-ahead scheduling plan with the core objective of minimizing the system's weighted average carbon emission intensity, based on load forecast data and renewable energy power forecast data, through an upper-level day-ahead optimization model. The second generation module is used to generate adjustment instructions for the output of each unit based on the day-ahead scheduling plan, combined with ultra-short-term forecast data and real-time system status data, through the lower-level intraday rolling optimization model. The optimization module is used to calculate the curtailment power of renewable energy based on the adjustment instructions, and optimize the consumption of renewable energy through a penalty cost mechanism; The coordination module is used to coordinate the charging and discharging strategies of the energy storage system based on the optimized unit output and renewable energy consumption plan, so as to achieve power balance and carbon emission control and obtain optimized dispatch instructions. The execution module is used to execute optimized scheduling instructions and update the system status through a closed-loop feedback mechanism to complete the scheduling of new energy and low-carbon operation across multiple time scales.

8. The system according to claim 7, characterized in that, The first generation module is specifically used for: Collect load forecast data, photovoltaic power forecast data, wind power forecast data, and grid electricity price for the next few hours to generate a multi-source forecast dataset; Based on the multi-source prediction dataset, an upper-level optimization objective function is constructed, which includes economic items such as coal-fired unit operating costs, electricity purchase and sale costs, and energy storage degradation costs, as well as low-carbon items such as the system weighted average carbon emission intensity, to generate a comprehensive optimization objective function. Based on the system equipment parameters, set power balance constraints, upper and lower limit constraints of unit output, unit ramp rate constraints, energy storage system power and energy constraints, and power interaction constraints with the main grid to generate a complete set of constraints. A mixed-integer linear programming solver is used to solve the comprehensive optimization objective function and the complete set of constraints, and outputs the day-ahead scheduling plan, including the output plan of each unit, the energy storage charging and discharging plan, the power purchase and sale plan, and the renewable energy consumption plan.

9. A storage medium, characterized in that, The storage medium stores a computer program, wherein the computer program is configured to execute the method of any one of claims 1-6 when it is run.

10. An electronic device comprising a memory and a processor, characterized in that, The memory stores a computer program, and the processor is configured to run the computer program to perform the method of any one of claims 1-6.

Images