Virtual power plant load peak clipping method based on mixed integer linear programming

Abstract

The invention discloses a virtual power plant load peak clipping method based on mixed integer linear programming, and belongs to the technical field of virtual power plant load peak clipping. According to the invention, load peak reduction is carried out through minimizing energy and demand cost by using a mixed integer linear programming method. Specifically, by optimizing embedded energy resources, such as distributed energy storage, distributed new energy and the like, the load demand in the peak period is reduced, so that the operation cost of the power system is saved. And demand response is applied in the peak time period, and use of equipment such as an electric water heater, a heat pump and a boiler of a user is limited. Results show that three load daily peak values can be reduced at most, safety and reliability are met, and economic benefits of electric power companies, power generation enterprises and users are improved. According to the method provided by the invention, the system load can be monitored in real time, resource matching is dynamically adjusted, and the load peak value is efficiently reduced.

Description

Technical Field

[0001] The invention belongs to the technical field of load peak shaving of virtual power plants, and in particular relates to a load peak shaving method of virtual power plants based on mixed integer linear programming. Background Art

[0002] Reducing the peak load of virtual power plants is a key means to ensure the security of the power system while improving economic efficiency. By implementing energy-saving and voltage-reducing technologies to work in conjunction with embedded energy resources, the power company can optimize the electricity cost during peak hours. Embedded energy resources include distributed energy resources, such as solar and diesel generators, and energy storage systems, such as commercial and household energy storage. Through appropriate intelligent scheduling strategies, the output of distributed energy resources and the charging and discharging of energy storage systems are dynamically adjusted to track changes in grid demand. That is, during peak hours when electricity demand surges, the dependence on high-cost peak power is alleviated by accurately reducing loads, which not only enhances the overall efficiency and reliability of the power system, but also significantly improves the utilization rate of clean energy, realizes the refined management of power resources and flexible response on the demand side, and lays the foundation for building a green and intelligent power grid system.

[0003] At present, the research on virtual power plant load peak shaving strategies mainly focuses on the demand-side management of power systems and the integration of renewable energy. Through intelligent control, energy storage systems, demand response, multi-energy integration, optimization algorithms, etc., while ensuring the stability and efficiency of the power system, it promotes the efficient use and sustainable development of renewable energy. However, existing studies have not discussed the ability to eliminate multiple peaks of virtual power plant loads in a day or conducted economic analysis. The reason for multiple peaks may be weather conditions: in cold weather, customers use a large number of devices such as electric water heaters, heat pumps, and electric blankets, resulting in multiple peaks; or the impact of the energy storage system: the energy storage system can reduce the peak demand load by discharging during peak hours, but if the energy storage system is charged at the same time during non-peak hours, it will cause load peaks. Therefore, the following problems still exist in the current complex work of eliminating multiple peaks of virtual power plant loads:

[0004] The universality of the control strategy that eliminates a single peak in a day to expand to multiple peaks in a day: In the design of control strategies, the strategy of eliminating a single peak in a day usually focuses on optimizing the load within a specific time window. To expand to multiple peaks in a day, it is necessary to consider the dynamic changes of the load curve. The control strategy needs to have stronger adaptability and predictive capabilities, monitor the grid load in real time, identify multiple peaks and adjust them quickly. At the same time, different types of loads need to be considered, such as weather-related loads (electric water heaters, heat pumps, barbecue grills, electric blankets, etc.) and the charging and discharging behavior of the energy storage system, effectively reducing multiple peaks while avoiding additional peaks caused by the simultaneous charging of the energy storage system.

[0005] The universality of the control strategy that eliminates multiple peaks in a day to expand to different seasons throughout the year: To expand the control strategy from eliminating multiple peaks in a day to different seasons throughout the year, the impact of seasonal load changes and weather conditions must be considered. Temperature and light conditions in different seasons will affect the efficiency of solar power generation and heat pumps, as well as the frequency of use of equipment such as electric water heaters and electric blankets. The control strategy must have predictive capabilities to predict load trends and adjust resource allocation based on historical data and weather forecasts. In addition, it is necessary to consider the impact of seasonal changes in the life and efficiency of the energy storage system, as well as the optimization of the charging and discharging strategy of the energy storage system in different seasons.

[0006] Is it more economical to cut multiple peaks in a day than to cut one peak? From an economic perspective, the strategy of cutting multiple peaks in a day is more economical than cutting a single peak. However, the control strategy needs to accurately calculate and balance the marginal cost and benefits of peak cutting to ensure that the total cost savings of the control strategy under cutting multiple peaks is greater than the additional cost of implementing the strategy, such as the accelerated aging caused by the frequent charging and discharging of the energy storage system.

[0007] Therefore, a new technical solution is urgently needed in the prior art to solve this problem. Summary of the invention

[0008] The technical problem to be solved by the present invention is to provide a virtual power plant load peak shaving method based on mixed integer linear programming to solve the technical problems that the current complex virtual power plant load multi-peak shaving work still has technical difficulties in terms of the universality of expanding to multiple peaks within a day, the universality of expanding to different seasons throughout the year, and the economy of reducing the number of peaks.

[0009] A virtual power plant load peak shaving method based on mixed integer linear programming includes the following steps, and the following steps are performed in sequence:

[0010] Step 1: Read the load forecast data for the next 30 days, track the highest peak in the load forecast data, and compare the highest peak with other peaks. If the highest peak exceeds 50% of other peaks, it is determined to be an important peak and needs to be reduced; if the highest peak does not exceed 50% of other peaks, it is determined to be a non-important peak;

[0011] The highest peak is an important peak and needs to be reduced. By turning on the diesel generator during the highest peak, the highest peak can be reduced to no more than 50% of other peaks and become a non-important peak.

[0012] If the highest peak value is a non-important peak value, the initial peak load reduction threshold is determined by taking the average value of the highest peak value of the historical load data and the maximum load peak data in the previous month, which is used for subsequent peak reduction. Load peaks exceeding this load threshold need to be reduced.

[0013] Step 2: Carry out load forecast for the next 24 hours to obtain the predicted load value;

[0014] If the predicted load value is lower than the set threshold, the distributed energy storage system will be charged;

[0015] If the predicted load value is higher than or equal to the set threshold, the load is reduced by reducing the use of electric water heaters, heat pumps, electric heaters and electric thermal energy storage equipment, specifically:

[0016] ① Setting the objective function: According to the electricity cost function, the objective function is set to minimize the electricity cost. The objective function of minimizing the load peak shaving electricity cost is transformed into a total electricity cost minimization function of load peak shaving that integrates energy costs and peak-valley electricity price differences through the mixed integer linear programming method;

[0017] ② Set resource constraints: Set operational constraints for all energy resources involved in peak reduction;

[0018] ③Perform constant voltage regulation and discharge of energy storage system: First, the constant voltage regulation method is applied to control the embedded energy resources to reduce the grid voltage, and then the energy storage system is discharged to reduce the load;

[0019] ④ Calculate photovoltaic power generation: Calculate the output power of photovoltaic power generation based on solar radiation data;

[0020] ⑤ Calculate energy cost: Calculate energy cost for each aggregated resource;

[0021] ⑥Optimize resource aggregation capacity: Virtual power plants share load peak information with customers and optimize resource aggregation capacity based on the lowest energy cost;

[0022] ⑦ Determine whether resource aggregation meets peak load shaving requirements:

[0023] If the resource aggregation capacity is less than the peak load reduction capacity, it is determined that resource aggregation cannot completely reduce the peak load, and distributed generation resources are used, and the load peak reduction objective function of distributed generation resources is set to complete the complete reduction of the peak load and proceed to the next step;

[0024] If the resource aggregation capacity is greater than or equal to the peak load reduction capacity, it is determined that resource aggregation can completely reduce the peak load, and the next step is directly performed;

[0025] ⑧ Based on the new peak update threshold: obtain the total electricity cost after reducing the peak load, and repeat the load forecast and peak load reduction for the next 24 hours according to the peak update load threshold after the reduction, and obtain the total electricity cost after the peak load reduction in this cycle;

[0026] ⑨ Billing cycle days: Set the billing cycle days to 30 days. If the number of repetitions of step ⑧ does not meet 30 times, continue to repeat the prediction and billing. If the number of repetitions of step ⑧ meets 30 times, proceed to the next step;

[0027] ⑩End: Complete the operations of the current cycle and enter the next billing cycle.

[0028] The total electricity cost minimization function of load peak shaving in step ①, which integrates energy costs and peak-valley electricity price differences, is:

[0029] MinC T (6);

[0030] C T =C E +C P (7);

[0031]

[0032] C P =α·P shave (9);

[0033] C T represents the total electricity cost of load peak shaving, C E Indicates energy cost C P represents the difference between peak and valley electricity prices; P Lshave and P shave are the target peak load shaving and maximum peak load shaving indicators respectively; ζ and α represent the energy rate and the peak-valley electricity price difference rate respectively; N represents the load forecast value for the next 24 hours.

[0034] The resource constraints in step ② are:

[0035]

[0036] In formula (11), P batt (t) is the dynamic output power function of energy storage, P DGi (t) is the generator capacity limitation function, is the maximum capacity of the generator, CVR(t) is the constant voltage regulation strategy capability function, CVR max is the maximum value of constant pressure regulation capability, P PV (t) is the photovoltaic capacity limitation function, Soc(t) is the energy storage charging function, P G (t) is the main power generation function of the power plant, is the maximum value of the main power generation of the power plant, Soc min 、Soc max are the minimum and maximum values ​​of energy storage charging, respectively. is the maximum photovoltaic capacity;

[0037]

[0038] Formula (12) represents the generator set P G (t), the remaining power generation resources P i (t) and load forecast P L (t) power balance state; Formula (13) is the minimization cost function of the remaining power generation resources, P EWH (t) is the water heater power, P HP (t) is the heat pump power, P BBH (t) is the power of electric heater, P ETS (t) is the electric thermal energy storage power,

[0039] Put each power generation resource into operation at the time of its lowest energy cost:

[0040]

[0041] In formula (14), γ c (t) represents the binary variable of distributed energy storage, γ c (t) = 0, then the charging and discharging are terminated, γ c (t) = 1, then the discharge is terminated and charging is started. M is the target capacity for charging or discharging, and P batt is the energy storage dynamic power;

[0042]

[0043] In formula (15), Peak is the peak load, P shave is the load peak reduction amount, P threshold It is the preset load peak shaving threshold.

[0044] The formula for the output power of photovoltaic power generation in step ④ is:

[0045] P PV (t) = I s (t)·A PV ·N PV ·η PV ·η t (1);

[0046] Among them, A PV and N PV Respectively represent the area and number of photovoltaic panels, η PV and η t Respectively represent the photovoltaic system efficiency and temperature coefficient, where η t The calculation formula is:

[0047] η t =1-[μ(T c-T stc )] (2);

[0048] In formula (2), T c and T stc They represent the actual operating temperature of the photovoltaic cell and the temperature under standard test conditions respectively, and μ is the temperature coefficient of the maximum output power of the photovoltaic panel.

[0049] The energy cost in step ⑤ is expressed by the following formula:

[0050] Cost T =Cost TInvest +Cost EPC (16);

[0051] In the formula, Cost T and Cost TInvest Represent the total cost and unit investment of aggregated resources, Cost EPC for energy costs;

[0052] Cost TInvest =Cost U +Cost Mis (17);

[0053] In the formula, Cost U and Cost Mis They represent the unit cost of aggregated resources and the remaining miscellaneous costs respectively;

[0054] Cost U =Cost UEWH +Cost UHP +Cost UBBH +Cost UETS (18);

[0055] In the formula, Cost UEWH Cost UHP Cost UBBH and Cost UETS represent the unit costs of electric water heaters, heat pumps, electric heaters, and electric thermal energy storage devices, respectively;

[0056] Cost Mis =Cost MisIns +Cost MisCont +Cost MisMain (19);

[0057] In the formula, Cost MisIns Cost MisCont and Cost MisMain They represent the installation cost, control cost and maintenance cost of aggregated energy, respectively;

[0058] Cost EPC =(β×P c ) (20);

[0059] In the formula, β and P c They represent the point price difference and power consumption of each aggregated energy at peak and valley times;

[0060]

[0061] In the formula, Cost Thour 、N U and Cost TInvest They are the total energy cost per hour, the number of devices, and the total investment cost of each aggregate energy source.

[0062] The load peak shaving objective function of the distributed generation resources in step ⑦ is represented by the load peak shaving objective function of the diesel generator:

[0063]

[0064] In the formula, Min represents the function minimization target, t is the power, and a i 、b i and c i is the fuel consumption cost coefficient, Represents the generating capacity of each generator, N gen represents the total number of generators, Indicates the remaining load after the diesel generator is put into use. Indicates the generator capacity range.

[0065] Through the above design scheme, the present invention can bring the following beneficial effects:

[0066] 1. The present invention proposes a virtual power plant load peak shaving control strategy method that fully considers safety and economy. A mixed integer linear programming method is used to reduce load peaks by minimizing energy and demand costs. Specifically, embedded energy resources, such as distributed energy storage or distributed new energy, are optimized to reduce load demand during peak hours, thereby saving power system operating costs. Demand response is applied during peak hours to limit the use of users' electric water heaters, heat pumps, boilers and other equipment. The results show that the present invention can reduce up to three daily load peaks, which not only meets safety and reliability, but also improves the economic benefits of power companies, power generation companies and users.

[0067] 2. The present invention manages the embedded energy resources within the virtual power plant in a refined manner, thereby eliminating load peaks while ensuring system safety, thereby improving the economy of system operation.

[0068] 3. The present invention fully considers the resource scheduling of distributed new energy and energy storage systems in virtual power plants, such as solar photovoltaics, diesel generators and distributed energy storage, to effectively reduce the load peak.

[0069] 4. The method proposed in the present invention can eliminate one load peak in a single day, and simultaneously eliminate multiple load peaks in a single day, which is more practical and flexible.

[0070] 5. The method proposed in the present invention can monitor the system load in real time, dynamically adjust resource matching, and achieve efficient reduction of load peak. BRIEF DESCRIPTION OF THE DRAWINGS

[0071] The present invention is further described below with reference to the accompanying drawings and specific embodiments:

[0072] Figure 1 This is a technical architecture diagram of a virtual power plant load peak shaving method based on mixed integer linear programming in the present invention.

[0073] Figure 2 It is a flow chart of a virtual power plant load peak shaving method based on mixed integer linear programming of the present invention;

[0074] Figure 3 A system topology diagram of an example of an embodiment of a virtual power plant load peak shaving method based on mixed integer linear programming in the present invention;

[0075] Figure 4 A peak shaving situation diagram of a single load peak occurring within one day in an embodiment of a virtual power plant load peak shaving method based on mixed integer linear programming of the present invention;

[0076] Figure 5 A peak shaving situation diagram of peak 1 among three load peaks occurring within one day in an embodiment of a virtual power plant load peak shaving method based on mixed integer linear programming of the present invention;

[0077] Figure 6 A peak shaving situation diagram of peak 2 among three load peaks occurring within one day in an embodiment of a virtual power plant load peak shaving method based on mixed integer linear programming of the present invention;

[0078] Figure 7 This is a peak shaving diagram of peak value 3 among three load peak values ​​occurring within one day in an embodiment of a virtual power plant load peak shaving method based on mixed integer linear programming of the present invention. DETAILED DESCRIPTION

[0079] The present invention proposes a virtual power plant load peak shaving control strategy method that fully considers safety and economy. A mixed integer linear programming method is used to reduce load peaks by minimizing energy and demand costs. Specifically, embedded energy resources, such as distributed energy storage or distributed new energy, are optimized to reduce load demand during peak hours, thereby saving the operating cost of the power system. Demand response is applied during peak hours to limit the use of users' electric water heaters, heat pumps, boilers and other equipment. The results show that the proposed control strategy can reduce up to three load day peaks, which meets safety and reliability while improving the economic benefits of power companies, power generation companies and users.

[0080] Figure 1 This is a technical architecture diagram of virtual power plant load peak shaving. Virtual power plants control several distributed power sources, such as local power generation equipment such as solar photovoltaics and diesel generators, through strategic scheduling. Virtual power plants determine peak loads and formulate strategies for peak shaving through different types of load forecasts. Subsequently, virtual power plants share load peak information with customers and formulate demand response strategies to reduce resource aggregation during load peaks.

[0081] Virtual power plants manage peak loads by setting load peak shaving thresholds, which are related to the duration of peak loads. The initial load threshold is set based on the average of the maximum values ​​of the same month in previous years and is updated based on the daily peak shaving situation. If the new peak value is equal to the threshold value, it remains unchanged, otherwise it is updated to the new peak value.

[0082] The solar photovoltaic model in the system structure can be expressed by formula (1), the output power P of the photovoltaic panel PV (t) and solar irradiance I s The relationship between (t) is:

[0083] P PV (t) = I s (t)·A PV ·N PV ·η PV ·η t (1);

[0084] Among them, A PV and N PV Respectively represent the area and number of photovoltaic panels, η PV and η t Respectively represent the photovoltaic system efficiency and temperature coefficient, where η t The calculation formula is:

[0085] η t =1-[μ t (T c -T stc )] (2);

[0086] In formula (2), T c and T stc They represent the actual operating temperature of the photovoltaic cell and the temperature under standard test conditions. stc =25℃ is the best working temperature. t is the temperature coefficient of the maximum output power of the photovoltaic panel.

[0087] Load peak shaving objective function for diesel generators It can be expressed as:

[0088]

[0089] In the above formula, Min represents the function minimization target, t is the power, and a i 、b i and c i is the fuel consumption cost coefficient, Represents the generating capacity of each generator, N gen represents the total number of generators, Indicates the remaining load after the diesel generator is put into use. Indicates the generator capacity range.

[0090] The objective function of minimizing the peak load shaving electricity cost is transformed into the control of embedded energy through the mixed integer linear programming method:

[0091] MinC T (6);

[0092] C T =C E +C P (7);

[0093]

[0094] C P =α·P shave (9);

[0095] Formulas (6)-(9) represent the total electricity cost C of load peak shaving T , including energy costs C E and peak-valley electricity price difference C P .P Lshave and P shave are the target peak load reduction and the maximum peak load reduction indicators, respectively. ζ and α represent the energy rate and the peak-valley electricity price difference rate, respectively. N represents the load forecast value for the next 24 hours.

[0096]

[0097] In formula (10), Soc(t) is the energy storage charging function, P batt(t) is the dynamic output power function of the energy storage, λ is the charging efficiency, μ is the discharging efficiency, and δ is the dynamic capacity parameter of the energy storage.

[0098]

[0099] In formula (11), P batt (t) is the dynamic output power function of energy storage, P DGi (t) is the generator capacity limitation function, is the maximum capacity of the generator, CVR(t) is the constant voltage regulation strategy capability function, CVR max is the maximum value of constant pressure regulation capability, P PV (t) is the photovoltaic capacity limitation function, Soc(t) is the energy storage charging function, P G (t) is the main power generation function of the power plant, Soc min 、Soc max They are the minimum and maximum values ​​of energy storage charging respectively; is the maximum value of the main power generation of the power plant, is the maximum photovoltaic capacity;

[0100]

[0101] Formula (12) represents the generator set P G (t), the remaining power generation resources P i (t) and load forecast P L (t) power balance state; Formula (13) is the minimization cost function of the remaining power generation resources, P EWH (t) is the water heater power, P HP (t) is the heat pump power, P BBH (t) is the power of electric heater, P ETS (t) is the electric thermal energy storage power. Formulas (16)-(21) can be used to put each power generation resource into operation at the time point with the lowest energy cost.

[0102]

[0103] In formula (14), γ c (t) represents the binary variable of distributed energy storage, γ c (t) = 0, then the charging and discharging are terminated, γ c (t) = 1, then the discharge is terminated and charging is started. M is the target capacity for charging or discharging, and P batt is the energy storage dynamic power;

[0104]

[0105] In formula (15), Peak is the peak load, P shaveis the load peak reduction amount, P threshold It is the preset load peak shaving threshold.

[0106] The energy cost of each power generation resource can be expressed as follows:

[0107] Cost T =Cost TInvest +Cost EPC (16);

[0108] In the formula, Cost T and Cost TInvest Represent the total cost and unit investment of aggregated resources, Cost EPC For energy costs.

[0109] Cost TInvest =Cost U +Cost Mis (17);

[0110] In the formula, Cost U and Cost Mis They represent the unit cost and remaining miscellaneous cost of aggregated resources respectively.

[0111] Cost U =Cost UEWH +Cost UHP +Cost UBBH +Cost UETS (18);

[0112] In the formula, Cost UEWH Cost UHP Cost UBBH and Cost UETS Represent the unit costs of electric water heaters, heat pumps, electric heaters and electric thermal energy storage equipment respectively.

[0113] Cost Mis =Cost MisIns +Cost MisCont +Cost MisMain (19);

[0114] In the formula, Cost MisIns Cost MisCont and Cost MisMain They represent the installation cost, control cost and maintenance cost of aggregated energy respectively.

[0115] Cost EPC =(β×P c ) (20);

[0116] In the formula, β and Pc They respectively represent the price difference and power consumption of each aggregated energy at peak and valley times.

[0117]

[0118] In the formula, Cost Thour 、N U and Cost TInvest are the total energy cost per hour, the number of devices, and the total investment cost of each aggregate energy source,

[0119] Figure 2 This is the flow chart of the entire control algorithm.

[0120] The algorithm starts by reading the load forecast data to predict future electricity demand.

[0121] The highest peak in the load data is tracked and compared with other peaks. If the detected peak exceeds 50% of other peaks, it is determined to be an important peak and needs to be reduced.

[0122] If it is an important peak, the diesel generator is started during the highest peak period. Considering that diesel generators usually have higher operating costs, the diesel generator set is only started during the highest peak period.

[0123] If it is not an important peak, an initial load threshold is determined based on historical data and the maximum load in previous months for subsequent peak reduction.

[0124] Conduct 24-hour load forecasts to get detailed information on upcoming power demand.

[0125] If the predicted load is lower than the set threshold, the distributed energy storage system is charged.

[0126] If the predicted load is higher than the set threshold, the load is reduced by reducing the use of electric water heaters, heat pumps, electric heaters and electric thermal energy storage devices, specifically:

[0127] Setting objective function: Based on the electricity cost function, set the objective function to minimize the electricity cost.

[0128] Set asset constraints: Set operational constraints for all energy resources participating in peak reduction.

[0129] Apply constant voltage regulation and discharge of energy storage system: First, constant voltage regulation technology is applied, and then the energy storage system is discharged to reduce the load. The discharge capacity of the energy storage system is optimized and defined by a mixed integer linear programming model. The specific discharge capacity depends on the constraints, formulas (6)-(11).

[0130] Calculate photovoltaic power generation: Calculate the output power of photovoltaic power generation based on solar radiation data.

[0131] Calculate Energy Cost: Calculate the energy cost for each aggregate resource to optimize its operation.

[0132] Optimize resource aggregate capacity: Optimize the capacity of resource aggregates based on the lowest energy cost.

[0133] Resource aggregation cannot meet peak load reduction requirements: If resource aggregation cannot completely reduce peak load, distributed generation resources are used.

[0134] Update thresholds based on new peak values: Update the load thresholds based on the reduced peak values ​​for use in the next cycle.

[0135] Billing cycle days: Set the billing cycle days to 30 days.

[0136] End: Complete the operation of the current cycle and enter the next cycle.

[0137] The entire flowchart embodies a dynamic, real-time optimization process that aims to effectively reduce peak loads, reduce electricity costs, and improve the stability and reliability of the power grid by comprehensively considering various energy resources and demand response measures.

[0138] Example

[0139] The virtual power plant is set to include 2 solar photovoltaic stations, 1 power plant, 3 centralized energy storage stations, and 4 distributed energy storage systems. Figure 3 This is the topology diagram of the example system.

[0140] Figure 4 The peak shaving performance of the proposed strategy method for a single load peak occurring within 1 day is demonstrated. From the enlarged detail diagram, it can be seen that the load peak lasting 4 hours can be completely shaving.

[0141] Figure 5 to Figure 7 The peak shaving performance of the strategy method proposed in the present invention for three load peaks in one day is shown. The first load peak occurs from 14:00 to 20:00. From the enlarged detail diagram, it can be seen that the load peak is not completely shaving. The second load peak occurs from 20:30 to 22:00. From the enlarged detail diagram, it can be seen that the load peak is completely shaving. The third short-term load peak occurs after 23:30. From the enlarged detail diagram, it can be seen that the load peak is completely shaving.

Claims

1. A virtual power plant load peak shaving method based on mixed integer linear programming, characterized by: The method comprises the following steps, and the following steps are performed in sequence: Step 1: Read the load forecast data for the next 30 days, track the highest peak in the load forecast data, and compare the highest peak with other peaks. If the highest peak exceeds 50% of other peaks, it is determined to be an important peak and needs to be reduced; if the highest peak does not exceed 50% of other peaks, it is determined to be a non-important peak; The highest peak is an important peak and needs to be reduced. By turning on the diesel generator during the highest peak, the highest peak can be reduced to no more than 50% of other peaks and become a non-important peak. If the highest peak value is a non-important peak value, the initial peak load reduction threshold is determined by taking the average value of the highest peak value of the historical load data and the maximum load peak data in the previous month, which is used for subsequent peak reduction. Load peaks exceeding this load threshold need to be reduced. Step 2: Carry out load forecast for the next 24 hours to obtain the predicted load value; If the predicted load value is lower than the set threshold, the distributed energy storage system will be charged; If the predicted load value is higher than or equal to the set threshold, the load is reduced by reducing the use of electric water heaters, heat pumps, electric heaters and electric thermal energy storage equipment, specifically: ① Setting the objective function: According to the electricity cost function, the objective function is set to minimize the electricity cost. The objective function of minimizing the load peak shaving electricity cost is transformed into a total electricity cost minimization function of load peak shaving that integrates energy costs and peak-valley electricity price differences through the mixed integer linear programming method; ② Set resource constraints: Set operational constraints for all energy resources involved in peak reduction; ③Perform constant voltage regulation and discharge of energy storage system: First, the constant voltage regulation method is applied to control the embedded energy resources to reduce the grid voltage, and then the energy storage system is discharged to reduce the load; ④ Calculate photovoltaic power generation: Calculate the output power of photovoltaic power generation based on solar radiation data; ⑤ Calculate energy cost: Calculate energy cost for each aggregated resource; ⑥Optimize resource aggregation capacity: Virtual power plants share load peak information with customers and optimize resource aggregation capacity based on the lowest energy cost; ⑦ Determine whether resource aggregation meets peak load shaving requirements: If the resource aggregation capacity is less than the peak load reduction capacity, it is determined that resource aggregation cannot completely reduce the peak load, and distributed generation resources are used, and the load peak reduction objective function of distributed generation resources is set to complete the complete reduction of the peak load and proceed to the next step; If the resource aggregation capacity is greater than or equal to the peak load reduction capacity, it is determined that resource aggregation can completely reduce the peak load, and the next step is directly performed; ⑧ Based on the new peak update threshold: obtain the total electricity cost after reducing the peak load, and repeat the load forecast and peak load reduction for the next 24 hours according to the peak update load threshold after the reduction, and obtain the total electricity cost after the peak load reduction in this cycle; ⑨ Billing cycle days: Set the billing cycle days to 30 days. If the number of repetitions of step ⑧ does not meet 30 times, continue to repeat the prediction and billing. If the number of repetitions of step ⑧ meets 30 times, proceed to the next step; ⑩End: Complete the operations of the current cycle and enter the next billing cycle.

2. The method for load peak shaving of a virtual power plant based on mixed integer linear programming according to claim 1 is characterized in that: The total electricity cost minimization function of load peak shaving in step ①, which integrates energy costs and peak-valley electricity price differences, is: MinC T (6); C T =C E +C P (7); C P =α·P shave (9); C T represents the total electricity cost of load peak shaving, C E Indicates energy cost C P represents the difference between peak and valley electricity prices; P Lshave and P shave are the target peak load shaving and maximum peak load shaving indicators respectively; ζ and α represent the energy rate and the peak-valley electricity price difference rate respectively; N represents the load forecast value for the next 24 hours.

3. The method for load peak shaving of a virtual power plant based on mixed integer linear programming according to claim 1 is characterized in that: The resource constraints in step ② are: In formula (11), P batt (t) is the dynamic output power function of energy storage, P DGi (t) is the generator capacity limitation function, is the maximum capacity of the generator, CVR(t) is the constant voltage regulation strategy capability function, CVR max is the maximum value of constant pressure regulation capability, P PV (t) is the photovoltaic capacity limitation function, Soc(t) is the energy storage charging function, P G (t) is the main power generation function of the power plant, is the maximum value of the main power generation of the power plant, Soc min , Soc max are the minimum and maximum values ​​of energy storage charging, respectively. is the maximum photovoltaic capacity; Formula (12) represents the generator set P G (t), the remaining power generation resources P i (t) and load forecast P L (t) power balance state; Formula (13) is the minimization cost function of the remaining power generation resources, P EWH (t) is the water heater power, P HP (t) is the heat pump power, P BBH (t) is the power of electric heater, P ETS (t) is the electric thermal energy storage power, Put each power generation resource into operation at the time of its lowest energy cost: In formula (14), γ c (t) represents the binary variable of distributed energy storage, γ c (t) = 0, then the charging and discharging are terminated, γ c (t) = 1, then the discharge is terminated and charging is started. M is the target capacity for charging or discharging, and P batt is the energy storage dynamic power; In formula (15), Peak is the peak load, P shave is the load peak reduction amount, P threshold It is the preset load peak shaving threshold.

4. The method for load peak shaving of a virtual power plant based on mixed integer linear programming according to claim 1 is characterized in that: The formula for the output power of photovoltaic power generation in step ④ is: P PV (t)=I s (t)·A PV ·N PV ·or PV ·or t (1); Among them, A PV and N PV Respectively represent the area and number of photovoltaic panels, η PV and η t Respectively represent the photovoltaic system efficiency and temperature coefficient, where η t The calculation formula is: or t =1-[μ(T c -T stc )] (2); In formula (2), T c and T stc They represent the actual operating temperature of the photovoltaic cell and the temperature under standard test conditions respectively, and μ is the temperature coefficient of the maximum output power of the photovoltaic panel.

5. The method for load peak shaving of a virtual power plant based on mixed integer linear programming according to claim 1 is characterized in that: The energy cost in step ⑤ is expressed by the following formula: Cost T =Cost TInvest +Cost EPC (16); In the formula, Cost T Cost TInvest Represent the total cost and unit investment of aggregated resources, Cost EPC for energy costs; Cost TInvest =Cost U +Cost Mis (17); In the formula, Cost U Cost Mis They represent the unit cost of aggregated resources and the remaining miscellaneous costs respectively; Cost U =Cost UEWH +Cost UHP +Cost UBBH +Cost UETS (18); In the formula, Cost UEWH Cost UHP Cost UBBH Cost UETS represent the unit costs of electric water heaters, heat pumps, electric heaters, and electric thermal energy storage devices, respectively; Cost Mis =Cost MisIns +Cost MisCont +Cost MisMain (19); In the formula, Cost MisIns Cost MisCont Cost MisMain They represent the installation cost, control cost and maintenance cost of aggregated energy, respectively; Cost EPC =(β×P c ) (20); In the formula, β and P c They represent the point price difference and power consumption of each aggregated energy at peak and valley times; In the formula, Cost Thour 、N U Cost TInvest They are the total energy cost per hour, the number of devices, and the total investment cost of each aggregate energy source.

6. The method for load peak shaving of a virtual power plant based on mixed integer linear programming according to claim 1 is characterized in that: The load peak shaving objective function of the distributed generation resources in step ⑦ is represented by the load peak shaving objective function of the diesel generator: In the formula, Min represents the function minimization target, t is the power, and a i 、b i and c i is the fuel consumption cost coefficient, Represents the generating capacity of each generator, N gen represents the total number of generators, Indicates the remaining load after the diesel generator is put into use. Indicates the generator capacity range.

Images