An energy optimization scheduling method and system taking into account the uncertainty of wind and solar power

Abstract

The present invention relates to the field of energy dispatching technology, and discloses an energy optimization dispatching method and system taking into account the uncertainty of wind and photovoltaic power, including calculating the average daily forecast relative deviation value according to the day-ahead forecast data and the actual output data, classifying the uncertainty scenarios of the historical data set, calculating the initial probability distribution corresponding to each scenario level, randomly selecting scenario representatives from each scenario level, and constructing a discrete uncertainty scenario set; establishing an equipment operation model, based on the discrete uncertainty scenario set, taking the optimal dispatching resource allocation efficiency as the objective function, setting constraints according to the equipment operation model, and constructing a two-layer distributed optimization dispatching model; solving the two-layer distributed optimization dispatching model to obtain the optimal dispatching strategy. The present invention effectively improves the system's new energy consumption level, can meet the real-time reliable energy supply of the integrated energy system, and improves the robustness of the system's comprehensive operation and resource allocation efficiency under uncertain scenarios.

Description

Technical Field

[0001] The present invention relates to the technical field of energy scheduling, and in particular to an energy optimization scheduling method and system taking into account the uncertainty of wind and solar power. Background Art

[0002] With the continuous development of my country's economy and society, the energy production and consumption patterns are undergoing major changes. The energy industry is shouldering new missions such as improving energy efficiency, ensuring energy security, promoting the consumption of new energy and promoting environmental protection. The traditional energy system is mainly based on the vertical extension of a single system, with less physical interconnection and information interaction between energy systems. The integrated energy system uses advanced technology and management models to integrate various energy resources such as oil, coal, natural gas and electricity in the region, and realizes coordinated planning, optimized operation, collaborative management, interactive response and mutual complementarity between multiple heterogeneous energy subsystems. While meeting diversified energy needs, it effectively improves energy utilization efficiency, and improves the absorption capacity of clean energy through energy transposition and cascade utilization, thereby promoting sustainable energy development. However, the actual output of wind and photovoltaic power is affected by many factors such as wind speed, wind direction, temperature, humidity, etc., and it is difficult for existing prediction technologies to eliminate the existence of errors. Its uncertainty has an unignorable impact on the stability and security of energy supply and the absorption capacity of clean energy in the integrated energy system.

[0003] At present, most of the research on power system uncertainty focuses only on the uncertainty of wind power, and lacks research on photovoltaic uncertainty. In fact, due to the different environmental influencing factors of wind power and photovoltaic power generation, the uncertainties of the two are inconsistent. The current research on power system uncertainty cannot take into account the uncertainty of wind power and photovoltaic uncertainty at the same time, and thus cannot meet the scheduling needs of the actual operation of the power system. Summary of the invention

[0004] In order to solve the above technical problems, the present invention provides an energy optimization scheduling method and system taking into account the uncertainty of wind and solar power. By constructing a multi-scenario wind and solar power uncertainty data set and a two-layer optimization scheduling model, the optimal scheduling strategy is obtained, thereby achieving the effect of improving the system's new energy absorption level and the robustness of the system's comprehensive operation under uncertain scenarios.

[0005] In a first aspect, the present invention provides an energy optimization scheduling method taking into account the uncertainty of wind and solar power, the method comprising:

[0006] Acquire historical output data of wind and photovoltaic devices in the integrated energy system, and construct a historical data set based on the historical output data, wherein the historical output data includes day-ahead forecast data and actual output data;

[0007] According to the day-ahead forecast data and the actual output data, a daily average forecast relative deviation value is calculated, and according to the daily average forecast relative deviation value, the historical data set is graded into uncertainty scenarios, and an initial probability distribution corresponding to each scenario grade is calculated, and a scenario representative is randomly selected from each scenario grade, and a discrete uncertainty scenario set is constructed according to the scenario representative and the initial probability distribution;

[0008] Based on the discrete uncertainty scenario set, taking the optimal efficiency of scheduling resource allocation in the integrated energy system as the objective function and the equipment operation constraints in the integrated energy system as the constraint conditions, a two-layer distributed optimization scheduling model is constructed, wherein the two-layer distributed optimization scheduling model includes an upper-layer day-ahead scheduling optimization model and a lower-layer daytime scheduling optimization model;

[0009] The double-layer distributed optimization scheduling model is solved by using a row-column generation algorithm to obtain an optimal scheduling strategy for the integrated energy system, wherein the optimal scheduling strategy includes a day-ahead scheduling optimization strategy and a daytime scheduling optimization strategy;

[0010] According to the optimal scheduling strategy, energy scheduling is performed on the integrated energy system.

[0011] In a second aspect, the present invention provides an energy optimization scheduling system taking into account the uncertainty of wind and solar power, the system comprising:

[0012] A data acquisition module is used to acquire historical output data of wind and photovoltaic devices in the integrated energy system, and to construct a historical data set based on the historical output data, wherein the historical output data includes day-ahead forecast data and actual output data;

[0013] A scenario set construction module, used to calculate the average daily forecast relative deviation value according to the day-ahead forecast data and the actual output data, classify the historical data set into uncertainty scenarios according to the average daily forecast relative deviation value, calculate the initial probability distribution corresponding to each scenario level, randomly select scenario representatives from each scenario level, and construct a discrete uncertainty scenario set according to the scenario representatives and the initial probability distribution;

[0014] A model building module is used to build a two-layer distributed optimization scheduling model based on the discrete uncertainty scenario set, taking the optimal scheduling resource allocation efficiency in the integrated energy system as the objective function and taking the equipment operation constraints in the integrated energy system as the constraint conditions, wherein the two-layer distributed optimization scheduling model includes an upper-layer day-ahead scheduling optimization model and a lower-layer daytime scheduling optimization model;

[0015] A strategy generation module, used to solve the two-layer distributed optimization scheduling model by using a row-column generation algorithm to obtain an optimal scheduling strategy for the integrated energy system, wherein the optimal scheduling strategy includes a day-ahead scheduling optimization strategy and a daytime scheduling optimization strategy;

[0016] The strategy execution module is used to perform energy scheduling for the integrated energy system according to the optimal scheduling strategy.

[0017] The present invention provides an energy optimization scheduling method and system taking into account the uncertainty of wind and photovoltaic power. Through the method, the optimal scheduling for the day and the optimal scheduling strategy for the day under the discrete representative scenario that adapts to the floating probability distribution of multiple uncertain scenarios can be obtained, thereby effectively improving the system's new energy consumption level and the robustness of the system's comprehensive operation under uncertain scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 It is a flow chart of an energy optimization scheduling method taking into account the uncertainty of wind and photovoltaic power generation in an embodiment of the present invention;

[0019] Figure 2 is a representative data curve of a wind and photovoltaic scene selected for example analysis in an embodiment of the present invention;

[0020] Figure 3 The output of each unit in the optimal scheduling plan during the day in the example analysis of the embodiment of the present invention;

[0021] Figure 4 The output of the renewable energy generator set in the example analysis in the embodiment of the present invention in different time periods;

[0022] Figure 5 is the Pareto optimal frontier under different robust fluctuation values ​​in the example analysis in the embodiment of the present invention;

[0023] Figure 6 It is a structural schematic diagram of an energy optimization scheduling system taking into account the uncertainty of wind and photovoltaic power in an embodiment of the present invention. DETAILED DESCRIPTION

[0024] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0025] See also Figure 1The first embodiment of the present invention proposes an energy optimization scheduling method taking into account the uncertainty of wind and solar power, which includes steps S10 to S50:

[0026] Acquire historical output data of wind and photovoltaic devices in the integrated energy system, and construct a historical data set based on the historical output data, wherein the historical output data includes day-ahead forecast data and actual output data;

[0027] According to the day-ahead forecast data and the actual output data, a daily average forecast relative deviation value is calculated, and according to the daily average forecast relative deviation value, the historical data set is graded into uncertainty scenarios, and an initial probability distribution corresponding to each scenario grade is calculated, and a scenario representative is randomly selected from each scenario grade, and a discrete uncertainty scenario set is constructed according to the scenario representative and the initial probability distribution;

[0028] Based on the discrete uncertainty scenario set, taking the optimal efficiency of scheduling resource allocation in the integrated energy system as the objective function and the equipment operation constraints in the integrated energy system as the constraint conditions, a two-layer distributed optimization scheduling model is constructed, wherein the two-layer distributed optimization scheduling model includes an upper-layer day-ahead scheduling optimization model and a lower-layer daytime scheduling optimization model;

[0029] The double-layer distributed optimization scheduling model is solved by using a row-column generation algorithm to obtain an optimal scheduling strategy for the integrated energy system, wherein the optimal scheduling strategy includes a day-ahead scheduling optimization strategy and a daytime scheduling optimization strategy;

[0030] According to the optimal scheduling strategy, energy scheduling is performed on the integrated energy system.

[0031] In the present invention, the integrated energy system is an electrically coupled integrated energy system, and the equipment in the system mainly includes power equipment, natural gas equipment and electrical energy coupling equipment, among which the main power equipment includes: conventional coal-fired power generation units, photovoltaic power generation systems, wind power generation systems, power loads and power energy storage equipment (batteries); the main natural gas equipment includes: natural gas well intake units, gas loads and natural gas energy storage equipment (gas tanks); the main electrical energy coupling equipment includes: power-to-gas equipment (hydrogen production by water electrolysis) and gas-to-power equipment (gas turbines).

[0032] In order to facilitate the analysis of the energy dispatch process of the integrated energy system, firstly, based on the steady-state energy flow architecture of the electrically coupled integrated energy system, the energy coupling and conversion model of the system is constructed. The equipment in the energy coupling and conversion model consists of the above-mentioned power equipment, natural gas equipment and electrical energy coupling equipment, and is connected to the power network and natural gas pipeline network of the integrated energy system. Among them, the power network can be connected to the external power grid, purchase electricity from the power sales company, or adopt an isolated operation mode; the natural gas network purchases natural gas from the natural gas company through the intake unit for consumption by the natural gas supply system. The input end of the water electrolysis hydrogen production equipment is connected to the power supply node of the power network, and the output end is connected to the corresponding intake node of the natural gas network; the input end of the gas turbine is connected to the gas supply node of the natural gas network, and the output end is connected to the corresponding power generation node of the power network.

[0033] Then, based on the energy coupling and conversion model, the energy production, conversion, consumption and storage mechanism of typical equipment in the integrated energy system are considered to build the equipment operation model, where the typical equipment here includes photovoltaic generators, wind turbines, gas turbines, water electrolysis hydrogen production equipment, batteries and gas tanks. The following describes the operation model of each device:

[0034] 1. Photovoltaic generator set

[0035] The operation mathematical model of the photovoltaic power generation unit includes the day-ahead predicted power generation curve, the actual operation maximum power curve and the actual power generation curve. Each curve uses the time scale of the hourly average power generation data. The actual operation maximum power curve may be greater than or less than the day-ahead predicted power generation curve, while the actual power generation curve shall not be greater than the actual operation maximum power curve. Its operation model is as follows:

[0036]

[0037] In the formula, It indicates the photovoltaic power generation power predicted at time t. represents the actual maximum power generation of the photovoltaic unit at time t, represents the actual power generation of the photovoltaic unit at time t, It represents the deviation between the day-ahead prediction and the actual maximum operating power of the photovoltaic unit at time t. represents the amount of abandoned light in actual operation at time t, where That is, the predicted absolute deviation of photovoltaic output, which is used to evaluate the severity of uncertainty scenarios in the future. It is then optimized in the subsequent objective function to improve the consumption of clean energy in the scheduling strategy.

[0038] 2. Wind turbines

[0039] The operation mathematical model of the wind turbine generator set includes the day-ahead predicted power curve, the actual maximum power curve and the actual power curve. Each curve uses the time scale of the hourly average power generation data. The actual maximum power curve may be greater than or less than the day-ahead predicted power curve, while the actual power curve must not be greater than the actual maximum power curve. Its operation model is as follows:

[0040]

[0041] In the formula, represents the predicted wind power generation at time t, represents the actual maximum power generation of the wind turbine at time t, represents the actual power generation of the wind turbine at time t, It represents the deviation between the day-ahead prediction and the actual maximum operating power of the wind turbine at time t, that is, the predicted absolute deviation of photovoltaic output, which is subsequently used to evaluate the severity of uncertain scenarios. It represents the amount of wind abandoned under actual operation at time t, which is optimized in the subsequent objective function to improve the consumption of clean energy in the scheduling strategy.

[0042] 3. Gas turbine

[0043] Gas turbine is the key equipment supporting multi-energy synergy and complementarity in the electrical coupling integrated energy system. The mathematical model of its power generation is as follows:

[0044]

[0045] In the formula, represents the actual power generation of the gas turbine at time t, represents the natural gas consumption flow of the gas turbine at time t, represents the combustion calorific value of natural gas at time t, represents the operating efficiency of the gas turbine at time t, and They represent the minimum operating power and maximum operating power of the gas turbine at time t, and They represent the lower and upper limits of the ramp rate of the gas turbine at time t respectively. Subsequently, these two sets of inequalities restrict the scheduling of the gas turbine in the constraints of the optimization model.

[0046] 4. Hydrogen production equipment by water electrolysis

[0047] The water electrolysis hydrogen production equipment is also a key equipment supporting multi-energy synergy and complementarity in the electrical coupling integrated energy system. The mathematical model of its natural gas production is as follows:

[0048]

[0049] In the formula, represents the actual power consumption of the water electrolysis hydrogen production equipment at time t, It represents the natural gas flow rate of the water electrolysis hydrogen production equipment at time t, represents the combustion calorific value of natural gas at time t, represents the operating efficiency of the water electrolysis hydrogen production equipment at time t, and They represent the minimum operating power and maximum operating power of the water electrolysis hydrogen production equipment at time t respectively. Subsequently, this inequality restricts the scheduling of the water electrolysis hydrogen production equipment in the constraints of the optimization model.

[0050] 5. Battery

[0051] Scheduling the charging and discharging of batteries in different time periods can effectively improve the clean energy consumption capacity of the integrated energy system and increase the flexibility of system operation. The operation model of power storage and release is as follows:

[0052]

[0053] In the formula, represents the amount of electricity currently stored in the battery at time t, and They represent the charging efficiency and discharging efficiency of the battery at time t, respectively. Indicates the time difference, and They represent the charging power and discharging power of the battery at time t, respectively. and They represent the lower and upper limits of the battery storage capacity at time t, respectively. and They represent the lower and upper limits of the battery’s charging power at time t, respectively. and They represent the lower and upper limits of the battery's discharge power at time t respectively. Subsequently, these three sets of inequalities restrict the battery's scheduling in the constraints of the optimization model.

[0054] 6. Gas tank

[0055] The scheduling of natural gas in and out of gas storage tanks at different time periods, combined with the use of energy coupling equipment, can also effectively improve the clean energy consumption capacity of the integrated energy system and improve the flexibility of system operation. The mathematical model of natural gas storage and release is as follows:

[0056]

[0057] In the formula, represents the amount of natural gas currently stored in the gas tank at time t, and They represent the current natural gas inlet and outlet flow rates of the gas storage tank at time t, respectively. and They represent the lower and upper limits of the storage capacity of the gas tank at time t, respectively. and They represent the lower and upper limits of the air intake flow rate of the air storage tank at time t, respectively. and They represent the lower limit and upper limit of the gas outlet flow rate of the gas tank at time t respectively. Subsequently, these three groups of inequalities restrict the scheduling of the gas tank in the constraints of the optimization model.

[0058] The above-mentioned operation model is mainly used for the subsequent construction of the optimization scheduling model, which will be described in detail later. While establishing the operation model, in this embodiment, the uncertainty of wind and photovoltaic power output is also graded based on the severity of the forecast according to the real historical data set of the day-ahead forecast data and actual output data of the wind and photovoltaic equipment. The specific steps include:

[0059] Dividing the historical data set into a photovoltaic data set and a wind power data set, and calculating the prediction deviation amount at a preset time scale according to the day-ahead prediction data and the actual output data in each data set;

[0060] Normalizing and averaging the forecast deviation to obtain a daily average forecast relative deviation value;

[0061] According to the average daily forecast relative deviation value, the uncertainty scenarios in the historical data set are graded, and according to the amount of scenarios in each scenario grade and the total sample size of the historical data set, the initial probability distribution of each scenario grade is calculated.

[0062] In this embodiment, the daily average forecast relative deviation of wind and solar power output is first calculated. Specifically, for each set of historical day-ahead forecast data and actual output data of wind and solar power output, the forecast deviation on an hourly time scale is calculated. At the same time, due to the different baselines of each set of historical data, the absolute value of the forecast deviation cannot accurately reflect the severity of the uncertainty of the scene. Therefore, it is necessary to normalize the forecast deviation based on the actual power generation data, calculate the forecast relative deviation and solve its daily average value, so as to quantitatively describe the severity of uncertainty of each historical data scene. The calculation formula is as follows:

[0063]

[0064]

[0065] In the formula, and They represent the daily average forecast relative deviation of wind power output and photovoltaic output at time t, and They represent the predicted absolute deviation of wind power output and photovoltaic output at time t respectively. The predicted absolute deviation value can be calculated according to the corresponding formula in the equipment operation model. and They represent the actual maximum output of wind power and photovoltaic power respectively.

[0066] Then, based on the above-mentioned daily average forecast relative deviation, the severity of the uncertainty scenario is graded. Specifically, based on the calculated daily average forecast relative deviation of historical wind and photovoltaic data, the data set is divided into multiple uncertainty levels, and the initial probability distribution of each level is calculated. Since wind power and photovoltaic power generation are affected by different environmental factors, their uncertainties are inconsistent, so independent grade division and probability distribution calculation are required, but the calculation formula can use the same formula. The calculation formula is as follows:

[0067]

[0068] In the formula, represents the initial probability distribution of the mth scene level, It represents the average daily forecast relative deviation corresponding to the mth scenario level. and They represent the lower and upper bounds of the daily average forecast relative deviation corresponding to the mth scenario level, K represents the total number of samples in the wind power / photovoltaic data set, and They represent the minimum and maximum values ​​of the daily average forecast relative deviation respectively, M represents the total number of wind power / photovoltaic uncertainty scenarios, and for all M uncertainty levels, the sum of their initial probability distributions should be equal to 1.

[0069] Then, according to the above scenario classification and probability distribution, a discrete uncertainty scenario set is constructed. The specific steps are as follows:

[0070] Randomly select scene representatives from each scene level corresponding to each data set, and construct an uncertainty scene set according to the scene representatives and the initial probability distribution, wherein the uncertainty scene set includes a wind power uncertainty scene set and a photovoltaic uncertainty scene set;

[0071] The wind power uncertainty scenario set and the photovoltaic uncertainty scenario set are fused to obtain a discrete uncertainty scenario set that takes into account the uncertainties of wind and photovoltaic power.

[0072] In this embodiment, after the scene division is completed, a set of data is randomly sampled from each scene level as a scene representative to form a discrete uncertainty scene set. , the number of samples in this scene set is the sum of the total number of wind power uncertainty scene levels and the total number of photovoltaic uncertainty scene levels. Due to the relative independence of wind and photovoltaic uncertainties, sample data sampling should also be performed independently in each wind and photovoltaic scene level. Here, assuming that the wind power scene is divided into M uncertainty levels and the photovoltaic power generation scene is divided into N uncertainty levels, the discrete uncertainty scene set is The size should be MN, that is:

[0073]

[0074] In the formula, Indicates a scenario set where wind power is at the mth scenario level and photovoltaic power is at the nth scenario level. Indicates that wind power is the scene set of the mth scene level, Indicates the scene set where optoelectronics is the nth scene level, represents the initial probability distribution when wind power is at the mth scenario level and photovoltaic power is at the nth scenario level, represents the initial probability distribution when wind power is at the mth scenario level, Represents the initial probability distribution when the optoelectronics is at the nth scene level. For all MN comprehensive uncertainty levels, the sum of their initial probability distributions should be equal to 1.

[0075] The purpose of performing scenario classification in this embodiment is to introduce the characteristics of scenario probability distribution on the basis of reasonably using the risk and volatility resistance of the robust planning model. On the one hand, a scenario representative can be selected for each uncertainty scenario level to form a discrete scenario data group, which solves the problem that the worst scenarios are difficult to traverse due to too many samples. On the other hand, it prevents the optimization scheduling results from being overly conservative due to excessive consideration of low-probability worst scenarios.

[0076] Based on the above equipment operation model and discrete uncertainty scenario set, the optimal scheduling resource allocation efficiency in the integrated energy system is taken as the objective function, and the power balance constraints, energy production equipment constraints, energy coupling equipment constraints, energy storage equipment constraints, etc. in each scenario are taken as constraints. A two-layer distributed optimization scheduling model is constructed, which specifically includes:

[0077] Taking the optimal utilization of day-ahead non-flexible resources as the upper-level objective function, the day-ahead scheduling strategy of non-flexible resources as the upper-level decision variable, and the power balance constraint, energy production equipment constraint, energy coupling equipment constraint and energy storage equipment constraint as the upper-level constraint conditions, an upper-level day-ahead scheduling optimization module is constructed.

[0078] The lower-level daytime scheduling optimization model is constructed by taking the maximization of the expected efficiency of the optimal daytime scheduling energy loss under the worst-case probability distribution under the discrete uncertainty scenario set as the lower-level objective function, the real-time scheduling strategy of flexible resources and the floating probability distribution function of each level of scenarios in the discrete uncertainty scenario set as the lower-level determining variables, and the daytime scheduling constraints of energy equipment and the scenario probability distribution constraints as the lower-level constraint conditions.

[0079] In this embodiment, the two-layer distributed optimization scheduling model includes an upper-layer day-ahead scheduling optimization model and a lower-layer daytime scheduling optimization model. The upper-layer model and the lower-layer model are described in detail below:

[0080] (1) Upper-level day-ahead scheduling optimization model

[0081] The first layer of the two-layer distributed optimization model considers the optimal scheduling of day-ahead resources based on the day-ahead forecast data of wind and solar power, mainly including the hourly power generation plan, start-stop plan and spare capacity reserve plan of conventional coal-fired power generation units, the hourly gas purchase plan of natural gas well input, and the hourly operation plan of energy coupling equipment. After the first layer of optimal scheduling is completed, these variables will participate in the daytime real-time optimal scheduling of the second layer as immutable parameters. Specifically, the objective function of the upper-layer day-ahead scheduling optimization model is constructed with the optimal day-ahead non-flexible resource utilization. In this embodiment, the resource utilization is characterized by resource consumption. Therefore, the objective function of the upper-layer optimization scheduling model is expressed by minimizing the total consumption of day-ahead scheduling resources. The resource consumption here includes the power generation loss of conventional coal-fired power generation units, the start-stop loss of units and the spare capacity reserve, the natural gas purchase volume, and the operating loss and start-stop loss of gas turbines. The formula description is as follows:

[0082]

[0083] In the formula, Indicates the power generation loss of coal-fired power generation units, Indicates the start-stop loss of coal-fired power generation units, Indicates the spare capacity reserve of coal-fired generating units, Indicates the natural gas purchase volume, represents the operating loss of the gas turbine, Indicates the start-stop loss of the gas turbine.

[0084] The calculation formulas for the above resource consumption can be expressed as:

[0085]

[0086] In the formula, , and are the 0th, 1st and 2nd order parameters of the power generation loss of the i-th conventional coal-fired unit, is the power generation capacity of the i-th conventional coal-fired unit at time t, and are the startup loss and shutdown loss of the i-th conventional coal-fired unit, represents the spare capacity reserve of the i-th conventional coal-fired unit, is the spare capacity reserve configured for the i-th conventional coal-fired unit at time t, is the gas purchase volume parameter of natural gas, represents the natural gas flow rate of the i-th water electrolysis hydrogen production equipment at time t, is the operating loss parameter of the i-th gas turbine, is the actual power generation of the i-th gas turbine at time t, and are the startup loss and shutdown loss of the i-th gas turbine, A Boolean variable representing the start / stop status of the i-th conventional coal-fired unit, and are the Boolean variables for the startup and shutdown of the i-th conventional coal-fired unit, A Boolean variable representing the start / stop status of the i-th gas turbine, and are the operational Boolean variables for starting and stopping the i-th gas turbine, respectively; G, Gas, and GT represent the coal-fired unit set, the water electrolysis hydrogen production equipment set, and the gas turbine set, respectively.

[0087] For the constraints of the day-ahead dispatch optimization model, this embodiment mainly considers the power balance constraints, energy production equipment constraints, energy coupling equipment constraints, energy storage equipment constraints and other constraints under the day-ahead predicted wind and solar power output scenarios and all uncertainty scenarios, so as to ensure that the day-ahead dispatch can meet the basic energy supply demand under the harsh scenarios taking into account the uncertainty of wind and solar power. The specific constraints are described in detail below:

[0088] 1. Overall power balance constraints of integrated energy systems

[0089] Assuming that the operation model of the power network is a DC power flow model without power loss, In the scenario of , at any time t, the total power generation of the integrated energy system should be equal to the total power consumption, that is:

[0090]

[0091] In the formula, , and They represent the power generation of the g-th conventional coal-fired unit at time t, the actual power generation of the standby capacity, and the start / stop state Boolean variable, respectively. and They represent the actual power generation of wind turbines and photovoltaic generators at time t, and They represent the current discharge power and charging power of the battery at time t, respectively. and They represent the actual power generation and start / stop status Boolean variables of the gas turbine at time t, represents the actual power consumption of the water electrolysis hydrogen production equipment at time t, It represents the energy consumption of the power load l at time t. It should be noted that, since the present invention is aimed at the optimal scheduling taking into account the uncertainty of wind and solar power, the load uncertainty scenario set is not considered on the power consumption side, and the power consumption in the above formula is not graded by scenario.

[0092] 2. Power balance constraints of integrated energy system network

[0093] Assuming that the operation model of the power network is a DC power flow model without power loss, In the scenario of and its connected busbars , the corresponding total input power should be equal to the total output power, that is:

[0094]

[0095] In the formula, In uncertain scenarios In the example, the bus connected to node b at time t The power flowing through the node. This value can be positive or negative. Its sign indicates the direction of power flow relative to node b. A positive value indicates that power flows to node b, and a negative value indicates that power flows out of node b. Indicates busbar The power capacity is determined by the length, cross-sectional area, material and winding method of the wire, which is a fixed parameter of the model. The variables in the form of represent the relevant power parameters of the corresponding energy equipment connected to node b, and their parameter meanings can refer to the parameter meanings of the above constraints.

[0096] 3. Battery equipment restrictions

[0097] Based on the battery operation model, In the scenario, at any time t, for any battery , its operation scheduling should meet the following two constraints: 1) the real-time charge / discharge power does not exceed the maximum charge / discharge power limit of the equipment; 2) the storage power does not exceed the upper and lower limits of the equipment storage power, that is:

[0098]

[0099] 4. Operation constraints of conventional coal-fired power generation units

[0100] Since coal-fired generating units have large operating power and large regulation inertia, in the day-ahead dispatch strategy, at any time t, for any coal-fired generating unit , its operation scheduling should meet the following three constraints: 1) the real-time power generation does not exceed the upper and lower limits of the unit's power generation; 2) the increase / decrease in power generation between adjacent times does not exceed the upper and lower ramp power limits of the equipment; 3) the power generation matches the start-stop operation plan of the equipment in the day-ahead scheduling, that is:

[0101]

[0102] In the formula, is the power generation capacity of the g-th coal-fired unit at time t, and are the lower and upper limits of the power generation capacity of coal-fired unit g, and are the lower and upper limits of the ramp power of coal-fired unit g, is the Boolean variable of the start / stop state of the coal-fired unit g at time t. By adding this variable, the scheduling of the unit's power generation is matched with the start / stop operation plan, that is, when the unit is scheduled to stop operating ( ), the power generation synchronization limit is 0.

[0103] 5. Start-stop control constraints of conventional coal-fired power generation units

[0104] At any time t in the day-ahead dispatch strategy, for any coal-fired unit G, the start-stop state variable of the unit is determined by its start-stop state at the previous moment and the start-stop operation instruction at this moment, that is:

[0105]

[0106] In the formula, and They are Boolean variables for the start and stop of the coal-fired unit g at time t. A value of 0 indicates that the start / stop operation is not performed, and a value of 1 indicates that the start / stop operation is performed. The sum of their values ​​should be less than 1, indicating that the start / stop instructions cannot be executed simultaneously for the same coal-fired unit at any time.

[0107] 6. Reserve capacity constraints of conventional coal-fired units

[0108] Since conventional coal-fired generating units are subject to upper and lower limits of power generation, they cannot be arbitrarily dispatched for power generation adjustment in actual operation. Therefore, it is necessary to comprehensively consider the deviation of uncertain scenarios in the day-ahead dispatch strategy and reasonably set the adjustable spare capacity reserve for each period. , so that in each uncertainty scenario, the daytime operation scheduling They can all make real-time adjustments to the unit output according to the spare capacity reserve plan set up a few days ago.

[0109] Therefore, at any time t in the day-ahead dispatch strategy, for any coal-fired unit G, the operation and dispatch of its spare capacity should meet the following constraints: 1) The spare capacity reserve plan for the day-ahead dispatch should comply with the unit power generation plan and the unit power generation limit; 2) The spare capacity reserve plan for the day-ahead dispatch should comply with the unit start-stop state limit; 3) The spare capacity investment for daytime dispatch should follow the spare capacity reserve plan for the day-ahead dispatch; 4) The spare capacity investment for daytime dispatch should meet the unit up and down ramp power limit, that is:

[0110]

[0111] In the formula, In uncertain scenarios In the equation, the actual power generation of the coal-fired unit g put into standby capacity at time t can be positive or negative, which means that the coal-fired unit adjusts power generation up and down based on the day-ahead plan, but still follows the up and down ramp limits of the unit;

[0112] in addition, A Boolean variable representing the start / stop status of the g-th conventional coal-fired unit at time t. It is used to adapt the start / stop plan of the unit scheduled on the day before. That is, the units that are planned to be shut down or have been shut down cannot temporarily start the spare capacity to generate electricity, so as to avoid high waste of equipment resources, human resources, management resources and other uncontrollable factors.

[0113] 7. Wind / solar curtailment constraints

[0114] In the actual operation of the integrated energy system, in some uncertain scenarios, due to the limitations of network energy supply and demand balance, some wind / solar power generation needs to be abandoned to ensure the normal consumption of energy. In the scenario, at any time t, for any wind turbine and any photovoltaic generator PV, its abandoned wind / solar Should not exceed the actual maximum operating power of the unit in the current state , that is:

[0115]

[0116] 8. Overall flow balance constraints of natural gas system

[0117] Considering that the pipeline resistance in the natural gas network is small, the pressure loss of natural gas during transmission is ignored, and the gas flows steadily. Therefore, in the steady-state operation model of the natural gas pipeline network, In the scenario of, at any time t, the total gas input flow of the natural gas pipeline network should be equal to the total gas output flow, that is:

[0118]

[0119] In the formula, represents the gas flow rate of natural gas well g at time t, It represents the natural gas consumption of power load l at time t. Since the gas purchasing plan of the natural gas company belongs to the medium- and long-term trading mode, it does not have the function of real-time daytime trading. Therefore, the natural gas input flow does not set the spare capacity adjustment variable. The uncertainty factors in the natural gas system are absorbed by the real-time inlet and outlet gas scheduling of the gas storage tank during the day. In addition, based on the consideration of the power system, the model does not perform uncertainty scenario classification for the natural gas load.

[0120] 9. Natural gas system network flow balance constraints

[0121] Assuming that the natural gas pipeline network is a steady-state model with no pressure loss, In the scenario of And its connected pipelines , the corresponding total natural gas input flow should be equal to the total output flow, that is:

[0122]

[0123] In the formula, In uncertain scenarios In the pipeline connected to node b at time t The natural gas flow rate flowing through the node b can be positive or negative. The positive and negative signs indicate the flow direction of the natural gas relative to the node b. A positive value indicates that the natural gas flows into the node b, and a negative value indicates that the natural gas flows out of the node b. The variables in the form of represent the natural gas flow-related parameters of the corresponding energy equipment connected to node b.

[0124] 10. Natural gas system network node pressure constraints

[0125] Unlike the power system bus, the natural gas pipeline usually does not restrict the natural gas flow, but the nodes of the natural gas pipeline network have pressure restrictions, and in the steady-state flow model, the node pressure is related to the natural gas flow. In the scenario of , its natural gas flow should meet the requirements of its two end nodes The node pressure limit is:

[0126]

[0127] In the formula, Indicates natural gas pipeline The pipeline parameters are determined by the length, cross-sectional area, material and other factors of the pipeline and are fixed parameters of the model; Representing uncertainty scenarios Under this condition, the pressure of the natural gas pipeline network node b at time t is: and Respectively represent the upper and lower limits of the pressure at node b.

[0128] 11. Gas tank equipment constraints

[0129] In each In the scenario, at any time t, for any gas tank , its operation scheduling should meet the following two constraints: 1) the real-time inlet / outlet gas flow rate does not exceed the maximum inlet / outlet gas flow rate limit of the equipment; 2) the storage of natural gas does not exceed the upper and lower limits of the equipment's storage gas volume, that is:

[0130]

[0131] 12. Natural gas well inlet flow constraints

[0132] Unlike power generation units, the inertia of the gas well inlet flow is small, so the gas well inlet flow does not have up and down ramp restrictions, nor is it subject to the day-ahead start and stop schedule. Therefore, at any time t in the day-ahead scheduling strategy, for any gas well , ensuring that the air intake flow does not exceed the upper and lower limits of the gas well input flow, that is:

[0133]

[0134] In the formula, and They respectively represent the upper and lower limits of the gas intake flow rate of the natural gas well g.

[0135] 13. Constraints on water electrolysis hydrogen production (P2G) equipment

[0136] At any time t in the day-ahead scheduling strategy, for any water electrolysis hydrogen production equipment , its operating power should meet the upper and lower limits of the equipment operating power, that is:

[0137]

[0138] 14. Gas turbine equipment constraints

[0139] Unlike water electrolysis hydrogen production equipment, gas turbines have a large regulation inertia similar to conventional coal-fired units. Therefore, at any time t in the day-ahead scheduling strategy, for any gas turbine equipment , its operation scheduling should meet the following constraints: 1) the real-time operating power does not exceed the upper and lower limits of the unit's operating power; 2) the increase / decrease of the operating power between adjacent times does not exceed the up and down ramp power limit of the equipment; 3) the operating power matches the start and stop operation plan of the equipment in the day-ahead scheduling. That is:

[0140]

[0141] In the formula, and are the upper and lower limits of the operating power of the gas turbine gt, and are the up and down ramp power limits of the gas turbine gt, is the Boolean variable of the start / stop state of the gas turbine gt at time t. By adding this variable, the scheduling of the unit's operating power matches the start / stop operation plan, that is, when the unit is scheduled to stop operating ( ), the operating power synchronization limit is 0.

[0142] 15. Gas turbine equipment start-stop control constraints

[0143] At any time t in the day-ahead scheduling strategy, for any gas turbine equipment , the start-stop state variables of the unit are determined by its start-stop state at the previous moment and the start-stop operation instructions at this moment, that is:

[0144]

[0145] middle, and They are respectively the start / stop operation Boolean variables of the gas turbine gt at time t. The value of 0 indicates that the start / stop operation is not executed, and 1 indicates that the start / stop operation is executed. The sum of their values ​​should be less than 1, indicating that the start / stop instructions cannot be executed simultaneously for the same gas turbine at any time.

[0146] (2) Lower-level intraday scheduling optimization model

[0147] The second layer of the two-layer distributed optimization model is more complicated. First, based on the results of the first layer day-ahead optimization scheduling, the set of known uncertain scenarios is selected. In the optimization scheduling of the uncertain flexibility resources in the upper model, the resource allocation efficiency of the real-time operation of the integrated energy system during the day in each scenario is met, so as to obtain the optimal daytime scheduling result of each scenario under the current day-ahead scheduling strategy; on top of this, the probability distribution of each scenario is optimized and adjusted to meet the needs of the model optimization scheduling robustness. Its significance lies in: since historical data cannot fully cover all possible situations that may occur in reality, there is a deviation between the sampling frequency of the scenario and the actual probability of the scenario. The optimization adjustment of the probability distribution is essentially to find the "maximum expected efficiency" based on the optimal daytime scheduling efficiency of each scenario by simulating reasonable probability deviations, that is, the "worst case" of system scenario uncertainty, thereby realizing the significance of robust optimization. Therefore, the optimization object of the second-level daytime scheduling optimization model not only includes the real-time scheduling strategies of energy equipment in the integrated energy system under various uncertain scenarios, such as the hourly standby capacity scheduling of conventional coal-fired power generation units, the hourly wind / light abandonment scheduling of wind / solar power generation units, and the hourly operation scheduling of energy storage equipment (P2G / GT), but also includes the probability distribution variables of various uncertain scenarios. These variables are integrated with the results of the first-level optimization model to form a complete day-ahead + daytime optimal scheduling strategy for the integrated energy system that takes into account the uncertainty of wind and solar power and considers the worst case.

[0148] In fact, according to the above description, the lower-level daytime scheduling optimization model itself is a simple two-level optimization model. The lower level determines the optimal daytime scheduling solution for uncertain scenarios based on the day-ahead scheduling strategy, and the upper level finds the probability distribution with the highest comprehensive expected efficiency based on the optimal efficiency. The model can be simplified into the following "Max-Min" mathematical model:

[0149] The model can be simplified into the following "Max-Min" mathematical model:

[0150]

[0151] In the formula, A and B are coefficients, Representing uncertainty scenarios The probability distribution variable, Representing uncertainty scenarios The probability distribution of represents the day-ahead scheduling variable determined by the first-layer optimization model, Representation scene The daytime real-time dispatch variables of the flexibility resources of the integrated energy system; represents the loss parameter of the flexibility resource of the integrated energy system, Representation scene The actual maximum power generation of wind and photovoltaic power under the scenario is used as a constant in the lower optimization model; the constraints in the lower model include: 1) energy and equipment constraints related to the first-level day-ahead scheduling strategy; 2) energy and equipment constraints not related to the first-level day-ahead scheduling strategy; In addition, the scenario probability distribution variables cannot be arbitrarily optimized without the initial probability distribution. In order to ensure that they are more in line with the actual operation data, it is necessary to use the 1-norm ( ) and the ∞-norm ( ), and limit the probability distribution value of the scene. The specific objective function and constraint settings are as follows:

[0152] 1. Lower layer objective function

[0153] As shown above, the optimization goal of the lower-level daytime scheduling is to find the optimal daytime scheduling energy loss for each uncertainty scenario based on the day-ahead scheduling strategy. The probability distribution of the worst scenario The expected maximum efficiency is as follows:

[0154]

[0155] In the formula, Optimal daytime scheduling energy loss, represents the worst-case probability distribution, Indicates an uncertain scenario. represents a set of discrete uncertainty scenarios, , and Represents uncertainty scenarios The amount of energy storage equipment loss, wind and solar power abandonment, and relative reserve capacity loss of coal-fired units caused by daytime scheduling.

[0156] It should be noted that the relative reserve capacity loss of coal-fired units It is usually a negative number, which means that the reserve capacity reserved in the day-ahead dispatch plan may not be fully put into power generation in actual operation, and the loss of the reserved reserve capacity has been calculated in the objective function of the first-level optimization model. Therefore, it is necessary to deduct the loss of unused reserve capacity in the second-level optimization model. The negative value can be understood as the scene The loss reduced by not fully utilizing the reserve capacity of coal-fired units for real-time dispatch; and Respectively represent the operating loss parameters of the battery and gas tank, and They represent the penalty parameters for curtailing wind / solar power respectively.

[0157] 2. Constraints

[0158] The constraints of the lower-level daytime optimization scheduling model include the daytime scheduling constraints of energy equipment and the probability distribution constraints of scenarios. Specifically, for the daytime scheduling of flexible energy equipment in the integrated energy system, the variable settings and constraint settings are exactly the same as the mathematical models of constraints 1 to 15 in the upper-level daytime optimization scheduling model, that is, all the variables with superscripts The variable indicates that it is the decision variable of the second-level daytime optimization model, and all the optimization variables of the first level become immutable constants. For example, for the overall power balance constraint of the power system in the upper level, its mathematical description in the lower level model is as follows:

[0159]

[0160] In the formula, , , , , as well as are known constants, while the other variables are model optimization variables.

[0161] The scene probability distribution constraint is divided into the scene probability distribution value constraint and the scene probability distribution value scheduling constraint in the lower model. Among them, for the scene probability distribution value constraint, the optimization value of each scene probability distribution should be restricted to a certain extent around its initial probability distribution value. This invention adopts 1-norm and -norm fusion constraints, therefore, the optimization of the scene probability distribution should meet the following three constraints: 1) the scene probability distribution should meet the basic rules of the probability distribution function; 2) the scene probability distribution should meet the 1-norm constraint; 3) the scene probability distribution should meet -norm constraint. Its mathematical description is as follows:

[0162]

[0163] In the formula, and Represents uncertainty scenarios probability distribution and initial probability distribution, and are respectively in 1-norm and -The probability allowable deviation limit under the norm constraint condition, under the above constraints, The following confidence levels are met:

[0164]

[0165] Where MN is the number of discrete uncertainty scenarios, i.e., the set The size of K is the total number of samples in the historical data set. The right-hand side expressions in the above formula are defined as 1-norm and -Uncertainty probability confidence under the norm and , and are all positive numbers less than or equal to 1, which can be defined by the user. Generally speaking, the larger the probability confidence, the smaller the probability allowable deviation limit, the stronger the adjustment constraint on the probability distribution value, and the smaller the expected maximum loss in the worst scenario (constraints are limited and cannot approach the worst point). After defining the probability confidence, the probability allowable deviation limit can be calculated by the following formula:

[0166]

[0167] For the scheduling constraints of the scenario probability distribution value, in order to simplify the solution process of the optimization model, it is necessary to establish the above probability distribution optimization process as a mixed integer optimization model, and its mathematical description is as follows:

[0168]

[0169] In the formula, and The uncertainty scenarios The upper and lower adjustment values ​​of the probability distribution value are all non-negative numbers, and are based on -norm constraint, its upper and lower adjustment range should be less than -Probability allowable deviation limit of norm constraint ; and For the scene The Boolean variable for adjusting the probability distribution value up or down. 1 represents upward / downward adjustment, and 0 represents no upward / downward adjustment. When no adjustment is made, the corresponding adjustment amount is constrained to 0, and the sum of the two Boolean variables should be less than or equal to 1, that is, the scenario The probability distribution value is adjusted upward or downward, or not adjusted; the significance of the Boolean variable setting here is to convert the 1-norm and -The absolute value calculation in the norm constraint condition is transformed into a simpler variable sum calculation, so the sum of the adjustment amount in all scenarios is less than or equal to That is, the 1-norm constraint is satisfied.

[0170] For the above two-layer distributed optimization module, in this embodiment, a row-column generation algorithm is used for solving, and the specific solving steps include:

[0171] Initializing performance parameters, the performance parameters including an upper performance limit, a lower performance limit and a performance feedback parameter;

[0172] Calculate the upper layer day-ahead scheduling optimization model, obtain the optimal day-ahead scheduling strategy, and update the efficiency lower bound;

[0173] According to the optimal day-ahead scheduling strategy, the optimal daytime scheduling energy loss and the worst-case scenario probability distribution under various uncertainty scenarios are calculated;

[0174] Calculate the maximum expected efficiency according to the optimal daytime scheduling energy loss and the worst-case scenario probability distribution, and update the efficiency upper bound;

[0175] The worst-case scenario probability distribution and the maximum expected efficiency are fed back to the upper-level day-ahead scheduling optimization model and iteratively calculated until a preset termination condition is met to obtain an optimal scheduling strategy. The preset termination condition includes that the difference between the efficiency upper bound and the efficiency lower bound is less than an efficiency threshold.

[0176] In this embodiment, the row and column generation algorithm C&CG refers to the idea of ​​adopting a two-layer iterative optimization, in which the results obtained by the lower-layer optimization model are fed back upward by "adding corresponding variables and constraints in the first-layer optimization model", so that the first-layer optimization model can consider the information fed back by the second-layer optimization in the subsequent iteration process. Therefore, the number of variables and constraints of the first-layer optimization model will increase with the number of iterations, so it is called the "row and column generation" algorithm.

[0177] In order to better explain the working principle of the C&CG algorithm, in this embodiment, the two-layer distributed optimization model is simplified into the following abstract mathematical model.

[0178] Among them, the simplified mathematical description of the upper-level day-ahead optimization scheduling model is as follows:

[0179]

[0180] In the formula, A, B, H and j represent coefficients, represents the day-ahead scheduling variable determined by the first-layer optimization model, and They represent the second-layer dispatching variables and wind and solar power output under various uncertain scenarios, represents the loss parameter of the flexibility resource of the integrated energy system, Represents the loss parameter of non-flexible resources in the integrated energy system; the objective function in the formula is the mathematical model of the objective function established in the upper-level optimization model, and the first two sets of constraint expressions are the energy system operation constraints described in the upper-level optimization model, including constraints related to multi-scenario daytime scheduling variables and constraints unrelated to daytime scheduling.

[0181] In addition, L in the formula represents the efficiency constant transmitted by the second-layer optimization model to the upper-layer model, that is, the "maximum expected efficiency of daytime optimization under the current day-ahead scheduling strategy, considering the probability distribution of scenario fluctuations" described in the lower-layer optimization model. Since the second-layer optimization model is looking for the "worst case", this constant is actually the "upper bound" of the daytime scheduling efficiency. Therefore, the efficiency constant receiving the second-layer feedback is And the corresponding worst probability distribution (is an immutable constant in the first-level optimization), the first-level optimization model searches for a new day-ahead scheduling strategy to ensure that the daytime scheduling plan based on the new day-ahead scheduling strategy does not exceed the original upper bound (that is, the third set of constraint expressions in the corresponding formula) and achieves overall efficiency optimization.

[0182] However, it should be noted that the "worst case" returned by the lower-level model is not the "worst case" in the global sense, but the "worst case" for the current day-ahead scheduling strategy. Therefore, this "worst case" cannot be compared with the "worst case" found in the previous iteration, but forms a "worst case set". ”, where each of the “worst case” elements Represents a probability distribution that may correspond to a discrete uncertainty scenario set , and the set The size of is the same as the number of iterations of the two-level optimization model, that is, each iteration adds a new "worst case". Therefore, in the first-level optimization model, it is necessary to consider not only the "worst case set" currently returned "The daytime scheduling efficiency should be less than the constant L. It should also be considered that in each of the other "worst case sets" found in the past, it should be less than the current maximum expected efficiency upper limit, that is, the daytime scheduling efficiency constraint. Otherwise, in other "worst cases", the daytime scheduling loss may be too high. Once this "worst case" occurs, the system operation loss will increase greatly, and the meaning of the optimal robust scheduling of comprehensive efficiency will be lost. Therefore, the number of variables and constraints in the upper model increases with the number of iterations. For example, after the nth iteration, the simplified mathematical description of the model is as follows:

[0183]

[0184] In the formula, all superscript forms are Variables / constants of the form represent the nth worst case scenario The scheduling variables and probability distribution constants under .

[0185] It should be noted that, since the second-level optimization model has not yet run calculations in the first iteration, there is no returned efficiency constant and worst probability distribution. Therefore, L is set to 0 in the first iteration and the third set of constraint expressions is canceled.

[0186] The simplified mathematical description of the lower-level daytime optimization scheduling model is as follows:

[0187]

[0188] The settings of all variables, objective functions and constraints in the formula are the same as those explained in the above lower-level optimization scheduling model. Its goal is to find the maximum expected performance of the "worst case" under the current day-ahead scheduling strategy.

[0189] Based on the above simplified model, the iterative solution process of the two-layer distributed optimization model is as follows:

[0190] Step 1: Set the upper bound UB of the comprehensive performance to , the lower bound LB is 0, and the feedback parameter L is 0;

[0191] Step 2: Calculate the upper-level optimization scheduling model to obtain the optimal day-ahead scheduling strategy , and update the lower bound of the comprehensive performance ;

[0192] Step 3: Based on the optimal day-ahead scheduling strategy , calculate the optimal scheduling energy consumption under each uncertainty scenario , optimize the scenario probability distribution value and obtain the "worst case" probability distribution And the maximum expected energy loss in this case , update the upper bound of the comprehensive loss ;

[0193] Step 4: Optimize the worst probability distribution of the lower layer model and the maximum expected loss Feedback to the upper model to add the corresponding “worst case” ", and add corresponding optimization variables and constraints, and update to the maximum expected loss ;

[0194] Step 5: Determine the termination condition. If it is met, return to the optimal scheduling result; otherwise, return to step 2 and re-optimize the two-layer model.

[0195] The termination condition of iterative optimization in the solution process is: the result of the upper model of the two-layer distributed optimization model determines the lower bound of the comprehensive efficiency, while the result of the lower model determines the upper bound of the comprehensive efficiency. This is because the upper model focuses on the day-ahead scheduling strategy of the day-ahead non-flexible resource allocation efficiency, while the lower model focuses on possible uncertainty scenarios and their worst probability distribution. Generally speaking, in the iterative process of the two-layer model, the upper model has more variables and constraints than the previous step, and the feasible domain is narrowed, so the optimization loss will gradually increase, so the lower bound of the model will gradually increase; while in the lower model, because the day-ahead scheduling strategy of the first layer is gradually more comprehensive, it is more difficult to find a worse probability distribution in the uncertainty scenario, and the maximum expected loss will gradually decrease, so the upper bound of the model will gradually decrease. Therefore, the termination condition of the model is: when the difference between the upper and lower bounds of the updated comprehensive efficiency is less than the threshold At this time, the comprehensive operation and dispatching loss of the energy system is controlled within an acceptable small range, and the model is considered to have been optimized. Its mathematical description is as follows:

[0196]

[0197] In the formula Indicates the acceptable deviation range of the comprehensive operation and scheduling loss of the integrated energy system.

[0198] In another preferred embodiment, the present invention further comprises:

[0199] Determine whether the wind and solar power output scenario of the integrated energy system belongs to the scenario in the discrete uncertainty scenario set, and if so, perform energy dispatch on the integrated energy system according to the optimal dispatch strategy;

[0200] If not, the day-ahead energy scheduling of the integrated energy system is performed according to the day-ahead scheduling optimization strategy, and after the scheduling is completed, the lower-layer daytime scheduling optimization model is updated according to the actual output data of the wind and photovoltaic equipment;

[0201] The updated lower-layer daytime scheduling optimization model is calculated to obtain the updated daytime scheduling optimization strategy, and the integrated energy system is scheduled for daytime energy according to the updated daytime scheduling optimization strategy.

[0202] In this embodiment, the day-ahead scheduling optimization strategy and the daytime scheduling optimization strategy can be obtained through the above steps, wherein the day-ahead scheduling optimization strategy represents a scheduling strategy that can be executed in a real integrated energy system. The plan ensures the benefits of the system's day-ahead resource scheduling allocation, and ensures that when wind and solar power uncertainty scenarios similar to those that have appeared in historical data sets occur, the system has daytime real-time scheduling capabilities to ensure a reliable supply of energy; and the daytime scheduling optimization strategy obtained by the present invention is not an actually executable daytime real-time scheduling strategy, it only represents the daytime scheduling strategy that the system should execute when discrete wind and solar power uncertainty scenarios sampled in historical data sets occur.

[0203] However, since the historical data set is limited, and the wind power output is affected by complex environmental factors such as temperature and light, it belongs to a chaotic uncertain system. The historical data set is difficult to cover all possible situations that may occur in reality. Therefore, in actual operation, the actual wind power output has the following possibilities: 1) Belongs to the discrete scene set obtained by sampling 2) belongs to the scene set sampled from the historical data set ; 3) It does not belong to any scenario K in the historical data set. In most cases, the actual wind and solar power output belongs to scenario 3). Therefore, in this embodiment, when the actual wind and solar power output belongs to the first scenario, the daytime resource scheduling output strategy for the corresponding scenario obtained by the present invention can be directly used. For other situations, after executing the day-ahead scheduling strategy, the optimization process of the lower-level optimization scheduling model can be re-executed according to the actual wind and solar power output to obtain the optimal daytime scheduling strategy for the electric power integrated energy system in the real scenario.

[0204] That is to say, the day-ahead scheduling strategy obtained by the method provided by the present invention is an actually executable scheduling plan, which ensures that when the actual wind and solar power uncertainty is similar to the description of the historical data set, the system has the ability to reliably schedule energy in real time, and the system operation scheduling resource allocation is highly efficient; and the obtained daytime scheduling strategy is a reference scheduling plan corresponding to a discrete sampling scenario set, which proves the system's robust operation capability and resource allocation efficiency under the consideration of wind and solar power uncertainty, and when applied in actual scenarios, it is necessary to re-optimize the daytime scheduling based on the actual wind and solar power output data.

[0205] The following example analysis is based on the method provided by the present invention, assuming that the integrated energy system includes a solar photovoltaic unit and a wind turbine unit, and selecting the cool period in winter. Set the maximum deviation of the actual power generation of the REG generator set in each period to ±30% of the predicted power generation the day before. The scheduling period is set to 15 minutes each. In addition, the penalty coefficient for REG output uncertainty is 10. Perform scenario analysis on the predicted data and actual operating data in the historical data set of wind and solar power output. Based on the average daily relative deviation between the predicted data the day before and the actual operating data, divide the historical data set into multiple uncertainty scenario levels and randomly select a group of data as the scenario representative. Figure 2 It is the representative data curve of the selected wind and solar power scene.

[0206] The C&CG algorithm is used to iteratively update the optimization model to obtain the optimal day-ahead scheduling that adapts to the floating probability distribution of multiple uncertain scenarios and the optimal daytime scheduling plan under discrete representative scenarios. The output of each unit is shown in Figure 3 .according to Figure 3 From the calculation results in , it can be seen that in the early morning, that is, before 6:00, due to insufficient wind and low light irradiance, diesel generators and energy storage generate electricity at a higher power to support the load demand. As time goes by, solar irradiation gradually increases, the ambient wind gradually increases, the output of renewable energy gradually increases, CAES is buffered, and charging begins. In the period from 12:00 to 16:00 at noon, renewable energy maintains a high output state. Entering the evening stage, as the sun gradually sets, the photovoltaic output gradually returns to zero, and the diesel generator becomes the main output unit. At the same time, the wind power generation increases, and the two jointly bear the high load demand at night. After midnight, the energy storage system begins to increase its output to fill the gap in power demand. In general, based on the model solution calculation of the optimization algorithm proposed in this paper, each unit in the system follows the power generation law to meet the load demand of each period, and the obtained multi-objective optimization scheme takes into account both robustness and low carbon, which meets the scheduling expectations.

[0207] In actual scheduling production scenarios, it is often necessary to adjust the probability of violating the spinning reserve constraint according to on-site needs, that is, it is necessary to select different robust fluctuation values. Here, the robust fluctuation value is selected. The values ​​of are 0, 1, and 2 respectively, and photovoltaic units and wind turbine units are selected as the research objects. Under the optimization calculation of the C&CG algorithm, the output of renewable energy generators in different time periods is as follows: Figure 4 According to the different values ​​of the robust fluctuation value, the corresponding probability of violation of the spinning reserve constraint and the operating loss of the system are shown in Table 1 below.

[0208] Table 1 Probability of violating spinning reserve constraints and system operation losses

[0209]

[0210] In Table 1 and Figure 4 In the above example, when the robust fluctuation value is equal to zero, it is an ideal scheduling scenario that does not consider the output uncertainty of renewable energy generators. As the robustness requirements increase, the output power volatility of renewable energy generators approaches the preset maximum fluctuation threshold. This phenomenon is accompanied by a decrease in the risk of constraint violation, but correspondingly, the overall scheduling loss of the system will also increase.

[0211] Figure 5 is the Pareto optimal frontier under different robust fluctuation values. This trend reveals the inherent contradictory relationship between robustness and low carbon in the power system, that is, improving the system's ability to resist uncertainty is often accompanied by an increase in operating losses. In actual power dispatching operations, if faced with high dispatching loss pressure, decision makers can use moderate relaxation of the probability requirements for constraint satisfaction as a response strategy. This approach can reduce dispatching losses to a certain extent, but it will also weaken the robustness of the system accordingly. Therefore, in order to achieve a compromise between low carbon and robustness in the power system, it is necessary to flexibly set the robustness loss parameters according to the operating requirements in specific scenarios to seek the best balance between the two. This process reflects the complex decision-making considerations in the optimal dispatching of the power system and emphasizes the importance of dynamically adjusting strategies under different operating scenarios.

[0212] This embodiment provides an energy optimization scheduling method that takes into account the uncertainty of wind and solar power. By constructing a wind and solar power uncertainty data set for multiple scenarios and a two-layer optimization scheduling model, the optimal scheduling on the day-ahead that adapts to the floating probability distribution of multiple uncertainty scenarios and the optimal daytime scheduling strategy under discrete representative scenarios are obtained, which effectively improves the system's new energy absorption level. When the actual output of wind and solar power deviates greatly from the predicted value, the coordinated operation of natural gas resources, energy storage equipment and energy coupling equipment is flexibly scheduled to meet the real-time reliable energy supply of the integrated energy system, thereby improving the robustness of the system's comprehensive operation under uncertain scenarios and the resource allocation efficiency.

[0213] See also Figure 6 Based on the same inventive concept, the second embodiment of the present invention proposes an energy optimization scheduling system taking into account the uncertainty of wind and solar power, including:

[0214] The data acquisition module 10 is used to acquire the historical output data of the wind and photovoltaic equipment in the integrated energy system, and to construct a historical data set based on the historical output data, wherein the historical output data includes the day-ahead forecast data and the actual output data;

[0215] A scenario set construction module 20 is used to calculate the average daily forecast relative deviation value according to the day-ahead forecast data and the actual output data, classify the historical data set into uncertainty scenarios according to the average daily forecast relative deviation value, calculate the initial probability distribution corresponding to each scenario level, randomly select scenario representatives from each scenario level, and construct a discrete uncertainty scenario set according to the scenario representatives and the initial probability distribution;

[0216] A model building module 30 is used to build a two-layer distributed optimization scheduling model based on the discrete uncertainty scenario set, taking the optimal scheduling resource allocation efficiency in the integrated energy system as the objective function and taking the equipment operation constraints in the integrated energy system as the constraint conditions, wherein the two-layer distributed optimization scheduling model includes an upper-layer day-ahead scheduling optimization model and a lower-layer daytime scheduling optimization model;

[0217] A strategy generation module 40 is used to solve the two-layer distributed optimization scheduling model by using a row-column generation algorithm to obtain an optimal scheduling strategy for the integrated energy system, wherein the optimal scheduling strategy includes a day-ahead scheduling optimization strategy and a daytime scheduling optimization strategy;

[0218] The strategy execution module 50 is used to perform energy scheduling for the integrated energy system according to the optimal scheduling strategy.

[0219] The technical features and technical effects of the energy optimization scheduling system taking into account the uncertainty of wind and solar power proposed in the embodiment of the present invention are the same as the method proposed in the embodiment of the present invention, and will not be repeated here. Each module in the above-mentioned energy optimization scheduling system taking into account the uncertainty of wind and solar power can be implemented in whole or in part by software, hardware and a combination thereof. The above-mentioned modules can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory of the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above modules.

[0220] In summary, an energy optimization scheduling method and system taking into account the uncertainty of wind and photovoltaic power generation proposed in an embodiment of the present invention, the method obtains the historical output data of wind and photovoltaic power generation equipment in the integrated energy system, and constructs a historical data set based on the historical output data, wherein the historical output data includes day-ahead forecast data and actual output data; calculates the average daily forecast relative deviation value based on the day-ahead forecast data and the actual output data, and divides the historical data set into levels of uncertainty scenarios based on the average daily forecast relative deviation value, and calculates the initial probability distribution corresponding to each scenario level, randomly selects scenario representatives from each scenario level, and calculates the probability distribution corresponding to the scenario representatives and the initial Probability distribution, construct a discrete uncertainty scenario set; based on the discrete uncertainty scenario set, take the optimal scheduling resource allocation efficiency in the integrated energy system as the objective function, take the equipment operation constraints in the integrated energy system as the constraint conditions, construct a two-layer distributed optimization scheduling model, the two-layer distributed optimization scheduling model includes an upper day-ahead scheduling optimization model and a lower daytime scheduling optimization model; adopt a row and column generation algorithm to solve the two-layer distributed optimization scheduling model to obtain the optimal scheduling strategy of the integrated energy system, the optimal scheduling strategy includes a day-ahead scheduling optimization strategy and a daytime scheduling optimization strategy; according to the optimal scheduling strategy, perform energy scheduling for the integrated energy system. The present invention obtains the day-ahead optimal scheduling and the daytime optimal scheduling strategy under discrete representative scenarios that adapt to the floating probability distribution of multiple uncertainty scenarios by constructing a multi-scenario wind and photovoltaic uncertainty data set and a two-layer optimization scheduling model, effectively improving the system's new energy consumption level, and in the case of a large deviation between the actual wind and photovoltaic output and the predicted value, by flexibly scheduling the coordinated operation of natural gas resources, energy storage equipment and energy coupling equipment, to meet the real-time reliable energy supply of the integrated energy system, and improve the robustness of the system's comprehensive operation under uncertain scenarios and the resource allocation efficiency.

[0221] Each embodiment in this specification is described in a progressive manner, and the same or similar parts of each embodiment can be directly referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the system embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment. It should be noted that the technical features of the above-mentioned embodiments can be combined arbitrarily. In order to make the description concise, all possible combinations of the technical features in the above-mentioned embodiments are not described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0222] The above-mentioned embodiments only express several preferred implementation modes of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the invention patent. It should be pointed out that, for ordinary technicians in the technical field, several improvements and substitutions can be made without departing from the technical principles of the present invention, and these improvements and substitutions should also be regarded as the protection scope of the present application. Therefore, the protection scope of the patent of the present application shall be based on the protection scope of the claims.

Claims

1. An energy optimization scheduling method taking into account the uncertainty of wind and solar power, characterized in that: include: Acquire historical output data of wind and photovoltaic devices in the integrated energy system, and construct a historical data set based on the historical output data, wherein the historical output data includes day-ahead forecast data and actual output data; According to the day-ahead forecast data and the actual output data, a daily average forecast relative deviation value is calculated, and according to the daily average forecast relative deviation value, the historical data set is graded into uncertainty scenarios, and an initial probability distribution corresponding to each scenario grade is calculated, and a scenario representative is randomly selected from each scenario grade, and a discrete uncertainty scenario set is constructed according to the scenario representative and the initial probability distribution; Based on the discrete uncertainty scenario set, taking the optimal efficiency of scheduling resource allocation in the integrated energy system as the objective function and the equipment operation constraints in the integrated energy system as the constraint conditions, a two-layer distributed optimization scheduling model is constructed, wherein the two-layer distributed optimization scheduling model includes an upper-layer day-ahead scheduling optimization model and a lower-layer daytime scheduling optimization model; The double-layer distributed optimization scheduling model is solved by using a row-column generation algorithm to obtain an optimal scheduling strategy for the integrated energy system, wherein the optimal scheduling strategy includes a day-ahead scheduling optimization strategy and a daytime scheduling optimization strategy; Performing energy dispatching on the integrated energy system according to the optimal dispatching strategy; The step of randomly selecting scene representatives from each scene level and constructing a discrete uncertainty scene set according to the scene representatives and the initial probability distribution includes: Randomly select scene representatives from each scene level corresponding to each data set, and construct an uncertainty scene set according to the scene representatives and the initial probability distribution, wherein the uncertainty scene set includes a wind power uncertainty scene set and a photovoltaic uncertainty scene set; Performing scene fusion on the wind power uncertainty scene set and the photovoltaic uncertainty scene set to obtain a discrete uncertainty scene set that takes into account wind and photovoltaic uncertainty; The steps of constructing a two-layer distributed optimization scheduling model based on the discrete uncertainty scenario set, taking the optimal scheduling resource allocation efficiency in the integrated energy system as the objective function, and taking the equipment operation constraints in the integrated energy system as the constraint conditions include: Taking the optimal utilization of day-ahead non-flexible resources as the upper-level objective function, the day-ahead scheduling strategy of non-flexible resources as the upper-level decision variable, and the power balance constraint, energy production equipment constraint, energy coupling equipment constraint and energy storage equipment constraint as the upper-level constraint conditions, an upper-level day-ahead scheduling optimization module is constructed. Taking the expected efficiency maximization of the optimal daytime scheduling energy loss under the discrete uncertainty scenario set under the worst scenario probability distribution as the lower-level objective function, the real-time scheduling strategy of flexible resources and the floating probability distribution function of each level of scenarios in the discrete uncertainty scenario set as the lower-level decision variables, and the daytime scheduling constraints of energy equipment and the scenario probability distribution constraints as the lower-level constraints, a lower-level daytime scheduling optimization model is constructed; The upper layer objective function is expressed by the following formula: In the formula, Indicates the power generation loss of coal-fired power generation units, Indicates the start-stop loss of coal-fired power generation units, Indicates the spare capacity reserve of coal-fired generating units, Indicates the natural gas purchase volume, represents the operating loss of the gas turbine, Indicates the start-stop loss of the gas turbine; The lower layer objective function is expressed by the following formula: In the formula, Optimal daytime scheduling energy loss, represents the worst-case probability distribution, Indicates an uncertain scenario. represents a set of discrete uncertainty scenarios, , and Represents uncertainty scenarios The amount of energy storage equipment loss, wind and solar power abandonment, and relative reserve capacity loss of coal-fired units caused by daytime scheduling.

2. The energy optimization scheduling method taking into account the uncertainty of wind and photovoltaic power according to claim 1 is characterized in that: The step of calculating the average daily forecast relative deviation value according to the day-ahead forecast data and the actual output data, classifying the historical data set into levels of uncertainty scenarios according to the average daily forecast relative deviation value, and calculating the initial probability distribution corresponding to each scenario level includes: Dividing the historical data set into a photovoltaic data set and a wind power data set, and calculating the prediction deviation amount at a preset time scale according to the day-ahead prediction data and the actual output data in each data set; Normalizing and averaging the forecast deviation to obtain a daily average forecast relative deviation value; According to the average daily forecast relative deviation value, the uncertainty scenarios in the historical data set are graded, and according to the amount of scenarios in each scenario grade and the total sample size of the historical data set, the initial probability distribution of each scenario grade is calculated.

3. The energy optimization scheduling method taking into account the uncertainty of wind and photovoltaic power according to claim 1 is characterized in that: The daily average predicted relative deviation value includes the daily average predicted relative deviation value of wind power output and the daily average predicted relative deviation value of photovoltaic output; The following formula is used to express the daily average forecast relative deviation value of the wind power output: The following formula is used to express the daily average forecast relative deviation of the photovoltaic output: In the formula, and They represent the predicted absolute deviation of wind power output and photovoltaic output at time t, and Respectively represent the actual maximum output of wind power output and photovoltaic output; The initial probability distribution is expressed by the following formula: In the formula, represents the initial probability distribution of the mth scene level, It represents the average daily forecast relative deviation corresponding to the mth scenario level. and They represent the lower and upper bounds of the daily average forecast relative deviation corresponding to the mth scenario level, K represents the total number of data set samples, and They represent the minimum and maximum values ​​of the daily average forecast relative deviation, and M represents the total number of levels of uncertainty scenarios; The discrete uncertainty scenario set is expressed by the following formula: In the formula, Indicates a scenario set where wind power is at the mth scenario level and photovoltaic power is at the nth scenario level. Indicates that wind power is the scene set of the mth scene level, Indicates the scene set where optoelectronics is the nth scene level, represents the initial probability distribution when wind power is at the mth scenario level and photovoltaic power is at the nth scenario level, represents the initial probability distribution when wind power is at the mth scenario level, Represents the initial probability distribution when photoelectricity is the nth scene level.

4. The energy optimization scheduling method taking into account the uncertainty of wind and photovoltaic power according to claim 1 is characterized in that: The step of using a row-column generation algorithm to solve the two-layer distributed optimization scheduling model to obtain the optimal scheduling strategy for the integrated energy system includes: Initializing performance parameters, the performance parameters including an upper performance limit, a lower performance limit and a performance feedback parameter; Calculate the upper layer day-ahead scheduling optimization model, obtain the optimal day-ahead scheduling strategy, and update the efficiency lower bound; According to the optimal day-ahead scheduling strategy, the optimal daytime scheduling energy loss and the worst-case scenario probability distribution under various uncertainty scenarios are calculated; Calculate the maximum expected efficiency according to the optimal daytime scheduling energy loss and the worst-case scenario probability distribution, and update the efficiency upper bound; The worst-case scenario probability distribution and the maximum expected efficiency are fed back to the upper-level day-ahead scheduling optimization model and iteratively calculated until a preset termination condition is met to obtain an optimal scheduling strategy. The preset termination condition includes that the difference between the efficiency upper bound and the efficiency lower bound is less than an efficiency threshold.

5. The energy optimization scheduling method taking into account the uncertainty of wind and photovoltaic power generation according to claim 4 is characterized in that The step of feeding back the worst-case scenario probability distribution and the maximum expected performance to the upper-layer day-ahead scheduling optimization model and performing iterative calculation includes: According to the worst-case probability distribution of iterative feedback, a worst-case set is obtained. According to the worst-case set, the worst-case constraint conditions are added to the upper-level constraint conditions of the upper-level day-ahead scheduling optimization model, so that the optimal day-ahead scheduling strategy meets the daytime scheduling efficiency constraints under the worst-case set.

6. The energy optimization scheduling method taking into account the uncertainty of wind and photovoltaic power according to claim 1 is characterized in that: The step of performing energy dispatching on the integrated energy system according to the optimal dispatching strategy comprises: Determine whether the wind and solar power output scenario of the integrated energy system belongs to the scenario in the discrete uncertainty scenario set, and if so, perform energy dispatch on the integrated energy system according to the optimal dispatch strategy; If not, the day-ahead energy scheduling of the integrated energy system is performed according to the day-ahead scheduling optimization strategy, and after the scheduling is completed, the lower-layer daytime scheduling optimization model is updated according to the actual output data of the wind and photovoltaic equipment; The updated lower-layer daytime scheduling optimization model is calculated to obtain the updated daytime scheduling optimization strategy, and the integrated energy system is scheduled for daytime energy according to the updated daytime scheduling optimization strategy.

7. An energy optimization dispatching system taking into account the uncertainty of wind and solar power, characterized in that: include: A data acquisition module is used to acquire historical output data of wind and photovoltaic devices in the integrated energy system, and to construct a historical data set based on the historical output data, wherein the historical output data includes day-ahead forecast data and actual output data; A scenario set construction module, used to calculate the average daily forecast relative deviation value according to the day-ahead forecast data and the actual output data, classify the historical data set into uncertainty scenarios according to the average daily forecast relative deviation value, calculate the initial probability distribution corresponding to each scenario level, randomly select scenario representatives from each scenario level, and construct a discrete uncertainty scenario set according to the scenario representatives and the initial probability distribution; The step of randomly selecting scene representatives from each scene level and constructing a discrete uncertainty scene set according to the scene representatives and the initial probability distribution comprises: Randomly select scene representatives from each scene level corresponding to each data set, and construct an uncertainty scene set according to the scene representatives and the initial probability distribution, wherein the uncertainty scene set includes a wind power uncertainty scene set and a photovoltaic uncertainty scene set; Performing scene fusion on the wind power uncertainty scene set and the photovoltaic uncertainty scene set to obtain a discrete uncertainty scene set that takes into account wind and photovoltaic uncertainty; A model building module is used to build a two-layer distributed optimization scheduling model based on the discrete uncertainty scenario set, taking the optimal scheduling resource allocation efficiency in the integrated energy system as the objective function and taking the equipment operation constraints in the integrated energy system as the constraint conditions, wherein the two-layer distributed optimization scheduling model includes an upper-layer day-ahead scheduling optimization model and a lower-layer daytime scheduling optimization model; The steps of constructing a two-layer distributed optimization scheduling model based on the discrete uncertainty scenario set, taking the optimal scheduling resource allocation efficiency in the integrated energy system as the objective function, and taking the equipment operation constraints in the integrated energy system as the constraint conditions include: Taking the optimal utilization of day-ahead non-flexible resources as the upper-level objective function, the day-ahead scheduling strategy of non-flexible resources as the upper-level decision variable, and the power balance constraint, energy production equipment constraint, energy coupling equipment constraint and energy storage equipment constraint as the upper-level constraint conditions, an upper-level day-ahead scheduling optimization module is constructed. Taking the expected efficiency maximization of the optimal daytime scheduling energy loss under the discrete uncertainty scenario set under the worst scenario probability distribution as the lower-level objective function, the real-time scheduling strategy of flexible resources and the floating probability distribution function of each level of scenarios in the discrete uncertainty scenario set as the lower-level decision variables, and the daytime scheduling constraints of energy equipment and the scenario probability distribution constraints as the lower-level constraints, a lower-level daytime scheduling optimization model is constructed; The upper layer objective function is expressed by the following formula: In the formula, Indicates the power generation loss of coal-fired power generation units, Indicates the start-stop loss of coal-fired power generation units, Indicates the spare capacity reserve of coal-fired generating units, Indicates the natural gas purchase volume, represents the operating loss of the gas turbine, Indicates the start-stop loss of the gas turbine; The lower layer objective function is expressed by the following formula: In the formula, Optimal daytime scheduling energy loss, represents the worst-case probability distribution, Indicates an uncertain scenario. represents a set of discrete uncertainty scenarios, , and Represents uncertainty scenarios The energy storage equipment loss, wind and solar power abandonment and relative reserve capacity loss of coal-fired units caused by daytime dispatch; A strategy generation module, used to solve the two-layer distributed optimization scheduling model by using a row-column generation algorithm to obtain an optimal scheduling strategy for the integrated energy system, wherein the optimal scheduling strategy includes a day-ahead scheduling optimization strategy and a daytime scheduling optimization strategy; The strategy execution module is used to perform energy scheduling for the integrated energy system according to the optimal scheduling strategy.

Images