Wind-solar-thermal storage combined scheduling method based on distributed robust optimization

Abstract

The invention belongs to the field of power system optimization scheduling, and provides a wind-solar-thermal storage combined scheduling method based on distribution robust optimization, which comprises the steps of initial scene probability distribution acquisition, distribution robust optimization model construction, initial scene probability distribution and optimization scheduling scheme iteration and scheduling scheme output. According to the method, the distribution robust optimization model based on 1-norm and infinity-norm constraints is constructed, and the column and constraint generation algorithm is adopted to carry out iterative solution on the main problem and the sub-problem, so that the problem of determining the optimal scheduling strategy in the worst scene is solved, the absorptivity of wind, light and electricity is improved, and the robustness of the system is guaranteed.

Description

Technical Field

[0001] The present invention relates to the field of power system optimization and dispatching, and in particular to a wind, solar, thermal and energy storage joint dispatching method based on distributed blue stick optimization. Background Art

[0002] In recent years, with the transformation of the global energy structure, the installed capacity of renewable energy, especially wind power and photovoltaic power generation, has grown rapidly. However, its intermittent and volatile nature makes the process of high-proportion access to the grid full of challenges. In the current power system, thermal power units are still an important force for peak load regulation. Deep peak load regulation not only requires thermal power units to flexibly adjust their output during high load fluctuations, but also to maintain system stability during low load operation. However, this process is usually accompanied by high coal costs, equipment wear and carbon emissions. At the same time, the flexibility and rapid response capabilities of energy storage equipment make it show great potential in absorbing excess wind and solar power and balancing the load of the power grid. However, due to the high cost of energy storage, how to optimize the use of energy storage equipment and maximize its economic benefits is also a hot topic in current research. Traditional optimization scheduling methods mostly use stochastic optimization or robust optimization.

[0003] Although the stochastic optimization method can handle the uncertainty of renewable energy output, it is highly dependent on the prediction model; although the robust optimization method can ensure the stability of the system, it is too conservative and often leads to a decline in economic efficiency. Summary of the invention

[0004] In order to overcome the shortcomings of the prior art, the purpose of the present invention is to provide a wind, solar, thermal and energy storage joint scheduling method based on distributed robust optimization to balance the robustness and economy of the system.

[0005] To achieve the above object, the present invention provides the following solutions:

[0006] A wind, solar, thermal and energy storage joint dispatching method based on distributed blue stick optimization, comprising:

[0007] The Copula method is used to combine the wind-solar scenario generation algorithm and the K-means clustering algorithm to process historical wind energy data, photovoltaic power generation data, and load data to obtain a dynamic initial scenario probability distribution.

[0008] Combining 1-norm and ∞-norm constraints to construct a distributed robust optimization model based on the probability distribution of the initial scene;

[0009] The initial scenario probability distribution and the optimization scheduling scheme are updated by iteratively updating the main problem and sub-problems of the distributed robust optimization model using a column and constraint generation algorithm; the main problem is used to minimize the total operating cost of the system under the worst scenario of wind and solar power output fluctuations; the sub-problems are used to maximize the utilization of wind and solar power under a given scenario and control the robustness of the system under the worst scenario;

[0010] The dispatching scheme is completed by optimizing the joint dispatching of wind, solar, thermal and energy storage.

[0011] Preferably, the Copula method is used in conjunction with the wind-solar scenario generation algorithm and the K-means clustering algorithm to process the historical wind energy data, photovoltaic power generation data, and load data to obtain a dynamic initial scenario probability distribution, including:

[0012] Preprocessing the historical wind energy data, the photovoltaic power generation data and the load data; the preprocessing includes: noise removal, missing value filling and standardization;

[0013] Generate multiple wind-solar load combination scenarios using the Copula method;

[0014] Using the K-means clustering algorithm to perform cluster analysis on all the wind-solar load combination scenarios to obtain clustering results;

[0015] The probability value of each wind-solar load combination scenario is counted according to the clustering result to obtain the initial scenario probability distribution.

[0016] Preferably, the distributed robust optimization model is:

[0017]

[0018] in,

[0019]

[0020] Among them, p i is the probability value that needs to be updated for the i-th scenario; represents the 1-norm; represents the ∞-norm; θ 1 and θ ∞ Respectively represent the maximum deviation value of the probability of 1-norm and ∞-norm; α 1 and α ∞ They represent the confidence of the probability distribution values ​​of 1-norm and ∞-norm respectively; K represents the number of discrete scenarios; M represents the number of sample scenarios.

[0021] Preferably, the distributed robust optimization objective function of the distributed robust optimization model includes:

[0022] min{C stage1 +max(p T+1 minC s ' tage2 )}and

[0023]

[0024] Wherein, T≥1; LB represents the upper bound of the distributed robust optimization objective function; UB represents the lower bound of the distributed robust optimization objective function; p T represents the probability value of the Tth generation scene; C stage1 is the electricity purchase cost of the system; C s ' tage2 Represents the operating cost of the unit during peak load regulation; η is greater than p T C s ' tage2 A constant; p 0 is the initial probability value.

[0025] Preferably, the optimization objectives of the main problem include: the deep peak regulation cost of thermal power units, the equipment life loss cost, the charging and discharging cost of energy storage equipment, and the penalty cost of wind and solar power abandonment.

[0026] Preferably, the constraints of the distributed robust optimization model include:

[0027] C emi =C emi.g +C emi.b

[0028]

[0029] P g1 +P g2 +P g3 +P pv +P wd +P ES.dis +P buy =P L +P ES.dis ,

[0030]

[0031] Where i = 1, 2, 3; C emi , C emi.g , C emi.b They represent the total carbon emissions of the power system, the carbon emissions of thermal power units during the peak load regulation process, and the equivalent carbon emissions of the system purchasing electricity from the power grid; α i , β i , i They represent the carbon emission characteristic function parameters of unit i respectively; H represents the carbon emission characteristic function parameter of equivalent electricity purchase; Pgi Indicates the actual output of the thermal power unit; P pv , P wd Respectively represent the actual output of wind power and photovoltaic power; P buy Indicates the amount of electricity purchased from the external power grid; P ES.dis , P ES.cha Respectively represent the charging and discharging amount of energy storage; P L Indicates load; Indicates the real-time capacity of the electric energy storage; Respectively represent the upper and lower limits of the electric energy storage capacity; α H.cha Represents the charging state parameter of the electric energy storage; α H.dis Indicates the discharge state parameter of the electric energy storage; Respectively represent the upper and lower limits of the energy storage charging power; Respectively represent the upper and lower limits of the energy storage discharge power; Indicates the energy storage charging efficiency; Indicates the energy storage discharge efficiency.

[0032] Preferably, the calculation formula of the deep peak load regulation cost is:

[0033]

[0034] in,

[0035]

[0036] C 3 =Q oil S oil ;

[0037] Among them, C 1 represents the coal consumption cost of the i-th thermal power unit at time t; P gi.t Indicates the actual output power of the thermal power unit; C coal represents the price of coal purchased from thermal power plants; a i , b i 、c i represent the first coal consumption characteristic parameter, the second coal consumption characteristic parameter and the third coal consumption characteristic parameter of the i-th thermal power unit respectively; C 2 P represents the shaft life cost of the i-th thermal power unit at time t during deep peak regulation; gi.t Indicates the actual output power of the thermal power unit; N f (P gi,t ) represents the function of the rotor fracture cycle number with respect to the actual output power; C unit Represents the construction cost of a unit thermal power unit; C 3represents the oil cost of the i-th thermal power unit at time t during deep peak regulation; Q oil Indicates the fuel consumption of the thermal power unit during the DRO stage; S oil Indicates the unit oil price of oil investment.

[0038] Preferably, the energy storage device includes: any one or more of lithium batteries, flywheel energy storage, super capacitors, and hydrogen energy storage.

[0039] The present invention discloses the following technical effects:

[0040] The present invention provides a wind, solar, thermal and storage joint scheduling method based on distributed robust optimization. By constructing a distributed robust optimization model based on 1-norm and ∞-norm constraints and using a column and constraint generation algorithm to iteratively solve the main problem and sub-problems, the problem of determining the optimal scheduling strategy under the worst scenario is solved, and the robustness and economy of the wind, solar and thermal storage system are improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0042] Figure 1 A schematic diagram of a wind-solar-thermal-storage joint dispatching process based on distributed blue-rod optimization provided in an embodiment of the present invention;

[0043] Figure 2 A system architecture diagram provided for an embodiment of the present invention;

[0044] Figure 3 A scheduling flow chart provided for an embodiment of the present invention;

[0045] Figure 4 A schematic diagram of a deep peak modulation mode provided by an embodiment of the present invention;

[0046] Figure 5 A typical daily load and output curve provided by an embodiment of the present invention, Figure 5 (a) is the load curve, Figure 5 (b) is the output curve;

[0047] Figure 6 The simulation result diagram provided by the embodiment of the present invention is as follows: Figure 6 (a) is the output power statistics of unit 1 under different scenarios. Figure 6 (b) is the output power statistics of unit 2 in different scenarios. Figure 6(c) is the output power statistics of unit 3 under different scenarios. DETAILED DESCRIPTION

[0048] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only 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.

[0049] The purpose of the present invention is to provide a wind, solar, thermal and energy storage joint scheduling method based on distributed robust optimization to balance the robustness and economy of the system.

[0050] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0051] Figure 1 A schematic diagram of a wind-solar-thermal-storage joint dispatching process based on distributed blue-rod optimization provided in an embodiment of the present invention, Figure 2 A system architecture diagram provided for an embodiment of the present invention, Figure 3 The scheduling flow chart provided by the embodiment of the present invention is as follows: Figures 1 to 3 As shown, the present invention provides a wind, solar, thermal and energy storage joint scheduling method based on distributed blue rod optimization, including:

[0052] Step 100: Use the Copula method to combine the wind-solar scenario generation algorithm and the K-means clustering algorithm to process the historical wind energy data, photovoltaic power generation data and load data to obtain a dynamic initial scenario probability distribution;

[0053] Step 200: constructing a distributed robust optimization model based on the initial scene probability distribution by combining 1-norm and ∞-norm constraints;

[0054] Step 300: Using the column and constraint generation algorithm, the initial scenario probability distribution and the optimization scheduling scheme are iteratively updated by the main problem and sub-problems of the distributed robust optimization model; the main problem is used to minimize the total operating cost of the system under the worst scenario of wind and solar power output fluctuations; the sub-problems are used to maximize the utilization of wind and solar power under a given scenario and control the robustness of the system under the worst scenario;

[0055] Step 400: Output the optimized dispatching plan for the joint dispatching of wind, solar, thermal and energy storage.

[0056] Furthermore, the Copula method is used to combine the wind-solar scenario generation algorithm and the K-means clustering algorithm to process the historical wind energy data, photovoltaic power generation data, and load data to obtain the dynamic initial scenario probability distribution, including:

[0057] Preprocess historical wind energy data, photovoltaic power generation data and load data; preprocessing includes: noise removal, missing value filling and standardization;

[0058] Use the Copula method to generate multiple wind and solar load combination scenarios;

[0059] Use K-means clustering algorithm to perform cluster analysis on all wind and solar load combination scenarios and obtain clustering results;

[0060] According to the clustering results, the probability value of each wind-solar load combination scenario is counted to obtain the initial scenario probability distribution.

[0061] Specifically, the distributed robust optimization model is:

[0062]

[0063] in,

[0064]

[0065] Among them, p i is the probability value that needs to be updated for the i-th scenario; represents the 1-norm; represents the ∞-norm; θ 1 and θ ∞ Respectively represent the maximum deviation value of the probability of 1-norm and ∞-norm; α 1 and α ∞ They represent the confidence of the probability distribution values ​​of 1-norm and ∞-norm respectively; K represents the number of discrete scenarios; M represents the number of sample scenarios.

[0066] Furthermore, the distributed robust optimization objective function of the distributed robust optimization model includes:

[0067] min{C stage1 +max(p T+1 minC s ' tage2 )}and

[0068]

[0069] Wherein, T≥1; LB represents the upper bound of the distributed robust optimization objective function; UB represents the lower bound of the distributed robust optimization objective function; p T represents the probability value of the Tth generation scene; C stage1is the electricity purchase cost of the system; C s ' tage2 Represents the operating cost of the unit during peak load regulation; η is greater than p T C s ' tage2 A constant; p 0 is the initial probability value, which is obtained by processing historical wind energy data, photovoltaic power generation data and load data using the Copula method combined with the wind-solar scene generation algorithm and the K-means clustering algorithm.

[0070] Specifically, the optimization objectives of the main problem include: the deep peak regulation cost of thermal power units, the equipment life loss cost, the charging and discharging cost of energy storage equipment, and the penalty cost for wind and solar power abandonment.

[0071] Furthermore, the constraints of the distributed robust optimization model include:

[0072] C emi =C emi.g +C emi.b

[0073]

[0074] P g1 +P g2 +P g3 +P pv +P wd +P ES.dis +P buy =P L +P ES.dis ,

[0075]

[0076] Where i = 1, 2, 3; C emi , C emi.g , C emi.b They represent the total carbon emissions of the power system, the carbon emissions of thermal power units during the peak load regulation process, and the equivalent carbon emissions of the system purchasing electricity from the power grid; α i , β i , i They represent the carbon emission characteristic function parameters of unit i respectively; H represents the carbon emission characteristic function parameter of equivalent electricity purchase; P gi Indicates the actual output of the thermal power unit; P pv , P wd Respectively represent the actual output of wind power and photovoltaic power; P buy Indicates the amount of electricity purchased from the external power grid; P ES.dis , P ES.cha Respectively represent the charging and discharging amount of energy storage; P L Indicates load; Indicates the real-time capacity of the electric energy storage; Respectively represent the upper and lower limits of the electric energy storage capacity; α H.cha Represents the charging state parameter of the electric energy storage; α H.dis Indicates the discharge state parameter of the electric energy storage; Respectively represent the upper and lower limits of the energy storage charging power; They represent the upper and lower limits of the energy storage discharge power respectively; Indicates the energy storage charging efficiency; Indicates the energy storage discharge efficiency.

[0077] Specifically, the calculation formula for deep peak load regulation cost is:

[0078]

[0079] in,

[0080]

[0081] C 3 =Q oil S oil ;

[0082] Among them, C 1 represents the coal consumption cost of the i-th thermal power unit at time t; P gi.t Indicates the actual output power of the thermal power unit; C coal represents the price of coal purchased from thermal power plants; a i 、b i 、c i represent the first coal consumption characteristic parameter, the second coal consumption characteristic parameter and the third coal consumption characteristic parameter of the i-th thermal power unit respectively; C 2 P represents the shaft life cost of the i-th thermal power unit at time t during deep peak regulation; gi.t Indicates the actual output power of the thermal power unit; N f (P gi,t ) represents the function of the rotor fracture cycle number with respect to the actual output power; C unit Represents the construction cost of a unit thermal power unit; C 3 represents the oil cost of the i-th thermal power unit at time t during deep peak regulation; Q oil Indicates the fuel consumption of the thermal power unit during the DRO stage; S oil Indicates the unit oil price of oil investment.

[0083] Optionally, the energy storage device includes: any one or more of: lithium batteries, flywheel energy storage, super capacitors, and hydrogen energy storage.

[0084] Preferably, the original data generated by the initial scenario probability distribution is historical operating data within 180 days.

[0085] Furthermore, the Copula method is used to establish the joint distribution between wind power, photovoltaic power generation and load. The Copula model can effectively capture the dependency structure between these variables, thereby generating multiple wind-solar load combination scenarios. This process ensures that the correlation between wind power and photovoltaic power generation is accurately reflected and takes into account the dynamic changes in load demand.

[0086] Specifically, the generated multi-dimensional scene data is clustered using the K-means clustering algorithm. The algorithm divides the data into K clusters, each cluster representing a scene with similar features. In this embodiment, K=10 is selected so that 10 typical scenes are extracted from the generated scenes.

[0087] Furthermore, based on the clustering results, the frequency of occurrence of each scenario is counted to calculate the probability value of each scenario. The probability value reflects the frequency of occurrence of each scenario in the historical data and serves as the probability distribution of the initial scenario.

[0088] Specifically, the specific constraints of 1-norm and ∞-norm are as follows: taking the initial probability distribution as the center, and taking 1-norm and ∞-norm as constraints, the probability distribution value of the discrete scene is constrained, and its feasible domain is Ω respectively. 1 and Ω ∞ .

[0089] Optionally, the distributed robust optimization objective function constructed in this embodiment is as follows:

[0090] min{C stage1 +max(p T+1 minC s ' tage2 )}

[0091] The above objective function can be decomposed into the following form:

[0092]

[0093] Since the fuzzy set based on norm distance contains the discrete probability distribution of random variables, it can be decomposed into a "min" main problem and a "max-min" sub-problem based on the C&CG algorithm. According to the inner "min" problem of the sub-problem, the discrete value scenarios of the random variables can be decoupled and independently calculated, and the "max-min" sub-problem can be efficiently solved without dual transformation. Therefore, based on the above conclusions, this paper divides the model into the main problem (Main Problem MP), solving LB and two sub-problems (Subproblem1SP1) and (Subproblem2SP2). MP solves LB in the objective function, and SP1 solves max(p T+1 minC s ' tage2 ) s ' tage2 , while in SP2 {p k} Then by max(p T+1 minC s ' tage2 ) and comprehensive norm constraints to solve. Among them, LB is the main problem and UB is the sub-problem. After decomposing the objective function into main and sub-problems, the column and constraint generation algorithm can update the scenario probability distribution and optimize the scheduling plan by iteratively optimizing between the main problem and the sub-problems.

[0094] Furthermore, based on the optimization results of the main problem and sub-problems, the algorithm will dynamically adjust the probability distribution of the scenarios to optimize the worst-case scheduling solution. By increasing the probability of the scenarios with better performance and reducing the probability of the scenarios with poor performance, the algorithm will iteratively optimize until the probability distribution of the scenarios and the scheduling solution converge, and output the optimal solution for the joint scheduling of wind, solar, thermal and energy storage, maximizing the wind and solar power absorption rate and reducing the system operation cost.

[0095] Optionally, the optimization objective of the main problem also includes: the coal burning cost of the thermal power unit.

[0096] Specifically, the deep peak-shaving mode of thermal power units includes the following operating conditions: oil-free peak-shaving mode, light-load peak-shaving mode, and overload peak-shaving mode at peak load. Figure 4 .

[0097] Preferably, the charging and discharging strategy of the energy storage device is to charge during off-peak load periods to absorb excess wind and solar power; and discharge during peak load periods to balance the load and reduce the peak load pressure of thermal power units.

[0098] Specifically, the scenario probability distribution update rule of the distributed robust optimization model is:

[0099] In each iteration, the scenario probability is adjusted through the optimization result of the main problem, so that the scheduling scheme in the worst case under the current scenario is optimal. In each iteration, the scenario probability is updated through the formula, and the scenario probability is inversely proportional to the impact of the scenario on the objective function. The importance of scenes with greater impact will be emphasized by increasing their probabilities to ensure that the scheduling scheme can provide robustness in these worst cases. In this way, the system can gradually adjust the scenario probability distribution so that the scheduling scheme can adapt to the uncertain environment and ensure the stability and economy of the system.

[0100] Furthermore, the application of the column and constraint generation algorithm includes: generating new constraints in the main problem; and updating the optimization variables in the sub-problems to gradually approach the optimal solution.

[0101] Preferably, the wind, solar, thermal and energy storage joint dispatching method of this embodiment can be used for intraday dispatching of large power systems and real-time optimal dispatching of regional power grids or microgrids.

[0102] Specifically, the environmental constraints of the optimization model include: carbon emission constraints of thermal power units, load balance constraints of the system, and capacity and charge and discharge rate constraints of energy storage equipment.

[0103] Furthermore, the carbon emissions of thermal power units are approximately:

[0104] C emi =C emi.g +C emi.b

[0105]

[0106] The load balancing constraints of the system:

[0107] P g1 +P g2 +P g3 +P pv +P wd +P ES.dis +P buy =P L +P ES.dis

[0108] Capacity and charge / discharge rate constraints of energy storage devices:

[0109]

[0110] Preferably, a wind, solar, thermal and storage joint dispatching system based on distributed robust optimization includes a typical scenario generation module, a distributed robust optimization model, a main problem optimization module, a sub-problem robustness evaluation module, a thermal power deep peak regulation control module, an energy storage dynamic response module and a wireless monitoring module. The distributed robust optimization model constructs a fuzzy scenario set by combining the 1-norm and the ∞-norm; the main problem optimization module is responsible for the optimization of economic objectives, including coal burning costs, energy storage charging and discharging costs and carbon emission costs; the sub-problem robustness evaluation module is responsible for the robustness evaluation under different scenarios, and updates the scenario probability distribution and optimizes the scheduling scheme through the column and constraint generation algorithm. The typical scenario generation module is used to generate probability distributions of multiple scenarios based on historical data, and uses the k-means clustering algorithm to divide the output of multiple scenarios to form several typical scenarios as input data for the optimization model. The distributed robust optimization model seeks the global optimal solution under the worst scenario conditions through cyclic iteration of the main problem and sub-problems. The thermal power deep peak regulation control module is combined with the energy storage dynamic response module. The energy storage device absorbs excess wind and solar power during the off-peak period, releases electricity during the peak period to balance the load demand, and reduces the peak regulation pressure of the thermal power unit.

[0111] Specifically, the K-means clustering analysis process. After generating the joint scenarios, the K-means clustering algorithm is used to classify these multidimensional scenarios. The K-means algorithm is an unsupervised learning method widely used in data clustering. It can divide the data set into several clusters so that the samples in the same cluster are as similar as possible, while the samples in different clusters are as different as possible. The specific process is as follows:

[0112] S1: Select the number of clusters K

[0113] Before clustering, it is necessary to determine the number of clusters K. By analyzing the distribution characteristics of the data and domain requirements, K = 10 is selected as the initial cluster number. The K value at this time can be adjusted according to actual needs and clustering results to ensure the optimal clustering effect.

[0114] S2: Initialize cluster centers

[0115] The K-means algorithm first randomly selects K samples as the initial cluster centers. These initial cluster centers represent the "typical" scenarios of each cluster, and these centers are subsequently adjusted through iterative updates.

[0116] S3: Assign samples to the nearest cluster center

[0117] All wind energy, photovoltaic power generation and load scenario data are assigned to the cluster center closest to them according to the distance calculation method (usually using Euclidean distance). Each sample will be assigned to a cluster so that the scenarios within the cluster are similar in terms of wind energy, photovoltaic power generation and load.

[0118] S4: Update cluster center

[0119] After all samples are redistributed, the mean of all samples in each cluster is calculated as the new cluster center, which reflects the aggregation trend of samples in the current cluster.

[0120] S5: Repeat S3 and S4 until the change of the cluster center is less than the set threshold or the maximum number of iterations is reached. At this point, the clustering process is completed and the final K clusters and their corresponding cluster centers are obtained.

[0121] S6: Scene classification and label assignment

[0122] Through clustering results, each wind power, photovoltaic power generation and load scenario is assigned to a specific cluster. Each cluster represents a scenario type, and the probability value of the cluster can be calculated by the frequency of samples it contains.

[0123] S7: Representative selection and probability calculation of scenarios

[0124] For each cluster, the cluster center is selected as the representative of the scenario, and the probability value of the scenario is calculated as part of the initial probability distribution. The scenario probability value reflects the frequency of occurrence of the scenario in historical data, that is, the probability that the scenario may occur in the future.

[0125] Furthermore, a method for implementing a wind-solar-thermal-storage joint scheduling method based on distributed blue stick optimization includes the following steps:

[0126] S1: Generate multi-scenario probability distribution based on historical data and form typical scenario input through k-means clustering algorithm;

[0127] S2: The scheduling framework is constructed by using the distributed robust optimization model, and the column and constraint generation algorithm is used to iteratively optimize the main problem and sub-problems;

[0128] S3: Optimize coal burning cost, energy storage charging and discharging cost, carbon emission cost and equipment wear cost in the main problem optimization module;

[0129] S4: Dynamically evaluate the robustness of different scenarios in the sub-problem robustness evaluation module, and continuously approach the optimal solution by updating the scenario probability distribution;

[0130] S5: Combine the deep peak regulation of thermal power with the dynamic response strategy of energy storage equipment to improve the absorption capacity of wind and solar power and reduce the phenomenon of wind and solar power abandonment;

[0131] S6: Output optimized dispatching plan, including wind and solar power output allocation, thermal power operation mode and charging and discharging strategy of energy storage equipment;

[0132] S7: Download operating parameters through the wireless monitoring module, including system scheduling results, equipment performance data and carbon emission data, to support remote control and operation optimization.

[0133] Furthermore, in the specific implementation process, the typical daily load curve (reference Figure 5 (a)) and the output curve of each device (reference Figure 5 (b)) has the following key functions:

[0134] 1) System design and parameter setting:

[0135] The typical daily load curve is used to design the capacity and equipment configuration of the energy system to ensure that the energy supply capacity meets the peak load demand. The output curve of each device is used to optimize the operating parameters of the energy equipment and improve the system operation efficiency.

[0136] 2) Scheduling and dynamic optimization:

[0137] Based on the time characteristics of the load curve, wind power, photovoltaic and other volatile energy sources are coordinated with energy storage equipment. During high-load periods, energy storage devices or backup power sources are mobilized to ensure supply and demand balance; during low-load periods, surplus clean energy is used to produce hydrogen or store energy.

[0138] 3) Economic and environmental benefit assessment:

[0139] By analyzing the matching of load and equipment output, the system's clean energy consumption rate, carbon emission reduction effect and economic operation benefits on a typical day are evaluated to provide data support for system optimization.

[0140] 4) Actual operation verification:

[0141] The load curve and output curve are applied to the operation verification of the demonstration project, and the scheduling strategy and optimization model are adjusted through data comparison to provide data basis for future promotion.

[0142] Specifically, the wind, solar, thermal and energy storage joint dispatch in a regional power grid:

[0143] 1) System configuration:

[0144] Thermal power generation unit: It includes three units, namely 200MW, 300MW and 600MW. The 600MW unit supports deep peak regulation.

[0145] Energy storage equipment: lithium battery with a capacity of 800MWh and a charge and discharge efficiency of 0.95.

[0146] Wind power and photovoltaic power: average daily output is 200MW and 150MW respectively.

[0147] 2) Scheduling process:

[0148] Data input: Collect actual wind and solar power output data and forecast values ​​for the past 180 days; predict the power demand for the next 24 hours.

[0149] Model solution: The initial scenario probability distribution is generated through historical data and k-means clustering. The column and constraint generation algorithm is used to solve the distributed robust optimization model and obtain the optimal scheduling strategy.

[0150] Dispatching results: The wind and solar power absorption rate increased to more than 95%; the thermal power units entered a deep peak-shaving state during peak load, while reducing output at low load to optimize coal consumption.

[0151] 3) Performance evaluation:

[0152] Compared with traditional methods, dispatch costs are reduced by 13% and carbon emissions are reduced by 8.9%.

[0153] Optionally, simulation test and result analysis:

[0154] 1) Simulation environment: Test platform: MATLAB 2021b, combined with Gurobi optimizer; Scenario setting: Configure 10 different output and load scenarios, each covering 24 hours.

[0155] 2) Simulation steps:

[0156] Initial scheduling: Generate a preliminary scheduling plan based on historical scenarios.

[0157] Iterative optimization: Through iterative optimization of the main problem and sub-problems, the scenario probability is updated to optimize the economy and robustness.

[0158] Output: Record the power balance, cost and carbon emissions under different scenarios. Figure 6 (a) to Figure 6 (c).

[0159] 3) Simulation results:

[0160] Energy storage equipment charging and discharging diagram: shows the charging of energy storage during peak hours and the discharging during off-peak hours. Thermal power unit output curve: shows the output changes of different units in different time periods.

[0161] The beneficial effects of the present invention are as follows:

[0162] The present invention solves the problem of determining the optimal scheduling strategy under the worst scenario by constructing a distributed robust optimization model based on 1-norm and ∞-norm constraints and adopting a column and constraint generation algorithm to iteratively solve the main problem and sub-problems, thereby improving the absorption rate of wind and solar power and ensuring the robustness of the system.

[0163] The various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the various embodiments can be referenced to each other.

[0164] The principles and implementation methods of the present invention are described in this article using specific examples. The description of the above embodiments is only used to help understand the method and core idea of ​​the present invention. At the same time, for those skilled in the art, according to the idea of ​​the present invention, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting the present invention.

Claims

1. A wind, solar, thermal and energy storage joint dispatching method based on distributed blue bar optimization, characterized in that: include: The Copula method is used to combine the wind-solar scenario generation algorithm and the K-means clustering algorithm to process historical wind energy data, photovoltaic power generation data, and load data to obtain a dynamic initial scenario probability distribution. Combining 1-norm and ∞-norm constraints to construct a distributed robust optimization model based on the probability distribution of the initial scene; The initial scenario probability distribution and the optimization scheduling scheme are updated by iteratively updating the main problem and sub-problems of the distributed robust optimization model using a column and constraint generation algorithm; the main problem is used to minimize the total operating cost of the system under the worst scenario of wind and solar power output fluctuations; the sub-problems are used to maximize the utilization of wind and solar power under a given scenario and control the robustness of the system under the worst scenario; The dispatching scheme is completed by optimizing the joint dispatching of wind, solar, thermal and energy storage.

2. The wind-solar-thermal-storage joint scheduling method based on distributed blue-rod optimization according to claim 1 is characterized in that: The Copula method is used to combine the wind-solar scenario generation algorithm and the K-means clustering algorithm to process historical wind energy data, photovoltaic power generation data, and load data to obtain a dynamic initial scenario probability distribution, including: Preprocessing the historical wind energy data, the photovoltaic power generation data and the load data; the preprocessing includes: noise removal, missing value filling and standardization; Generate multiple wind-solar load combination scenarios using the Copula method; Using the K-means clustering algorithm to perform cluster analysis on all the wind-solar load combination scenarios to obtain clustering results; The probability value of each wind-solar load combination scenario is counted according to the clustering result to obtain the initial scenario probability distribution.

3. The wind-solar-thermal-storage joint scheduling method based on distributed blue-rod optimization according to claim 1 is characterized in that: The distribution robust optimization model is: in, Among them, p i is the probability value that needs to be updated for the i-th scenario; represents the 1-norm; represents the ∞-norm; θ1 and θ ∞ Respectively represent the maximum deviation values ​​of the probability of 1-norm and ∞-norm; α1 and α ∞ They represent the confidence of the probability distribution values ​​of 1-norm and ∞-norm respectively; K represents the number of discrete scenarios; M represents the number of sample scenarios.

4. The wind-solar-thermal-storage joint scheduling method based on distributed blue-rod optimization according to claim 1 is characterized in that: The distributed robust optimization objective function of the distributed robust optimization model includes: min{C stage1 +max(p T+1 minC s ' tage2 )}and Wherein, T≥1; LB represents the upper bound of the distributed robust optimization objective function; UB represents the lower bound of the distributed robust optimization objective function; p T represents the probability value of the Tth generation scene; C stage1 is the electricity purchase cost of the system; C s ' tage2 Represents the operating cost of the unit during peak load regulation; η is greater than p T C s ' tage2 A constant; p 0 is the initial probability value.

5. The wind-solar-thermal-storage joint scheduling method based on distributed robust optimization according to claim 1 is characterized in that: The optimization objectives of the main problem include: the deep peak regulation cost of thermal power units, the equipment life loss cost, the charging and discharging cost of energy storage equipment, and the penalty cost of wind and solar power abandonment.

6. The wind-solar-thermal-storage joint scheduling method based on distributed blue-rod optimization according to claim 1 is characterized in that: The constraints of the distributed robust optimization model include: C emi =C emi.g +C emi.b P g1 +P g2 +P g3 +P pv +P wd +P ES.dis +P buy =P L +P ES.dis 、 Where i = 1, 2, 3; C emi , C emi.g , C emi.b They represent the total carbon emissions of the power system, the carbon emissions of thermal power units during the peak load regulation process, and the equivalent carbon emissions of the system purchasing electricity from the power grid; α i , β i , i They represent the carbon emission characteristic function parameters of unit i respectively; H represents the carbon emission characteristic function parameter of electricity purchase equivalent; P gi Indicates the actual output of the thermal power unit; P pv , P wd Respectively represent the actual output of wind power and photovoltaic power; P buy Indicates the amount of electricity purchased from the external power grid; P ES.dis , P ES.cha Respectively represent the charging and discharging amount of energy storage; P L Indicates load; Indicates the real-time capacity of the electric energy storage; Respectively represent the upper and lower limits of the electric energy storage capacity; α H.cha Represents the charging state parameter of the electric energy storage; α H.dis Indicates the discharge state parameter of the electric energy storage; Respectively represent the upper and lower limits of the energy storage charging power; They represent the upper and lower limits of the energy storage discharge power respectively; Indicates the energy storage charging efficiency; Indicates the energy storage discharge efficiency.

7. The wind-solar-thermal-storage joint scheduling method based on distributed blue-rod optimization according to claim 5 is characterized in that: The calculation formula of the deep peak load regulation cost is: in, C3=Q oil S oil ; Where C1 represents the coal consumption cost of the i-th thermal power unit at time t; P gi.t Indicates the actual output power of the thermal power unit; C coal represents the price of coal purchased from thermal power plants; a i , b i 、c i They represent the first, second and third coal consumption characteristic parameters of the i-th thermal power unit respectively; C2 represents the shaft life cost of the i-th thermal power unit at time t during deep peak regulation; P gi.t Indicates the actual output power of the thermal power unit; N f (P gi,t ) represents the function of the rotor fracture cycle number with respect to the actual output power; C unit represents the construction cost of a unit thermal power unit; C3 represents the oil cost of the ith thermal power unit at time t during deep peak regulation; Q oil Indicates the fuel consumption of the thermal power unit during the DRO stage; S oil Indicates the unit oil price of oil investment.

8. The wind-solar-thermal-storage joint scheduling method based on distributed blue-rod optimization according to claim 5 is characterized in that: The energy storage device includes: any one or more of lithium batteries, flywheel energy storage, super capacitors, and hydrogen energy storage.

Images