A wind-solar-thermal storage distributed robust optimization scheduling method considering variable operating condition characteristics of pumped storage units

Abstract

The present application relates to the technical field of comprehensive energy system application, in particular to a wind-solar-thermal-pumped distributed robust optimization scheduling method considering variable working condition characteristics of pumped storage units. In view of the problems that system power balance is difficult under high proportion of new energy access, and traditional model is difficult to accurately solve the non-convex nonlinear characteristics of pumped storage units, the present application firstly generates typical scenarios and initial probability based on clustering; the core lies in introducing a model linearization processing strategy, processing the water level-storage capacity curve by piecewise linearization, and using McCormick envelope method to convex relaxation the head-flow-power coupling characteristics of the unit, and constructing a mixed integer linear distributed robust optimization model based on comprehensive norm fuzzy set. On this basis, by using column and constraint generation algorithm for solving, the optimal balance of system operation cost minimization and scheduling robustness is realized while coping with source and load uncertainty.

Description

Technical Field

[0001] This invention relates to the field of integrated energy system application technology, specifically to a wind-solar-thermal-storage split-brush optimal scheduling method that takes into account the variable operating characteristics of pumped-storage units. Background Technology

[0002] Wind-solar-thermal-storage complementary systems are an important means to enhance the absorption capacity of new energy sources. However, in actual dispatching, due to the uncertainty of wind and solar power output, how to construct an accurate probability distribution fuzzy set is the key to sub-Bruker optimization. Existing sub-Bruker optimization methods usually use a single norm constraint to construct the fuzzy set. A single norm is difficult to simultaneously take into account both the "central probability distribution deviation" and the "tail extreme scenario": the 1-norm focuses on overall statistical characteristics and easily ignores extreme risks; Norms focus on single-point fluctuations, which can lead to overly conservative results. This limitation makes it difficult for scheduling strategies to achieve optimal economic efficiency while meeting the system's power exchange plan. Therefore, a wind-solar-thermal-storage split-rod optimization scheduling method that considers the variable operating characteristics of pumped-storage units is needed. This method should accurately describe uncertainties through multiple constraints and improve the scheduling capability of complementary systems. Summary of the Invention

[0003] To address the shortcomings of existing technologies, the present invention aims to provide a wind-solar-thermal-storage split-rod optimization scheduling method that takes into account the variable operating characteristics of pumped-storage units.

[0004] To achieve the above objectives, the technical solution adopted by the present invention is as follows: a wind-solar-thermal-storage split-rod optimal scheduling method considering the variable operating characteristics of pumped-storage units, comprising the following steps: Step (1): Cluster typical scenarios and initial probabilities based on historical wind and light data; Step (2): Construct a comprehensive norm fuzzy set with dual constraints; Step (3): Establish a multi-norm fuzzy set-based wind-solar-thermal-storage complementary system sub-Bruker optimal scheduling model; Step (4) linearizes the nonlinear constraints in the comprehensive norm fuzzy set wind-solar-thermal-storage complementary system split-bar optimization scheduling model; Step (5): Based on the column and constraint generation algorithm, solve the BLU bar optimization scheduling model of the wind-solar-thermal-storage complementary system with linearized comprehensive norm fuzzy set.

[0005] The typical scenario and initial probability for clustering based on historical wind and solar data are as follows: The K-means clustering algorithm was used to reduce the scenes based on historical wind and light data to five typical scenes. Euclidean distance was used as the similarity index for iterative clustering. After clustering, the curves of each cluster center were taken as typical scenes, and the initial probability of each typical scene was determined according to the proportion of historical samples contained in that scene. , which serves as the benchmark parameter for constructing comprehensive norm fuzzy sets.

[0006] In step (2), the construction of the dual-constraint comprehensive norm fuzzy set is specifically as follows: in, They are respectively Norm sets Norm sets, composite norm sets; For the first The actual probability distribution of each scenario; For the first The initial probability distribution of each scenario; The number of typical scenarios selected from historical data samples; The number of historical data samples is replaced by the generated spatiotemporal correlation scenarios. These represent the confidence levels of uncertainty; They are respectively The allowable deviation of the probability distribution under norm constraints The allowable deviation of the probability distribution under norm constraints.

[0007] In step (3), the objective function of the integrated norm fuzzy set wind-solar-thermal-storage complementary system sub-bar optimization scheduling model is divided into a first-stage objective function and a second-stage objective function; 4.1) First-stage objective function; in, , These are the number of thermal power units and pumped storage units, respectively. , These are the start-up cost and shutdown cost of thermal power units, respectively. , These are the start-up cost and the shutdown cost of the pumped-storage generator unit, respectively. , These are the start-up cost and the shutdown cost of a pumped-storage pumped unit, respectively. , thermal power units exist Constant operation variables, thermal power units exist Variables that require constant shutdown operation; , Pumped storage generator units exist Variable operation of pumped storage generator unit exist Variables that require constant shutdown operation; , These are pumped storage pumping units. exist Constantly operating variables, pumped storage pumping unit exist Variables that require constant shutdown operation; 4.2) Two-stage objective function; in, , , , , These are respectively: gas cost of thermal power units, curtailment penalty cost, load deficit penalty cost, spinning reserve capacity cost of pumped storage units, and spinning reserve capacity cost of thermal power units; for Thermal power units in the scenario exist Power generation during a given time period; , These are the gas consumption coefficient and cost coefficient of thermal power, respectively. , , They are respectively In the scene wind curtailment power during certain periods In the scene Time-of-use power, In the scene Power deficit during specific time periods; , , These are respectively the wind curtailment penalty coefficient, the solar curtailment penalty coefficient, and the load deficit penalty coefficient; , They are respectively In the scene Periodic pumped storage generator unit Increase in reserve capacity In the scene Periodic pumped storage generator unit The reduction of reserve capacity; , They are respectively In the scene Periodic pumped storage pumping units Increase in reserve capacity In the scene Periodic pumped storage pumping units The reduction of reserve capacity; , They are respectively In the scene Periodic thermal power units Increase in reserve capacity In the scene Periodic thermal power units The reduction of reserve capacity; , These are the costs of increasing the reserve capacity of pumped-storage generator sets and the costs of decreasing the reserve capacity of pumped-storage generator sets, respectively. , These are the costs of increasing the standby capacity of pumped storage pumping units and the costs of decreasing the standby capacity of pumped storage pumping units, respectively. , These represent the costs of increasing reserve capacity for thermal power units and the costs of decreasing reserve capacity for thermal power units, respectively.

[0008] In step (4), the nonlinear constraints in the model are linearized, as follows: 5.1) Fuzzy set linearization; Introducing 0-1 auxiliary variables in the comprehensive norm fuzzy set with dual constraints , , , The original absolute value expression is transformed into: in, , They are respectively Norm set relatively Positive offset marker variables and negative offset marker variables; , They are respectively Norm set relatively Positive offset marker variables and negative offset marker variables; , Each is a set of comprehensive norms. relatively Excess and deficiency; 5.2) Linearization of the water level-reservoir capacity curve function; The reservoir capacity range is divided into piecewise linearization. , Discretize into M subintervals: in, , These are the upper limit and lower limit of the reservoir's capacity, respectively. For the reservoir capacity Subintervals and the first The boundary points of each sub-interval; for The corresponding reservoir water level; This is the function representing the relationship between the water level and storage capacity of the upper reservoir. By introducing 0-1 variables And satisfy the following constraints: in, As an indicator variable, when the reservoir is The storage capacity during this period is in the first When there are multiple storage capacity sub-intervals, the value is 1; otherwise, it is 0. As an auxiliary variable; The reservoir water level at the end of the time period for each reservoir capacity sub-interval is: in, , These are the reservoir capacity boundary points. Corresponding reservoir water level and reservoir capacity boundary point The corresponding reservoir water level; for Reservoir water level at the end of the time period; 5.3) Linearization of the unit's dynamic characteristic curve function; Using the McCormick envelope, the generator set dynamic characteristics are linearized as follows: in, , These are the lower limit and the upper limit of the power generation flow of pumped storage units, respectively. Pumped storage unit j exist t Power generation during a given time period; This is the output coefficient under power generation conditions; Pumped storage unit j exist t Power generation flow during a given time period; for t Water head height during the period; These are the lower and upper limits of the water head height, respectively; The dynamic characteristics of the pumping unit are linearized as follows: in, , These are the lower limit and the upper limit of the pumping flow rate of the pumped storage unit, respectively. Pumped storage unit j exist t Pumping power during a given time period; This is the output coefficient under pumping conditions.

[0009] In step (5), the column and constraint generation algorithm is divided into a main problem and sub-problems: 6.1) The main problem; in, The relaxation amount for the subproblem; This represents the number of iterations. For the first Auxiliary variables related to the subproblem introduced in each iteration; For the first The probability of the worst-case scenario for wind and solar load obtained by solving the sub-problem in the next iteration; 6.2) Subproblems; in, The solution to the main problem; 6.3) Specific process of alternating iterative solution using the algorithm: (1) Model initialization: Given an initial scene probability distribution, set a lower bound. and the Upper Realm Number of iterations ; (2) Solve the main problem to obtain the optimal solution to the main problem. At the same time, update the lower bound value. ; (3) By substituting into the subproblems and solving them, we can obtain the optimal value for each discrete scenario. The optimal solution to the subproblem is obtained. Update the upper bound value ; (4) Verify whether the convergence condition is met. If the condition is met, stop iterating; otherwise, update the worst-case probability distribution of the main problem. Add new variables to the main problem and the constraints associated with the new variables; (5) Order Then jump to S2 until the algorithm converges.

[0010] The beneficial effects of this invention are: 1. Compared with existing technologies, this invention introduces a piecewise linearization strategy to process the water level-reservoir capacity curve and uses the McCormick envelope method to perform convex relaxation on the strongly coupled terms of head-flow-power, transforming the originally complex nonlinear programming into a standard mixed-integer linear programming (MILP) model. This method adapts the model to the C&CG algorithm, significantly improves computational efficiency, and ensures the global optimality of the scheduling strategy.

[0011] 2. Compared with existing technologies, this invention accurately considers the head effect and variable operating condition characteristics of pumped-storage units within a distributed model framework. This high-precision modeling can more realistically reflect the unit's regulation capacity, effectively avoiding the risk of wind and solar curtailment or load shedding due to model deviations. It achieves the optimal balance between minimizing operating costs and system robustness when dealing with source and load uncertainties. Attached Figure Description

[0012] Figure 1 This is a flowchart of the present invention; Figure 2 This is a structural diagram of the present invention; Figure 3 This is a flowchart of the column and constraint generation algorithm of the present invention; Figure 4 This is a schematic diagram of the wind-solar-thermal-storage split-rod optimization scheduling method of the present invention, which takes into account the variable operating conditions of pumped storage units. Figure 5This is a diagram of the optimized scheduling process using a multi-bar algorithm in the embodiment. Figure 6 This is a schematic diagram of the backup capacity of the complementary system in the embodiment; (a) shows the increase in backup capacity; (b) shows the decrease in backup capacity. Detailed Implementation

[0013] The present invention will now be described in further detail with reference to the accompanying drawings and technical solutions.

[0014] This invention studies an energy base in Northwest China. Based on historical power output data from wind and solar power plants in the region, a K-means clustering algorithm is used to generate typical scenarios. The specific steps are as follows: Historical wind and solar power output time-series data from the past year in the region is selected as the sample set; the number of typical scenarios is set to K=5; Euclidean distance is used as a similarity index to cluster the samples; and finally, the curve at the center of each cluster is selected as the typical wind and solar power output scenario. The initial probability of each typical scenario is determined based on the proportion of samples after clustering. When the installed capacity of wind power is 7000MW, the installed capacity of photovoltaic power is 19000MW, the installed capacity of pumped storage hydropower is 9100MW, and the installed capacity of thermal power is 1000MW, Norm confidence =0.5、 Norm confidence =0.9, using the comprehensive norm set as the fuzzy set for optimization scheduling, the scheduling result is as follows. Figure 5 As shown.

[0015] Figure 5 The model demonstrates the 24-hour scheduling process of a wind-solar-thermal-storage complementary system. It shows that the total output of the complementary energy sources closely matches the load demand and the rigid tie-line exchange schedule. During off-peak electricity prices or when wind and solar output is strong, pumped-storage units pump water, converting electrical energy into potential energy for storage; during peak electricity prices or when wind and solar output is insufficient, they release electricity, achieving a peak-shaving and valley-filling strategy of low storage and high output. Simultaneously, the model strictly adheres to the upper and lower reservoir water level constraints, ensuring that the water level returns to its initial level at the end of the scheduling cycle (T=24), guaranteeing the sustainability of the pumped-storage power station's operation the following day. Furthermore, despite the optimized scheduling, a small load deficit still exists, indicating that considering the system's reserve capacity has a certain impact on the optimized scheduling.

[0016] To ensure the reliability and stability of power supply, and to cope with unpredictable load fluctuations, the intermittency and volatility of renewable energy output such as wind and solar power, equipment failures, maintenance, or other unforeseen circumstances, this invention maintains stable system frequency and voltage. The invention considers reserve capacity constraints during the dispatching process. Figure 6This invention demonstrates the reserve capacity provided by various energy sources during the scheduling process. It uses 10% of the load demand as the system's reserve capacity and adjusts the reserve capacity accordingly. Figure 6 In (a), the system's increased reserve capacity consists of increased reserve capacity from pumped storage power generation, decreased reserve capacity from pumped hydro power generation, and increased reserve capacity from thermal power generation. When increased reserve capacity is needed, it indicates an increase in system load demand or a failure in the supply of a certain type of energy source, requiring more electricity to meet the power shortage. Pumped storage reserve capacity is divided into generation and pumping states. Because pumped storage and pumping do not operate simultaneously, when pumped storage is in generation state, it increases reserve capacity to boost power generation, while when in pumping state, it decreases reserve capacity to reduce pumping power. Therefore, increased system reserve capacity includes decreased reserve capacity from pumped hydro power generation. Decreased reserve capacity Figure 6 (b) represents the opposite situation.

[0017] Sensitivity analysis of key parameters affecting the total operating cost of the system in optimized scheduling, including different optimization models, fuzzy sets, confidence levels, and the number of scenario samples, can help us better understand how these parameters affect the total operating cost of the system and verify the effectiveness of the data-driven sub-bar optimization scheduling model.

[0018] Table 1 Comparison of Results from Different Optimization Models

[0019] The optimization results of the three models are shown in Table 1. The cost of the sub-Brutal optimization model falls between that of stochastic optimization and robust optimization. Sub-Brutal optimization uses fuzzy sets to describe the probability distribution range of uncertain variables. It does not completely rely on a single distribution, nor is it limited to fixed boundaries. Instead, it allows the probability to fluctuate within a certain confidence interval. By balancing the characteristics of the two models, it achieves a balance between economy and safety. Specifically, while robust optimization can meet rigid external transmission requirements, excessive reserves lead to a surge in costs. Stochastic optimization has the lowest cost, but it is prone to high penalties due to insufficient reserves under extreme risks. The sub-Brutal optimization proposed in this invention accurately obtains the key distribution through comprehensive norm fuzzy sets to achieve a better overall system performance. Specifically, to ensure... Robust optimization, which can meet strict tie-line power transmission constraints under any extreme wind and solar fluctuations, retains excessive thermal power and energy storage reserves, resulting in a significant increase in costs. While stochastic optimization has the lowest cost, it is prone to insufficient reserves and faces high deficit penalties when wind and solar power changes suddenly and the transmission plan must be enforced. In contrast, distributed robust optimization, through comprehensive norm fuzzy sets, accurately obtains the key probability distributions affecting transmission security. At a cost only 0.9% higher than stochastic optimization, it achieves high reliability assurance for the predetermined tie-line exchange plan and is the optimal solution under tie-line balance constraints.

[0020] exist , When they change separately, Norm fuzzy sets The results of the split-bar optimization scheduling under three different sets: norm fuzzy set and comprehensive norm fuzzy set.

[0021] Table 2 differs Result Comparison Table

[0022] Table 3 differs Result Comparison Table

[0023] Table 2 shows... When fixed, exist Table 3 shows the changes in scheduling results under different fuzzy sets when the interval changes. When fixed, exist The changes in scheduling results under different fuzzy sets when the interval varies. The total system operating cost of the combined norm fuzzy set is higher than that of the individual fuzzy sets. Norm fuzzy set sum Norm fuzzy sets have lower costs, and comprehensive norm constraints are achieved through joint... Norm and Norm constructs composite constraints that control both overall distribution deviation and limit extreme fluctuations at single points. This avoids the shortcomings of single-norm models that neglect instantaneous power mutations due to excessive focus on overall deviation, or sacrifice economic efficiency due to excessive limitation of peak deviation. It dynamically adjusts the model's sensitivity to "average risk" and "extreme risk" to adapt to the operating characteristics of different power grids.

[0024] Table 4 Comparison of Results for Different Total Number of Scenes

[0025] Table 4 shows the confidence levels. , When the total number of scenarios remains constant, the impact of changes in the total number of scenarios on the model results under the influence of the comprehensive norm fuzzy set. Under the influence of the comprehensive norm fuzzy set, the increase in the total number of scenarios manifests as a significant economic optimization effect on the total system cost. As the total number of scenarios increases, the fuzzy set parameter describing the range of uncertainty, i.e., the deviation value... The deviation value will gradually shrink as the data richness increases. The shrinkage of the deviation value directly reduces the conservatism of the split bar model, thereby reducing costs.

[0026] The results obtained by this invention fall between stochastic optimization and robust optimization, and other costs also conform to this trend, indicating that stochastic robust optimization scheduling can achieve a balance between economy and robustness.

[0027] To verify the effectiveness of the McCormick envelope and piecewise linearization strategy proposed in this invention, a comparative experiment was conducted in this embodiment.

[0028] 1. Experimental setup: Select a typical daily load and wind / solar output data, and use the following three methods to optimize the scheduling of the same wind-solar-thermal-storage system: Method 1: Use a general energy storage model, ignore the head effect of pumped storage units, and assume that the charging and discharging efficiency is a fixed constant; Method 2: Establish an accurate non-convex nonlinear model, do not perform linearization processing, and directly use a nonlinear solver to solve it; Method 3: Use the mixed integer linear programming model based on McCormick envelope and piecewise linearization proposed in this invention.

[0029] 2. Comparison of Solution Efficiency and Results Under the same computing environment (Intel Core i7-10700 CPU @2.90GHz, 32GB RAM), the solution performance and scheduling results of the three methods are compared in Table 5.

[0030] Table 5. Performance and Results Comparison of Different Solution Methods (Constant Efficiency)

[0031] 3. Results Analysis As shown in Table 5, although Method 1 has the fastest solution speed, it neglects the impact of pumped storage unit head changes on output under different reservoir capacities, leading to a deviation in the model's prediction of peak-shaving capacity and resulting in the highest total system cost. Furthermore, the cost of power curtailment penalty increases by approximately 44.5% compared to the present invention. Method 2, due to complex non-convex nonlinear constraints, is difficult to converge within a given time and often gets stuck in local optima, with a cost of RMB 383.564 million and computation timeout. The present invention, through McCormick linearization, controls the solution time to the minute level, meeting the timeliness requirements of intraday scheduling, while having the lowest total system cost of RMB 381.421 million. This verifies that the present invention ensures high-precision modeling while also taking into account extremely high computational efficiency.

Claims

1. A wind-solar-thermal-storage split-rod optimal scheduling method considering the variable operating characteristics of pumped-storage units, characterized in that, Includes the following steps: Step (1): Cluster typical scenarios and initial probabilities based on historical wind and light data; Step (2): Construct a comprehensive norm fuzzy set with dual constraints; Step (3): Establish a multi-norm fuzzy set-based wind-solar-thermal-storage complementary system sub-Bruker optimal scheduling model; Step (4) linearizes the nonlinear constraints in the comprehensive norm fuzzy set wind-solar-thermal-storage complementary system split-bar optimization scheduling model; Step (5): Based on the column and constraint generation algorithm, solve the BLU bar optimization scheduling model of the wind-solar-thermal-storage complementary system with linearized comprehensive norm fuzzy set.

2. The wind-solar-thermal-storage split-rod optimal scheduling method for pumped-storage units considering variable operating conditions, as described in claim 1, is characterized in that... The typical scenario and initial probability for clustering based on historical wind and solar data are as follows: The K-means clustering algorithm was used to reduce the scenes based on historical wind and light data to five typical scenes. Euclidean distance was used as the similarity index for iterative clustering. After clustering, the curves of each cluster center were taken as typical scenes, and the initial probability of each typical scene was determined according to the proportion of historical samples contained in that scene. , which serves as the benchmark parameter for constructing comprehensive norm fuzzy sets.

3. The wind-solar-thermal-storage split-rod optimal scheduling method for pumped-storage units considering variable operating conditions, as described in claim 1, is characterized in that... In step (2), the construction of the dual-constraint comprehensive norm fuzzy set is specifically as follows: in, They are respectively Norm sets Norm sets, composite norm sets; For the first The actual probability distribution of each scenario; For the first The initial probability distribution of each scenario; The number of typical scenarios selected from historical data samples; The number of historical data samples is replaced by the generated spatiotemporal correlation scenarios. These represent the uncertainty confidence levels; They are respectively The allowable deviation of the probability distribution under norm constraints The allowable deviation of the probability distribution under norm constraints.

4. The wind-solar-thermal-storage split-rod optimal scheduling method for pumped-storage units considering variable operating conditions, as described in claim 1, is characterized in that... In step (3), the objective function of the integrated norm fuzzy set wind-solar-thermal-storage complementary system sub-bar optimization scheduling model is divided into a first-stage objective function and a second-stage objective function; 4.1) First-stage objective function; in, , These are the number of thermal power units and pumped storage units, respectively. , These are the start-up cost and shutdown cost of thermal power units, respectively. , These are the start-up cost and the shutdown cost of the pumped-storage generator unit, respectively. , These are the start-up cost and the shutdown cost of a pumped-storage pumped unit, respectively. , thermal power units exist Constant operation variables, thermal power units exist Variables that require constant shutdown operation; , Pumped storage generator units exist Variable operation of pumped storage generator unit exist Variables that require constant shutdown operation; , These are pumped storage pumping units. exist Constantly operating variables, pumped storage pumping unit exist Variables that require constant shutdown operation; 4.2) Two-stage objective function; in, , , , , These are respectively: gas cost of thermal power units, curtailment penalty cost, load deficit penalty cost, spinning reserve capacity cost of pumped storage units, and spinning reserve capacity cost of thermal power units; for Thermal power units in the scenario exist Power generation during a given time period; , These are the gas consumption coefficient and cost coefficient of thermal power, respectively. , , They are respectively In the scene Time-based wind curtailment power, In the scene Time-of-use power, In the scene Power deficit during specific time periods; , , These are respectively the wind curtailment penalty coefficient, the solar curtailment penalty coefficient, and the load deficit penalty coefficient; , They are respectively In the scene Periodic pumped storage generator unit Increase in reserve capacity In the scene Periodic pumped storage generator unit The reduction of reserve capacity; , They are respectively In the scene Periodic pumped storage pumping units Increase in reserve capacity In the scene Periodic pumped storage pumping units The reduction of reserve capacity; , They are respectively In the scene Periodic thermal power units Increase in reserve capacity In the scene Periodic thermal power units The reduction of reserve capacity; , These are the costs of increasing the reserve capacity of pumped-storage generator sets and the costs of decreasing the reserve capacity of pumped-storage generator sets, respectively. , These are the costs of increasing the standby capacity of pumped storage pumping units and the costs of decreasing the standby capacity of pumped storage pumping units, respectively. , These represent the costs of increasing reserve capacity for thermal power units and the costs of decreasing reserve capacity for thermal power units, respectively.

5. The wind-solar-thermal-storage split-rod optimal scheduling method for pumped-storage units considering variable operating conditions, as described in claim 1, is characterized in that... In step (4), the nonlinear constraints in the model are linearized, as follows: 5.1) Fuzzy set linearization; Introducing 0-1 auxiliary variables in the comprehensive norm fuzzy set with dual constraints , , , The original absolute value expression is transformed into: in, , They are respectively Norm sets relatively Positive offset marker variables and negative offset marker variables; , They are respectively Norm sets relatively Positive offset marker variables and negative offset marker variables; , Each is a set of comprehensive norms. relatively Excess and deficiency; 5.2) Linearization of the water level-reservoir capacity curve function; The reservoir capacity range is divided into piecewise linearization. , Discretize into M subintervals: in, , These are the upper limit and lower limit of the reservoir's capacity, respectively. For the reservoir capacity Subintervals and the first The boundary points of each sub-interval; for The corresponding reservoir water level; This is the function representing the relationship between the water level and storage capacity of the upper reservoir. By introducing 0-1 variables And satisfy the following constraints: in, As an indicator variable, when the reservoir is The storage capacity during this period is in the first When there are multiple storage capacity sub-intervals, the value is 1; otherwise, it is 0. As an auxiliary variable; The reservoir water level at the end of the time period for each reservoir capacity sub-interval is: in, , These are the reservoir capacity boundary points. Corresponding reservoir water level and reservoir capacity boundary point The corresponding reservoir water level; for Reservoir water level at the end of the time period; 5.3) Linearization of the unit's dynamic characteristic curve function; Using the McCormick envelope, the generator set dynamic characteristics are linearized as follows: in, , These are the lower limit and the upper limit of the power generation flow of pumped storage units, respectively. Pumped storage unit j exist t Power generation during a given time period; This is the output coefficient under power generation conditions; Pumped storage unit j exist t Power generation flow during a given time period; for t Water head height during the period; These are the lower and upper limits of the water head height, respectively; The dynamic characteristics of the pumping unit are linearized as follows: in, , These are the lower limit and the upper limit of the pumping flow rate of the pumped storage unit, respectively. Pumped storage unit j exist t Pumping power during a given time period; This is the output coefficient under pumping conditions.

6. The wind-solar-thermal-storage split-rod optimal scheduling method for pumped-storage units considering variable operating conditions, as described in claim 1, is characterized in that... In step (5), the column and constraint generation algorithm is divided into a main problem and sub-problems: 6.1) The main problem; in, The relaxation amount for the subproblem; This represents the number of iterations. For the first Auxiliary variables related to the subproblem introduced in each iteration; For the first The probability of the worst-case scenario for wind and solar load obtained by solving the sub-problem in the next iteration; 6.2) Subproblems; in, The solution to the main problem; 6.3) Specific process of alternating iterative solution using the algorithm: (1) Model initialization: Given an initial scene probability distribution, set a lower bound. and the Upper Realm Number of iterations ; (2) Solve the main problem to obtain the optimal solution to the main problem. At the same time, update the lower bound value. ; (3) By substituting into the subproblems and solving them, we can obtain the optimal value for each discrete scenario. The optimal solution to the subproblem is obtained. Update the upper bound value ; (4) Verify whether the convergence condition is met. If the condition is met, stop iterating; otherwise, update the worst-case probability distribution of the main problem. Add new variables to the main problem and the constraints associated with the new variables; (5) Order Then jump to S2 until the algorithm converges.

Images