Water-light-storage day-ahead optimal scheduling research considering photovoltaic uncertainty

Abstract

The invention aims at the goal of carbon peak reaching and carbon neutralization at present, and the acceleration of the development of new energy is imperative. As a representative of new energy, photovoltaic has serious randomness, volatility and intermittency problems, the invention provides a water-light complementation day-ahead optimization scheduling model containing energy storage, and the impact of the uncertainty of photovoltaic output on a power grid is reduced by utilizing hydroelectric generation and energy storage smoothing photovoltaic output with flexible adjustment capability. The method comprises the following steps: firstly, analyzing the uncertainty of photovoltaic power generation according to different photovoltaic output conditions in sunny days, cloudy days, rainy days and four seasons; then, photovoltaic scene generation and reduction are realized based on a Monte Carlo simulation method and K-means clustering, and a typical solar photovoltaic output curve is obtained by using a probability weighting method; and finally, taking the minimum economic cost of the complementary system as a target, considering photovoltaic power stations, hydropower stations and energy storage related constraint conditions, and carrying out day-ahead optimization scheduling to meet power grid scheduling requirements. Three scenes are set for comparison, a CPLEX solver is adopted to respectively obtain the optimal economic cost of the model and the optimal output of each device, and the result shows that the scene 3 can realize photovoltaic consumption while reducing the economic cost, the average light abandoning rate is only 2.15%, and the economy and effectiveness of the model provided by the invention are proved.

Description

Technical Field

[0001] The present invention relates to the technical field of source-network-load-storage Background Technique

[0002] With the development of new energy, clean energy represented by wind power, photovoltaic power, etc. has developed rapidly, greatly meeting the electricity demand of users, alleviating the pressure on the power grid, and at the same time reducing the output of traditional power generation units mainly based on thermal power units, thereby reducing environmental pollution. In order to achieve the grand goal of "carbon peak and carbon neutrality" as soon as possible, the country has accelerated the development of new energy, continuously taken effective measures to increase the penetration rate of new energy, and paid more attention to environmental protection issues while taking into account economic efficiency. China is rich in wind and light resources, and wind and light, as representatives of clean energy, have become indispensable energy carriers for building a new power system. Taking photovoltaic power as an example, by the end of 2022, the newly installed photovoltaic capacity in China was 87,410 MW, and the cumulative installed capacity was 395,240 MW. The growth rate of the newly installed capacity was 59%, and the growth rate of the total installed capacity was 28%. As a representative of the northwest region, due to the influence of the region and light intensity, the photovoltaic industry in Xinjiang has developed rapidly. By the end of 2022, the newly installed photovoltaic capacity in Xinjiang was 1,095 MW, and the cumulative installed capacity was 14,509 MW. Due to the instability of photovoltaic output caused by meteorological factors, it poses many threats to the safe operation of the power grid. Therefore, the research on the multi-energy complementary system is of great significance.

[0003] At present, the research on the multi-energy complementary system by scholars mainly focuses on the coordination of wind, light, water, and fire. Among them, many studies have been carried out on the water-light complementary system, which are mainly summarized from two aspects: operation characteristics and dispatching models. Ye Haojie, Han Liu, etc. analyzed the output characteristics of wind power and photovoltaic power and the complementary regulation ability by taking wind-fire and wind-light-water-fire as the research objects respectively; Zhu Ye and Xi Yisha established the optimal dispatching models of wind-light-storage and water-light-storage respectively. The results show that the multi-energy complementary system can reduce the phenomenon of abandoned wind and abandoned light and suppress the output fluctuation of new energy.

[0004] The research on the complementary optimal scheduling of hydropower stations and photovoltaic power stations mainly consists of three parts. The first is the analysis of the uncertainty of new energy output, including wind power and photovoltaic power. Currently, the research methods for uncertainty analysis are mainly divided into stochastic optimization, interval optimization, and robust optimization. Stochastic optimization generally uses chance constraints to handle uncertain variables. Bai Kaifeng et al. conducted uncertainty modeling for wind power, photovoltaic power, and load, generated and reduced scenarios based on the wind speed obeying the Weibull distribution, the light intensity obeying the Beta distribution, and the load probability distribution, and then transformed the uncertain variables into deterministic variables. Interval optimization represents uncertain variables in the form of intervals. Song Xiaofang et al. proposed an interval optimal scheduling method for power systems based on the uncertainty of power sources and loads. When the predicted values of wind power and load are known, their uncertainty is characterized as the fluctuation interval of the prediction error. Robust optimization obtains the optimal solution under the worst conditions and is more reliable than other methods, but less economical. Wu Mengxue et al. proposed a distributionally robust method for integrated electricity, heat, and hydrogen energy systems based on the distributionally robust conditional value-at-risk method, considering the uncertainty of wind and light, which can significantly reduce the economic losses caused by the uncertainty of wind and light. The second is the establishment of a multi-energy complementary optimal scheduling model to achieve the optimal objective function under the constraint conditions of various power generation equipment. Song Ke et al. constructed a water-light complementary optimal scheduling model with the objectives of minimizing the total output fluctuation and maximizing the overall stability. Guo Xiaoya et al. constructed multiple objective functions of maximizing power generation, highest power generation guarantee rate, minimum comprehensive risk rate, and minimum standard deviation of the remaining load, and established a medium- and long-term optimal scheduling model for water-light complementarity. Li D established a hydropower-photovoltaic optimal scheduling model with complementary scheduling cycles of year, month, and day. This model uses a simulation optimization algorithm and introduces a photovoltaic absorption measurement method to achieve the maximum power generation benefit of the system and finally determine the cascade hydropower scheduling method. Zhu Yanmei et al. considered two objectives of power generation and output fluctuation of the water-light complementary system and established an optimal scheduling model, relying on the flexible regulation ability of hydropower to reduce the impact of photovoltaic output fluctuation on the power grid. The third is the selection of model solution methods, usually using intelligent optimization algorithms for solution. The optimization algorithm can convert multiple objectives into a single objective for solution in some ways, or use the CPLEX solver for solution. Luo Bin et al. considered the uncertainty of photovoltaic output and proposed a short-term optimal scheduling model for the cascade water-light complementary system with the objective of maximizing the expected value of the overall absorbable electricity, and used the CPLEX solver for solution. Liu L et al. proposed a short-term optimization method for the integrated hydropower system, deduced the objective function and constraint conditions of the maximum power output, complementary power, and minimum change of the original power of the integrated system based on the hydropower calculation principle. Based on the optimization model composed of the objective function and constraints, the genetic algorithm was applied to the modeling of the optimization problem to obtain the optimal solution for the operation scheduling of the integrated system.

[0005] In summary, regarding the research on the multi - energy complementary system, scholars have done a lot of work in terms of operating characteristics and scheduling models. They mainly study how the multi - energy complementary system coordinates power generation to maximize the absorption of curtailed wind power and minimize the economic cost, but rarely mention the treatment of uncertain power sources such as wind power and photovoltaic power. Regarding the research on the water - light complementary system, scholars generally consider the uncertainty of photovoltaic output. Generally, they tend to establish multi - objective functions such as maximizing the peak - shaving capacity, maximizing the power generation of the complementary power generation system, and minimizing the cost, and use intelligent optimization algorithms to solve the model. However, since the algorithm is prone to falling into local optima, the optimal result cannot be determined. The present invention focuses on analyzing the impact of the uncertainty of photovoltaic output on the optimization scheduling result. First, it analyzes the photovoltaic power generation characteristics under different meteorological conditions such as sunny days, cloudy days, rainy days, and four seasons. Then, based on the Monte Carlo simulation method and K - means clustering, scenario generation and reduction are realized, and the probability - weighted method is used to obtain the typical - day photovoltaic output curve to participate in the day - ahead optimization scheduling of the water - light complementary system. Finally, by setting three scenarios for comparison, with the goal of minimizing the economic cost, combined with the relevant constraints of hydropower stations, photovoltaic power stations, and energy storage devices, a water - light complementary optimization scheduling model is established, and the minimum cost of the model and the optimal output of each device are solved through the CPLEX solver. The simulation results verify the economy and effectiveness of the scenario - three model proposed in the present invention. Summary of the Invention

[0006] To solve the problems existing in the above - mentioned prior art, the purpose of the present invention is to provide a research method for the day - ahead optimization scheduling of water - light - storage considering photovoltaic uncertainty.

[0007] To achieve the above - mentioned purpose, the technical solution of the present invention consists of the following steps:

[0008] The research on the day - ahead optimization scheduling of water - light - storage considering photovoltaic uncertainty consists of the following steps:

[0009] S1 Water - light complementary power generation system

[0010] S1.1 Structure of the water - light complementary power generation system

[0011] The coordinated scheduling of the water - light complementary power generation system mainly relies on the flexible adjustment ability of the hydropower station to suppress the volatility of photovoltaic output. Different scales of the complementary system undertake different tasks in the power grid, but the main goal is to maximize the consumption of new energy, reduce the economic cost, and reduce environmental pollution while meeting the power consumption and peak - shaving requirements of the power grid.

[0012] The designed water-light complementary system of the present invention aims to reduce the phenomenon of light curtailment, while meeting the grid dispatching instructions, and introduces pumped storage to better utilize the power generation resources. When the complementary system generates more electricity and the grid cannot receive more electrical energy, in order to reduce light curtailment, the excess photovoltaic power generation is stored through energy storage devices. When the complementary power generation system cannot meet the grid dispatching plan, energy storage can compensate for part of the deficit power through discharging. The water-light complementary system with energy storage can not only reduce the fluctuation of photovoltaic output, realize the transfer of electrical energy, but also obtain greater economic benefits. Since electrical energy storage can respond to load changes in a timely manner and has a high charge-discharge efficiency, it also participates in the energy management and day-ahead optimal dispatching of the system.

[0013] S1.2 Photovoltaic uncertainty analysis

[0014] The photovoltaic power generation of three typical days, namely sunny days, cloudy days and rainy days, is compared and analyzed. The photovoltaic output has obvious daily characteristics, that is, during the day, the output is affected by meteorological factors and is relatively large. In the morning, it is in the rising stage of output, and in the afternoon, it is in the falling stage of output. The photovoltaic output fluctuates greatly. At night, due to the absence of light, the output is zero. The photovoltaic output is mainly affected by light intensity, temperature and weather. The output is the largest on sunny days, the curve is relatively smooth, and the peak time is relatively fixed. The output on cloudy days is the second, and the output on rainy days is the smallest. However, the output on cloudy and rainy days shows obvious fluctuations and the peak time is not fixed and unique. Generally, the photovoltaic output will show typical seasonal characteristics of being higher in summer and autumn and lower in spring and winter. However, due to the unique geographical location and climatic conditions in Xinjiang, the change of light intensity throughout the year is not significant, so the change of photovoltaic power generation is not obvious, and the seasonal characteristics are also weak.

[0015] S1.3 Photovoltaic scenario generation and reduction

[0016] S1.3.1 Monte Carlo scenario generation

[0017] There may be various different scenarios for the output of a photovoltaic power station, and it is difficult for the general processing methods of photovoltaic output to take this into account, which will affect the reliability of the operation of the water-light complementary system. In order to consider the photovoltaic output under various extreme weather conditions, 2160 photovoltaic output data of a certain photovoltaic power station in 12 months of 2018 are used to generate representative photovoltaic output prediction scenarios by using the Monte Carlo simulation method.

[0018] S1.3.2 K-means clustering

[0019] Common clustering methods include K-means clustering, Gaussian mixture model clustering, density clustering, hierarchical clustering, and spectral clustering. These methods have different advantages and applicability in different data situations. Selecting an appropriate clustering method requires considering factors such as data type, data distribution, and clustering objectives. Although representative photovoltaic output prediction scenarios are obtained by the Monte Carlo simulation method, in actual engineering, various other factors and related constraints often need to be considered, and the model complexity is relatively high. The most commonly used K-means clustering can minimize the model complexity, thereby reducing a large number of typical scenarios previously obtained by the Monte Carlo simulation method. The K-means clustering algorithm classifies and partitions data units with similarity and convergence to determine the similarity between each data object. Four typical scenarios are obtained by reduction through K-means clustering, and the typical daily photovoltaic predicted output is obtained by weighted summation according to the probability generated for each scenario.

[0020] K-means clustering first divides the data set into K non-overlapping clusters, and each data point is the center of the cluster closest to it. It is an iterative and unsupervised clustering algorithm commonly used for numerical data. When each of the K clusters is relatively dense and the differences between the clusters are obvious, the default algorithm converges. This method can not only reduce the number of scenarios but also reflect the actual distribution process of photovoltaic prediction, further reducing the number of input variables in the model, thereby reducing the model complexity and improving the model solution efficiency. In the K-means clustering method, the silhouette coefficient method is usually used to determine the optimal value of K. Suppose the value of K is k 0 , first introduce the within-cluster dissimilarity d 1 (i), which represents the average distance between this sample and other samples within the cluster; d 2 (i) is called the between-cluster dissimilarity, which represents the minimum value of the average distance between this sample and all samples in other clusters. Then the silhouette coefficient ε(i) of this sample can be obtained by the following formula:

[0021]

[0022] That is:

[0023]

[0024] S1.4 Hydropower Station

[0025] The peak load period of the power grid often coincides with the peak output period of the photovoltaic power station. Applying the photovoltaic power station to power grid peak shaving has natural advantages. Generally, large reservoirs have strong regulation performance. When the photovoltaic power station generates a high amount of electricity, the hydropower station will store a part of the water resources in the reservoir. During the period when the photovoltaic output is low or there is no power generation, the hydropower station will generate electricity again to effectively reduce the peak value of the power grid load.

[0026] S2 day-ahead optimization scheduling of hydro-photovoltaic hybrid with energy storage

[0027] S2.1 Objective function

[0028] For small and medium-sized hydro-photovoltaic complementary systems, the power grid pre-sets reasonable dispatching instructions based on their power generation capacity, and requires the hydro-photovoltaic complementary systems to jointly output in each period of the dispatching period to track the plan given by the power grid. At the same time, in order to improve the energy utilization rate of the hydro-photovoltaic complementary system, on the basis of ensuring the safe and stable operation of the power grid, the objective function is to minimize the cost of the hydro-photovoltaic complementary system, while considering the operating costs of photovoltaic power stations, hydropower stations and energy storage, the penalty cost of abandoned light, the operation and maintenance costs of photovoltaic power stations and hydropower stations, and the income from the access to the grid of photovoltaic power stations and hydropower stations.

[0029] min F=f run +f aban +f om -f sell (3)

[0030] Where f run is the operating cost of photovoltaic power station, hydropower station and energy storage, f aban is the penalty cost for abandoning light, f om is the operation and maintenance cost of photovoltaic power station and hydropower station, f sell Income from photovoltaic power generation and hydropower grid connection.

[0031] (1) Operating costs

[0032]

[0033] Where, λ run_pv 、λ run_w 、λ run_es are the operating cost coefficients of photovoltaic power station, hydropower station and energy storage, is the photovoltaic power generation power in period t, MW, is the hydropower power in period t, MW, Respectively, the energy storage charging and discharging power in period t, MW.

[0034] (2) Abandonment penalty cost

[0035]

[0036] Where, λ aban is the penalty cost coefficient for abandoned light, λ om_w Forecasted photovoltaic power in period t, MW.

[0037] (3) Operation and maintenance costs

[0038]

[0039] Where, λ om_pv ​​​​​​, λ om_w are the operation and maintenance cost coefficients of the photovoltaic power station and the hydropower station respectively.

[0040] (4) Income from selling electricity back to the grid of the photovoltaic power station and the hydropower station

[0041]

[0042] In the formula, λ grid_pv is the on-grid electricity price of photovoltaic power generation, and λ grid_w is the on-grid electricity price of hydropower.

[0043] Constraint conditions

[0044] This model takes the hydropower unit, the photovoltaic power station, and the energy storage as the basic dispatching units, and takes 24 hours a day as the dispatching period, and mainly considers the following constraint conditions:

[0045] (1) Power balance constraint

[0046]

[0047] In the formula, is the grid dispatching plan, MW.

[0048] (2) Hydropower generation flow constraint

[0049]

[0050] In the formula, η w is the power generation efficiency of the hydropower station, is the power generation flow, m 3 / s, and H w is the head height of the hydropower station for power generation, m.

[0051] (3) Upper and lower limit constraints on the output of hydropower units

[0052]

[0053] In the formula, P w_max is the maximum installed capacity of the hydropower unit, MW.

[0054] (4) Output constraint of the photovoltaic power station

[0055]

[0056] (5) Energy storage related constraints

[0057] Energy storage capacity constraint

[0058]

[0059] In the formula, are the energy storage capacities at time t and time t - 1 respectively, MW, and ηes For the charging and discharging efficiency of energy storage, E s_max and E s_min are respectively the upper and lower limits of the energy storage capacity, MW.

[0060] Energy storage initial and final capacity constraints

[0061]

[0062] In the formula, are respectively the capacities of the energy storage at the 1st period and the 24th period, MW.

[0063] Energy storage charge and discharge power constraints

[0064]

[0065] In the formula, μ c and μ dis are respectively the charge and discharge states of the energy storage. The energy storage cannot charge and discharge simultaneously.

[0066] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0067] To actively respond to the national policies of "carbon peak and carbon neutrality", reduce the impact of the uncertainty of photovoltaic power output on the power grid, and reduce environmental pollution, the present invention establishes a coordinated optimization scheduling model for water-light complementary considering the uncertainty of photovoltaic power output, and conducts simulation analysis to verify the effectiveness of the proposed model. The conclusions are as follows:

[0068] 1) First, the uncertainty of photovoltaic power output is analyzed, and the Monte Carlo simulation method is used for scenario generation. Four typical power output scenarios are obtained through K-means clustering. The typical daily photovoltaic power output is obtained by weighted summing the four typical daily power outputs and their scenario probabilities, and it is used as the photovoltaic prediction data to participate in the day-ahead optimization scheduling.

[0069] 2) Taking the minimum economic cost of the water-light complementary system as the objective function, considering the relevant constraint conditions of the photovoltaic power station, hydropower station and energy storage equipment, a day-ahead optimization scheduling model for the complementary system is constructed. The model is solved by the CPLEX solver, and the optimal economic cost is 410,990 yuan. Three scenarios are set for comparative analysis, which proves that the model proposed in the present invention has better economy. The economic costs of scenario three are 84.6% and 5.7% less than those of scenario one and scenario two respectively, and the curtailment of light is effectively reduced. The average curtailment rate of light is 2.85% less than that of scenario two.

[0070] 3) The simulation results show that hydropower and energy storage have good regulation capabilities and energy transfer functions, can compensate for some power shortages, make the output of the water-light complementary system smoother, and meet the requirements of the power grid dispatching plan. Description of the Drawings

[0071] Figure 1 Schematic diagram of the water-light-storage complementary system;

[0072] Figure 2 Photovoltaic output characteristic curves under different weather conditions; Figure 3 Photovoltaic output characteristic curves in different seasons;

[0073] (a) Photovoltaic output curve in spring;

[0074] (b) Photovoltaic output curve in summer;

[0075] (c) Photovoltaic output curve in autumn;

[0076] (d) Photovoltaic output curve in winter;

[0078] Figure 4 Four typical daily output scenarios of a photovoltaic power station;

[0079] Figure 5 Photovoltaic uncertainty output diagram;

[0080] Figure 6 Diagram of the day-ahead scheduling result;

[0081] Figure 7 Diagram of the charge-discharge power of the energy storage and the change trend of the energy storage capacity;

[0082] Figure 8 Typical daily predicted output and actual output curves of the photovoltaic;

[0083] Figure 9 Diagram of the relationship between the hydropower power and the generation flow; Figure 10 Diagrams of the scheduling results and the photovoltaic accommodation curves for Scenarios 1 and 2;

[0084] (a) Scheduling result diagram for Scenario 1;

[0085] (b) Photovoltaic accommodation curve for Scenario 1;

[0086] (c) Scheduling result diagram for Scenario 2;

[0087] (d) Photovoltaic accommodation curve for Scenario 2; Specific implementation manners

[0089] The technical solution of the present invention will be further described in detail below in conjunction with the accompanying drawings and specific implementation manners:

[0090] The research method for the day-ahead optimal scheduling of the water-light-storage considering the photovoltaic uncertainty consists of the following steps:

[0091] S1 Water-light complementary power generation system

[0092] S1.1 Structure of the hydropower and PV hybrid power generation system

[0093] The coordinated dispatching of the hydropower and PV hybrid power generation system mainly relies on the flexible regulation ability of the hydropower station to suppress the volatility of PV output. Different scales of the hybrid system undertake different tasks in the power grid, but the main goal is to maximize the consumption of new energy, reduce economic costs and environmental pollution while meeting the power consumption and peak shaving requirements of the power grid.

[0094] The hydropower and PV hybrid system constructed in the present invention aims to reduce the PV curtailment phenomenon and meet the power grid dispatching instructions at the same time. Pumped storage is introduced to better utilize the power generation resources. When the hybrid system generates more electricity and the power grid cannot receive more electrical energy, in order to reduce PV curtailment, the excess PV power generation is stored through energy storage devices. When the hybrid power generation system cannot meet the power grid dispatching plan, energy storage can compensate for part of the power deficit through discharging. The hydropower and PV hybrid system with energy storage can not only reduce the fluctuation of PV output and realize the transfer of electrical energy, but also obtain greater economic benefits. Since the electrical energy storage can respond to the load change in a timely manner and has a relatively high charge-discharge efficiency, it also participates in the energy management and day-ahead optimal dispatching of the system.

[0095] S1.2 Analysis of PV uncertainty

[0096] The PV power generation under three typical days of sunny, cloudy and rainy days is compared and analyzed. As Figure 2 shown, the PV output has obvious daily characteristics, that is, the output is larger during the day affected by meteorological factors, in the rising stage in the morning and in the falling stage in the afternoon. The PV output fluctuates greatly, and the output is zero at night due to the absence of light. The PV output is mainly affected by light intensity, temperature and weather. The output is the largest on sunny days, the curve is relatively smooth, and the peak time is relatively fixed. The output on cloudy days is the second, and the output on rainy days is the smallest. However, the output on cloudy and rainy days shows obvious fluctuations and the peak time is not fixed and unique. Generally, the PV output will show typical seasonal characteristics of being higher in summer and autumn and lower in spring and winter. However, due to the unique geographical location and climate conditions in Xinjiang, the change of light intensity throughout the year is not significant, so the change of PV power generation is not obvious and the seasonal characteristics are also weak. As Figure 3 shown.

[0097] S1.3 PV scenario generation and reduction

[0098] S1.3.1 Monte Carlo scenario generation

[0099] There may be multiple different scenarios for the output of a PV power station, and it is difficult for the general processing methods of PV output to take this into account, which will affect the reliability of the operation of the hydropower and PV hybrid system. In order to consider the PV output under various extreme weather conditions, the Monte Carlo simulation method is used to generate representative PV output prediction scenarios.

[0100] S1.3 Photovoltaic Scenario Generation and Reduction

[0101] S1.3.1 Monte Carlo Scenario Generation

[0102] There may be various different scenarios for the power output of a photovoltaic power station, and it is difficult for general methods of dealing with photovoltaic power output to take this into account, which will affect the reliability of the operation of the water-light complementary system. In order to consider the photovoltaic power output under various extreme weather conditions, this paper uses 2,160 photovoltaic power output data for 12 months of the whole year of 2018 of a certain photovoltaic power station to generate representative photovoltaic power output prediction scenarios by using the Monte Carlo simulation method.

[0103] S1.3.2 K-Means Clustering

[0104] Common clustering methods include K-means clustering, Gaussian mixture model clustering, density clustering, hierarchical clustering, and spectral clustering. These methods have different advantages and applicability in different data situations. Selecting an appropriate clustering method requires considering factors such as data type, data distribution, and clustering objectives. Although representative photovoltaic power output prediction scenarios are obtained by using the Monte Carlo simulation method, in actual engineering, various other factors and related constraints often need to be considered, and the model complexity is relatively high. The most commonly used K-means clustering at present can minimize the model complexity, so as to reduce a large number of typical scenarios obtained by the Monte Carlo simulation method before. The K-means clustering algorithm is to classify and divide data units with similarity and convergence, so as to determine the similarity between each data object. Using K-means clustering to achieve scenario reduction finally obtains four typical scenarios as shown in Figure 4 shown, and the weighted sum is carried out according to the probability of each typical scenario generated to obtain the predicted photovoltaic power output of a typical day, as shown in Figure 5 shown.

[0105] K-means clustering first divides the data set into K non-overlapping clusters, and each data point is the center of the cluster closest to it. It is an iterative and unsupervised clustering algorithm, often used for numerical data. When each of the K clusters is relatively dense and the differences between the clusters are obvious, the default algorithm converges. This method can not only reduce the number of scenarios, but also reflect the actual distribution process of photovoltaic prediction, further reduce the number of input variables in the model, thereby reducing the model complexity and improving the model solution efficiency. In the K-means clustering method, the silhouette coefficient method is usually used to determine the optimal value of K. Assuming that the value of K is k 0 , first introduce the within-cluster dissimilarity d 1 (i), which represents the average distance between this sample and other samples in the cluster; d 2 (i) is called the between-cluster dissimilarity, which represents the minimum value of the average distance between this sample and all samples in other clusters. Then the silhouette coefficient ε(i) of this sample can be obtained by the following formula:

[0106]

[0107] That is:

[0108]

[0109] S1.4 Hydropower Station

[0110] The peak load period of the power grid often coincides with the peak output period of the photovoltaic power station. Applying the photovoltaic power station to power grid peak shaving has natural advantages. Generally, large reservoirs have strong regulation capabilities. When the photovoltaic power station generates a high amount of electricity, the hydropower station stores a part of the water resources in the reservoir. During the period when the photovoltaic output is low or there is no power generation, the hydropower station generates electricity again to effectively reduce the peak value of the power grid load.

[0111] S2 Day-ahead Optimal Scheduling of Hydropower and Photovoltaic Complementary with Energy Storage

[0112] S2.1 Objective Function

[0113] For small and medium-sized hydropower and photovoltaic complementary systems, the power grid gives reasonable scheduling instructions in advance according to their power generation capabilities, and requires the hydropower and photovoltaic complementary system to jointly output power in each period of the scheduling period to track the given plan of the power grid. At the same time, to improve the energy utilization rate of the hydropower and photovoltaic complementary system, on the basis of ensuring the safe and stable operation of the power grid, the objective function is to minimize the cost of the hydropower and photovoltaic complementary system, while considering the operation costs of the photovoltaic power station, hydropower station and energy storage, the penalty cost for curtailment of light, the operation and maintenance costs of the photovoltaic power station and hydropower station, and the grid connection benefits of the photovoltaic power station and hydropower station.

[0114] min F = f run + f aban + f om - f sell (3)

[0115] In the formula, f run is the operation cost of the photovoltaic power station, hydropower station and energy storage, f aban is the penalty cost for curtailment of light, f om is the operation and maintenance cost of the photovoltaic power station and hydropower station, f sell is the grid connection benefit of photovoltaic power generation and hydropower.

[0116] (1) Operation Cost

[0117]

[0118] In the formula, λ run_pv , λ run_w , λ run_es are the operation cost coefficients of the photovoltaic power station, hydropower station and energy storage respectively, is the photovoltaic power generation power at time t, MW, is the hydropower power in period t, MW,

[0119] are the charge and discharge powers of energy storage in period t, MW, respectively.

[0120] (2) Penalty cost for curtailment of photovoltaic power

[0121]

[0122] In the formula, λ aban is the penalty cost coefficient for curtailment of photovoltaic power, and λ om_w is the predicted photovoltaic power in period t, MW.

[0123] (3) Operation and maintenance cost

[0124]

[0125] In the formula, λ om_pv and λ om_w are the operation and maintenance cost coefficients of the photovoltaic power station and the hydropower station, respectively.

[0126] (4) Revenue from the connection to the grid of the photovoltaic power station and the hydropower station

[0127]

[0128] In the formula, λ grid_pv is the on-grid electricity price of photovoltaic power generation, and λ grid_w is the on-grid electricity price of hydropower.

[0129] Constraint conditions

[0130] This model takes the hydropower unit, the photovoltaic power station, and the energy storage as the basic dispatching units, and takes 24 hours a day as the dispatching period, and mainly considers the following constraint conditions:

[0131] (1) Power balance constraint

[0132]

[0133] In the formula, is the grid dispatching plan, MW.

[0134] (2) Hydropower generation flow constraint

[0135]

[0136] In the formula, η w is the power generation efficiency of the hydropower station, is the power generation flow, m 3 / s, and H w is the head height of the hydropower station for power generation, m.

[0137] (3) Constraints on the upper and lower limits of the output of hydropower units

[0138]

[0139] In the formula, P w_max is the maximum installed capacity of the hydropower unit, in MW.

[0140] (4) Constraints on the output of the photovoltaic power station

[0141]

[0142] (5) Constraints related to energy storage

[0143] Energy storage capacity constraint

[0144]

[0145] In the formula, are the energy storage capacities at time t and time t - 1 respectively, in MW, and η es is the charging and discharging efficiency of the energy storage, and E s_max and E s_min are the upper and lower limits of the energy storage capacity respectively, in MW.

[0146] Initial and final energy storage capacity constraint

[0147]

[0148] In the formula, are the energy storage capacities at time 1 and time 24 respectively, in MW.

[0149] Energy storage charging and discharging power constraint

[0150]

[0151] In the formula, μ c and μ dis are the charging and discharging states of the energy storage respectively, and the energy storage cannot charge and discharge simultaneously.

[0152] S3.1 Parameter setting

[0153] The main parameters of the present invention are shown in Table 1.

[0154] Table 1 Related parameters

[0155]

[0156] S3.2 Result analysis

[0157] In the combined operation system of hydropower and photovoltaic power generation, when the photovoltaic output power is not balanced with the load demand power, the hydropower station can promptly respond to the power change, release water to drive the water turbine to generate electricity, and supplement the deficit power in the power grid.

[0158] Figure 6 The following is the scheduling result for a certain day. Due to the influence of meteorological factors, the output of the photovoltaic power generation shows the characteristic of "generating power during the day and stopping at night" with the change of light intensity. The output reaches its peak at around 11:00 noon, which is 587.415 MW, while the photovoltaic output is zero from 20:00 in the evening to 4:00 in the early morning of the next day. After 17:00, the photovoltaic output weakens significantly, while the output of the grid scheduling plan remains relatively high. At this time, hydropower with flexible adjustment ability is needed to supplement the deficit power in a timely manner. However, limited by the installed capacity and economic cost of the hydropower station, energy storage is required to discharge and compensate for part of the electric energy. When the photovoltaic power generation is greater than the output of the grid scheduling plan, the excess electric energy is stored through energy storage charging. Through the coordinated scheduling of power generation sources, the combined power generation of the hydropower-photovoltaic-energy storage complementary system closely follows the grid scheduling plan, reducing the impact on the grid caused by the uncertainty of photovoltaic output.

[0159] As Figure 7 can be seen, the initial capacity of the energy storage is 20 MW. When the energy storage discharges, its capacity gradually decreases. Due to the existence of the energy storage capacity constraint, the discharge stops when it reaches a certain depth. When the energy storage charges, its capacity gradually increases. Since the energy storage discharges during the two periods from 15:00 to 16:00, the energy storage capacity starts to decrease back to the initial capacity until the end of the scheduling period.

[0160] As Figure 8 can be seen, the actual grid-connected power of the photovoltaic can better track the predicted output of the typical photovoltaic day. There is only partial curtailment of photovoltaic power from 12:00 to 13:00 in the photovoltaic power station, and full consumption is achieved in the remaining periods. The average curtailment rate is 2.15%.

[0161] As Figure 9 can be seen, there is an obvious linear relationship between the hydropower power and the generation flow rate, and they have the same change trend.

[0162] Figure 10 (a) and (b) are respectively the scheduling result of Scenario 1 and the photovoltaic consumption curve. Due to the influence of light intensity, there is obvious intermittency in the photovoltaic power generation, that is, "generating power during the day and stopping at night". During the two periods from 12:00 to 13:00, the photovoltaic power generation is greater than the output of the grid scheduling plan, and the load shedding value is negative. In the remaining stages, due to the lack of sufficient power generation sources, the demand of the grid scheduling plan cannot be met, and there is a large risk of load shedding. Therefore, the cost is relatively high, which is 2,663,381 yuan, proving that the economy of Scenario 1 is relatively poor.

[0163] Figure 10 (c) and (d) are respectively the scheduling result of Scenario 2 and the photovoltaic consumption curve. Compared with Scenario 3, there is no energy storage in this complementary system. Therefore, only hydropower is used to compensate for the deficit power, and the photovoltaic power cannot be fully grid-connected during the high-generation period of photovoltaic. Therefore, there is a curtailment phenomenon, and the average curtailment rate is 5%. The economic cost is 435,936 yuan.

[0164] As can be seen from Table 2, in Scenario 1, due to only photovoltaic power generation, there is a serious risk of load shedding, so the cost is relatively high. The cost of Scenario 3 is the lowest, which is 84.6% and 5.7% less than that of Scenario 1 and Scenario 2 respectively. And the average curtailment rate of Scenario 3 is relatively low, which is 2.85% lower than that of Scenario 2. It can be seen that photovoltaic power generation alone cannot meet the grid dispatching plan, and there is a large risk of load shedding. Hydropower can be used as a regulating power source to compensate for the shortage of power, significantly reducing the economic cost; the on-site energy storage of the photovoltaic power station can effectively regulate the output of each device through charge and discharge, thereby reducing the cost and the curtailment rate.

[0165] Table 2 Comparative analysis table of three scenarios

[0166]

[0167] S4 Conclusion

[0168] In order to actively respond to the national policies of "carbon peak and carbon neutrality", reduce the impact of the uncertainty of photovoltaic output on the power grid, and reduce environmental pollution, the present invention establishes a coordinated optimal dispatching model of water-light complementary considering the uncertainty of photovoltaic output, and conducts simulation analysis to verify the effectiveness of the proposed model. The conclusions are as follows:

[0169] 1) First, the uncertainty of photovoltaic output is analyzed, and the Monte Carlo simulation method is used to generate scenarios. Four typical output scenarios are obtained through K-means clustering. The typical daily output of photovoltaic is obtained by weighted summation of the four typical daily outputs and their scenario probabilities, and it is used as the photovoltaic prediction data to participate in the day-ahead optimal dispatching.

[0170] 2) Taking the minimum economic cost of the water-light complementary system as the objective function, considering the relevant constraints of the photovoltaic power station, hydropower station and energy storage equipment, a day-ahead optimal dispatching model of the complementary system is constructed. The model is solved by the CPLEX solver, and the optimal economic cost is 410,990 yuan. Three scenarios are set for comparative analysis to prove that the model proposed in the present invention has good economy. The economic cost of Scenario 3 is 84.6% and 5.7% less than that of Scenario 1 and Scenario 2 respectively, and it effectively reduces the curtailment. The average curtailment rate is 2.85% less than that of Scenario 2.

[0171] 3) The simulation results show that hydropower and energy storage have good regulation ability and energy transfer function, can compensate for part of the shortage of power, make the output of the water-light complementary system smoother, and meet the requirements of the grid dispatching plan.

Claims

1. The present invention designs a water-solar-storage day-ahead optimization scheduling study considering photovoltaic uncertainty, which consists of the following steps: Step 1: Based on the uncertainty analysis of photovoltaic output, the scenario is generated through the Monte Carlo simulation method. Step 2: Use K-means clustering to reduce the scenarios and obtain four typical photovoltaic output scenarios. The typical daily photovoltaic output is obtained by weighted summation of the output of the four scenarios and their probabilities, and participates in the day-ahead optimization scheduling of hydro-photovoltaic complementarity. Step 3: Taking the minimum economic cost of the complementary system as the objective function, considering the relevant constraints of photovoltaic power stations, hydro-photovoltaic complementarity and energy storage, a hydro-photovoltaic complementarity day-ahead optimization scheduling model with energy storage is established. The CPLEX solver is used to obtain the optimal economic cost of the model and the optimal output of each device.

2. The method according to claim 1, characterized in that In step 1, there may be many different scenarios for the output of photovoltaic power stations, and it is difficult to take this into account in general for the treatment of photovoltaic output, which will affect the reliability of the operation of the water-photovoltaic complementary system. In order to consider the photovoltaic output under various extreme weather conditions, the Monte Carlo simulation method is used to generate representative photovoltaic output prediction scenarios.

3. The method according to claim 1, characterized in that In step 2, K-means clustering first divides the data set into K non-overlapping clusters, and each data point is the center of the cluster closest to it. It is an iterative, unsupervised clustering algorithm, often used for numerical data. When each cluster in the K clusters is relatively dense and the clusters are clearly distinguished, the default algorithm converges. This method can not only reduce the number of scenarios, but also reflect the actual distribution process of photovoltaic prediction, further reduce the number of input variables in the model, thereby reducing the complexity of the model and improving the efficiency of model solution. The silhouette coefficient method is usually used in the K-means clustering method to determine the optimal K value. Assuming that the value of K is k0, first introduce its intra-cluster dissimilarity d1(i), which represents the average distance between the sample and other samples in the cluster; d2(i) is called inter-cluster dissimilarity, which represents the minimum value of the average distance between the sample and all samples in other clusters. The silhouette coefficient ε(i) of the sample can be obtained by the following formula: That is:

4. According to the objective function described in claim 1, corresponding to step 3, for small and medium-sized hydro-photovoltaic complementary systems, the power grid pre-sets reasonable dispatching instructions based on their power generation capacity, and requires the hydro-photovoltaic complementary systems to jointly output and track the plan given by the power grid during each period of the dispatch period. At the same time, in order to improve the energy utilization rate of the hydro-photovoltaic complementary system, on the basis of ensuring the safe and stable operation of the power grid, the cost of the hydro-photovoltaic complementary system is minimized as the objective function, while considering the operating costs of photovoltaic power stations, hydropower stations and energy storage, the penalty cost of abandoned light, the operation and maintenance costs of photovoltaic power stations and hydropower stations, and the income from the access to the grid of photovoltaic power stations and hydropower stations. min F=f run +f aban +f om -f sell (3) In the formula, f run is the operating cost of photovoltaic power station, hydropower station and energy storage, f aban is the penalty cost for abandoning light, f om is the operation and maintenance cost of photovoltaic power station and hydropower station, f sell The income from photovoltaic power generation and hydropower access to the grid. (1) Operating costs In the formula, λ run_pv , run_w , run_es are the operating cost coefficients of photovoltaic power stations, hydropower stations and energy storage, is the photovoltaic power generation power in period t, MW, is the hydropower power in period t, MW, They are the energy storage charging and discharging power in period t, MW. (2) Penalty cost for abandoned light In the formula, λ aban is the penalty cost coefficient for abandoned light, λ om_w The predicted photovoltaic power for period t, MW. (3) Operation and maintenance costs In the formula, λ om_pv , om_w They are the operation and maintenance cost coefficients of photovoltaic power stations and hydropower stations respectively. (4) Income from photovoltaic power stations and hydropower stations In the formula, λ grid_pv is the photovoltaic power generation grid-connected electricity price, λ grid_w The on-grid electricity price for hydropower. Constraints This model uses hydropower units, photovoltaic power stations, and energy storage as basic scheduling units, with a scheduling cycle of 24 hours a day, and mainly considers the following constraints: (1) Power balance constraints In the formula, is the grid dispatch plan, MW. (2) Hydropower flow constraints Where η w The power generation efficiency of the hydropower station, is the power generation flow, m 3 / s,H w is the generating head height of the hydropower station, m. (3) Upper and lower limits of hydropower unit output Where P w_max is the maximum installed capacity of the hydropower unit, MW. (4) PV power station output constraints (5) Energy storage related constraints Energy storage capacity constraints In the formula, are the energy storage capacity in period t and period t-1, MW, η es is the energy storage charging and discharging efficiency, E s_max 、E s_min They are the upper and lower limits of energy storage capacity, MW respectively. Energy storage capacity constraints In the formula, They are the energy storage capacity for 1 period and 24 periods respectively, MW. Energy storage charging and discharging power constraints In the formula, μ c , μ dis They are the charging and discharging states of the energy storage respectively. The energy storage cannot be charged and discharged at the same time.

Images