Wind and light storage planning method considering long-term and short-term time sequence fluctuation characteristics of new energy

Abstract

The invention relates to the technical field of power system planning, in particular to a wind and light storage planning method considering long-term and short-term time sequence fluctuation characteristics of new energy. According to the method, a basic data set is predicted based on the daily output of a wind turbine generator and a photovoltaic unit, the basic data set is clustered through a clustering algorithm, clustering center data is used as a typical scene, and an initial representative scene set is established; taking other data as non-typical scenes and establishing a non-representative scene set; based on an iterative optimization strategy, screening out atypical scenes and incorporating the atypical scenes into an initial representative scene set; and based on the relative root-mean-square error of the new energy continuous output curve, selecting the optimal typical scene set and representative scene set to reconstruct a whole-year time sequence, and finally solving the wind-light-storage collaborative planning model based on the reconstructed whole-year time sequence to obtain a wind-light-storage planning scheme. According to the method, the model solving efficiency can be remarkably improved on the premise of ensuring the model solving precision.

Description

Technical Field

[0001] The present invention relates to the technical field of power system planning, and specifically discloses a wind, solar and energy storage planning method that takes into account the long-term and short-term temporal fluctuation characteristics of new energy. Background Art

[0002] As renewable energy generation accounts for an increasing proportion of power systems, the randomness and uncertainty of renewable energy output pose new challenges to power planning. Traditional operational simulation methods based on continuous load curves struggle to account for the demands for flexibility in power system operations caused by fluctuations in renewable energy output, such as peak shaving, frequency regulation, ramping, and standby. Furthermore, wind and photovoltaic output are significantly affected by natural conditions, exhibiting strong volatility and randomness on short-term timescales and seasonality on longer-term timescales. Therefore, power planning must balance both short-term power supply and demand with medium- and long-term seasonality, while also addressing the goals of ensuring power supply, consuming electricity, and reducing carbon emissions.

[0003] At the same time, short-term energy storage, such as electrochemical energy storage, has the ability to balance electricity supply and demand, smoothing renewable energy output and participating in frequency regulation and intraday peak load regulation. Long-term energy storage, such as pumped storage and compressed air storage, has the ability to adjust power across seasons, addressing the temporal and spatial mismatch between renewable energy output and load demand, and improving the system's absorption capacity. In this context, achieving efficient and coordinated deployment of wind, solar, and storage is crucial.

[0004] To adapt to the development requirements of power systems with a high proportion of renewable energy access, there is an urgent need to build a power planning model that establishes high spatiotemporal resolution and the coordination of wind, solar, and storage. This model can rationally allocate renewable energy and multiple types of energy storage, ensuring the reliability of power supply while increasing the absorption rate of renewable energy. However, short-term energy storage generally requires completing daily charging and discharging cycles while smoothing the output of renewable energy, while long-term energy storage requires achieving energy balance over a yearly time scale. The multi-timescale temporal coupling constraints of wind, solar, and storage lead to an exponential increase in computational complexity. How to achieve efficient model solutions while ensuring modeling accuracy is a key challenge facing wind, solar, and storage coordinated planning. Summary of the Invention

[0005] The purpose of this invention is to provide a wind, solar, and energy storage planning method that considers the long-term and short-term temporal fluctuation characteristics of renewable energy, and to solve the problem in existing solutions where the multi-time-scale temporal coupling constraints of wind, solar, and energy storage lead to an exponential increase in computational complexity. The specific solution is as follows:

[0006] Provides a wind, solar, and energy storage planning method that takes into account the long-term and short-term temporal fluctuation characteristics of renewable energy, including:

[0007] Based on the daily power forecast data of wind turbines and photovoltaic units, a basic data set is established. The basic data set is clustered using a clustering algorithm, and the cluster center data is used as a typical scene to form an initial representative scene set. The remaining data is used as atypical scenes to form a non-representative scene set.

[0008] Based on an iterative optimization strategy, the Euclidean distance between each atypical scene and each representative scene is obtained, and the minimum Euclidean distance between each atypical scene and the set of representative scenes is recorded. The atypical scene corresponding to the Euclidean distance with the largest median of the minimum Euclidean distance is then taken as the new representative scene, and extracted from the set of non-representative scenes and included in the set of representative scenes.

[0009] Based on the representative scenario sets iteratively obtained with different initial typical scenario numbers, the relative root mean square error of the renewable energy continuous output curve is calculated, and the representative scenario set with the relative root mean square error less than the threshold ε and the least number of representative scenarios is selected to reconstruct the full-year time series;

[0010] A wind, solar and storage collaborative planning model is established, and the solution is obtained based on the reconstructed annual time series to obtain the wind, solar and storage planning scheme.

[0011] Furthermore, the basic data set includes hourly output forecast data of wind turbines and photovoltaic units throughout the year. The expression of the basic data set is as follows:

[0012]

[0013] Where W represents wind power output data; PV represents photovoltaic output data; N w Indicates the number of wind turbines; N pv Indicates the number of photovoltaic units; T s Indicates the number of time steps of scene s.

[0014] Furthermore, the clustering algorithm includes K-Medoids, K-Means, and DBSCAN clustering algorithms, which cluster the basic data set through the clustering algorithm, and use the cluster center of each cluster as a typical scene; different initial representative scene sets composed of typical scenes under different cluster numbers are obtained.

[0015] Furthermore, the execution condition of the iterative optimization strategy is to stop the iteration when the relative root mean square error of the new energy continuous output curve calculated based on the representative scenario set and the number of typical scenarios of the current iteration is less than a threshold ε.

[0016] Furthermore, the expression for taking the atypical scene corresponding to the Euclidean distance with the largest median minimum Euclidean distance as the new representative scene is as follows:

[0017] s * =arg maxd(s,SR )

[0018] Among them, s * Denotes a new representative scene, S = {s1, s2, ..., s D} represents the full set of basic scenarios; represents the selected representative scene set; d(s,S R ) represents the scene s and the representative scene set S R The minimum Euclidean distance between them is expressed as follows:

[0019] d(s,S m )=min‖ss r ‖

[0020] Among them, ‖ss r ‖ represents scene s and representative scene s r The Euclidean distance between .

[0021] Furthermore, the expression for calculating the relative root mean square error of the new energy continuous output curve is as follows:

[0022]

[0023] Among them, ω w,t 、ω pv,t They contribute to the original scenery respectively; are the wind and solar power outputs reconstructed based on the new representative scene set; T is the total number of time steps throughout the year; k is the number of different typical scenes;

[0024] The representative scene set with the minimum number of representative scenes and the relative root mean square error less than the threshold ε is selected as follows:

[0025]

[0026] Among them, k * represents the number of typical scenarios used to reconstruct the full-year time series, Represents the representative set of scenarios used to reconstruct the full-year time series.

[0027] Furthermore, the reconstruction of the full-year time series is specifically as follows: the mapping function is defined as follows:

[0028]

[0029] The mapping function is used to map each basic scene s to its closest scene s in the representative scene set. r , the mapping is established by the minimum Euclidean distance, as follows:

[0030]

[0031] Based on the function χ N (s), the full-year time series is expressed as:

[0032]

[0033] in, Represented by the representative scene set S R Reconstructed full-year time series.

[0034] Furthermore, the wind, solar and energy storage collaborative planning model takes minimizing the system's total investment and operating costs as its objective function, and is embedded with safe and stable operation constraints for units and energy storage equipment under various representative scenarios.

[0035] Furthermore, the wind-solar-storage collaborative planning model is expressed as follows:

[0036]

[0037] Among them, c g , γ g They represent the annualized installed cost per unit capacity and the power generation cost per unit power of the unit or energy storage device g respectively; ω s represents the weight of scenario s; ψ represents the set of units and energy storage devices; T s Indicates the number of time steps of scene s; O g Indicates the installed capacity of the unit and energy storage equipment; P g,s,t represents the output of unit g at time t in scenario s; f(O g ,p(O g ,s,t)) indicates that there is no timing coupling constraint, h(O g ,q(O g ,s,t,t-1)) indicates that there is a constraint with timing coupling.

[0038] Furthermore, the method further includes: establishing an intraday component and an inter-day component for the state of charge of the energy storage; the intraday component is used to characterize the smoothing effect of the energy storage on the short-term fluctuations of renewable energy, and the inter-day component is used to characterize the energy transfer effect of long-term energy storage on the seasonal fluctuations of renewable energy output across weeks or months;

[0039] Energy storage is modeled as follows:

[0040]

[0041] Among them, SOC b,d,t is the charge of energy storage device b at time t on day d; is the daytime component of the SOC of energy storage device b on day d; is the intraday component of the SOC of the energy storage device b at time t in scenario s; s = χ S(d) is the mapping of day d to the representative scene set S R After that, it belongs to scene s;

[0042] The expression of the intraday component is as follows:

[0043]

[0044] in, are the charging efficiency and discharging efficiency of energy storage device b respectively; are the charging power and discharging power of energy storage device b at time t-1 in scenario s;

[0045] The expression of the daytime component is as follows:

[0046]

[0047] Where v is the initial SOC ratio of the energy storage device; is the rated SOC of energy storage device b;

[0048] The intraday component and the interday component follow the following constraints:

[0049]

[0050] Where D represents the total number of natural days; ψ short , ψ long represent the collection of short-term energy storage devices and long-term energy storage devices respectively;

[0051] The energy storage device complies with the following constraints:

[0052] The upper and lower limits of energy storage charging and discharging power are:

[0053]

[0054] in, is the rated charge and discharge power of energy storage device b;

[0055] The upper and lower limits of energy storage capacity are:

[0056]

[0057] Beneficial effects of the present invention:

[0058] Fully consider the various typical and extreme operating scenarios that the power system may face under the high proportion of renewable energy penetration in the future, select representative scenarios from massive scenarios to reduce the decision variables and constraints of the model; and establish a wind, solar and storage collaborative planning model that can capture the long-term and short-term temporal fluctuation characteristics of renewable energy, accurately depict the collaborative optimization operation mode of renewable energy and multi-type energy storage at different time scales, and thus provide technical support for decisions such as new energy base planning and multi-type energy storage configuration. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] Figure 1 This is a schematic diagram of the wind, solar and energy storage planning method that takes into account the long-term and short-term temporal fluctuation characteristics of new energy. DETAILED DESCRIPTION

[0060] To make the objectives, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions of the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings of 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. Generally, the components of the embodiments of the present invention described and shown in the drawings herein can be arranged and designed in various different configurations.

[0061] Since short-term energy storage is usually required to complete the daily charge and discharge cycle while smoothing the output of renewable energy, and long-term energy storage requires energy balance within the annual time scale, the multi-time-scale temporal coupling constraints of wind, solar and storage will lead to an exponential increase in the amount of calculation. Therefore, how to achieve efficient solution of the model while ensuring the accuracy of modeling is an important difficulty faced by the coordinated planning of wind, solar and storage. In response to this difficulty, the present invention proposes a wind, solar and storage planning method that takes into account the long-term and short-term temporal fluctuation characteristics of renewable energy. On the one hand, it fully considers various typical and extreme operating scenarios that the power system may face under the high proportion of renewable energy penetration in the future, selects representative scenarios from massive scenarios to reduce the decision variables and constraints of the model; on the other hand, it establishes a wind, solar and storage coordinated planning model that can capture the long-term and short-term temporal fluctuation characteristics of renewable energy, accurately depicts the coordinated optimization operation mode of renewable energy and multi-type energy storage at different time scales, and thus provides technical support for decisions such as new energy base planning and multi-type energy storage configuration.

[0062] The specific plan is as follows:

[0063] like Figure 1 As shown in the figure, the wind, solar and storage planning method considering the long-term and short-term fluctuation characteristics of renewable energy includes:

[0064] Based on the daily power forecast data of wind turbines and photovoltaic units, a basic data set is established; the basic data set is clustered using a clustering algorithm, and the cluster center data is used as the typical scene to form an initial representative scene set; the remaining data is used as atypical scenes to form a non-representative scene set.

[0065] The basic data set includes hourly output forecast data for wind turbines and photovoltaic units throughout the year. The expression of the basic data set is as follows:

[0066]

[0067] Where W represents wind power output data; PV represents photovoltaic output data; N w Indicates the number of wind turbines; N pv Indicates the number of photovoltaic units; T s Represents the number of time steps in scenario s. It should be noted that the scenario here is a natural day. The basic data set in the solution of the present invention is the annual output forecast data of the unit, which means that it includes a total of 365 scenarios, and the number of time steps in each scenario is 24, that is, the time interval is 1 hour.

[0068] Clustering algorithms such as K-Medoids, K-Means, and DBSCAN can be used. The K-Medoids algorithm is particularly recommended. It is a partition-based clustering algorithm similar to the K-Means algorithm, but differs in its approach to cluster center selection. Its advantage lies in its robustness to noise and outliers. The K-Medoids algorithm selects actual points in the dataset as cluster centers (medoids), rather than using the mean of the data points within a cluster as the center, as the K-Means algorithm does. This makes the K-Medoids algorithm less sensitive to noise and outliers, as the cluster centers are actual data points and are not overly influenced by a few extreme values, providing a more stable representation of the characteristics of each cluster. Furthermore, because the cluster centers selected by the K-Medoids algorithm are actual points in the dataset, these points have clear characteristics and attributes, making the clustering results easier to understand and interpret. Users can directly view the data points represented by each cluster center to understand the typical characteristics of each cluster.

[0069] The basic data set is clustered by the K-Medoids clustering algorithm, and the cluster center of each cluster is used as the typical scene; different initial representative scene sets composed of typical scenes are obtained under different numbers of clusters. It should be noted that the cluster center is the actual data point with the typical characteristics of the cluster. When the initial clustering parameters before each clustering are different, the number of cluster centers obtained for each clustering will be different, so a different number of clusters can be obtained; and thus different initial representative scene sets can be obtained; that is, the initial representative scene set is composed of the typical scenes represented by the cluster centers after clustering. Alternatively, clustering can be performed again based on the same conditions based on different basic data sets, and a different number of clusters can be obtained; and thus different initial representative scene sets can be obtained.

[0070] It should be further explained that the basic data set in this embodiment is mainly the hourly output forecast data of wind turbines and photovoltaic units throughout the year; in order to make the final planning more accurate, node load, irradiation intensity, wind speed and other data can also be included in the basic data set. The more data in the basic data set, the higher the complexity of the final solution; therefore, in order to take into account the convenience of solution, this embodiment preferably uses the hourly output forecast data of wind turbines and photovoltaic units throughout the year as the basic data set, provided that the planning requirements are met.

[0071] Based on the iterative optimization strategy, the Euclidean distance between each atypical scene and each representative scene is obtained, and the minimum Euclidean distance between each atypical scene and the representative scene is recorded; then the atypical scene corresponding to the Euclidean distance with the largest median of the minimum Euclidean distance is taken as the new representative scene, and extracted from the non-representative scene set and included in the representative scene set.

[0072] Since we performed cluster analysis on the basic dataset, each data point has its own position in the cluster space. Based on this, the data point corresponding to the cluster center is the typical scene, and the other data points are atypical scenes. The Euclidean distance between each atypical scene and each typical scene (i.e., representative scene) is then obtained, and only the distance value with the smallest Euclidean distance is recorded or retained. The atypical scene corresponding to the largest Euclidean distance value among multiple (i.e., the same number of atypical scenes) minimum Euclidean distances is then selected as the new representative scene. The new representative scene is included in the initial representative scene set to update the representative scene set. It should be noted that each iteration selects an atypical scene as the new representative scene. That is, in the first iteration, the new representative scene is included in the initial representative scene set to update the representative scene set. After the first iteration, the new representative scene is included in the representative scene set obtained after the previous iteration to update the representative scene set. The purpose of screening atypical scenes and including them in the representative scene set here is to make the reconstructed full-year time series closer to reality and to make the final solution more accurate.

[0073] To better understand the solution, this embodiment uses the atypical scenario corresponding to the Euclidean distance with the largest median minimum Euclidean distance as the new representative scenario, as expressed as follows:

[0074] s * =arg maxd(s,S R )

[0075] Among them, s * Denotes a new representative scene, S = {s1, s2, ..., s D} represents the full set of basic scenarios; represents the selected representative scene set; d(s,S R ) represents the scene s and the representative scene set S R The minimum Euclidean distance between them is expressed as follows:

[0076] d(s,S R )=min‖ss r ‖

[0077] Among them, ‖ss r ‖ represents scene s and representative scene s r The Euclidean distance between .

[0078] After each iteration, the relative root mean square error of the renewable energy continuous output curve is further calculated based on the representative scenario set of the current iteration and the initial number of typical scenarios, and an evaluation is made as to whether the root mean square error is less than the threshold ε. If so, the iteration is stopped; otherwise, the iteration will continue.

[0079] Since the same clustering parameter may not ultimately be able to iterate out a representative scene set that can make the root mean square error less than the threshold ε, the clustering parameter can be adjusted to obtain a different number of initial typical scenes.

[0080] Therefore, based on the representative scene sets obtained by iteration with different initial typical scene numbers, the relative root mean square error of the continuous output curve of the new energy is calculated, and the representative scene set with the relative root mean square error less than the threshold ε and the least number of representative scenes is selected to reconstruct the full-year time series; in order to understand the scheme more clearly, the following explanation is made: when the algorithm parameters are adjusted during clustering, the cluster center or the number of cluster centers will change; such as the K-Medoids clustering algorithm that is more recommended when implementing the scheme of the present invention, when the K value is different, the number of data points selected as the initial medoid will be different, and the number of representative scenes in the obtained initial representative scene set will be different; that is, the aforementioned iteration based on different initial typical scene numbers. That is to say, in this scheme, there will be multiple groups of initial representative scene sets obtained based on different K values, the number of representative scenes in each group is related to the K value, and each group will be iterated, and the termination condition of the iteration is that the iteration is stopped when the relative root mean square error of the continuous output curve of the new energy is less than the threshold ε based on the representative scene set of the current iteration and the number of typical scenes obtained during clustering.

[0081] The expression for calculating the relative root mean square error of the new energy continuous output curve is as follows:

[0082]

[0083] Among them, ω w,t 、ω pv,tThey contribute to the original scenery respectively; are the wind and solar power outputs reconstructed based on the new representative scene set; T is the total number of time steps throughout the year; k is the number of different typical scenes;

[0084] The representative scene set with the relative root mean square error less than the threshold ε and the least number of representative scenes is selected as follows:

[0085]

[0086] Among them, k * represents the number of typical scenarios used to reconstruct the full-year time series, Represents the representative set of scenarios used to reconstruct the full-year time series.

[0087] Reconstructing the full-year time series is as follows: Define the mapping function as follows:

[0088]

[0089] The mapping function is used to map each base scene s to its closest scene s in the set of representative scenes. r , the mapping is established by the minimum Euclidean distance, as follows:

[0090]

[0091] Based on the function χ N (s), the full-year time series is expressed as:

[0092]

[0093] in, Represented by the representative scene set S R Reconstructed full-year time series.

[0094] After completing the reconstruction of the full-year time series for solution, it is necessary to further establish a wind, solar and storage collaborative planning model. Specifically, a wind, solar and storage planning model that can capture the long-term and short-term time series fluctuation characteristics of new energy is established. Its compact form is expressed as follows:

[0095]

[0096] Among them, c g , γ g They represent the annualized installed cost per unit capacity and the power generation cost per unit power of the unit or energy storage device g respectively; ω s represents the weight of scenario s; ψ represents the set of units and energy storage devices; T s Indicates the number of time steps of scene s; O g Indicates the installed capacity of the unit and energy storage equipment; P g,s,t Indicates that unit g is in scene s at time t

[0097] Indicates the existence of timing coupled constraints.

[0098] The aforementioned wind, solar, and energy storage collaborative planning model takes minimizing the total system investment and operating costs as its objective function and incorporates built-in constraints for the safe and stable operation of generators and energy storage equipment under various representative scenarios. Specifically, the objective function consists of two components: investment cost and operating cost. Among the constraints, the first is independent of timing coupling, including requirements that the output of generators and equipment must not exceed their total installed capacity, node power balance, requirements for the proportion of renewable energy output, and load shedding limits. The second is constraints that do involve timing coupling, such as output ramping constraints, minimum on / off duration constraints, and energy storage SOC constraints.

[0099] Regarding the energy storage SOC constraint, it is necessary to establish intraday and interday components for the energy storage state of charge; the intraday component is used to characterize the smoothing effect of energy storage on short-term fluctuations in renewable energy, and the interday component is used to characterize the cross-weekly or inter-monthly energy transfer effect of long-term energy storage on seasonal fluctuations in renewable energy output. In this wind, solar, and storage joint planning model that can capture the long-term and short-term time series fluctuation characteristics of renewable energy, the energy storage SOC is modeled as follows:

[0100]

[0101] Among them, SOC b,d,t is the charge of energy storage device b at time t on day d; is the daytime component of the SOC of energy storage device b on day d; is the intraday component of the SOC of the energy storage device b at time t in scenario s; s = X S (d) is the mapping of day d to the representative scene set S R After that, it belongs to scene s;

[0102] The expression of intraday component is as follows:

[0103]

[0104] in, are the charging efficiency and discharging efficiency of energy storage device b respectively; are the charging power and discharging power of energy storage device b at time t-1 in scenario s;

[0105] The expression for the daytime component is as follows:

[0106]

[0107] Where v is the initial SOC ratio of the energy storage device; is the rated SOC of energy storage device b.

[0108] For short-term energy storage such as electrochemical energy storage, the charge and discharge cycle must be completed within a day. For seasonal energy storage, the charge and discharge cycle must be completed throughout the year. Therefore, the intraday component and the daytime component also follow the following constraints:

[0109]

[0110] Wherein, D represents the total number of natural days. In this embodiment, annual planning is taken as an example, so D is 365; short , ψ long represent the collection of short-term energy storage devices and long-term energy storage devices respectively;

[0111] Energy storage devices follow the following constraints:

[0112] The upper and lower limits of energy storage charging and discharging power are:

[0113]

[0114] in, is the rated charge and discharge power of energy storage device b;

[0115] The upper and lower limits of energy storage capacity are:

[0116]

[0117] By using the established wind, solar, and storage collaborative planning model and solving it based on the reconstructed annual time series, we can obtain the optimal wind, solar, and storage planning solution. The optimal solution here is the one with the lowest investment and operating costs. The solver can be one of the CPLEX solver, IPOPT solver, or CPLEX solver.

[0118] The above-mentioned embodiment provides an efficient wind, solar and storage planning method that takes into account the long-term and short-term time series fluctuation characteristics of new energy. It can comprehensively consider the impact of typical scenarios and extreme scenarios on the planning of new energy installed capacity and energy storage configuration schemes, and ensure the safety and stability of the system. Secondly, the established wind, solar and storage joint planning model takes into account the power balance principle of short-term energy storage and long-term energy storage, and can fully capture the fluctuation characteristics of new energy output at different time scales, thereby improving the system's new energy absorption capacity. Finally, based on a simplified set of representative scenarios, the annual time series is reconstructed and planning solutions are performed, which can improve the solution efficiency by 1 to 2 orders of magnitude while ensuring the accuracy of the model solution.

[0119] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Those skilled in the art will readily appreciate that various modifications and variations of the present invention are possible. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of the present invention shall be included within the scope of protection of the present invention.

Claims

1. A wind, solar and energy storage planning method that considers the long-term and short-term temporal fluctuation characteristics of new energy is characterized by: include: Establish a basic data set based on the daily power forecast data of wind turbines and photovoltaic units; The basic data set is clustered using a clustering algorithm, and the cluster center data is used as the typical scene to form the initial representative scene set; the remaining data is used as the atypical scene to form the non-representative scene set; Based on an iterative optimization strategy, the Euclidean distance between each atypical scene and each representative scene is obtained, and the minimum Euclidean distance between each atypical scene and the set of representative scenes is recorded. The atypical scene corresponding to the Euclidean distance with the largest median of the minimum Euclidean distance is then taken as the new representative scene, and extracted from the set of non-representative scenes and included in the set of representative scenes. Based on the representative scenario sets iteratively obtained with different initial typical scenario numbers, the relative root mean square error of the renewable energy continuous output curve is calculated, and the representative scenario set with the relative root mean square error less than the threshold ε and the least number of representative scenarios is selected to reconstruct the full-year time series; A wind, solar and storage collaborative planning model is established, and the solution is obtained based on the reconstructed annual time series to obtain the wind, solar and storage planning scheme.

2. The wind, solar and energy storage planning method considering the long-term and short-term temporal fluctuation characteristics of new energy as claimed in claim 1 is characterized in that: The basic data set includes hourly output forecast data of wind turbines and photovoltaic units throughout the year. The expression of the basic data set is as follows: Where W represents wind power output data; PV represents photovoltaic output data; N w Indicates the number of wind turbines; N pv Indicates the number of photovoltaic units; T s Indicates the number of time steps of scene s.

3. The wind, solar and energy storage planning method considering the long-term and short-term temporal fluctuation characteristics of new energy as claimed in claim 1 is characterized in that: The clustering algorithms include K-Medoids, K-Means, and DBSCAN clustering algorithms. The basic data set is clustered by the clustering algorithm, and the cluster center of each cluster is used as a typical scene; different initial representative scene sets composed of typical scenes under different cluster numbers are obtained.

4. The wind, solar and energy storage planning method considering the long-term and short-term time series fluctuation characteristics of new energy as claimed in claim 1 is characterized in that: The execution condition of the iterative optimization strategy is to stop the iteration when the relative root mean square error of the new energy continuous output curve calculated based on the representative scenario set and the number of typical scenarios of the current iteration is less than a threshold ε.

5. The wind, solar and energy storage planning method considering the long-term and short-term temporal fluctuation characteristics of new energy as claimed in claim 4 is characterized in that: The expression for taking the atypical scene corresponding to the Euclidean distance with the largest median minimum Euclidean distance as the new representative scene is as follows: s * =arg maxd(s,S R ) Among them, s * Denotes a new representative scene, s={s1,s2,……,s D } represents the full set of basic scenarios; represents the selected representative scene set; d(s,S R ) represents the scene s and the representative scene set s R The minimum Euclidean distance between them is expressed as follows: d(s,S R )=min‖s-s r ‖ Among them, ‖ss r ‖ represents scene s and representative scene s r The Euclidean distance between .

6. The wind, solar and energy storage planning method considering the long-term and short-term temporal fluctuation characteristics of new energy as claimed in claim 5 is characterized in that: The expression for calculating the relative root mean square error of the new energy continuous output curve is as follows: Among them, ω w,t 、ω pv,t They contribute to the original scenery respectively; are the wind and solar power outputs reconstructed based on the new representative scene set; T is the total number of time steps throughout the year; k is the number of different typical scenes; The representative scene set with the minimum number of representative scenes and the relative root mean square error less than the threshold ε is selected as follows: Among them, k * represents the number of typical scenarios used to reconstruct the full-year time series, Represents the representative set of scenarios used to reconstruct the full-year time series.

7. The wind, solar and energy storage planning method considering the long-term and short-term time series fluctuation characteristics of new energy as claimed in claim 6 is characterized in that: The reconstruction of the full-year time series is specifically as follows: the mapping function is defined as follows: The mapping function is used to map each basic scene s to its closest scene s in the representative scene set. r , the mapping is established by the minimum Euclidean distance, as follows: Based on the function χ N (s), the full-year time series is expressed as: in, Represented by the representative scene set S R Reconstructed full-year time series.

8. The wind, solar and energy storage planning method considering the long-term and short-term time series fluctuation characteristics of new energy as claimed in claim 1 is characterized in that: The wind, solar and energy storage collaborative planning model takes the minimization of the system's investment and total operating costs as its objective function, and is embedded with safe and stable operation constraints for units and energy storage equipment under various representative scenarios.

9. The wind, solar and energy storage planning method considering the long-term and short-term time series fluctuation characteristics of new energy as claimed in claim 8 is characterized in that: The expression of the wind-solar-storage collaborative planning model is as follows: Among them, c g , γ g They represent the annualized installed cost per unit capacity and the power generation cost per unit power of the unit or energy storage device g respectively; ω s represents the weight of scenario s; Ψ represents the set of units and energy storage devices; T s Indicates the number of time steps of scene s; O g Indicates the installed capacity of the unit and energy storage equipment; P g,s,t represents the output of unit g at time t in scenario s; f(O g ,p(O g ,s,t)) indicates that there is no timing coupling constraint, h(O g ,q(O g ,s,t,t-1)) indicates that there is a constraint with timing coupling.

10. The wind, solar and energy storage planning method considering the long-term and short-term time series fluctuation characteristics of new energy as claimed in claim 9 is characterized in that: Also includes: Establish intraday component and interday component for the state of charge of energy storage; The intraday component is used to characterize the smoothing effect of energy storage on short-term fluctuations in renewable energy, and the interday component is used to characterize the cross-weekly or cross-monthly energy transfer effect of long-term energy storage on seasonal fluctuations in renewable energy output. Energy storage is modeled as follows: Among them, SOC b,d,t is the charge of energy storage device b at time t on day d; is the daytime component of the SOC of energy storage device b on day d; is the intraday component of the SOC of the energy storage device b at time t in scenario s; s = χ S (d) is the mapping of day d to the representative scene set S R After that, it belongs to scene s; The expression of the intraday component is as follows: in, are the charging efficiency and discharging efficiency of energy storage device b respectively; are the charging power and discharging power of energy storage device b at time t-1 in scenario s; The expression of the daytime component is as follows: Where v is the initial SOC ratio of the energy storage device; is the rated SOC of energy storage device b; The intraday component and the interday component follow the following constraints: Where D represents the total number of natural days; ψ short , ψ long represent the collection of short-term energy storage devices and long-term energy storage devices respectively; The energy storage device complies with the following constraints: The upper and lower limits of energy storage charging and discharging power are: in, is the rated charge and discharge power of energy storage device b; The upper and lower limits of energy storage capacity are:

Images