A method and system for generating wind and solar power scenarios taking into account normal and extreme outputs

The generation of wind and light power scenarios through soft-DTW based K-medoids algorithm and LOF algorithm is solved, and the problem of output prediction of wind and light generation extremes is improved, the accuracy and interpretability of scene generation are enhanced, and the stability and scheduling efficiency of the power system are enhanced.

CN118734113BActive Publication Date: 2024-11-08SOUTHEAST UNIV
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411214817.9
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-09-02
Publication Date
2024-11-08
Estimated Expiration
2044-09-02

AI Technical Summary

Technical Problem

The prior art is difficult to accurately predict and simulate extreme output conditions of wind and light power generation, especially in conventional and extreme climate conditions, resulting in the impact of power system stability and the traditional methods lack interpretability.

Method used

The soft-DTW based K-medoids algorithm is used to cluster the weekly scene data to generate conventional scenery output scenarios, and the daily scene data is outlier detected through the LOF algorithm to generate extreme scenery output scenarios, and then replace the extreme scenes. In some conventional scenarios, a wind and light power scenario is formed that comprehensively considers conventional and extreme output.

Benefits of technology

It improves the accuracy and interpretability of wind and light power scenario generation, can effectively deal with extreme output situations, and enhances the scheduling efficiency and risk prevention capabilities of the power system.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118734113B_ABST
    Figure CN118734113B_ABST
Patent Text Reader

Abstract

The invention discloses a method and system for generating a wind-solar power scenario taking into account conventional and extreme outputs, belonging to the field of new energy. The method comprises: obtaining original wind-solar output data and reconstructing it, and obtaining weekly and daily scenario data; clustering the weekly scenario data by using a soft-DTW based K-medoids algorithm to generate a conventional weekly wind-solar output scenario; performing outlier detection on the daily scenario data by using a LOF algorithm to generate an extreme daily wind-solar output scenario; replacing part of the days in the conventional weekly wind-solar output scenario with the extreme daily wind-solar output scenario to generate a wind-solar power scenario taking into account conventional and extreme outputs; the scenario takes into account both conventional and extreme characteristics of wind-solar output, and provides scenario support for the optimization decision of the power system.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention belongs to the field of new energy, and in particular relates to a method and system for generating wind and solar power scenarios taking into account conventional and extreme outputs. Background Art

[0002] With the continuous development of renewable energy technology, wind power generation and photovoltaic power generation have become important sources of electricity. However, wind and solar power generation are random and intermittent, and are greatly affected by climate and environmental conditions. For example, changes in wind speed and light intensity can cause fluctuations in wind and photovoltaic power generation, which in turn affects the stability of the power system. Therefore, how to accurately predict and simulate the power output of wind and solar power generation, especially under normal and extreme output conditions, has become an important topic in power system research.

[0003] At present, there are many methods for generating wind and solar power scenarios, including statistical methods, physical modeling methods, and machine learning-based methods. These methods can simulate the power output of wind and solar power generation to a certain extent, but they are often insufficient when dealing with output conditions under extreme climate conditions. For example, statistical methods rely on historical data, and when encountering extreme conditions that have never occurred in history, the prediction accuracy will drop significantly; although physical modeling methods can consider more physical factors, the model is complex and the amount of calculation is large; machine learning methods require a large amount of high-quality data for training, and in the absence of data, the effect is poor and lacks a theoretical basis.

[0004] The safe and stable operation of the power system places higher demands on the accuracy of power forecasting. In particular, with the increasing proportion of wind and solar power generation, traditional load forecasting methods can no longer meet the demand. For application scenarios such as power dispatching, load management, and new energy grid connection, it is of great practical significance to be able to generate wind and solar power scenarios covering both conventional and extreme output conditions. This can not only improve the dispatching efficiency of the power system, but also effectively prevent power system risks caused by extreme weather.

[0005] Given that traditional methods lack consideration for extreme wind and solar power output, or lack a theoretical basis to support them, it is necessary to further develop wind and solar power scenario generation methods that can comprehensively consider conventional and extreme output conditions, while ensuring the accuracy of the scenario and maximizing the interpretability of the scenario generation method.

[0006] For example, Chinese patent CN202010340376.2 proposes a new energy scenario generation method based on an adversarial learning model, which can overcome the problems of traditional methods of generating single output parameters, inaccurate results, and the need for a large number of simplified assumptions, and optimize the new energy output parameters. However, this method does not consider extreme scenarios, and the use of artificial intelligence methods lacks interpretability.

[0007] Another example is Chinese patent CN202311146819.4, which discloses a new energy scene generation method based on normalized flow. This method makes full use of the deep structure of the normalized flow model, so that the generated scene can more accurately approximate the distribution of actual observation data, improving the realism of the generated scene. However, this method also does not consider extreme scenarios and lacks interpretability.

[0008] For example, Chinese patent CN202311369375.0 proposes a method for generating extreme wind power scenarios based on a conditional generative adversarial network, which uses a conditional generative adversarial network to generate scenarios under extreme conditions, making the generated scenarios more comprehensive. However, this method is still based on artificial intelligence and lacks interpretability. Summary of the invention

[0009] In view of the deficiencies in the prior art, the purpose of the present invention is to provide a method and system for generating wind and solar power scenarios taking into account conventional and extreme outputs, thereby solving the problems in the prior art.

[0010] The purpose of the present invention can be achieved through the following technical solutions:

[0011] A method for generating a wind and solar power scenario taking into account normal and extreme outputs comprises the following steps:

[0012] Obtain the original data of wind and solar power output and reconstruct it to obtain weekly and daily scene data;

[0013] The soft-DTW based K-medoids algorithm is used to cluster the weekly scene data to generate a conventional weekly wind and solar output scene;

[0014] The LOF algorithm is used to detect outliers on daily scene data to generate extreme daily wind and solar output scenes;

[0015] The extreme daily wind-solar output scenario is used to replace some days in the conventional weekly wind-solar output scenario to generate a wind-solar power scenario that takes into account conventional and extreme outputs.

[0016] Furthermore, the steps of generating a conventional weekly wind and solar output scenario include:

[0017] S21, serially connecting the wind and solar power output data at the same time section in the weekly scene data to generate a series of bi-weekly scene data;

[0018] S22, uses the soft-DTW method to measure the distance between two-way scene data;

[0019] S23, the distance calculated by the soft-DTW method replaces the Euclidean distance in the traditional K-medoids algorithm, clusters the bi-weekly scene data, and generates a conventional weekly wind and solar output scene.

[0020] Furthermore, the calculation formula for the distance between the two-week scene data is:

[0021]

[0022]

[0023] in, is the cumulative distance matrix of the soft-DTW algorithm, For point i and Point j The distance between i and j are the indexes of the two time series respectively. is a hyperparameter used to control the smoothness of the distance matrix. n is the total number of time series, a i For the i Time series.

[0024] Furthermore, the steps of generating an extreme day-level wind and solar output scenario include:

[0025] S31, serially connecting the scenery and solar data in the same time section in the daily scene data to generate a series of double-day scene data;

[0026] S32, using the LOF algorithm to calculate the local outlier factors of a series of double-day-level scene data;

[0027] S33, setting a threshold, and screening out scenes where the local outlier factor exceeds the threshold as extreme day-level wind and solar output scenes.

[0028] Furthermore, the calculation formula of the local outlier factor is:

[0029]

[0030] in, For point P No. k The local outlier factor of the distance, For point P The k-th distance neighbor of For point P No. k All points in the neighborhood P The average reachable distance, For point O No.k All points in the neighborhood O The average reachable distance, P and O All are sample points.

[0031] Further, in S33, set the k In the neighborhood k The local outlier factor of each scene is 48, and the top 15 scenes with the largest local outlier factors are selected as the extreme day-level wind and solar output scenes, and the local outlier factor corresponding to the 15th scene is the threshold.

[0032] A wind and solar power scenario generation system taking into account normal and extreme outputs, comprising:

[0033] Data reconstruction module: obtains the original data of wind and solar power output and reconstructs it to obtain weekly and daily scene data;

[0034] Conventional scenario generation module: a soft-DTW based K-medoids algorithm is used to cluster the weekly scenario data to generate conventional weekly wind and solar output scenarios;

[0035] Extreme scenario generation module: uses the LOF algorithm to detect outliers on daily scenario data and generate extreme daily wind and solar output scenarios;

[0036] And, a scene fusion module: replaces some days in the conventional weekly wind and solar output scene with the extreme daily wind and solar output scene to generate a wind and solar power scene that takes into account conventional and extreme outputs.

[0037] A computer storage medium stores a readable program, which can execute the above-mentioned method for generating a wind and solar power scene taking into account conventional and extreme outputs when the program is running.

[0038] An electronic device, comprising: a processor, a memory, a communication interface and a communication bus, wherein the processor, the memory and the communication interface communicate with each other through the communication bus;

[0039] The memory is used to store at least one executable instruction, and the executable instruction enables the processor to perform operations corresponding to the above-mentioned method for generating a wind and solar power scenario taking into account conventional and extreme outputs.

[0040] A computer program product includes computer instructions, wherein the computer instructions instruct a computing device to perform operations corresponding to the above-mentioned method for generating a wind-solar power scenario taking into account conventional and extreme outputs.

[0041] Beneficial effects of the present invention:

[0042] 1. This paper proposes a conventional scene generation method based on the soft-DTW based K-medoids algorithm, which effectively improves the universality and authenticity of conventional scene generation;

[0043] 2. The present invention proposes an extreme scene generation method based on the LOF algorithm, which effectively ensures the theoretical and interpretability of extreme scene generation;

[0044] 3. The present invention can generate wind and solar power scenarios with conventional and extreme outputs, taking into account both the conventional and extreme characteristics of wind and solar outputs. The method is simple, practical, and easy to expand, providing scenario support for the optimization decision-making of the power system. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0046] Figure 1 is a flow chart of a method for generating wind and solar power scenarios taking into account conventional and extreme outputs of the present invention;

[0047] Figure 2 It is a flow chart of the conventional wind and solar power scenario generation method;

[0048] Figure 3 It is a flow chart of the method for generating extreme wind and solar power scenarios;

[0049] Figure 4 This is a conventional wind and solar power scenario generation diagram;

[0050] Figure 5 This is the map generated by the extreme wind and solar power scenario. DETAILED DESCRIPTION

[0051] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0052] Example 1

[0053] like Figure 1 As shown, a method for generating a wind and solar power scenario taking into account normal and extreme outputs includes the following steps:

[0054] S1, obtain the original data of wind and solar power output and reconstruct it to obtain weekly scene data and daily scene data;

[0055] The raw data of wind and solar power output can be obtained by predicting meteorological factors, such as calculating photovoltaic output through historical data such as solar radiation and temperature, and calculating wind turbine output through historical data such as wind speed and wind direction.

[0056] The acquired raw data of wind and solar power output are split and reconstructed at weekly and daily levels respectively, so as to facilitate the subsequent generation of conventional and extreme scenarios.

[0057] S2, using the soft-DTW based K-medoids algorithm to cluster the weekly scene data and generate conventional weekly wind and solar output scenarios;

[0058] Since conventional scenarios need to characterize the universal characteristics of wind and solar output, and need to ensure the temporal consistency between wind and solar output, the use of conventional K-means and K-medoids algorithms will ignore the authenticity of the original data; therefore, a method that can fully characterize the universality and authenticity of wind and solar data is needed; such as Figure 2 As shown, the present invention proposes a conventional weekly wind and solar output scene generation method based on a soft-DTWbased K-medoids algorithm, which specifically includes the following steps:

[0059] S21, serially connecting the wind and solar power output data at the same time section in the weekly scene data to generate a series of bi-weekly scene data;

[0060] In order to ensure the temporal consistency of wind and solar data, it is necessary to connect the wind and solar data in the same time section in series to construct a series of biweekly data containing wind and solar data in the same time section.

[0061] S22, uses the soft-DTW method to measure the distance between two-way scene data;

[0062] The soft-DTW method uses a smoother DTW distance on the traditional DTW distance to measure the similarity between time series to enhance the robustness of the algorithm and its adaptability to noisy data. The specific calculation method is:

[0063]

[0064]

[0065] in, is the cumulative distance matrix of the soft-DTW algorithm, For point i and Point j The distance betweeni and j are the indexes of the two time series respectively. is a hyperparameter used to control the smoothness of the distance matrix. n is the total number of time series, a i For the i Time series;

[0066] S23, the distance calculated by the soft-DTW method replaces the Euclidean distance in the traditional K-medoids algorithm, clusters the bi-weekly scene data, and generates a conventional weekly wind and solar output scene;

[0067] The traditional K-medoids algorithm uses Euclidean distance to measure the similarity between sample points. However, this is not a good clustering effect for time series data. The distance calculated by the soft-DTW method is used instead of the Euclidean distance for clustering, and the number of clusters is specified as 1. The best cluster center is generated through iteration, which is a regular weekly scenario.

[0068] S3, uses the LOF algorithm to detect outliers on the daily scene data and generate extreme daily wind and solar output scenes;

[0069] Because the proportion of wind and solar extreme output scenarios is relatively low in all output scenarios, but the impact is relatively large, directly selecting scenes with low wind and solar output cannot guarantee the "extremeness" of the scenes, and this method is not feasible when the number of scenes is large; while the use of artificial intelligence methods depends on the selection of parameters, and lacks theoretical basis and generalization. Therefore, an extreme scene generation algorithm with a theoretical basis is needed. Figure 3 As shown, the present invention proposes a method for generating an extreme day-level wind and solar output scene based on the LOF algorithm, which specifically includes the following steps:

[0070] S31, serially connecting the scenery and solar data in the same time section in the daily scene data to generate a series of double-day scene data;

[0071] In order to ensure the temporal consistency of wind and solar data, it is necessary to connect the wind and solar data in the same time section in series to construct a series of double-day scene data containing wind and solar data in the same time section.

[0072] S32, using the LOF algorithm to calculate the local outlier factors of a series of double-day-level scene data;

[0073] The LOF algorithm is a density-based outlier detection method. Its main idea is to assign a discrete degree to each data point, use a threshold to define the distance, and determine whether it is an outlier. The calculation method of the local outlier factor is:

[0074]

[0075] in, For point P No. k The local outlier factor of the distance, For point P The k-th distance neighbor of For point P No. k All points in the neighborhood P The average reachable distance, For point O No. k All points in the neighborhood O The average reachable distance, P and O All are sample points.

[0076] S33, setting a threshold, screening out scenes where the local outlier factor exceeds the threshold as extreme day-level wind and solar output scenes;

[0077] Considering the massive data of the double-day scenery scene, the k In the neighborhood k The value is 48, and the top 15 scenes with the largest local outlier factors of each scene are selected as extreme scenes. The local outlier factor corresponding to the 15th scene is used as the threshold, and the 15 scenes are selected according to the number of extreme scenes actually required.

[0078] S4, replaces some days in the conventional weekly wind-solar output scenario generated by S2 with the extreme daily wind-solar output scenario generated by S3, and constructs a wind-solar power scenario that takes into account both conventional and extreme outputs.

[0079] The selected conventional weekly wind and solar power output scenarios and the required number of extreme daily wind and solar power output scenarios require reverse reconstruction of wind and solar data. After reverse reconstruction, they can be split back into two types of wind and solar power scenarios, each of which includes conventional and extreme output scenarios.

[0080] Example 2

[0081] In this embodiment, specific examples of a conventional weekly wind-solar output scenario and an extreme daily wind-solar output scenario are respectively given to verify the effectiveness of the scenario generation method proposed in the present invention.

[0082] First, the 87,600-hour wind and solar power data for 10 years from 2010 to 2020 are obtained through the meteorological simulation platform renewables.ninja. Secondly, the 87,600-hour data is reconstructed into weekly and daily data, and the time series consistency between wind and solar power is maintained to construct the corresponding bi-weekly and bi-daily wind and solar power scenarios. Thirdly, the number of clusters is specified as 1, and the soft-DTW based K-medoids algorithm is used to generate typical weekly wind and solar output scenarios for the bi-weekly data, such as Figure 4 Finally, set the k In the neighborhood k The local outlier factor is 48. The LOF algorithm is used to calculate the local outlier factor for the double-day data. The top 15 scenarios with the largest local outlier factors are selected as extreme scenarios, such as Figure 5 shown.

[0083] Depend on Figure 4 It can be seen that the conventional weekly photovoltaic output scenario has typical periodic characteristics, with a one-day cycle and similar change trends every day, while the conventional weekly wind power output scenario does not have periodic characteristics, reflecting the uncertainty of wind power output. It also shows the effectiveness of the proposed conventional weekly wind and solar output scenario generation method; Figure 5 It can be found that extreme day-level wind-solar output scenarios are mainly divided into two categories: extreme high output and extreme low output. They can be freely selected according to the required extreme scene type, highlighting the effectiveness and interpretability of the proposed extreme day-level wind-solar output scenario generation method.

[0084] The scene generation method of the present invention may be implemented in hardware, firmware, or as software or computer code that can be stored in a recording medium (such as a CD ROM, RAM, floppy disk, hard disk, or magneto-optical disk), or as computer code that is originally stored in a remote recording medium or a non-temporary machine-readable medium downloaded over a network and will be stored in a local recording medium, so that the method described herein can be stored in such software processing on a recording medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware (such as an ASIC or FPGA). It is understood that a computer, a processor, a microprocessor controller, or programmable hardware includes a storage component (e.g., RAM, ROM, flash memory, etc.) that can store or receive software or computer code, and when the software or computer code is accessed and executed by a computer, a processor, or hardware, the method described herein is implemented. In addition, when a general-purpose computer accesses the code for implementing the method shown herein, the execution of the code converts the general-purpose computer into a dedicated computer for executing the method shown herein.

[0085] The above shows and describes the basic principles, main features and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments, and the above embodiments and descriptions are only for explaining the principles of the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, and these changes and improvements all fall within the scope of the present invention to be protected.

Claims

1. A method for generating wind and solar power scenarios taking into account normal and extreme outputs, characterized in that: The following steps are involved: Obtain the original data of wind and solar power output and reconstruct it to obtain weekly and daily scene data; The soft-DTW based K-medoids algorithm is used to cluster the weekly scene data to generate a conventional weekly wind and solar output scene; The LOF algorithm is used to detect outliers on daily scene data to generate extreme daily wind and solar output scenes; Substituting the extreme daily wind-solar output scenario for some days in the conventional weekly wind-solar output scenario to generate a wind-solar power scenario taking into account conventional and extreme outputs; The steps to generate a conventional weekly wind and solar output scenario include: S21, serially connecting the wind and solar power output data at the same time section in the weekly scene data to generate a series of bi-weekly scene data; S22, uses the soft-DTW method to measure the distance between two-way scene data; S23, the distance calculated by the soft-DTW method replaces the Euclidean distance in the traditional K-medoids algorithm, clusters the bi-weekly scene data, and generates a conventional weekly wind and solar output scene; The calculation formula for the distance between the two-week scene data is: in, is the cumulative distance matrix of the soft-DTW algorithm, For point i and Point j The distance between i and j are the indexes of the two time series respectively. is a hyperparameter used to control the smoothness of the distance matrix. n is the total number of time series, a i For the i Time series.

2. A method for generating wind and solar power scenarios taking into account normal and extreme outputs according to claim 1, characterized in that: The steps to generate an extreme day-level wind and solar output scenario include: S31, serially connecting the scenery and solar data in the same time section in the daily scene data to generate a series of double-day scene data; S32, using the LOF algorithm to calculate the local outlier factors of a series of double-day-level scene data; S33, setting a threshold, and screening out scenes where the local outlier factor exceeds the threshold as extreme day-level wind and solar output scenes.

3. A method for generating wind and solar power scenarios taking into account normal and extreme outputs according to claim 2, characterized in that: The calculation formula of the local outlier factor is: in, For point P No. k The local outlier factor of the distance, For point P The k-th distance neighbor of For point P No. k All points in the neighborhood P The average reachable distance, For point O No. k All points in the neighborhood O The average reachable distance, P and O All are sample points.

4. A method for generating wind and solar power scenarios taking into account normal and extreme outputs according to claim 2, characterized in that: In S33, set k Neighborhood k The local outlier factor of each scene is 48, and the top 15 scenes with the largest local outlier factors are selected as the extreme day-level wind and solar output scenes, and the local outlier factor corresponding to the 15th scene is the threshold.

5. A wind and solar power scenario generation system taking into account normal and extreme outputs, characterized in that: include: Data reconstruction module: obtains the original data of wind and solar power output and reconstructs it to obtain weekly and daily scene data; Conventional scenario generation module: a soft-DTW based K-medoids algorithm is used to cluster the weekly scenario data to generate conventional weekly wind and solar output scenarios; Extreme scenario generation module: uses the LOF algorithm to detect outliers on daily scenario data and generate extreme daily wind and solar output scenarios; And, a scene fusion module: replacing some days in the conventional weekly wind-solar output scene with the extreme daily wind-solar output scene to generate a wind-solar power scene taking into account conventional and extreme outputs; The steps to generate a conventional weekly wind and solar output scenario include: S21, serially connecting the wind and solar power output data at the same time section in the weekly scene data to generate a series of bi-weekly scene data; S22, uses the soft-DTW method to measure the distance between two-way scene data; S23, the distance calculated by the soft-DTW method replaces the Euclidean distance in the traditional K-medoids algorithm, clusters the bi-weekly scene data, and generates a conventional weekly wind and solar output scene; The calculation formula for the distance between the two-week scene data is: in, is the cumulative distance matrix of the soft-DTW algorithm, For point i and Point j The distance between i and j are the indexes of the two time series respectively. is a hyperparameter used to control the smoothness of the distance matrix. n is the total number of time series, a i For the i Time series.

6. A computer storage medium storing a readable program, characterized in that: When the program is running, it can execute the method for generating a wind and solar power scenario taking into account normal and extreme outputs as described in any one of claims 1 to 4.

7. An electronic device, characterized in that: include: A processor, a memory, a communication interface and a communication bus, wherein the processor, the memory and the communication interface communicate with each other via the communication bus; The memory is used to store at least one executable instruction, and the executable instruction enables the processor to perform operations corresponding to a method for generating a wind and solar power scenario taking into account conventional and extreme outputs as described in any one of claims 1-4.

8. A computer program product, characterized in that The method comprises computer instructions, wherein the computer instructions instruct a computing device to perform operations corresponding to a method for generating a wind / solar power scenario taking into account normal and extreme output as described in any one of claims 1 to 4.