New energy power generation power prediction method and system based on artificial intelligence

Abstract

The application provides a new energy power generation power prediction method and system based on artificial intelligence, and the method comprises the following steps: based on the installed capacity data of each new energy power station type in each new energy power cluster, and combined with historical power supply operation data, an initial feed-in power curve corresponding to different new energy power station types is determined; based on the geographical position distribution data of each new energy power cluster, a collaborative correlation analysis is performed to obtain a power collaborative complementary relationship between each new energy power cluster; based on the initial feed-in power curve corresponding to different new energy power station types and the power collaborative complementary relationship between each new energy power cluster, a cluster collaborative feed-in power curve of each new energy power cluster is determined; and based on the cluster collaborative feed-in power curve of each new energy power cluster and power grid load prediction data of a region within a future preset time length, a new energy power generation power prediction result is determined. The application improves the new energy consumption capacity and power grid operation stability.

Description

Technical Field

[0001] This invention relates to the field of computer technology, and in particular to a method and system for predicting the power generation of new energy sources based on artificial intelligence. Background Technology

[0002] In the field of renewable energy power generation, power generation forecasting is a crucial link in ensuring the stable operation of the power grid and improving the renewable energy absorption capacity. Currently, the mainstream renewable energy power generation forecasting method is the single-station independent forecasting method. This method takes a single renewable energy power station as an independent forecasting unit, and based on the historical power generation data of the power station and local meteorological data monitored within the station (such as wind speed and solar intensity), it completes the forecasting of the power generation of a single power station for a future period of time through preset forecasting rules.

[0003] However, due to the highly volatile and intermittent nature of renewable energy generation, the power output of a single power station is easily affected by factors such as sudden local weather changes and instantaneous equipment failures. Single-station independent forecasting methods rely solely on local data from a single power station, failing to consider the synergistic relationships between different renewable energy power stations and the complementary characteristics of different types of power sources (such as wind power, photovoltaics, and energy storage) within a region. This results in poor accuracy and stability of the forecast results. When the predicted power output deviates significantly from the actual power output, the grid dispatch center cannot formulate reasonable dispatching plans based on accurate forecasts, leading to problems such as excess renewable energy power that cannot be absorbed or insufficient power supply, affecting the renewable energy absorption capacity and the stability of grid operation. Summary of the Invention

[0004] This invention provides a new energy power generation prediction method and system based on artificial intelligence, aiming to improve the new energy absorption capacity and grid operation stability.

[0005] In a first aspect, the present invention provides a new energy power generation prediction method based on artificial intelligence, comprising: For each region to be predicted, based on the installed capacity data of each type of new energy power station within each new energy power generation cluster, and combined with the historical power operation data of the region, the initial feed-in power curve corresponding to different types of new energy power stations within each new energy power generation cluster is determined. Based on the geographical location distribution data of each new energy power generation cluster, a collaborative correlation analysis is conducted to obtain the power synergy and complementarity relationship between the various new energy power generation clusters. Based on the initial feed-in power curves corresponding to different types of new energy power plants within each new energy power generation cluster, and combined with the power synergy and complementarity relationship between each new energy power generation cluster, the cluster synergy feed-in power curves of each new energy power generation cluster are determined. Based on the cluster collaborative feed-in power curves of various new energy power generation clusters and the grid load forecast data of the region within a preset time period, the new energy power generation forecast results for each region to be predicted are determined.

[0006] Optionally, the steps for determining the predicted renewable energy power generation for each region to be predicted include: Based on the time-period load demand in the power grid load forecast data, the minimum operating load baseline sequence of the power grid in the region at a preset time resolution is determined; the minimum operating load baseline sequence of the power grid represents the minimum power supply level at which the regional power grid can maintain safe and stable operation at each time point. Based on the actual feed-in power value of each cluster at each time point, as represented by the cluster collaborative feed-in power curve of each new energy power generation cluster, the total feed-in power sequence of new energy in the region under the same time resolution is determined; the total feed-in power sequence of energy represents the physical maximum output capacity of the new energy power generation system at each time point. Based on the minimum operating load baseline sequence of the power grid and the total feed-in power sequence of new energy sources, the prediction results of new energy power generation for each region to be predicted are determined.

[0007] Optionally, determining the renewable energy power generation forecast result for each region to be predicted based on the minimum operating load baseline sequence of the power grid and the total feed-in power sequence of renewable energy includes: Based on the minimum operating load baseline sequence of the power grid and the total feed-in power sequence of new energy sources, determine the feasibility assessment result of new energy feed-in in the region at each time point; Based on the feasibility assessment results of the new energy feed-in, it is determined whether the new energy power generation is within the acceptable operating range of the power grid at the corresponding time point, and the scheduling responsibility period within the future preset time period is obtained. Based on the total renewable energy feed-in power sequence and the minimum operating load baseline sequence of the power grid during each scheduling responsibility period, the feed-in power sub-sequence for each scheduling responsibility period is determined. By splicing and integrating the feed-in power subsequences of each scheduling responsibility period in chronological order, the predicted power generation of new energy in each region to be predicted within a preset time period is obtained.

[0008] Optionally, the scheduling responsibility period includes a full acceptance period, a limited acceptance period, and an unacceptable period; the full acceptance period indicates that the total renewable energy feed-in power does not exceed the grid load demand and is not lower than the minimum operating load baseline; the limited acceptance period indicates that the total renewable energy feed-in power exceeds the grid load demand and there is a risk of exceeding the limit upwards; the unacceptable period indicates that the total renewable energy feed-in power is lower than the grid minimum operating load baseline and there is a risk of insufficient downward support.

[0009] Optionally, based on the total renewable energy feed-in power sequence within each scheduling responsibility period, the feed-in power subsequence for each scheduling responsibility period is determined, including: For periods when full acceptance is possible, the power input subsequence of renewable energy generation during the periods when full acceptance is possible is determined based on the power value of the total renewable energy input power sequence during the periods when full acceptance is possible.

[0010] Optionally, based on the total renewable energy feed-in power sequence and the minimum grid operating load baseline sequence for each scheduling responsibility period, the feed-in power sub-sequence for each scheduling responsibility period is determined, including: For the restricted access period, the maximum allowable feed-in power of new energy generation during the restricted access period is determined based on the upper limit of load demand during the restricted access period, which is based on the minimum operating load baseline sequence of the power grid. Based on the maximum allowable feed-in power limit, the power value of the total feed-in power sequence of the new energy sources during the restricted acceptance period is restricted, thus obtaining the feed-in power subsequence of new energy power generation during the restricted acceptance period.

[0011] Optionally, based on the total renewable energy feed-in power sequence and the minimum grid operating load baseline sequence for each scheduling responsibility period, the feed-in power sub-sequence for each scheduling responsibility period is determined, including: For unacceptable periods, the minimum guaranteed feed-in power limit for new energy power generation is determined based on the preset system safety lower limit in the minimum operating load baseline sequence of the power grid during the unacceptable periods; the minimum guaranteed feed-in power limit is equal to the minimum operating load baseline value of the power grid for the corresponding period. The power value of the total feed-in power sequence of the new energy sources during the unacceptable period is increased to no less than the minimum guaranteed feed-in power limit, thus obtaining the feed-in power subsequence of the new energy generation during the unacceptable period.

[0012] Optionally, the steps for determining the cluster-coordinated feed-in power curves of each new energy power generation cluster include: Based on the time-series power change characteristics of the initial feed-in power curves corresponding to different types of new energy power plants within each new energy power generation cluster, a multi-type power source time-series feed-in behavior sequence is determined for each new energy power generation cluster; the multi-type power source time-series feed-in behavior sequence characterizes the power output fluctuation mode of different types of new energy power plants within each new energy power generation cluster and their phase distribution characteristics in the time dimension. Based on the power synergy and complementarity relationship between various new energy power generation clusters, the trend relationship of power output trend between each pair of new energy power generation clusters at the same point in time is determined. Based on the trend relationship, determine whether there is power output complementarity potential between two new energy power generation clusters at any point in time, and obtain the complementarity potential results of each pair of new energy power generation clusters at any point in time. Based on the complementary potential results of each pair of new energy power generation clusters and the time-series feed-in behavior sequence of multiple types of power sources in each new energy power generation cluster, the cluster collaborative feed-in power curve of each new energy power generation cluster is determined.

[0013] Optionally, the step of determining the cluster collaborative feed-in power curve of each new energy power generation cluster based on the complementary potential results of each pair of new energy power generation clusters and the multi-type power source timing feed-in behavior sequence of each new energy power generation cluster includes: Based on the time-series feed-in behavior sequences of various new energy power generation clusters and the complementary potential results of each pair of new energy power generation clusters, power synergy cluster combinations with conditions for coordinated operation in the time dimension are determined; each power synergy cluster combination includes two or more new energy power generation clusters with conditions for coordinated operation in the time dimension, and any two new energy power generation clusters have complementary power output potential in at least one time period. Based on the time-series feed-in behavior sequence of multiple types of power sources in each new energy power generation cluster within each power coordination cluster combination, the joint power fluctuation profile at each time point is determined. Based on the joint power fluctuation profile of each power coordination cluster combination, the coordination fluctuation characteristics used to characterize the power output stability under the combined action of multiple new energy power generation clusters within the cluster combination are determined. Based on the complementary structural relationship and cooperative fluctuation characteristics among the various new energy power generation clusters within each power cooperative cluster combination, the cluster cooperative feed-in power curve of each new energy power generation cluster is determined.

[0014] Optionally, determining the cluster collaborative feed-in power curve of each new energy power generation cluster based on the complementary structural relationship and collaborative fluctuation characteristics among the various new energy power generation clusters within each power collaborative cluster combination includes: For each power synergy cluster combination, based on the synergy fluctuation characteristics and the complementary structural relationship between each new energy power generation cluster, the power synergy role of each new energy power generation cluster in the power synergy cluster combination is determined; the power synergy role includes power-dominant clusters and power-auxiliary clusters. Based on the time-series feed-in behavior sequence of various new energy power generation clusters, the power supply period undertaken by the power-dominant cluster in the power-coordinated cluster combination is determined, and the compensation time window for the power-assisted cluster to perform power compensation outside the power supply period is determined based on the power supply period. Based on the compensation time window and the joint power fluctuation profile, the power adjustment target range of the power auxiliary cluster within its respective compensation time window is determined, and the power value of the corresponding time period in the original multi-type power supply timing feed behavior sequence of the power auxiliary cluster is adjusted based on the power adjustment target range to obtain the adjusted multi-type power supply timing feed behavior sequence. By integrating the original multi-type power source timing behavior sequences of the power-dominant cluster and the adjusted multi-type power source timing behavior sequences of the power-assisted cluster, the cluster collaborative feed-in power curves of each new energy power generation cluster are obtained.

[0015] Secondly, the present invention also provides an artificial intelligence-based new energy power generation prediction system for implementing the artificial intelligence-based new energy power generation prediction method as described in the first aspect; the artificial intelligence-based new energy power generation prediction system includes: The power curve prediction module is used to determine the initial feed-in power curve corresponding to different types of new energy power plants in each new energy power generation cluster for each region to be predicted, based on the installed capacity data of each type of new energy power plant in each new energy power generation cluster and combined with the historical power operation data in the region. The collaborative correlation analysis module is used to perform collaborative correlation analysis based on the geographical location distribution data of each new energy power generation cluster to obtain the power synergy and complementarity relationship between each new energy power generation cluster. The power curve optimization module is used to determine the cluster collaborative feed-in power curve of each new energy power generation cluster based on the initial feed-in power curves corresponding to different types of new energy power plants within each new energy power generation cluster and the power synergy and complementarity relationship between each new energy power generation cluster. The power generation prediction module is used to determine the power generation prediction result of each region to be predicted based on the cluster collaborative feed-in power curve of each new energy power generation cluster and the grid load prediction data of the region within a preset time period.

[0016] Thirdly, the present invention also provides an electronic device, comprising: a memory for storing computer software programs; and a processor for reading and executing the computer software programs, thereby realizing the artificial intelligence-based new energy power generation prediction method as described above.

[0017] Fourthly, the present invention also provides a non-transitory computer-readable storage medium storing a computer software program, which, when executed by a processor, implements the artificial intelligence-based new energy power generation prediction method as described above.

[0018] Fifthly, the present invention also provides a computer program product, including a computer program that, when executed by a processor, implements the artificial intelligence-based new energy power generation prediction method as described above.

[0019] The artificial intelligence-based new energy power generation prediction method provided in this invention overcomes the limitation of single-station independent prediction relying solely on local data from a single power station by using initial feed-in power curves corresponding to different types of new energy power stations within each new energy power generation cluster. By integrating installed capacity data of similar power stations within the cluster with historical power operation data of the region, the power variation patterns of different power types within the cluster are clarified. Based on the geographical distribution data of each cluster, the power synergy and complementarity relationship between different new energy power generation clusters is obtained, clarifying the power correlation characteristics between different clusters and compensating for the problem of not considering inter-cluster synergy and complementarity. Based on the initial feed-in power curve and the power synergy and complementarity relationship, the cluster synergy feed-in power curve of each new energy power generation cluster is determined. The initial feed-in power curve is corrected using the power complementarity between clusters. Simultaneously, the power characteristics of different power types within the cluster are integrated, reducing the impact of power fluctuations from a single power source or a single cluster, and solving the problem of poor prediction stability caused by the lack of consideration for synergy and complementarity in single-station independent prediction methods. Based on the cluster collaborative feed-in power curve and the grid load forecast data within the preset time period of the region, the forecast results of renewable energy power generation in each region to be predicted are determined. This makes the renewable energy power generation forecast results integrate the characteristics of cluster collaboration and power complementarity. Therefore, the grid dispatch center can formulate dispatch plans based on the accurate renewable energy power generation forecast results, avoid the situation of renewable energy power surplus that cannot be absorbed or insufficient supply, and improve the renewable energy absorption capacity and grid operation stability. Attached Figure Description

[0020] Figure 1 This is a flowchart illustrating the artificial intelligence-based new energy power generation prediction method provided in an embodiment of the present invention. Figure 2 This is a schematic diagram of the structure of the artificial intelligence-based new energy power generation prediction system provided in an embodiment of the present invention; Figure 3 An embodiment diagram of the electronic device provided in this invention; Figure 4 An embodiment diagram of a computer-readable storage medium provided in accordance with the present invention. Detailed Implementation

[0021] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0022] In the description of this invention, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the stated features. In the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0023] In the description of this invention, the term "for example" is used to mean "used as an example, illustration, or description." Any embodiment described as "for example" in this invention is not necessarily to be construed as being more preferred or advantageous than other embodiments. The following description is provided to enable any person skilled in the art to make and use the invention. Details are set forth in the following description for purposes of explanation. It should be understood that those skilled in the art will recognize that the invention can be made without using these specific details. In other instances, well-known structures and processes will not be described in detail to avoid obscuring the description of the invention with unnecessary detail. Therefore, the invention is not intended to be limited to the embodiments shown, but is consistent with the broadest scope of the principles and features disclosed herein.

[0024] See Figure 1 , Figure 1 This is a flowchart illustrating the artificial intelligence-based new energy power generation prediction method provided by the present invention. In this embodiment, the execution entity of the artificial intelligence-based new energy power generation prediction method is a power prediction system. Therefore, the artificial intelligence-based new energy power generation prediction method includes: Step 10: For each region to be predicted, based on the installed capacity data of each type of new energy power station within each new energy power generation cluster, and combined with the historical power operation data of the region, determine the initial feed-in power curve corresponding to different types of new energy power stations within each new energy power generation cluster.

[0025] Optionally, for each region to be predicted, the power prediction system acquires basic data for each new energy power generation cluster within that region. The basic data specifically includes the installed capacity data of each type of new energy power station within each new energy power generation cluster, as well as historical power operation data for that region.

[0026] Among them, a new energy power generation cluster refers to a collection of power generation facilities composed of multiple new energy power plants of the same or different types, which are geographically concentrated or connected to the power grid. New energy power plant type refers to the type of power plant classified based on its energy source, including but not limited to photovoltaic power plants, wind power plants, hydropower plants (renewable hydropower within the new energy category), and biomass power plants. Installed capacity data refers to the maximum rated power value that a new energy power plant can continuously and stably output during its design and construction, expressed in kilowatts or megawatts. Historical power source operation data refers to the actual operation records of all new energy power plants and other types of power sources (if any) within the area to be predicted over a continuous period of time in the past. This includes at least time-series power generation data, time-series environmental data (such as environmental parameters related to new energy power generation, such as sunlight intensity, wind speed, and precipitation), and time-series equipment operation status data (such as whether it is operating normally or under maintenance). Time-series refers to continuous arrangement in chronological order, and the time interval can be set according to actual prediction needs, such as 15 minutes, 30 minutes, or 1 hour.

[0027] Furthermore, the power prediction system performs preprocessing operations on the acquired historical power source operation data. These preprocessing operations include data cleaning, data standardization, and data filtering. Data cleaning refers to removing outliers (such as power spikes due to equipment failure, missing or invalid values ​​due to data acquisition errors) and duplicate values ​​from the historical power source operation data. For missing values, interpolation of data from adjacent time periods or substitution with data from similar power plants during the same period can be used. Data standardization involves converting historical power source operation data with different dimensions (such as kilowatts of installed capacity, meters per second of environmental parameters, etc.) into data of the same dimension or dimensionless data to ensure data comparability. Data filtering involves selecting relevant historical operation data corresponding to each type of new energy power generation cluster in the area to be predicted. For example, for photovoltaic power plants, historical operation data containing time-series data of solar irradiance and photovoltaic power generation is selected; for wind power plants, historical operation data containing time-series data of wind speed and wind power generation is selected.

[0028] Furthermore, the power prediction system establishes a correlation between installed capacity data and historical power source operation data. Specifically, for each type of new energy power station within each new energy power generation cluster, the installed capacity data of that type of power station is matched with the corresponding historical power source operation data after filtering, to clarify the actual power output pattern of that type of power station under different time-series environmental conditions and equipment operating states at different installed capacity levels.

[0029] Furthermore, based on the aforementioned correlations, the power prediction system determines the initial feed-in power curves corresponding to different types of new energy power plants within each new energy power generation cluster. The initial feed-in power curve refers to the power output curve that reflects the time-varying characteristics of a specific type of new energy power plant, predicted based on historical operating patterns over a predetermined future period. The time-series power variation characteristics refer to the trend of power value changes over time, including the time nodes of power peaks and troughs, the rate of change during power increases, the rate of change during power decreases, and the duration of stable power output. The time interval of this curve is consistent with the time interval of historical power source operating data, and the duration of the curve is consistent with the predetermined future prediction period.

[0030] In one embodiment, it is assumed that a certain area to be predicted contains two new energy power generation clusters, namely New Energy Power Generation Cluster A and New Energy Power Generation Cluster B. New Energy Power Generation Cluster A includes both photovoltaic (PV) and wind power plants, with an installed capacity of 50 MW for PV plants and 80 MW for wind power plants. New Energy Power Generation Cluster B only includes biomass power plants, with an installed capacity of 30 MW. The power prediction system acquires historical power generation data for the past three years for the area to be predicted, with a data interval of one hour. This data includes time-series power generation data for each new energy power plant, time-series environmental data (solar intensity data for PV plants, wind speed data for wind power plants, and raw material supply data for biomass power plants), and time-series equipment operating status data. The historical power source operation data was preprocessed as follows: zero power surge data that occurred during a certain equipment failure at the photovoltaic power station in cluster A was removed, and power generation data corresponding to the light intensity of adjacent 2 hours were used for interpolation and supplementation; data such as light intensity (lux), wind speed (m / s), and installed capacity (megawatt) were standardized to dimensionless data in the range of [0, 1]; and historical operation data corresponding to the photovoltaic power station, wind power station and biomass power station in cluster A and cluster B were selected respectively.

[0031] Establishing correlations: For photovoltaic power station A, its 50 MW installed capacity data was matched with the filtered time-series data of irradiance and photovoltaic power generation. It was found that when the irradiance is between 8000-10000 lux (normalized to 0.8-1.0), the actual power generation of the photovoltaic power station can reach 80%-90% of the installed capacity (i.e., 40-45 MW); when the irradiance is below 2000 lux (normalized to 0.2), the actual power generation is only 5%-10% of the installed capacity (i.e., 2.5-5 MW). Based on this correlation, and combined with the projected 72-hour time-series environmental trends (referencing historical patterns of solar irradiance variation), the initial feed-in power curve for the photovoltaic power station in Cluster A is generated. This curve has a 1-hour time interval and a duration of 72 hours. The time-series power variation characteristics are as follows: power is in a rising and stable phase from 6:00 AM to 6:00 PM daily, reaching a peak power (42-45 MW) from 12:00 PM to 2:00 PM, and then in a declining and trough phase from 6:00 PM to 6:00 AM the following day, with a trough power of 2.5-5 MW. Similarly, the initial feed-in power curves for the wind power station in Cluster A and the biomass power station in Cluster B are generated.

[0032] Step 20: Based on the geographical location distribution data of each new energy power generation cluster, perform collaborative correlation analysis to obtain the power synergy and complementarity relationship between each new energy power generation cluster.

[0033] Optionally, the power prediction system acquires the geographical location distribution data of each new energy power generation cluster. This geographical location distribution data refers to the specific latitude and longitude coordinates, altitude, topographic features (such as plains, mountains, and coastlines) of each new energy power generation cluster, as well as the spatial distance between each cluster and other new energy power generation clusters. Topographic features refer to the surface morphology of the area where the new energy power generation cluster is located. Different topographic features affect the environmental conditions for new energy power generation (e.g., the wind speed distribution differs between mountainous and coastal terrains where wind power generation clusters are located), thus affecting the power output pattern.

[0034] Furthermore, the power prediction system conducts a collaborative correlation analysis on various new energy power generation clusters based on geographical location distribution data. Collaborative correlation analysis refers to the process of analyzing the impact of the geographical location correlation of different new energy power generation clusters on their power output, and exploring the power synergy and complementarity relationships between the clusters. Specifically, based on the latitude and longitude coordinates and spatial distance in the geographical location distribution data, geographically adjacent new energy power generation clusters are grouped (clusters within 50 kilometers are typically grouped together, but this can be adjusted according to the actual area). Then, combined with the topographic features of each cluster, the correlation of environmental conditions in different clusters is determined (e.g., two adjacent wind power generation clusters, if both located in coastal plains, may have correlated wind speed patterns; if one is in mountainous terrain and the other in coastal areas, their wind speed patterns may differ). Finally, combining the historical power output time-series data of different power station types within each cluster obtained in step 10, the correlation index of the power output trends between the clusters at the same time point is calculated. The correlation index is used to quantify the degree of correlation between the power output trends of two clusters, with a value range of -1 to 1.

[0035] Based on the aforementioned correlation indicators, the power synergy and complementarity relationships among various new energy power generation clusters are determined. The power synergy and complementarity relationship refers to the complementary and synergistic relationship in the power output over time between different new energy power generation clusters due to differences in environmental conditions caused by their geographical distribution. This relationship is used to determine the trend relationship of power output between each pair of new energy power generation clusters at the same point in time. The trend relationship includes consistency and inverse relationships. A consistency relationship means that when the correlation index is greater than 0.6 (this threshold can be adjusted according to actual prediction accuracy requirements), the power output trends of two new energy power generation clusters are the same at the same point in time; that is, when the power of one cluster increases, the power of the other cluster also increases synchronously; when the power of one cluster decreases, the power of the other cluster also decreases synchronously. An inverse relationship means that when the correlation index is less than -0.6 (this threshold can be adjusted according to actual prediction accuracy requirements), the power output trends of two new energy power generation clusters are opposite at the same point in time; that is, when the power of one cluster increases, the power of the other cluster decreases; and when the power of one cluster decreases, the power of the other cluster increases. When the correlation index is between -0.6 and 0.6, it is considered that there is no obvious power synergy or complementarity between the two new energy power generation clusters.

[0036] In one embodiment, the geographical location data of the A new energy power generation cluster in the area to be predicted are as follows: latitude and longitude coordinates (118°E, 32°N), altitude 10 meters, topography is a coastal plain, and the spatial distance between it and the B new energy power generation cluster is 30 kilometers; the geographical location data of the B new energy power generation cluster are as follows: latitude and longitude coordinates (118.3°E, 32.2°N), altitude 15 meters, topography is a coastal plain. A collaborative correlation analysis is performed on the two new energy power generation clusters A and B: since their spatial distance is 30 kilometers (less than 50 kilometers), they are classified as geographically adjacent clusters; secondly, both are located in coastal plain terrain, indicating a strong correlation between their environmental conditions (such as wind speed and precipitation); next, the historical power generation time series data (1-hour time interval) of the past year for cluster A (including photovoltaic power stations and wind power stations) and cluster B (including biomass power stations) obtained in step 10 are extracted, and the correlation index of their power output trends at the same time point is calculated.

[0037] Calculations revealed that the correlation index between the historical power generation time-series data of wind power plants in cluster A and biomass power plants in cluster B is -0.75 (less than -0.6), indicating an inverse relationship. The correlation index between the historical power generation time-series data of photovoltaic power plants in cluster A and biomass power plants in cluster B is 0.2 (between -0.6 and 0.6), indicating no significant power synergy or complementarity. The correlation index between the historical power generation time-series data of photovoltaic power plants in cluster A and wind power plants is 0.8 (greater than 0.6), indicating a consistent relationship. Therefore, the power synergy and complementarity relationships among the various new energy power generation clusters are as follows: wind power plants in cluster A and biomass power plants in cluster B have an inverse relationship; photovoltaic power plants in cluster A and wind power plants have a consistent relationship; other cluster combinations show no significant power synergy or complementarity.

[0038] Step 30: Based on the initial feed-in power curves corresponding to different types of new energy power plants within each new energy power generation cluster and the power synergy and complementarity relationship between each new energy power generation cluster, determine the cluster synergy feed-in power curves of each new energy power generation cluster.

[0039] Optionally, the power prediction system uses the initial feed-in power curves corresponding to different types of new energy power plants within each new energy power generation cluster as basic data, and combines the power synergy and complementarity relationships between the various new energy power generation clusters to construct a cluster collaborative power correction model. Through the cluster collaborative power correction model, the initial feed-in power curve of each new energy power generation cluster is collaboratively corrected. During the correction process, the influence of power output trends between clusters with consistent or opposing relationships is fully considered, resulting in cluster collaborative feed-in power curves for each new energy power generation cluster that reflect the synergy effect between clusters, as described in steps 301 to 304. The cluster collaborative feed-in power curve refers to a power curve that integrates the power synergy and complementarity relationships between clusters and can more accurately reflect the overall power output time-series characteristics of the new energy power generation cluster.

[0040] Step 40: Based on the cluster collaborative feed-in power curves of each new energy power generation cluster and the grid load forecast data of the region within a preset time period in the future, determine the new energy power generation forecast result for each region to be predicted.

[0041] Optionally, the power prediction system acquires grid load prediction data for the region to be predicted within a preset future time period. The preset future time period refers to the length of time for which new energy power generation prediction needs to be performed, which can be set to 1 day, 3 days, 7 days, etc., depending on the actual application scenario. The grid load prediction data refers to the data obtained by predicting the electricity demand of all power users (including industrial users, residential users, commercial users, etc.) within the region to be predicted within the preset future time period. The time-period load demand refers to the grid load demand value within each time interval after dividing the preset future time period into fixed time intervals (consistent with the time interval of the initial feed-in power curve).

[0042] Furthermore, the power prediction system uses the cluster collaborative feed-in power curves of each new energy power generation cluster as its core basis, and constructs a power balance prediction model in conjunction with grid load prediction data. This power balance prediction model takes the ability of new energy power generation to meet grid load demand as its core objective. By analyzing the matching relationship between the total time-series power output of the cluster collaborative feed-in power curves and the time-period load demand in the grid load prediction data, the collaborative feed-in power of each cluster is adjusted to determine the new energy power generation prediction result for each region to be predicted, as detailed in steps 401 to 403.

[0043] This invention, based on the initial feed-in power curves corresponding to different types of new energy power plants within each new energy power generation cluster, overcomes the limitation of single-station independent prediction relying solely on local data from a single power plant. By integrating installed capacity data of similar power plants within the cluster with historical power operation data for the region, the power variation patterns of different power types within the cluster are clarified. Based on the geographical distribution data of each cluster, the power synergy and complementarity relationship between various new energy power generation clusters is obtained, clarifying the power correlation characteristics between different clusters and addressing the issue of neglecting inter-cluster synergy and correlation. Based on the initial feed-in power curves and power synergy and complementarity relationships, the cluster synergy feed-in power curves for each new energy power generation cluster are determined. The power complementarity between clusters is used to correct the initial feed-in power curves, while integrating the power characteristics of different power types within the cluster. This reduces the impact of power fluctuations from a single power source or a single cluster, solving the problem of poor prediction stability caused by the lack of consideration for synergy and complementarity in single-station independent prediction methods. Based on the cluster collaborative feed-in power curve and the grid load forecast data within the preset time period of the region, the forecast results of renewable energy power generation in each region to be predicted are determined. This makes the renewable energy power generation forecast results integrate the characteristics of cluster collaboration and power complementarity. Therefore, the grid dispatch center can formulate dispatch plans based on the accurate renewable energy power generation forecast results, avoid the situation of renewable energy power surplus that cannot be absorbed or insufficient supply, and improve the renewable energy absorption capacity and grid operation stability.

[0044] Optionally, the processes of steps 301 to 304 include: Step 301: Based on the time-series power change characteristics of the initial feed-in power curves corresponding to different types of new energy power plants within each new energy power generation cluster, determine the time-series feed-in behavior sequence of multiple power sources for each new energy power generation cluster.

[0045] Optionally, the power prediction system uses the initial feed-in power curves corresponding to different types of new energy power plants within each new energy power generation cluster as the basic data, and extracts the time-series power change characteristics of each initial feed-in power curve.

[0046] Furthermore, for each new energy power generation cluster, the power prediction system integrates and analyzes the time-series power variation characteristics of different types of new energy power plants within it to determine the time-series feed-in behavior sequence of multiple power sources for the cluster. The time-series feed-in behavior sequence of multiple power sources refers to an ordered data set that records the power output fluctuation patterns of each type of new energy power plant within the cluster and the phase distribution characteristics of each fluctuation pattern in the time dimension, with time as the axis.

[0047] Among them, the power output fluctuation mode refers to the specific form of power change over time, including the steady fluctuation mode (absolute value of power change rate less than or equal to 5 MW per hour), the small fluctuation mode (absolute value of power change rate greater than 5 MW per hour and less than or equal to 15 MW per hour), and the large fluctuation mode (absolute value of power change rate greater than 15 MW per hour). The phase distribution characteristics in the time dimension refer to the relative positional relationship of the power peak and valley values ​​of different types of new energy power plants on the time axis, including in-phase (peak / valley occurrence time difference less than or equal to 1 hour), out-of-phase (peak / valley occurrence time difference greater than 1 hour and less than or equal to 6 hours), and no phase correlation (peak / valley occurrence time difference greater than 6 hours).

[0048] Optionally, the specific integration and analysis process of this invention is as follows: align the initial feed-in power curves of each type of new energy power station within each new energy power generation cluster with the same time axis to ensure that the time intervals and durations of each curve are completely consistent; secondly, extract the power output fluctuation patterns of each type of power station in each time period and mark the time interval corresponding to each fluctuation pattern; finally, analyze the time distribution of the power peak and valley values ​​of each type of power station, clarify its phase distribution characteristics, and combine the fluctuation pattern time series data and the phase distribution characteristic data in time order to form a multi-type power supply time series feed-in behavior sequence for the cluster.

[0049] In one embodiment, the area to be predicted includes two new energy power generation clusters: Cluster A and Cluster B. Cluster A contains both photovoltaic (PV) and wind power plants. Step 10 has already yielded the initial power feed curves for the PV plants (72-hour duration, 1-hour time intervals, time-series power variation characteristics: daily rise and stabilization phase from 6:00 AM to 6:00 PM, peak power of 42-45 MW from 12:00 PM to 2:00 PM, decline and trough phase from 6:00 PM to 6:00 AM the following day, trough power of 2.5-5 MW, power rise rate of 4-5 MW per hour, and decline rate of 3-4 MW per hour) and the initial power feed curves for the wind power plants. Power curve (72 hours in length, 1 hour in interval, time-series power variation characteristics: stable output phase from 8 pm to 8 am the next day, peak value 22-25 MW, fluctuating phase from 8 am to 8 pm, valley value 8-10 MW, power fluctuation rate 2-3 MW per hour); B New Energy Power Generation Cluster includes biomass power plants, initial feed-in power curve (72 hours in length, 1 hour in interval, time-series power variation characteristics: stable power throughout the day, fluctuation amplitude less than 1 MW per hour, power maintained at 25-27 MW).

[0050] The power prediction system integrates and analyzes the power input curves of the A new energy power generation cluster: aligning the initial input power curves of the photovoltaic (PV) and wind power stations along a 72-hour time axis, and extracting the fluctuation patterns for each time period: from 6:00 to 12:00, the PV power station exhibits a steady upward trend while the wind power station exhibits a fluctuating trend; from 12:00 to 14:00, the PV power station exhibits a steady peak trend while the wind power station exhibits a fluctuating trend; from 14:00 to 18:00, the PV power station exhibits a steady downward trend while the wind power station exhibits a fluctuating trend; from 18:00 to 20:00, the PV power station exhibits a rapid downward trend while the wind power station exhibits a fluctuating trend; from 20:00 to 6:00 the next day, the PV power station exhibits a steady low-trough trend while the wind power station exhibits a steady peak trend; and from 6:00 to 8:00, the PV power station exhibits a steady upward trend while the wind power station exhibits a fluctuating trend. Further analysis of the phase distribution characteristics reveals that the time difference between the peak values ​​of the PV power station (12:00-14:00) and the peak values ​​of the wind power station (20:00-8:00 the next day) is 6-8 hours, indicating no phase correlation. By combining the aforementioned fluctuation pattern time-series data with phase distribution characteristic data, a multi-type power source time-series feed-in behavior sequence is formed for the A new energy power generation cluster. Similarly, the B new energy power generation cluster contains only biomass power plants, and its multi-type power source time-series feed-in behavior sequence is a time-series data set with a stable fluctuation pattern throughout the day and no multi-type phase correlation.

[0051] Step 302: Based on the power synergy and complementarity relationship between various new energy power generation clusters, determine the trend relationship of power output trend between each pair of new energy power generation clusters at the same time point.

[0052] Optionally, the power output trend refers to the overall direction of change of the power value over time in a continuous time series, including an upward trend (the power value continues to increase over two or more consecutive time intervals), a downward trend (the power value continues to decrease over two or more consecutive time intervals), and a stable trend (the change in power value is less than a preset threshold over two or more consecutive time intervals, such as 2 megawatts per hour).

[0053] Furthermore, based on the aforementioned power synergy and complementarity, the power prediction system systematically analyzes each pair of new energy power generation clusters, clarifying the trend relationship between the power output trends of each pair of clusters at the same point in time. This trend relationship includes consistency and inverse relationships. A consistency relationship refers to the relationship when two clusters have the same power output trend type at the same point in time; that is, if one cluster is on an upward trend, the other is also on an upward trend, if one is on a downward trend, the other is also on a downward trend, and if one is on a stable trend, the other is also on a stable trend. An inverse relationship refers to the relationship when two clusters have opposite power output trend types at the same point in time; that is, if one cluster is on an upward trend, the other is on a downward trend, and vice versa (a stable trend does not constitute an inverse relationship with either an upward or downward trend).

[0054] Optionally, the specific process of this embodiment of the invention is as follows: First, the power synergy and complementarity relationship of each pair of new energy power generation clusters is decomposed according to the time dimension, and the validity of this relationship at each time point within a preset future time period is clarified (since the temporal changes of environmental conditions corresponding to geographical locations are stable, the power synergy and complementarity relationship is valid at each time point within the preset future time period). Second, based on the initial feed-in power curves of different types of power plants within each cluster, the power output trend type of each pair of clusters at each same time point within the preset future time period is determined. Finally, combining the attributes (consistency or reversal) of the power synergy and complementarity relationship, the trend relationship of each pair of clusters at each same time point is verified and clarified.

[0055] In one embodiment, step 20 has determined the power synergy and complementarity relationship: the wind power station in cluster A and the biomass power station in cluster B have an inverse relationship; the photovoltaic power station in cluster A and the wind power station have a consistent relationship; other cluster combinations do not have obvious power synergy and complementarity relationships. The preset duration is 72 hours, with a time interval of 1 hour. The trend relationship between the two clusters was analyzed pairwise: For the photovoltaic power station and the wind power station in cluster A (consistent relationship), the power output trend of their initial feed-in power curves at each time point within 72 hours was extracted. For example, from 6:00 to 12:00 every day, both showed an upward trend (the photovoltaic power station's power increased from 2.5 MW to 42 MW, and the wind power station's power increased from 8 MW to 15 MW), which is consistent with the relationship. From 12:00 to 14:00 every day, both showed a stable trend (the photovoltaic power station maintained 42-45 MW, and the wind power station maintained 15-18 MW), which is consistent with the relationship. From 18:00 to 20:00 every day, both showed a downward trend (the photovoltaic power station decreased from 40 MW to 5 MW, and the wind power station decreased from 18 MW to 10 MW), which is consistent with the relationship. Therefore, it was determined that the two had a consistent relationship at all the same time points within 72 hours.

[0056] For the A cluster wind power plant and the B cluster biomass power plant (inverse relationship), the trends of their initial feed-in power curves at various time points are extracted. For example, from 8 PM to 8 AM the next day, the A cluster wind power plant shows an upward trend (from 10 MW to 25 MW), while the B cluster biomass power plant shows a stable trend (maintaining 26 MW). Since a stable trend and an upward trend do not constitute an inverse relationship, and considering the effectiveness of the power synergy and complementarity relationship, it is determined that there is no inverse trend relationship between the two during this period. From 8 AM to 8 PM daily, the A cluster wind power plant shows a downward trend. The power output of the A cluster's wind power station decreased from 25 MW to 10 MW, showing a stable trend with no inverse trend. Upon verification, an inverse relationship was found only between the two during extreme weather simulation periods (e.g., between 12:00 and 18:00, the power output of the A cluster's wind power station decreased from 20 MW to 8 MW, showing a downward trend, while the power output of the B cluster's biomass power station increased from 26 MW to 27 MW, showing an upward trend). Therefore, it was determined that this pair of clusters had an inverse relationship at the same time point between 12:00 and 18:00 during extreme weather simulations; no inverse trend relationship was found at other time points. For the A cluster's photovoltaic power station and the B cluster's biomass power station (which have no obvious synergistic relationship), it is not necessary to determine a trend relationship.

[0057] Step 303: Based on the trend relationship, determine whether there is complementary power output potential between two new energy power generation clusters at any point in time, and obtain the complementary potential results of each pair of new energy power generation clusters at any point in time.

[0058] Optionally, the power prediction system uses the trend relationship between each pair of new energy power generation clusters at the same point in time as a basis, combined with the power output gap or surplus of each cluster at the corresponding point in time, to determine whether there is power output complementarity potential between the two new energy power generation clusters at any point in time, and obtains the complementarity potential result of each pair of clusters at any point in time. Here, power output complementarity potential refers to the possibility that the power output surplus of one new energy power generation cluster can make up for the power output gap of another cluster; power output gap refers to the difference between the actual predicted power of a cluster at a certain point in time and the preset power output benchmark value of that cluster, where the power output benchmark value is the average power output value of the cluster during the same period in history; power output surplus refers to the difference between the actual predicted power of a cluster at a certain point in time and the preset power output benchmark value of that cluster.

[0059] Optionally, the specific judgment process in this embodiment of the invention is as follows: A power output benchmark value is set for each pair of new energy power generation clusters exhibiting a trend relationship (consistent or inverse relationship). The benchmark value is determined by statistically analyzing the average power output value of the cluster at the same time point over the past three years. For any given time point, the power output gap or surplus value of each of the two clusters in each pair is calculated. Furthermore, based on trend relationships: if it is an inverse relationship, when one cluster has a power output gap, the other cluster has a power output surplus, and the surplus value is greater than or equal to the gap value, then it is determined that there is potential for complementary power output at that point in time.

[0060] If the relationship is consistent, when both clusters have power output gaps or surpluses at the same time, it is determined that there is no power output complementarity potential. Only when the gap of one cluster and the surplus of the other cluster exist at the same time and the values ​​match is it determined that there is complementarity potential (the probability of complementarity potential is low under the consistency relationship).

[0061] Finally, the judgment result of each pair of clusters at each time point is recorded as the complementary potential result, including whether there is complementary potential or not, and the corresponding time point, gap and surplus value are marked.

[0062] In one embodiment, two pairs of clusters with a trend relationship are selected: Cluster A photovoltaic power station and Cluster A wind power station (consistent relationship), and Cluster A wind power station and Cluster B biomass power station (reverse relationship). By statistically analyzing data from the same period over the past three years, a benchmark value for the power output of each cluster is set: the benchmark value for Cluster A photovoltaic power station is 25 MW, the benchmark value for Cluster A wind power station is 18 MW, and the benchmark value for Cluster B biomass power station is 26 MW.

[0063] Regarding the wind power plants in cluster A and the biomass power plants in cluster B (an inverse relationship), an analysis was conducted between 12:00 and 18:00 during an extreme weather simulation: At 12:00, the wind power output of cluster A was 20 MW, lower than the baseline value of 18 MW, with no deficit (20-18=2 MW, representing a surplus). The biomass power output of cluster B was 26 MW, equal to the baseline value, with neither a deficit nor a surplus, indicating no complementary potential. At 14:00, the wind power output of cluster A was 15 MW, lower than the baseline value of 18 MW, with a deficit of 3 MW (18-15=3). The biomass power output of cluster B... The biomass power plant has a power output of 27 MW, which is higher than the benchmark value of 26 MW, resulting in a surplus of 1 MW (27-26=1). The surplus value is less than the gap value, so it is determined that there is no complementary potential. At 16:00, the wind power plant in Cluster A has a power output of 10 MW, with a gap of 8 MW, and the biomass power plant in Cluster B has a power output of 28 MW, with a surplus of 2 MW. They are still mismatched and there is no complementary potential. At 18:00, the wind power plant in Cluster A has a power output of 8 MW, with a gap of 10 MW, and the biomass power plant in Cluster B has a power output of 29 MW, with a surplus of 3 MW. They are still mismatched and there is no complementary potential.

[0064] For the photovoltaic (PV) power station and wind power station cluster A (consistent relationship), analysis was conducted daily between 10:00 and 12:00: At 10:00, the PV power station in cluster A had a power output of 30 MW, with a surplus of 5 MW (30-25=5), while the wind power station had a power output of 15 MW, with a deficit of 3 MW (18-15=3). The surplus was greater than the deficit, indicating potential for complementarity. At 11:00, the PV power station had a power output of 35 MW, with a surplus of 10 MW, while the wind power station had a power output of 16 MW, with a deficit of 2 MW, again indicating potential for complementarity. At 12:00, both power outputs were higher than the baseline values ​​(PV power station 42 MW, wind power station 18 MW), indicating surplus, and therefore no potential for complementarity. The final result for the complementary potential of each cluster at any given time point was determined. For example, the PV power station and wind power station in cluster A showed complementary potential between 10:00 and 11:00, but no complementary potential at 12:00 and most other time points.

[0065] Step 304: Based on the complementary potential results of each pair of new energy power generation clusters and combined with the time-series feed-in behavior sequence of multiple types of power sources in each new energy power generation cluster, determine the cluster collaborative feed-in power curve of each new energy power generation cluster.

[0066] Optionally, the power prediction system determines the cluster collaborative feed-in power curve of each new energy power generation cluster based on the complementary potential results of each pair of new energy power generation clusters and the time-series feed-in behavior sequence of multiple types of power sources in each new energy power generation cluster, which can accurately reflect the synergistic complementary effect between clusters and the synergistic output characteristics of multiple types of power sources within the cluster, as in steps 3041 to 3044.

[0067] The cluster collaborative feed-in power curve obtained in this embodiment of the invention integrates the collaborative characteristics of multiple types of power sources within the cluster and the complementary effects between clusters, which greatly reduces the impact of power fluctuations in a single cluster on the overall power generation stability, improves the accuracy and reliability of new energy power generation prediction, and thus enhances the new energy absorption capacity and grid operation stability.

[0068] Optionally, the processes of steps 3041 to 3044 include: Step 3041: Based on the time-series feed-in behavior sequences of various new energy power generation clusters and the complementary potential results of each pair of new energy power generation clusters, determine the power coordination cluster combinations that have the conditions for coordinated operation in the time dimension.

[0069] Optionally, a power synergy cluster combination refers to a set consisting of two or more new energy power generation clusters, and any two new energy power generation clusters within the set must meet the condition that there is complementary power output potential within at least one time period; a time period refers to a time interval consisting of multiple consecutive time points, and the minimum time period contains no less than 3 time points (when the corresponding time interval is 1 hour, the minimum time period duration is no less than 3 hours); synergistic operation conditions refer to the basic conditions under which clusters can achieve overall power output stability improvement through power complementarity, and the determination basis is that there is complementary potential between clusters for at least one effective time period.

[0070] Optionally, the specific screening process in this embodiment of the invention is as follows: First, sort out all pairs of new energy power generation clusters and exclude pairs of clusters that are determined in step 303 to have no complementary potential throughout the entire time period; second, for the remaining pairs of clusters, check their complementary potential results, extract the time points with "complementary potential exists", integrate consecutive such time points into time periods, and determine whether there is an effective time period of not less than 3 hours. If so, the pairs of clusters are initially included in the candidate power coordination cluster combinations; third, for combinations containing three or more new energy power generation clusters, it is necessary to verify that there is at least one effective time period of complementary potential between any two clusters in the combination, or that there is a core cluster in the combination that has effective time period of complementary potential with all other clusters (the core cluster refers to the cluster with the longest complementary potential coverage time period in the combination); finally, combine the multi-type power supply time-series feed-in behavior sequences of each cluster to verify that the time axes of each cluster in the candidate combination are completely aligned (ensuring that the time-series data can be analyzed collaboratively), eliminate combinations whose time axes cannot be aligned, and finally determine the power coordination cluster combinations.

[0071] In one embodiment, the region to be predicted includes new energy power generation cluster A (including photovoltaic power stations and wind power stations) and new energy power generation cluster B (including biomass power stations), with the addition of new energy power generation cluster C (including wind power stations and hydropower stations). Step 301 has obtained the time-series feed behavior sequences of multiple power sources for clusters A, B, and C (each with a 72-hour time axis and a 1-hour time interval); Step 303 has obtained the complementary potential results of pairwise combinations: Cluster A and Cluster B have complementary potential during 10-11 am daily (2 hours, non-effective time period) and 3-7 pm during extreme weather simulation (4 hours, effective time period); Cluster A and Cluster C have complementary potential during 8-12 pm daily (4 hours, effective time period) and 8-11 pm daily (3 hours, effective time period); Cluster B and Cluster C have complementary potential during 2-5 pm daily (3 hours, effective time period); no complementary potential exists for the remaining time periods of other pairwise combinations.

[0072] The power prediction system performs the following screening: First, it analyzes pairwise combinations (AB, AC, BC, A-other, B-other, C-other), excluding combinations without any effective time-slot complementarity potential, leaving AB, AC, and BC as the remaining three pairwise combinations. Second, it verifies the effective time slots for each combination. The AB combination has an effective time slot of 15:00-19:00 (4 hours), the AC combination has effective time slots of 8:00-12:00 (4 hours) and 20:00-23:00 (3 hours), and the BC combination has an effective time slot of 14:00-17:00 (3 hours), all meeting the criteria. These three pairwise combinations are included as candidate combinations. Third, it considers the three-cluster combination ABC, verifying that any pairwise combination within the combination has effective time-slot complementarity potential (AB, AC, and BC all meet this criterion), and that the time axes of the three clusters are aligned (72-hour intervals and 1-hour intervals). Finally, it determines the power collaborative cluster combinations as: pairwise combinations AB, AC, and BC, and the three-cluster combination ABC.

[0073] Step 3042: Based on the time-series feed-in behavior sequence of multiple types of power sources in each new energy power generation cluster within each power coordination cluster combination, determine the joint power fluctuation profile at each time point.

[0074] Optionally, the power combined power fluctuation profile refers to the set of features that can characterize the overall fluctuation state of the power output of all new energy power generation clusters in the power coordination cluster combination at a certain point in time. The power combined power fluctuation profile includes the power output value of each cluster in the combination at that point in time, the proportion of each cluster's power to the total power of the combination, the total power value of the combination, and the difference between the maximum and minimum power output values ​​in the combination (i.e., the power fluctuation amplitude in the combination).

[0075] Optionally, the specific determination process in this embodiment of the invention is as follows: First, align the time-series feed-in behavior sequences of multiple power sources in each new energy power generation cluster within each power coordination cluster combination according to time points to ensure that the power data of all clusters at the same time point can be extracted synchronously; second, for each time point, extract the real-time power output value of each cluster within the combination (if there are multiple types of power plants within the cluster, take the sum of the power output values ​​of each type of power plant as the real-time power output value of the cluster); third, calculate the total power value of the combination at that time point, that is, the sum of the real-time power output values ​​of all clusters within the combination; then, calculate the proportion of each cluster's power to the total power of the combination, that is, the ratio of the real-time power output value of a single cluster to the total power value of the combination; next, find the maximum and minimum values ​​among the real-time power output values ​​of all clusters within the combination at that time point, calculate the difference between the two, and obtain the power fluctuation amplitude within the combination; finally, integrate the power output values ​​of each cluster, the power proportion of each cluster, the total power value of the combination, and the power fluctuation amplitude within the combination at the same time point to form the joint power fluctuation profile at that time point.

[0076] In one embodiment, a power coordination cluster combination AB (cluster A and cluster B) is selected. Within this combination, cluster A contains power output data from photovoltaic power plants and wind power plants, while cluster B contains power output data from biomass power plants. The time axis is 72 hours, with a time interval of 1 hour. Three time points are selected for analysis: 16:00-18:00 daily (belonging to the effective complementary time period of the AB combination, 15:00-19:00). At 16:00: Extract the real-time power output values ​​of cluster A (30 MW photovoltaic power station + 10 MW wind power station = 40 MW) and cluster B (28 MW biomass power station); calculate the total combined power value as 40 MW + 28 MW = 68 MW; calculate the power share of cluster A as 40 / 68 ≈ 58.8% and the power share of cluster B as 28 / 68 ≈ 41.2%; the maximum power value within the combination is 40 MW, the minimum power value is 28 MW, and the fluctuation range is 40-28=12 MW; integrate the above data to obtain the combined power fluctuation profile at 16:00: cluster A 40 MW (58.8%), cluster B 28 MW (41.2%), total power 68 MW, fluctuation range 12 MW.

[0077] At 5 PM: Real-time power output of Cluster A (25 MW photovoltaic power station + 9 MW wind power station = 34 MW), real-time power output of Cluster B (29 MW); total combined power 34 + 29 = 63 MW; Cluster A's share 34 / 63 ≈ 54.0%, Cluster B's share 29 / 63 ≈ 46.0%; fluctuation range 34 - 29 = 5 MW; combined power fluctuation profile: Cluster A 34 MW (54.0%), Cluster B 29 MW (46.0%), total power 63 MW, fluctuation range 5 MW.

[0078] At 18:00: Real-time power output of Cluster A (20 MW photovoltaic power station + 8 MW wind power station = 28 MW), real-time power output of Cluster B (30 MW); total combined power 28 + 30 = 58 MW; Cluster A's share 28 / 58 ≈ 48.3%, Cluster B's share 30 / 58 ≈ 51.7%; fluctuation range 30 - 28 = 2 MW; combined power fluctuation profile: Cluster A 28 MW (48.3%), Cluster B 30 MW (51.7%), total power 58 MW, fluctuation range 2 MW.

[0079] Step 3043: Based on the joint power fluctuation profile of each power coordination cluster combination, determine the coordinated fluctuation characteristics that characterize the power output stability under the combined action of multiple new energy power generation clusters within the cluster combination.

[0080] Optionally, time-series fluctuation characteristic analysis refers to longitudinally comparing the combined power fluctuation profiles at continuous time points to uncover the patterns of change in the combined total power and the power fluctuation amplitude within the combination over time. Cooperative fluctuation characteristics refer to core indicators that reflect the overall power output stability of the cluster combination. Cooperative fluctuation characteristics include the time-series fluctuation coefficient of the combined total power, the time-series mean of the power fluctuation amplitude within the combination, the difference between the peak and trough values ​​of the combined total power, and the duration of continuous stable operation of the combined total power.

[0081] The time-series fluctuation coefficient of the combined total power is the ratio of the standard deviation to the mean of the combined total power within a preset analysis period. It quantifies the relative degree of time-series fluctuation in total power; a smaller fluctuation coefficient indicates smoother fluctuations in total power. The preset analysis period is the effective complementary time period of the power synergy cluster combination. If multiple effective time periods exist, they are calculated separately and then averaged.

[0082] For the time-series mean of power fluctuation amplitude within the combination: the ratio of the sum of power fluctuation amplitude within the combination at each time point within the preset analysis period to the number of time points, which is used to characterize the average level of power difference among clusters within the combination. The smaller the mean, the smaller the power difference between clusters and the stronger the synergy.

[0083] For the difference between the peak and valley values ​​of the combined total power: the difference between the maximum and minimum values ​​of the combined total power within the preset analysis period is used to reflect the maximum fluctuation range of the total power. The smaller the difference, the better the stability.

[0084] For the duration of continuous stable operation of the combined total power: the longest interval within the preset analysis period where the change in the combined total power is continuously less than or equal to the preset stability threshold. The preset stability threshold is 3 megawatts per hour. The longer the duration, the better the stability.

[0085] Optionally, the specific determination process in this embodiment of the invention is as follows: For each power cooperative cluster combination, determine its preset analysis period (effective complementary period). Extract the joint power fluctuation profile at all time points within the preset analysis period, and extract the combined total power and power fluctuation amplitude data within the combination at each time point.

[0086] Furthermore, following the calculation methods of the aforementioned key indicators, the time-series fluctuation coefficient of the combined total power, the time-series mean of the power fluctuation amplitude within the combination, the difference between the peak and valley values ​​of the combined total power, and the duration of continuous stable operation of the combined total power are calculated in sequence to form the cooperative fluctuation characteristics of the power cooperative cluster combination.

[0087] In one embodiment, the power coordinated cluster combination AB has a preset analysis period of 15-19 hours (15:00, 16:00, 17:00, 18:00, and 19:00) for effective complementary periods, and a preset stability threshold of 3 MW per hour. The joint power fluctuation profile from 16:00 to 18:00 has been obtained, and data for 15:00 and 19:00 have been supplemented: at 15:00, cluster A has 45 MW and cluster B has 27 MW, for a total power of 72 MW, with a fluctuation range of 18 MW; at 19:00, cluster A has 25 MW and cluster B has 31 MW, for a total power of 56 MW, with a fluctuation range of 6 MW.

[0088] For the time-series fluctuation coefficient of the combined total power: the average value of the total power at 5 time points (72, 68, 63, 58, 56) is 63.4 MW; the standard deviation is calculated by taking the square root of the sum of the squares of the differences between each value and the average value divided by 5. The sum of the squares of the differences is 43.56 + 21.16 + 0.16 + 29.16 + 54.76 = 148.8, and the standard deviation is approximately 5.45 MW; the time-series fluctuation coefficient is approximately 5.45 / 63.4 ≈ 0.086.

[0089] For the time-series mean of the power fluctuation amplitude within the combination: the sum of the fluctuation amplitudes at 5 time points (18, 12, 5, 2, 6) is 43, and the mean is 43 / 5 = 8.6 MW.

[0090] The difference between the peak and valley values ​​of the total combined power is: peak value 72 MW, valley value 56 MW, difference = 72 - 56 = 16 MW.

[0091] For the duration of continuous stable operation of the combined total power: from 16:00 to 17:00, the total power decreased from 68 to 63 (the change range is 5 MW > 3 MW, which is not stable); from 17:00 to 18:00, it decreased from 63 to 58 (the change range is 5 MW > 3 MW, which is not stable); from 18:00 to 19:00, it decreased from 58 to 56 (the change range is 2 MW < 3 MW). There was only one time interval, corresponding to a duration of 1 hour. Therefore, the duration of continuous stable operation is 1 hour.

[0092] Therefore, the coordinated fluctuation characteristics of the AB combination are: time-series fluctuation coefficient 0.086, time-series average fluctuation amplitude 8.6 MW, total power peak-to-valley difference 16 MW, and continuous stable operation duration 1 hour.

[0093] Step 3044: Based on the complementary structural relationship and cooperative fluctuation characteristics among the various new energy power generation clusters within each power cooperative cluster combination, determine the cluster cooperative feed-in power curve of each new energy power generation cluster.

[0094] Optionally, the power prediction system determines the cluster collaborative feed-in power curve of each new energy power generation cluster based on the complementary structural relationship and collaborative fluctuation characteristics among the various new energy power generation clusters within each power collaborative cluster combination, as described in steps 30441 to 30444.

[0095] The cluster collaborative feed-in power curve obtained in this embodiment of the invention fully integrates the complementary effects between clusters and the overall stability requirements. Therefore, the final output cluster collaborative feed-in power curve can reduce the impact of power fluctuations of a single cluster on the overall power generation system, improve the stability and predictability of new energy power output, and thus improve the new energy absorption capacity and grid operation stability.

[0096] Optionally, the process of steps 30441 to 30444 includes: Step 30441: For each power coordination cluster combination, based on the coordination fluctuation characteristics and the complementary structural relationship between each new energy power generation cluster, determine the power coordination role of each new energy power generation cluster in the power coordination cluster combination.

[0097] Optionally, complementary structural relationships refer to the association structure formed between clusters within a power-coordinated cluster combination based on complementary potential distribution, power proportion, etc., including information such as core complementary cluster pairs, complementary time period distribution, and time-series changes in the power proportion of each cluster. Power coordination role refers to the functional positioning of a cluster during combined coordinated operation, specifically divided into power-dominant clusters and power-auxiliary clusters.

[0098] Among them, the power-dominant cluster refers to a new energy power generation cluster with a large power output scale and relatively high power output stability in the combination, and which assumes the main power supply responsibility during the core complementary period. The power-auxiliary cluster refers to a new energy power generation cluster that operates in conjunction with the power-dominant cluster in the combination, and achieves overall power stability of the combination through power compensation during periods when the power output of the power-dominant cluster is insufficient or fluctuates greatly.

[0099] Optionally, the specific determination process in this embodiment of the invention is as follows: extract the average power output value of each new energy power generation cluster within each power coordination cluster combination during a preset analysis period (effective complementary period), calculate the proportion of the average power of each cluster to the total average power of the combination, and select clusters with a proportion of not less than 50% as candidate power-dominant clusters. If there is no cluster with a proportion of not less than 50%, then select the two clusters with the highest proportions to proceed to the next step of determination.

[0100] Furthermore, by combining the combined total power time-series fluctuation coefficient in the cooperative fluctuation characteristics, the power fluctuation characteristics of each candidate cluster are analyzed (based on the power fluctuation patterns in its multi-type power supply time-series feed behavior sequence), and the candidate cluster with the smallest power fluctuation coefficient (the ratio of its own power standard deviation to its own average power) is selected as the initial power-dominant cluster.

[0101] Furthermore, the distribution of the initial power-dominant cluster within the core complementary periods of the complementary structure is examined to confirm whether it possesses continuous power output capability during these core complementary periods, and whether it has the highest matching degree of complementary potential with other clusters during these periods (i.e., its power gap / surplus matches the surplus / gap of other clusters the most times). Finally, clusters other than the power-dominant cluster within the portfolio are identified as power-auxiliary clusters. If the portfolio contains only two clusters, and their average power share difference is less than 10% and their fluctuation coefficients are similar, then based on the complementary direction in the complementary structure, the cluster providing power surplus during most complementary periods is identified as the power-dominant cluster, and the other is identified as the power-auxiliary cluster.

[0102] In one embodiment, a power synergy cluster combination AB is selected (A is a new energy power generation cluster including photovoltaic power stations and wind power stations; B is a new energy power generation cluster including biomass power stations). The preset analysis period for this combination is 15:00-19:00 (effective complementary time period). The synergistic fluctuation characteristics are: time series fluctuation coefficient 0.086, time series average fluctuation amplitude 8.6 MW, total power peak-valley difference 16 MW, and continuous stable operation duration 1 hour. The complementary structure relationship is: during the core complementary period of 15:00-19:00, cluster A and cluster B have multiple matching relationships of power gap and surplus during this period.

[0103] Extract the average power output values ​​of each cluster from 15 to 19: the average power of cluster A is (45+40+34+28+25) / 5=34.4 MW, the average power of cluster B is (27+28+29+30+31) / 5=29 MW; the combined average power is 34.4+29=63.4 MW.

[0104] Calculate the average power share of each cluster: Cluster A accounts for 34.4 / 63.4≈54.3%, Cluster B accounts for 29 / 63.4≈45.7%. Cluster A accounts for no less than 50%, thus becoming the candidate power-dominant cluster.

[0105] Analysis of power fluctuation coefficients: The standard deviation of the power data (45, 40, 34, 28, 25) of cluster A is approximately 7.98 MW, and the fluctuation coefficient is approximately 0.232 (7.98 / 34.4). The standard deviation of the power data (27, 28, 29, 30, 31) of cluster B is approximately 1.58 MW, and the fluctuation coefficient is approximately 0.054 (1.58 / 29).

[0106] Verification of complementary structural relationships: During the core complementary period from 3 PM to 7 PM, the power of Cluster A showed a continuous downward trend (from 45 MW to 25 MW), with a significant power gap (below its own baseline value of 18 MW corresponding to the total power demand of the cluster at multiple time points), while the power of Cluster B showed a continuous upward trend (from 27 MW to 31 MW), with a continuous power surplus. The complementary potential of the two is highly matched.

[0107] Comprehensive assessment of power coordination roles: Although cluster B has a smaller fluctuation coefficient, cluster A has a higher average power share (over 50%) and is the main power demander during the core complementary period (its gap needs to be filled by cluster B), bearing the basic responsibility for core power supply. Therefore, cluster A is determined to be the power-dominant cluster and cluster B is the power-auxiliary cluster.

[0108] Step 30442: Based on the multi-type power supply timing behavior sequence of each new energy power generation cluster, determine the power supply period undertaken by the power-dominant cluster in the power coordination cluster combination, and determine the compensation time window for the power auxiliary cluster to carry out power compensation outside the power supply period based on the power supply period.

[0109] Optionally, the power supply period refers to the time during which the power-dominant cluster can stably output power and the output power can meet the basic power demand within the portfolio. During this period, the power output fluctuation of the power-dominant cluster is less than a preset dominant fluctuation threshold (the preset dominant fluctuation threshold is 5 MW per hour). The compensation time window refers to the time during which the power output of the power-dominant cluster is lower than its own baseline power (the historical average power output value of the power-dominant cluster for the same period), or the power fluctuation is greater than the preset dominant fluctuation threshold. During this period, the power auxiliary cluster needs to output additional power to compensate and maintain the overall power stability of the portfolio.

[0110] The specific determination process in this embodiment of the invention is as follows: First, extract the time-series feed-in behavior sequences of multiple power sources in the power-dominant cluster, and filter out all time points where the power output fluctuation mode is a stable fluctuation mode (the absolute value of the power change rate is less than or equal to 5 MW per hour). Integrate the continuous stable fluctuation time points into an initial power supply period. Second, calculate the average power output value of the power-dominant cluster within the initial power supply period. If the average value is not lower than the combined basic power demand value (the combined basic power demand value is 80% of the combined historical average power output value for the same period), then the initial power supply period is determined as the final power supply period. If the average value is lower than the combined basic power demand value, then the period is split, and only the sub-periods with an average value not lower than the combined basic power demand value are retained as power supply periods. Next, analyze the power output data of the power-dominant cluster within the preset analysis period, identify the time points when the power output value is lower than its own baseline power, and the time points when the power fluctuation amplitude is greater than the preset dominant fluctuation threshold, and integrate these time points into continuous time periods; finally, remove the parts of these time periods that overlap with the power supply period, and the remaining continuous time periods are the compensation time windows of the power-assisted cluster. The duration of each compensation time window must be no less than 2 hours (to ensure the effectiveness and stability of compensation). If there are scattered time periods with a duration of less than 2 hours, they are merged into adjacent compensation time windows or discarded.

[0111] In one embodiment, the power-dominant cluster is cluster A, and the power-auxiliary cluster is cluster B. The multi-type power supply timing feed behavior sequence of cluster A includes power data (45, 40, 34, 28, 25 MW) from 15 to 19, with its own base power of 30 MW (the historical average power output value of cluster A for the same period), and a preset dominant fluctuation threshold of 5 MW per hour. The combined base power demand value is 80% of the combined historical average power output value of 63.4 MW, i.e., 50.72 MW.

[0112] Determine the power supply periods: Extract the power fluctuation pattern of Cluster A from 3 PM to 7 PM, and calculate the power change rate at each adjacent time point: 3 PM to 16 PM (45→40, change rate -5 MW / h, absolute value equal to 5 MW, belonging to stationary fluctuation), 4 PM to 17 PM (40→34, change rate -6 MW / h, absolute value greater than 5 MW, non-stationary fluctuation), 5 PM to 18 PM (34→28, change rate -6 MW / h, non-stationary fluctuation), 6 PM to 19 PM (28→25, change rate -3 MW / h, absolute value less than 5 MW, stationary fluctuation). The initial power supply periods are 3 PM to 16 PM and 6 PM to 19 PM. The average power during the initial power supply period is calculated as follows: From 3 PM to 4 PM, the average power is (45+40) / 2 = 42.5 MW, which is lower than the combined base power demand of 50.72 MW, requiring splitting. From 6 PM to 7 PM, the average power is (28+25) / 2 = 26.5 MW, also lower than 50.72 MW. Therefore, Cluster A has no suitable power supply period from 3 PM to 7 PM (i.e., Cluster A cannot stably meet the combined base power demand during this period).

[0113] Determine the compensation time window: Identify the time points when the power output of Cluster A is lower than its own baseline power by 30 MW: 17:00 (34 MW > 30 MW, not met), 18:00 (28 MW < 30 MW, met), 19:00 (25 MW < 30 MW, met); Identify the time points when the power fluctuation amplitude is greater than the preset dominant fluctuation threshold of 5 MW: 16:00-17:00 (fluctuation amplitude 6 MW), 17:00-18:00 (fluctuation amplitude 6 MW). Integrate these time points to form a continuous time period: 16:00-19:00. Since Cluster A has no power supply period from 15:00 to 19:00, there is no need to remove the overlapping part. Therefore, the compensation time window for Cluster B is determined to be 16:00-19:00 (duration of 4 hours, meeting the requirement of not less than 2 hours).

[0114] Step 30443: Based on the compensation time window and the joint power fluctuation profile, determine the power adjustment target range of the power auxiliary cluster within its respective compensation time window, and adjust the corresponding time period power values ​​in the original multi-type power supply timing feed behavior sequence of the power auxiliary cluster based on the power adjustment target range to obtain the adjusted multi-type power supply timing feed behavior sequence.

[0115] Optionally, the power adjustment target range refers to the range of power output values ​​that the power auxiliary cluster needs to achieve within the compensation time window to ensure stable overall power output. This range is a continuous range of power values, not a fixed value. The adjusted multi-type power supply timing feed behavior sequence refers to the new timing data set formed after adjusting the power values ​​corresponding to the compensation time window in the original sequence. Its time axis remains consistent with the original sequence, and the power data during non-compensation periods remain unchanged.

[0116] Optionally, the specific determination and adjustment process in this embodiment of the invention is as follows: Extract the combined total power data from the combined power fluctuation profile within the compensation time window, calculate the stable target value of the combined total power, and take the historical average power output value of the combined power during the same period within the compensation time window (this value reflects the normal power demand of the combined power during this period); if the historical average power data of the combined power during the same period is missing, then take the average of the combined average power output values ​​for 2 hours before and after the compensation time window as the stable target value. Secondly, based on the stable target value and the predicted power output data of the power-dominant cluster within the compensation time window (derived from its multi-type power supply timing feed-in behavior sequence), calculate the power compensation amount that the power auxiliary cluster needs to provide at each time point. Power compensation amount = stable target value - predicted power output value of the power-dominant cluster - predicted power output value of other power auxiliary clusters (if present); if the calculated power compensation amount is positive, it indicates that the power auxiliary cluster needs to increase power output to make up for the gap; if it is negative, it indicates that the power auxiliary cluster can appropriately reduce power output to avoid excessive combined power surplus. Next, combining the power auxiliary cluster's maximum generating capacity (based on its installed capacity data) and minimum stable generating power (historical minimum power output value for the same period), the power adjustment target sub-interval for each time point is determined. Then, the adjustment target sub-intervals for each time point within the compensation time window are integrated to form the power adjustment target interval for the entire compensation time window (taking the minimum value of each sub-interval as the lower limit of the interval and the maximum value of each sub-interval as the upper limit of the interval). Finally, the power values ​​corresponding to the compensation time window in the original multi-type power source time-series feed-in behavior sequence of the power auxiliary cluster are adjusted: if the original power value is within the adjustment target interval, it remains unchanged; if the original power value is lower than the lower limit of the interval, it is adjusted to the lower limit of the interval (it must not exceed its maximum generating capacity); if the original power value is higher than the upper limit of the interval, it is adjusted to the upper limit of the interval (it must not be lower than its minimum stable generating power). After the adjustment is completed, the original power data of the non-compensation period and the adjusted power data of the compensation period are integrated to form the adjusted multi-type power source time-series feed-in behavior sequence.

[0117] In one embodiment, the power auxiliary cluster is cluster B, and the compensation time window is from 16:00 to 19:00. The combined total power data for 16:00 to 19:00 in the joint power fluctuation profile are 68, 63, 58, and 56 MW. The combined historical average power output for the same period is 60 MW (stable target value). The predicted power output data for the power dominant cluster A within the compensation time window are 40, 34, 28, and 25 MW. Cluster B has no other power auxiliary clusters, and its installed capacity is 30 MW (maximum generating capacity of 30 MW). The historical minimum stable generating power for the same period is 25 MW.

[0118] Calculate the power compensation at each time point: 16:00 compensation = 60 - 40 = 20 MW (original power of Cluster B is 28 MW); 17:00 compensation = 60 - 34 = 26 MW (original power is 29 MW); 18:00 compensation = 60 - 28 = 32 MW (original power is 30 MW); 19:00 compensation = 60 - 25 = 35 MW (original power is 31 MW). Determine the adjustment target sub-intervals for each time point: 16:00, compensation 20 MW, combined with Cluster B's maximum of 30 MW and minimum of 25 MW, the sub-interval is 25-30 MW; 17:00 compensation 26 MW, sub-interval 26-30 MW; 18:00 compensation 32 MW, Cluster B's maximum of 30 MW, sub-interval 28-30 MW; 19:00 compensation 35 MW, sub-interval 29-30 MW. The power regulation target range for the B cluster compensation time window (16:00-19:00) is 25-30 MW.

[0119] The power values ​​from 16:00 to 19:00 in the original multi-type power supply timing sequence of Cluster B are adjusted as follows: 16:00 original 28 MW (within the 25-30 MW range, unchanged), 17:00 original 29 MW (within the 26-30 MW range, unchanged), 18:00 original 30 MW (within the 28-30 MW range, unchanged), 19:00 original 31 MW (higher than the upper limit of 30 MW, adjusted to 30 MW). The power data for the non-compensated periods (15:00 and 20:00-72:00) remain unchanged, resulting in the adjusted multi-type power supply timing sequence of Cluster B.

[0120] Step 30444: Based on the original multi-type power source timing behavior sequence of the power-dominant cluster and the adjusted multi-type power source timing behavior sequence of the power-assisted cluster, the cluster collaborative feed-in power curves of each new energy power generation cluster are integrated to obtain the cluster collaborative feed-in power curves of each new energy power generation cluster.

[0121] Optionally, time-axis calibration is performed on the original multi-type power source time-series feed-in behavior sequences of the power-dominant cluster and the adjusted multi-type power source time-series feed-in behavior sequences of each power auxiliary cluster to ensure that the time start, time interval (e.g., 1 hour), and total duration (e.g., 72 hours) of all sequences are completely consistent. Secondly, for each new energy power generation cluster, the core power output data (i.e., the total power output value at each time point, obtained by summing the power output values ​​of each type of power station within the cluster) is extracted from its corresponding time-series dataset. Combining the complementary structural relationship and coordinated fluctuation characteristics of the power cooperative cluster combination, the extracted core power output data is finally verified to confirm that within the compensation time window, the adjusted power value of the power auxiliary cluster can effectively compensate for the power gap of the power-dominant cluster, and the fluctuation amplitude of the combined total power output meets the preset stability requirements (the combined total power time-series fluctuation coefficient is less than or equal to 0.08). Finally, the verified core power output data of each cluster are arranged in chronological order to form the cluster coordinated feed-in power curve for each new energy power generation cluster.

[0122] The cluster collaborative feed-in power curve obtained in this embodiment of the invention fully integrates the basic supply capacity of the dominant cluster and the compensation and adjustment capacity of the auxiliary cluster, effectively solving the problems of large power fluctuations and unstable supply of a single cluster, improving the stability and predictability of the overall power output of the new energy power generation cluster, thereby improving the new energy absorption capacity and grid operation stability.

[0123] Optionally, the processes of steps 401 to 403 include: Step 401: Based on the time-period load demand in the power grid load forecast data, determine the minimum operating load baseline sequence of the power grid in the region at a preset time resolution.

[0124] Optionally, the grid minimum operating load baseline sequence refers to an ordered set of data consisting of the grid minimum operating load baseline values ​​corresponding to each time point within a future preset time period in chronological order. This sequence represents the minimum power supply level at which the regional power grid can maintain safe and stable operation at each time point.

[0125] Optionally, the minimum operating load baseline value of the power grid refers to the minimum power supply value that the power grid must maintain at a certain point in time in order to ensure the normal operation of core equipment in the power grid and the basic power demand of core power users (such as hospitals, transportation hubs, important industrial enterprises, etc.). This value shall not be lower than the sum of the no-load loss power of all core equipment in the power grid and the minimum power demand power of core power users.

[0126] Optionally, the specific determination process in this embodiment of the invention is as follows: the load demand in the power grid load forecast data is classified by time sequence, and divided into peak load periods (such as peak residential electricity consumption from 18:00 to 22:00, peak industrial electricity consumption from 8:00 to 18:00), flat load periods (such as 9:00 to 17:00 excluding industrial peak, 23:00 to 6:00 the next day), and low load periods (such as 0:00 to 5:00 in the morning), and the core power user range and electricity demand intensity remain consistent in different periods.

[0127] Furthermore, for each time period type, the minimum value of all time-by-time load demand within that time period is extracted as the initial load baseline reference value for that time period type. At the same time, core equipment parameter data of the power grid in the area to be predicted (including no-load loss power of core transformers, transmission lines, switching equipment, etc.) are collected, and the total no-load loss power of all core equipment is calculated. The minimum electricity demand data of core power users are collected, and the total minimum electricity demand power of all core power users is calculated. The sum of the two is used as the absolute minimum load threshold of the power grid.

[0128] Furthermore, the initial load baseline reference value for each time period type is compared with the absolute minimum load threshold of the power grid. If the initial load baseline reference value is greater than or equal to the absolute minimum load threshold of the power grid, then the initial load baseline reference value is used as the initial baseline value for all time points under the corresponding time period type; if the initial load baseline reference value is less than the absolute minimum load threshold of the power grid, then the absolute minimum load threshold of the power grid is used as the initial baseline value for all time points under the corresponding time period type.

[0129] Furthermore, the initial baseline values ​​at each time point are subjected to time-series smoothing correction. The correction method is to take the average of the initial baseline value at the current time point and the initial baseline values ​​at the two adjacent time points. If the corrected value is not lower than the absolute minimum load threshold of the power grid, the corrected value is used as the final baseline value; if the corrected value is lower than the absolute minimum load threshold of the power grid, the absolute minimum load threshold of the power grid is still used as the final baseline value, thus obtaining the minimum operating load baseline sequence of the power grid.

[0130] In one embodiment, the preset duration is 72 hours, the preset time resolution is 1 hour, and the power grid load forecast data includes 72 time-period load demands (megawatts). The total no-load loss power of the core power grid equipment is 200 megawatts, and the total minimum power demand power of core power users is 300 megawatts. Therefore, the absolute minimum load threshold of the power grid is 200 + 300 = 500 megawatts. The time-period load demands for the 72 hours are classified by time sequence: the peak load period is 18:00-22:00 daily (15 time points in total), with a minimum time-period load demand of 650 megawatts; the flat load period is 6:00-18:00 daily (excluding 18:00) and 22:00-23:00 daily (42 time points in total), with a minimum time-period load demand of 550 megawatts; the low load period is 0:00-5:00 daily (15 time points in total), with a minimum time-period load demand of 480 megawatts.

[0131] Compare the initial load baseline reference values ​​for each time period with the grid's absolute minimum load threshold of 500 MW: During peak load periods, the initial reference value is 650 MW > 500 MW, and the initial baseline value is 650 MW; during flat load periods, the initial reference value is 550 MW > 500 MW, and the initial baseline value is 550 MW; during off-peak load periods, the initial reference value is 480 MW < 500 MW, and the initial baseline value is 500 MW. The initial baseline values ​​at each time point are adjusted for time-series smoothing. Taking a certain time point during the off-peak load period (e.g., 3:00) as an example, the initial baseline value is 500 MW. The initial baseline value at the previous time (2:00) is 500 MW, and the initial baseline value at the next time (4:00) is 500 MW. The average value is 500 MW, and the adjusted baseline value is 500 MW. Taking the transition period from flat to peak load (17:00) as an example, the initial baseline value is 550 MW. The initial baseline value at the previous time (16:00) is 550 MW, and the next time (18:00) is 650 MW. The average value is (550+550+650) / 3=583.33 MW, which is greater than 500 MW. The adjusted baseline value is 583.33 MW. Arrange the final baseline values ​​of all time points within 72 hours in chronological order to obtain the minimum operating load baseline sequence of the power grid, such as 500 MW from 0:00 to 5:00, 550 MW from 6:00, ..., 583.33 MW from 17:00, 650 MW from 18:00 to 22:00, etc.

[0132] Step 402: Based on the actual feed-in power value of each cluster at each time point represented by the cluster collaborative feed-in power curve of each new energy power generation cluster, determine the total feed-in power sequence of new energy in the region at the same time resolution.

[0133] Optionally, the actual feedable power value refers to the actual power that a certain renewable energy power generation cluster can feed into the grid at a certain point in time, after deducting equipment losses and plant power consumption. This value is the difference between the power output value at the corresponding time point in the cluster's coordinated feedable power curve and the power loss of the cluster's equipment and the power consumption of the plant (if the cluster's coordinated feedable power curve has deducted the above losses, it is directly used as the actual feedable power value). Therefore, based on the extracted actual feedable power values ​​of each cluster, the power prediction system determines the sequence of total renewable energy feedable power in the region under the same preset time resolution. The sequence of total renewable energy feedable power refers to an ordered data set composed of the total renewable energy feedable power values ​​corresponding to each time point within a preset future time period, arranged in chronological order. This sequence represents the maximum physical output capacity of the renewable energy power generation system at each time point. The total renewable energy feedable power value refers to the sum of the power that all renewable energy power generation clusters in the region to be predicted can actually feed into the grid at a certain point in time, i.e., the cumulative result of the actual feedable power values ​​of each renewable energy power generation cluster at that time point.

[0134] Optionally, the specific determination process in this embodiment of the invention is as follows: the time axis of the cluster collaborative feed-in power curve of each new energy power generation cluster is calibrated to ensure that the time starting point, preset time resolution (consistent with the grid load forecast data), and total duration (future preset duration) of all curves are completely consistent, so as to avoid power value matching errors caused by time misalignment.

[0135] Furthermore, for each time point within the preset time period, the actual feed-in power value of each new energy power generation cluster at that time point is extracted one by one; if a certain new energy power generation cluster has an abnormal state such as equipment maintenance or failure at that time point (the information comes from the additional status identifier of the cluster collaborative feed-in power curve), its actual feed-in power value is calculated as 0, or calculated according to the actual output power under the maintenance / failure state (clear status-power correspondence data is required).

[0136] Furthermore, the actual feed-in power values ​​of all renewable energy power generation clusters at the same time point are summed to obtain the total feed-in power value of renewable energy at that time point. During the summation process, the validity of the actual feed-in power value of each cluster needs to be verified, and obvious outliers (such as power values ​​exceeding 110% of the cluster's installed capacity, power values ​​less than 0, where installed capacity refers to the sum of the installed capacities of all power plants within the cluster) are removed. Outliers are replaced with the average of the actual feed-in power values ​​of two adjacent normal time points before summation. Finally, the total feed-in power values ​​of renewable energy at all time points within a preset future time period are arranged in chronological order to form a sequence of total feed-in power of renewable energy.

[0137] In one embodiment, the area to be predicted includes three new energy power generation clusters, A, B, and C, all of which have obtained cluster collaborative feed-in power curves (with a future preset duration of 72 hours and a preset time resolution of 1 hour; the curves have deducted equipment losses and plant power consumption, and are directly used as the actual feed-in power values), and the time axes of the three curves have been calibrated to be consistent; Cluster A has an installed capacity of 130 MW (50 MW photovoltaic power station + 80 MW wind power station), Cluster B has an installed capacity of 30 MW (biomass power station), and Cluster C has an installed capacity of 120 MW (60 MW wind power station + 60 MW hydropower station).

[0138] Three typical time points (10:00, 18:00, and 2:00) within a 72-hour period were selected for analysis: At 10:00, the actual feed-in power of cluster A was 35 MW (<130 MW, effective), cluster B was 26 MW (<30 MW, effective), and cluster C was 50 MW (<120 MW, effective), for a total feed-in power of 35 + 26 + 50 = 111 MW; At 18:00, the actual feed-in power of cluster A was 28 MW (effective), cluster B was 30 MW (effective), and cluster C, due to temporary equipment maintenance, had an actual feed-in power of... Assuming the power value is 0, the total renewable energy feed-in power value = 28 + 30 + 0 = 58 MW; at time 2, the actual feed-in power value of cluster A is 3 MW (effective), cluster B is 25 MW (effective), and cluster C is 130 MW (exceeding the installed capacity by 120 MW, which is an abnormal value). Taking the average value of cluster C at adjacent time points (60 MW at time 1 and 55 MW at time 3) (60 + 55) / 2 = 57.5 MW to replace the abnormal value, the total renewable energy feed-in power value = 3 + 25 + 57.5 = 85.5 MW.

[0139] Calculate the total renewable energy feed-in power at all time points within 72 hours using the above method, and arrange them in chronological order to obtain the total renewable energy feed-in power sequence, such as 3+25+60=88 MW at 1 o'clock, 85.5 MW at 2 o'clock, ..., 111 MW at 10 o'clock, ..., 58 MW at 18 o'clock, etc.

[0140] Step 403: Based on the baseline sequence of minimum operating load of the power grid and the sequence of total feed-in power of new energy sources, determine the prediction results of new energy power generation for each region to be predicted.

[0141] Optionally, based on the minimum operating load baseline sequence of the power grid and the total feed-in power sequence of new energy sources, the prediction results of new energy power generation for each region to be predicted are determined, as detailed in steps 4031 to 4034.

[0142] The embodiments of the present invention achieve precise matching between the power supply capacity of new energy sources and the load demand of the power grid. This not only ensures the basic electricity needs of core power users and the safe and stable operation of the power grid, but also dynamically adjusts the output of new energy sources according to load changes, minimizes the curtailment rate of new energy sources, and improves the reliability of new energy power generation prediction, thereby enhancing the absorption capacity of new energy sources and the stability of power grid operation.

[0143] Optionally, the process of steps 4031 to 4034 includes: Step 4031: Based on the minimum operating load baseline sequence of the power grid and the total feed-in power sequence of new energy sources, determine the feasibility assessment result of new energy feed-in at each time point in the region.

[0144] Optionally, the minimum operating load baseline sequence of the power grid, the total feed-in power sequence of renewable energy sources, and the time-period load demand are precisely aligned along the time axis to ensure that the values ​​at the same point in time in the three datasets can be accurately matched. The standard for time axis alignment is that the time start point, the preset time resolution (e.g., 1 hour), and the total duration (a future preset duration, e.g., 72 hours) are completely consistent. For each time point within the future preset duration, the three core values ​​corresponding to that time point are extracted one by one: the minimum operating load baseline value of the power grid, the total feed-in power value of renewable energy sources, and the time-period load demand.

[0145] Optionally, a three-dimensional judgment rule is constructed: Rule 1, if the total renewable energy feed-in power is greater than or equal to the minimum operating load baseline of the power grid, and less than or equal to the load demand for each time period, it is initially judged as "feasible for feed-in"; Rule 2, if the total renewable energy feed-in power is greater than the load demand for each time period, it is initially judged as "risk of upward over-limit feed-in"; Rule 3, if the total renewable energy feed-in power is less than the minimum operating load baseline of the power grid, it is initially judged as "risk of insufficient downward support for feed-in".

[0146] Furthermore, the preliminary judgment results are verified a second time. The verification content is the numerical relationship between two adjacent time points before and after the current time point. If the judgment results of adjacent time points are consistent with the current time point, the preliminary judgment result of the current time point is confirmed as the final feasibility judgment result of new energy feed-in. If the judgment results of adjacent time points are inconsistent with the current time point, the accuracy of the extraction of the three core values ​​at that time point needs to be re-verified. After eliminating data extraction errors, the final judgment result is determined based on the verified numerical relationship.

[0147] In one embodiment, the preset duration is 72 hours, and the preset time resolution is 1 hour; the minimum operating load baseline sequence of the power grid (e.g., 500 MW from 0:00 to 5:00, 550 MW at 6:00, 583.33 MW at 17:00, and 650 MW from 18:00 to 22:00, etc.), the total feed-in power sequence of new energy sources (e.g., 88 MW at 1:00, 85.5 MW at 2:00, 111 MW at 10:00, and 58 MW at 18:00, etc.), and the load demand for some time periods in the power grid load forecast data are: 520 MW at 0:00, 510 MW at 2:00, 600 MW at 10:00, 620 MW at 17:00, and 680 MW at 18:00.

[0148] After aligning the three datasets along the timeline, five typical time points were selected for evaluation: At 1:00 AM: The minimum operating load baseline value of the power grid is 500 MW, the total feed-in power value of new energy sources is 88 MW, and the load demand for each time period is 520 MW. Since 88 MW < 500 MW, it is initially determined that "there is a risk of insufficient downward support for feed-in". Upon checking adjacent time points (11:00 PM and 1:00 AM), the determination result at 11:00 PM was "there is a risk of insufficient downward support for feed-in", and the determination result at 1:00 AM was "there is a risk of insufficient downward support for feed-in". The final determination result is confirmed to be "there is a risk of insufficient downward support for feed-in".

[0149] At 2 o'clock: the minimum operating load baseline value of the power grid is 500 MW, the total feed-in power value of new energy is 85.5 MW, and the load demand for each time period is 510 MW; 85.5 MW < 500 MW, and it is initially determined that "there is a risk of insufficient downward support for feed-in". The adjacent time points (1 o'clock and 3 o'clock) have the same judgment results, and the final result is confirmed.

[0150] At 10:00: The minimum operating load baseline value of the power grid is 550 MW, the total power that can be fed into the grid from new energy sources is 111 MW, and the load demand for each time period is 600 MW. Since 111 MW < 550 MW, it is initially determined that "there is a risk of insufficient downward support for feed-in". The adjacent time points (9:00 and 11:00) have the same determination results, and the final result is confirmed.

[0151] At 17:00: The minimum operating load baseline value of the power grid is 583.33 MW, the total power that can be fed into the grid from new energy sources is assumed to be 600 MW, and the load demand for each time period is 620 MW; 600 MW > 583.33 MW and < 620 MW, it is initially determined that "feasibility of feeding into the grid is available". The final result is confirmed at adjacent time points (16:00 and 18:00).

[0152] At 18:00: The minimum operating load baseline value of the power grid is 650 MW, the total power that can be fed into the grid from new energy sources is assumed to be 680 MW, and the load demand for each time period is 680 MW; 680 MW > 650 MW and = 680 MW, it is initially determined that "feasibility of feeding into the grid is available". The adjacent time points (17:00 and 19:00) have the same determination results, and the final result is confirmed.

[0153] Step 4032: Based on the feasibility assessment results of new energy feed-in, determine whether the new energy power generation is within the grid's acceptable operating range at the corresponding time point, and obtain the scheduling responsibility period within the future preset time.

[0154] Optionally, the grid-acceptable operating range refers to the power range within which, after the renewable energy generation power is connected to the grid, the grid's voltage, frequency, and other operating parameters can be kept stable within acceptable ranges (voltage deviation not exceeding ±7%, frequency deviation not exceeding ±0.2 Hz). The dispatch responsibility period refers to the continuous time interval corresponding to different dispatch strategies, divided according to the feasibility assessment results of renewable energy feed-in. The dispatch responsibility period includes the full acceptance period, the limited acceptance period, and the non-acceptable period.

[0155] Optionally, the full acceptance period is a continuous time interval that represents the total renewable energy power that can be fed into the grid, which does not exceed the grid load demand (per-period load demand) and is not lower than the minimum operating load baseline (grid minimum operating load baseline value). During this period, renewable energy power generation is within the grid's acceptable operating range, and the grid can fully accept renewable energy power generation.

[0156] Optionally, the restricted access period: represents a continuous time interval in which the total power that new energy sources can feed into exceeds the grid load demand (load demand per time period) and there is a risk of exceeding the limit. During this period, the power generation of new energy sources exceeds the upper limit of the grid's acceptable operating range, and the access of new energy power generation needs to be restricted through dispatching measures.

[0157] Optionally, the unacceptable period: characterized by a continuous time interval in which the total power that can be fed into the grid is lower than the minimum operating load baseline of the grid (the minimum operating load baseline value of the grid) and there is a risk of insufficient downward support. During this period, the power generation of new energy is lower than the lower limit of the operating range that the grid can accept, and cannot support the safe operation of the grid on its own. It is necessary to rely on other power sources (such as thermal power and hydropower) to supplement the power supply.

[0158] Optionally, the specific division process of this embodiment of the invention is as follows: the result of the feasibility determination of new energy feed-in at each time point is matched with the definition of the scheduling responsibility period: "feed-in feasible" corresponds to the time point of the full acceptance period, "feed-in has the risk of exceeding the upward limit" corresponds to the time point of the limited acceptance period, and "feed-in has the risk of insufficient downward support" corresponds to the time point of the unacceptable period.

[0159] Furthermore, consecutive intervals are integrated for time points with the same judgment result. Consecutive time points with consistent judgment results are grouped into a scheduling responsibility period. Each scheduling responsibility period must specify the start time, end time, and corresponding period type (full acceptance, restricted acceptance, and non-acceptance).

[0160] Furthermore, boundary verification is performed on the integrated scheduling responsibility periods to check the accuracy of the boundary time point determination results of two adjacent scheduling responsibility periods of different types, ensuring that the boundary division has no overlap or omission; if there is an isolated determination result for a single time point (different from the types of the preceding and following time periods), then the time point is merged into the adjacent scheduling responsibility period with a longer duration, or the assignment is adjusted after re-determination based on the numerical relationship of the time point, and the complete scheduling responsibility period division result within the future preset duration is output.

[0161] In one embodiment, the preset duration is 72 hours (3 days), the preset time resolution is 1 hour, and step 4031 has obtained the feasibility assessment results of new energy feed-in at each time point. The assessment results of the first day (0:00-23:00) are selected for the division of scheduling responsibility periods: from 0:00 to 16:00, the assessment results of all time points are "there is a risk of insufficient downward support for feed-in"; from 17:00 to 22:00, the assessment results of all time points are "feasible for feed-in"; at 23:00, the assessment result is "feasible for feed-in".

[0162] The period from 0:00 to 16:00 is considered a continuous time interval with "insufficient downward support risk for feed-in," starting at 0:00 and ending at 16:00, and is therefore deemed an unacceptable period. The period from 17:00 to 23:00 is considered a continuous time interval with "feed-in feasibility," starting at 17:00 and ending at 23:00, and is therefore deemed a period of full acceptance. Boundary verification is performed: the boundary time points between the unacceptable period (0:00-16:00) and the full acceptance period (17:00-23:00) are 16:00 and 17:00. 16:00 is deemed to have "insufficient downward support risk for feed-in," and 17:00 is deemed to have "feed-in feasibility." The judgment results are accurate, with no overlap or omissions in the boundaries.

[0163] Optionally, the typical time period of the second day can be further divided as follows: Assume that the judgment result for the second day from 6:00 to 16:00 is "there is a risk of insufficient downward support for feed-in" (unacceptable period, 6:00 to 16:00); the judgment result for the second day from 17:00 to 21:00 is "feed-in is feasible" (full acceptance period, 17:00 to 21:00); and the judgment result for the second day from 22:00 to 23:00 is "there is a risk of upward overshooting of feed-in limits" (restricted acceptance period, 22:00 to 23:00).

[0164] During boundary verification, the restricted admission period (22:00-23:00) consists of two consecutive time points with clear boundaries and no overlap with the adjacent full admission period (17:00-21:00). The judgment result is accurate and no merging or adjustment is required. The final result is the allocation of scheduling responsibility periods within the future preset duration, such as: Day 1 0:00-16:00 (unacceptable), 17:00-23:00 (full admission); Day 2 6:00-16:00 (unacceptable), 17:00-21:00 (full admission), 22:00-23:00 (restricted admission), etc.

[0165] Step 4033: Based on the total renewable energy feed-in power sequence and the minimum operating load baseline sequence of the power grid within each scheduling responsibility period, determine the feed-in power subsequence for each scheduling responsibility period.

[0166] Optionally, the power prediction system determines the feed-in power subsequence for each scheduling responsibility period based on the total renewable energy feed-in power sequence and the minimum operating load baseline sequence of the power grid within each scheduling responsibility period, as described in steps 40331 to 40333.

[0167] Step 4034: The feed-in power subsequences of each scheduling responsibility period are spliced ​​and integrated in chronological order to obtain the prediction result of new energy power generation in each region to be predicted within a preset time period in the future.

[0168] Optionally, the feed-in power subsequence refers to an ordered data set consisting of the optimal renewable energy feed-in power values ​​at each time point within a specific scheduling responsibility period. The time sequence refers to the natural flow of time from the start time to the end time of a future preset duration.

[0169] Optionally, the specific splicing and integration process in this embodiment of the invention is as follows: First, analyze all feed power subsequences for all scheduling responsibility periods, clarify the time range (start time, end time) corresponding to each subsequence, and sort all subsequences according to their chronological order, with the sorting criterion being the order of the start times of each subsequence. Second, connect the sorted feed power subsequences sequentially in chronological order, ensuring seamless connection between the end time of the previous subsequence and the start time of the next subsequence, guaranteeing that the spliced ​​sequence has no time breaks or time overlaps.

[0170] Furthermore, the spliced ​​complete sequence undergoes a time continuity check, verifying whether the power value at each time point is unique and whether the time interval meets the preset time resolution requirements. If there is a sudden change in power value at the splicing boundary (the power difference between two adjacent time points is greater than the preset change threshold, which is 50 MW per hour), the power value at the change point is smoothed by taking the average of the two power values ​​at the change point as the transition power value to ensure the temporal continuity of the sequence. Finally, the checked and smoothed complete sequence undergoes a final review to confirm that the sequence covers all time points of the preset future duration, has no missing data, and has no abnormal power values ​​(such as negative numbers or values ​​exceeding the maximum total feed-in power of new energy sources). Once the review is passed, it becomes the prediction result for the new energy power generation of the region to be predicted.

[0171] In one embodiment, the preset duration is 72 hours (3 days), the preset time resolution is 1 hour, and step 4033 has obtained the feed power subsequence for each scheduling responsibility period. Some subsequences are as follows: Unacceptable period (0:00-16:00 on the first day) power input subsequence: The power value at each time point from 0:00 to 16:00 is the total renewable energy input power value at the corresponding time point (e.g., 85 MW at 0:00, 85.5 MW at 2:00, 111 MW at 10:00, etc.). Full acceptance period (17:00-23:00 on the first day) feed-in power subsequence: The power value at each time point from 17:00 to 23:00 is the total renewable energy feed-in power value at the corresponding time point (e.g., 600 MW at 17:00, 680 MW at 18:00, etc.). Unacceptable period (6:00-16:00 the next day) feed-in power subsequence: The power value at each time point from 6:00 to 16:00 is the total renewable energy feed-in power value at the corresponding time point (e.g., 90 MW at 6:00, 115 MW at 10:00, etc.). Full acceptance period (17:00-21:00 on the second day) power input subsequence: The power value at each time point from 17:00 to 21:00 is the total renewable energy power that can be fed in at the corresponding time point (e.g., 610 MW at 17:00). Feed power subsequence during the restricted admission period (22:00-23:00 on the second day): The power value at each time point from 22:00 to 23:00 is the optimized restricted power value (e.g., 650 MW at 22:00, 640 MW at 23:00, etc.).

[0172] The subsequences are ordered chronologically as follows: Day 1 0:00-16:00 subsequence → Day 1 17:00-23:00 subsequence → Day 2 06:00-16:00 subsequence → Day 2 17:00-21:00 subsequence → Day 2 22:00-23:00 subsequence… The power values ​​at 16:00 (110 MW) and 17:00 (600 MW) of Day 1 are concatenated; the power values ​​at 16:00 (115 MW) and 17:00 (610 MW) of Day 2 are concatenated; and the power values ​​at 21:00 (610 MW) and 22:00 (650 MW) of Day 2 are concatenated. Continuity verification was performed: The power difference between 16:00 and 17:00 on the first day was 490 MW, which exceeded the preset abrupt change threshold of 50 MW. This abrupt change was smoothed out, and (110+600) / 2=355 MW was taken as the transition power value at 16:30 (because the preset time resolution is 1 hour, the power value at 17:00 was adjusted to 355 MW, and subsequent time points gradually transitioned to 600 MW); other splicing boundary differences were all less than 50 MW and required no processing. Finally, the optimal renewable energy power generation values ​​for each time point within 72 hours were obtained, and these values ​​were arranged in chronological order to form the renewable energy power generation prediction results for the region to be predicted.

[0173] The embodiments of the present invention, from feasibility assessment to time period division to accurate prediction of power generation, enable the output of new energy power generation prediction results to tap the potential of new energy power generation and improve the new energy absorption rate, while adhering to the constraints of safe and stable operation of the power grid, thereby improving the new energy absorption capacity and the stability of power grid operation.

[0174] Optionally, the processes of steps 40331 to 40333 include: Step 40331: Based on the power value of the total renewable energy feed-in power sequence during the period when it can be fully accepted, determine the feed-in power subsequence of renewable energy generation during the period when it can be fully accepted.

[0175] Optionally, from the scheduling responsibility period division results in step 4032, the start and end times of the fully acceptable period are extracted, and all time points included in the period are identified (divided according to a preset time resolution, such as one time point per hour). Next, the total renewable energy feed-in power sequence is called, and the total renewable energy feed-in power value corresponding to each time point is extracted one by one according to the time point range of the fully acceptable period to form an initial feed-in power dataset.

[0176] Furthermore, the initial feed-in power dataset is validated for validity. This validation includes: data completeness (ensuring a corresponding power value for each time point without missing values) and data rationality (power values ​​must be positive and not exceed the maximum value of the total feed-in power sequence from renewable energy sources, while also meeting the "feasibility of feed-in" criteria in step 4031, i.e., greater than or equal to the minimum operating load baseline value of the power grid at the corresponding time point, and less than or equal to the time-period load demand at the corresponding time point). Finally, the validated initial feed-in power dataset is arranged chronologically according to the time periods during which full feed-in is possible, forming a feed-in power subsequence for each period. If missing data is found during validation, it is supplemented using the average power value of two adjacent normal time points; if unreasonable data is found (e.g., exceeding constraints), the original data of the total feed-in power sequence from renewable energy sources is re-examined, and the final power value is determined after eliminating extraction errors.

[0177] In one embodiment, the preset duration is 72 hours, and the preset time resolution is 1 hour. Step 4032 obtains a fully absorbable period of 17:00-23:00 on the first day, with a total of 7 time points (17:00, 18:00, 19:00, 20:00, 21:00, 22:00, 23:00). In the total renewable energy feed-in power sequence obtained in step 402, the power values ​​corresponding to the above time points are: 600 MW at 17:00, 680 MW at 18:00, 670 MW at 19:00, and 200 MW at 23:00. At 0:00, the load is 660 MW; at 21:00, it is 650 MW; at 22:00, it is 640 MW; and at 23:00, it is 630 MW. The minimum operating load baseline values ​​of the power grid at the corresponding time points are 583.33 MW (17:00) or 650 MW (18:00-23:00). The load demand for each time period is 620 MW at 17:00, 680 MW at 18:00, 690 MW at 19:00, 680 MW at 20:00, 670 MW at 21:00, 660 MW at 22:00, and 650 MW at 23:00.

[0178] The power prediction system extracts the total feed-in power values ​​of new energy sources at the above 7 time points, forming an initial feed-in power dataset: [600, 680, 670, 660, 650, 640, 630]. Validity verification is performed: Regarding data completeness, all 7 time points have corresponding power values ​​with no missing values; regarding data rationality, all power values ​​are positive and meet the criteria of being "greater than or equal to the minimum operating load baseline of the power grid and less than or equal to the load demand for each time period" (e.g., 600 MW at 17:00 > 583.33 MW and < 620 MW, 680 MW at 18:00 > 650 MW and = 680 MW, etc.). After verification, the feed power subsequences for the fully accommodative time periods are arranged in chronological order from 5 PM to 11 PM: 600 MW at 5 PM, 680 MW at 6 PM, 670 MW at 7 PM, 660 MW at 8 PM, 650 MW at 9 PM, 640 MW at 10 PM, and 630 MW at 11 PM.

[0179] Step 40332: Based on the upper limit of load demand during the restricted acceptance period of the grid minimum operating load baseline sequence, determine the upper limit of the maximum allowable feed-in power of new energy power generation during the restricted acceptance period.

[0180] Optionally, the maximum allowable feed-in power limit refers to the maximum renewable energy generation power that the grid can safely accept at a certain point in time within the restricted acceptance period, which is equal to the upper limit of load demand at the corresponding point in time (load demand per time period). Therefore, from the scheduling responsibility period division results in step 4032, the start and end times of the restricted acceptance period are extracted to determine all time points included in the period.

[0181] Furthermore, by calling the grid load forecast data associated with the grid minimum operating load baseline sequence, the time-period load demand at each time point within the restricted acceptance period is extracted and determined as the upper limit of load demand at the corresponding time point, thus obtaining the maximum allowable feed-in power limit (i.e., the upper limit of load demand) at each time point. Next, by calling the total renewable energy feed-in power sequence, the original power values ​​at all time points within the corresponding restricted acceptance period are extracted from this sequence.

[0182] Furthermore, the original power value at each time point is compared with the maximum allowable feed power limit: if the original power value is less than or equal to the maximum allowable feed power limit, the original power value is retained as the candidate power value for that time point; if the original power value is greater than the maximum allowable feed power limit, the maximum allowable feed power limit is used as the candidate power value for that time point (i.e., an upper limit is imposed). Finally, the temporal continuity of the candidate power values ​​at all time points is checked. If the difference between the candidate power values ​​of two adjacent time points is greater than a preset abrupt change threshold (e.g., 50 MW per hour), the candidate power value at the abrupt change is smoothed (the average of the two power values ​​is taken as the transition power value); after the check passes, the candidate power values ​​are arranged in chronological order according to the restricted acceptance period to form the feed power subsequence for the restricted acceptance period.

[0183] In one embodiment, step 4032 divides a restricted access period into two time points: 22:00-23:00 on the second day. In the grid load forecast data, the time-period load demand (load demand ceiling) at the corresponding time points is 660 MW at 22:00 and 650 MW at 23:00. Therefore, the maximum allowable feed-in power ceilings for the two time points are 660 MW and 650 MW, respectively. In the total renewable energy feed-in power sequence, the original power values ​​at the corresponding time points are 700 MW at 22:00 and 680 MW at 23:00. The preset mutation threshold is 50 MW. After extracting the maximum allowable feed-in power ceiling and the original power values, the power forecast system compares the ceiling limits. At 22:00, the original power value was 700 MW, which is greater than the maximum allowable feed-in power limit of 660 MW, so the candidate power value was determined to be 660 MW; at 23:00, the original power value was 680 MW, which is greater than the maximum allowable feed-in power limit of 650 MW, so the candidate power value was determined to be 650 MW.

[0184] A time-series continuity check was performed: the difference between the candidate power values ​​at 22:00 and 23:00 was 10 MW (660-650), which is less than the preset abrupt change threshold of 50 MW, so no smoothing processing was required. Arranged in chronological order from 22:00 to 23:00, the feed power subsequence for the restricted reception period was obtained: 660 MW at 22:00 and 650 MW at 23:00.

[0185] Step 40333: Based on the preset system safety lower limit in the minimum operating load baseline sequence of the power grid, determine the minimum guaranteed feed-in power lower limit for renewable energy generation during the unacceptable period. Increase the power value of the total renewable energy feed-in power sequence during the unacceptable period to no less than the minimum guaranteed feed-in power lower limit, thus obtaining the feed-in power subsequence for renewable energy generation during the unacceptable period.

[0186] Optionally, the minimum guaranteed feed-in power limit refers to the minimum feed-in power value that renewable energy generation must reach at a certain point in time within an unacceptable period. It is equal to the minimum operating load baseline value of the power grid for the corresponding period. Its purpose is to prevent the power grid from being unable to maintain the operation of core equipment and meet the electricity demand of core users due to excessively low renewable energy feed-in power. Therefore, from the scheduling responsibility period division results in step 4032, the start and end times of the unacceptable period are extracted to clarify all time points included in the period.

[0187] Furthermore, the minimum operating load baseline sequence of the power grid is invoked, and the minimum operating load baseline value of the power grid at each time point within the unacceptable period is extracted and determined as the minimum guaranteed feed-in power lower limit at the corresponding time point (because the minimum guaranteed feed-in power lower limit is equal to the minimum operating load baseline value of the power grid at the corresponding time period).

[0188] Furthermore, the total feed-in power sequence of new energy sources is invoked, and the original power values ​​of all time points within the corresponding unacceptable period in the sequence are extracted.

[0189] Furthermore, the original power value at each time point is compared with the minimum guaranteed feed power lower limit: if the original power value is greater than or equal to the minimum guaranteed feed power lower limit, the original power value is retained as the candidate power value for that time point; if the original power value is less than the minimum guaranteed feed power lower limit, the minimum guaranteed feed power lower limit is used as the candidate power value for that time point (i.e., the lower limit is raised). Finally, the temporal continuity of the candidate power values ​​at all time points is checked. If the difference between the candidate power values ​​of two adjacent time points is greater than a preset abrupt change threshold (e.g., 50 MW per hour), the candidate power value at the abrupt change is smoothed (the average of the two power values ​​is taken as the transition power value); after the check passes, the candidate power values ​​are arranged in chronological order according to the unacceptable time period to form the feed power subsequence of the unacceptable time period.

[0190] In one embodiment, step 4032 divides a certain unacceptable period into the first day from 0:00 to 16:00, and selects three typical time points (0:00, 2:00, and 10:00) within this period for analysis; in the grid minimum operating load baseline sequence, the grid minimum operating load baseline value (minimum guaranteed feed-in power lower limit) at the corresponding time point is 500 MW; in the total renewable energy feed-in power sequence, the original power values ​​at the corresponding time points are 88 MW at 0:00, 85.5 MW at 2:00, and 111 MW at 10:00; the preset mutation threshold is 50 MW.

[0191] After extracting the minimum guaranteed feed-in power lower limit and the original power value, the power prediction system performs a lower limit improvement comparison: at 0:00, the original power value is 88 MW < the minimum guaranteed feed-in power lower limit of 500 MW, so the candidate power value is determined to be 500 MW; at 2:00, the original power value is 85.5 MW < 500 MW, so the candidate power value is determined to be 500 MW; at 10:00, the original power value is 111 MW < 500 MW, so the candidate power value is determined to be 500 MW. A time-series continuity check is performed: the candidate power values ​​at 0:00, 2:00, and 10:00 are all 500 MW, the difference is 0, which is less than the preset abrupt change threshold of 50 MW, so no smoothing processing is required. Arranged in chronological order, the feed-in power data for the above three time points are obtained: 0:00 500 MW, 2:00 500 MW, and 10:00 500 MW; and so on. The candidate power values ​​(all 500 MW) for all time points from 0:00 to 16:00 are arranged in chronological order to form the feed-in power subsequence for the unacceptable period.

[0192] The embodiments of the present invention achieve precise adaptation to different renewable energy input scenarios, so that the final input power subsequence for each time period not only fully explores the potential of renewable energy power generation, but also follows the core constraints of safe and stable grid operation, thereby accurately obtaining the renewable energy power generation prediction results, improving the renewable energy absorption capacity and grid operation stability.

[0193] Furthermore, the artificial intelligence-based new energy power generation prediction system provided by the present invention will be described below. The artificial intelligence-based new energy power generation prediction system described below can be referred to in correspondence with the artificial intelligence-based new energy power generation prediction method described above.

[0194] Optionally, refer to Figure 2 , Figure 2 This is a schematic diagram of the artificial intelligence-based new energy power generation prediction system provided by the present invention. The artificial intelligence-based new energy power generation prediction system includes: The power curve prediction module 210 is used to determine the initial feed-in power curve corresponding to different types of new energy power plants in each new energy power generation cluster based on the installed capacity data of each type of new energy power plant in each new energy power generation cluster and combined with the historical power operation data in the region. The collaborative correlation analysis module 220 is used to perform collaborative correlation analysis based on the geographical location distribution data of each new energy power generation cluster to obtain the power collaborative complementarity relationship between each new energy power generation cluster. The power curve optimization module 230 is used to determine the cluster collaborative feed-in power curve of each new energy power generation cluster based on the initial feed-in power curves corresponding to different types of new energy power plants in each new energy power generation cluster and the power collaborative and complementary relationship between each new energy power generation cluster. The power generation prediction module 240 is used to determine the power generation prediction result of each region to be predicted based on the cluster collaborative feed-in power curve of each new energy power generation cluster and the grid load prediction data of the region within a preset time period.

[0195] This invention, based on the initial feed-in power curves corresponding to different types of new energy power plants within each new energy power generation cluster, overcomes the limitation of single-station independent prediction relying solely on local data from a single power plant. By integrating installed capacity data of similar power plants within the cluster with historical power operation data for the region, the power variation patterns of different power types within the cluster are clarified. Based on the geographical distribution data of each cluster, the power synergy and complementarity relationship between various new energy power generation clusters is obtained, clarifying the power correlation characteristics between different clusters and addressing the issue of neglecting inter-cluster synergy and correlation. Based on the initial feed-in power curves and power synergy and complementarity relationships, the cluster synergy feed-in power curves for each new energy power generation cluster are determined. The power complementarity between clusters is used to correct the initial feed-in power curves, while integrating the power characteristics of different power types within the cluster. This reduces the impact of power fluctuations from a single power source or a single cluster, solving the problem of poor prediction stability caused by the lack of consideration for synergy and complementarity in single-station independent prediction methods. Based on the cluster collaborative feed-in power curve and the grid load forecast data within the preset time period of the region, the forecast results of renewable energy power generation in each region to be predicted are determined. This makes the renewable energy power generation forecast results integrate the characteristics of cluster collaboration and power complementarity. Therefore, the grid dispatch center can formulate dispatch plans based on the accurate renewable energy power generation forecast results, avoid the situation of renewable energy power surplus that cannot be absorbed or insufficient supply, and improve the renewable energy absorption capacity and grid operation stability.

[0196] Please see Figure 3 , Figure 3 An embodiment diagram of an electronic device provided in accordance with the present invention. For example... Figure 3 As shown, an embodiment of the present invention provides an electronic device 300, including a memory 310, a processor 320, and a computer program 311 stored in the memory 310 and executable on the processor 320. When the processor 320 executes the computer program 311, it performs the following steps: For each region to be predicted, based on the installed capacity data of each type of new energy power station within each new energy power generation cluster, and combined with the historical power operation data of the region, the initial feed-in power curve corresponding to different types of new energy power stations within each new energy power generation cluster is determined. Based on the geographical location distribution data of each new energy power generation cluster, a collaborative correlation analysis is conducted to obtain the power synergy and complementarity relationship between the various new energy power generation clusters. Based on the initial feed-in power curves corresponding to different types of new energy power plants within each new energy power generation cluster, and combined with the power synergy and complementarity relationship between each new energy power generation cluster, the cluster synergy feed-in power curves of each new energy power generation cluster are determined. Based on the cluster collaborative feed-in power curves of various new energy power generation clusters and the grid load forecast data of the region within a preset time period, the new energy power generation forecast results for each region to be predicted are determined.

[0197] Please see Figure 4 , Figure 4 An embodiment diagram of a computer-readable storage medium provided in accordance with an embodiment of the present invention is shown. Figure 4 As shown, this embodiment provides a computer-readable storage medium 400 on which a computer program 311 is stored. When the computer program 311 is executed by a processor, it performs the following steps: For each region to be predicted, based on the installed capacity data of each type of new energy power station within each new energy power generation cluster, and combined with the historical power operation data of the region, the initial feed-in power curve corresponding to different types of new energy power stations within each new energy power generation cluster is determined. Based on the geographical location distribution data of each new energy power generation cluster, a collaborative correlation analysis is conducted to obtain the power synergy and complementarity relationship between the various new energy power generation clusters. Based on the initial feed-in power curves corresponding to different types of new energy power plants within each new energy power generation cluster, and combined with the power synergy and complementarity relationship between each new energy power generation cluster, the cluster synergy feed-in power curves of each new energy power generation cluster are determined. Based on the cluster collaborative feed-in power curves of various new energy power generation clusters and the grid load forecast data of the region within a preset time period, the new energy power generation forecast results for each region to be predicted are determined.

[0198] On the other hand, the present invention also provides a computer program product, which includes a computer program that can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer is able to execute the artificial intelligence-based new energy power generation prediction method provided by the above methods, the method comprising: For each region to be predicted, based on the installed capacity data of each type of new energy power station within each new energy power generation cluster, and combined with the historical power operation data of the region, the initial feed-in power curve corresponding to different types of new energy power stations within each new energy power generation cluster is determined. Based on the geographical location distribution data of each new energy power generation cluster, a collaborative correlation analysis is conducted to obtain the power synergy and complementarity relationship between the various new energy power generation clusters. Based on the initial feed-in power curves corresponding to different types of new energy power plants within each new energy power generation cluster, and combined with the power synergy and complementarity relationship between each new energy power generation cluster, the cluster synergy feed-in power curves of each new energy power generation cluster are determined. Based on the cluster collaborative feed-in power curves of various new energy power generation clusters and the grid load forecast data of the region within a preset time period, the new energy power generation forecast results for each region to be predicted are determined.

[0199] The system embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0200] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0201] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A new energy power generation prediction method based on artificial intelligence, characterized in that, include: For each region to be predicted, based on the installed capacity data of each type of new energy power station within each new energy power generation cluster, and combined with the historical power operation data of the region, the initial feed-in power curve corresponding to different types of new energy power stations within each new energy power generation cluster is determined. Based on the geographical location distribution data of each new energy power generation cluster, a collaborative correlation analysis is conducted to obtain the power synergy and complementarity relationship between the various new energy power generation clusters. Based on the initial feed-in power curves corresponding to different types of new energy power plants within each new energy power generation cluster, and combined with the power synergy and complementarity relationship between each new energy power generation cluster, the cluster synergy feed-in power curves of each new energy power generation cluster are determined. Based on the cluster collaborative feed-in power curves of various new energy power generation clusters and the grid load forecast data of the region within a preset time period, the new energy power generation forecast results for each region to be predicted are determined.

2. The artificial intelligence-based new energy power generation prediction method according to claim 1, characterized in that, The steps for determining the predicted renewable energy power generation for each region to be predicted include: Based on the time-period load demand in the power grid load forecast data, the minimum operating load baseline sequence of the power grid in the region at a preset time resolution is determined; the minimum operating load baseline sequence of the power grid represents the minimum power supply level at which the regional power grid can maintain safe and stable operation at each time point. Based on the actual feed-in power value of each cluster at each time point, as represented by the cluster collaborative feed-in power curve of each new energy power generation cluster, the total feed-in power sequence of new energy in the region under the same time resolution is determined; the total feed-in power sequence of energy represents the physical maximum output capacity of the new energy power generation system at each time point. Based on the minimum operating load baseline sequence of the power grid and the total feed-in power sequence of new energy sources, the prediction results of new energy power generation for each region to be predicted are determined.

3. The artificial intelligence-based new energy power generation prediction method according to claim 2, characterized in that, The process of determining the renewable energy power generation forecast results for each region to be predicted, based on the minimum operating load baseline sequence of the power grid and the total feed-in power sequence of renewable energy sources, includes: Based on the minimum operating load baseline sequence of the power grid and the total feed-in power sequence of new energy sources, determine the feasibility assessment result of new energy feed-in in the region at each time point; Based on the feasibility assessment results of the new energy feed-in, it is determined whether the new energy power generation is within the acceptable operating range of the power grid at the corresponding time point, and the scheduling responsibility period within the future preset time period is obtained. Based on the total renewable energy feed-in power sequence and the minimum operating load baseline sequence of the power grid during each scheduling responsibility period, the feed-in power sub-sequence for each scheduling responsibility period is determined. By splicing and integrating the feed-in power subsequences of each scheduling responsibility period in chronological order, the predicted power generation of new energy in each region to be predicted within a preset time period is obtained.

4. The artificial intelligence-based new energy power generation prediction method according to claim 3, characterized in that, The scheduling responsibility period includes the full acceptance period, the limited acceptance period, and the non-acceptance period; The full acceptance period indicates that the total power that new energy sources can feed into does not exceed the grid load demand and is not lower than the minimum operating load baseline. The restricted acceptance period indicates that the total power that new energy sources can feed into exceeds the grid load demand, and there is a risk of exceeding the limit upwards; The unacceptable period indicates that the total feed-in power of new energy sources is lower than the minimum operating load baseline of the power grid, and there is a risk of insufficient downward support.

5. The artificial intelligence-based new energy power generation prediction method according to claim 4, characterized in that, Based on the total renewable energy feed-in power sequence within each scheduling responsibility period, the feed-in power sub-sequence for each scheduling responsibility period is determined, including: For periods when full acceptance is possible, the power input subsequence of renewable energy generation during the periods when full acceptance is possible is determined based on the power value of the total renewable energy input power sequence during the periods when full acceptance is possible.

6. The artificial intelligence-based new energy power generation prediction method according to claim 4, characterized in that, Based on the total renewable energy feed-in power sequence and the grid minimum operating load baseline sequence for each scheduling responsibility period, the feed-in power sub-sequence for each scheduling responsibility period is determined, including: For the restricted access period, the maximum allowable feed-in power of new energy generation during the restricted access period is determined based on the upper limit of load demand during the restricted access period, which is based on the minimum operating load baseline sequence of the power grid. Based on the maximum allowable feed-in power limit, the power value of the total feed-in power sequence of the new energy sources during the restricted acceptance period is restricted, thus obtaining the feed-in power subsequence of new energy power generation during the restricted acceptance period.

7. The artificial intelligence-based new energy power generation prediction method according to claim 4, characterized in that, Based on the total renewable energy feed-in power sequence and the grid minimum operating load baseline sequence for each scheduling responsibility period, the feed-in power sub-sequence for each scheduling responsibility period is determined, including: For unacceptable periods, the minimum guaranteed feed-in power limit for new energy power generation is determined based on the preset system safety lower limit in the minimum operating load baseline sequence of the power grid during the unacceptable periods; the minimum guaranteed feed-in power limit is equal to the minimum operating load baseline value of the power grid for the corresponding period. The power value of the total feed-in power sequence of the new energy sources during the unacceptable period is increased to no less than the minimum guaranteed feed-in power limit, thus obtaining the feed-in power subsequence of the new energy generation during the unacceptable period.

8. The artificial intelligence-based new energy power generation prediction method according to any one of claims 1 to 7, characterized in that, The steps for determining the cluster-coordinated feed-in power curves for each new energy power generation cluster include: Based on the time-series power change characteristics of the initial feed-in power curves corresponding to different types of new energy power plants within each new energy power generation cluster, a multi-type power source time-series feed-in behavior sequence is determined for each new energy power generation cluster; the multi-type power source time-series feed-in behavior sequence characterizes the power output fluctuation mode of different types of new energy power plants within each new energy power generation cluster and their phase distribution characteristics in the time dimension. Based on the power synergy and complementarity among various new energy power generation clusters, the trend relationship of power output trends between each pair of new energy power generation clusters at the same point in time is determined. Based on the trend relationship, determine whether there is complementary power output potential between two new energy power generation clusters at any point in time, and obtain the complementary potential results of each pair of new energy power generation clusters at any point in time. Based on the complementary potential results of each pair of new energy power generation clusters and the time-series feed-in behavior sequence of multiple types of power sources in each new energy power generation cluster, the cluster collaborative feed-in power curve of each new energy power generation cluster is determined.

9. The artificial intelligence-based new energy power generation prediction method according to claim 8, characterized in that, The method of determining the cluster collaborative feed-in power curve of each new energy power generation cluster based on the complementary potential results of each pair of new energy power generation clusters and the multi-type power source timing feed-in behavior sequence of each new energy power generation cluster includes: Based on the time-series feed-in behavior sequences of various new energy power generation clusters and the complementary potential results of each pair of new energy power generation clusters, power synergy cluster combinations with conditions for coordinated operation in the time dimension are determined; each power synergy cluster combination includes two or more new energy power generation clusters with conditions for coordinated operation in the time dimension, and any two new energy power generation clusters have complementary power output potential in at least one time period. Based on the time-series feed-in behavior sequence of multiple types of power sources in each new energy power generation cluster within each power coordination cluster combination, the joint power fluctuation profile at each time point is determined. Based on the joint power fluctuation profile of each power coordination cluster combination, the coordination fluctuation characteristics used to characterize the power output stability under the combined action of multiple new energy power generation clusters within the cluster combination are determined. Based on the complementary structural relationship and cooperative fluctuation characteristics among the various new energy power generation clusters within each power cooperative cluster combination, the cluster cooperative feed-in power curve of each new energy power generation cluster is determined.

10. The artificial intelligence-based new energy power generation prediction method according to claim 9, characterized in that, The determination of the cluster collaborative feed-in power curve for each new energy power generation cluster, based on the complementary structural relationship and collaborative fluctuation characteristics among the various new energy power generation clusters within each power collaborative cluster combination, includes: For each power synergy cluster combination, based on the synergy fluctuation characteristics and the complementary structural relationship between each new energy power generation cluster, the power synergy role of each new energy power generation cluster in the power synergy cluster combination is determined; the power synergy role includes power-dominant clusters and power-auxiliary clusters. Based on the time-series feed-in behavior sequence of various new energy power generation clusters, the power supply period undertaken by the power-dominant cluster in the power-coordinated cluster combination is determined, and the compensation time window for the power-assisted cluster to perform power compensation outside the power supply period is determined based on the power supply period. Based on the compensation time window and the joint power fluctuation profile, the power adjustment target range of the power auxiliary cluster within its respective compensation time window is determined, and the power value of the corresponding time period in the original multi-type power supply timing feed behavior sequence of the power auxiliary cluster is adjusted based on the power adjustment target range to obtain the adjusted multi-type power supply timing feed behavior sequence. By integrating the original multi-type power source timing behavior sequences of the power-dominant cluster and the adjusted multi-type power source timing behavior sequences of the power-assisted cluster, the cluster collaborative feed-in power curves of each new energy power generation cluster are obtained.

11. A new energy power generation prediction system based on artificial intelligence, characterized in that, The system is used to implement the artificial intelligence-based new energy power generation prediction method as described in any one of claims 1 to 10; the artificial intelligence-based new energy power generation prediction system includes: The power curve prediction module is used to determine the initial feed-in power curve corresponding to different types of new energy power plants in each new energy power generation cluster for each region to be predicted, based on the installed capacity data of each type of new energy power plant in each new energy power generation cluster and combined with the historical power operation data in the region. The collaborative correlation analysis module is used to perform collaborative correlation analysis based on the geographical location distribution data of each new energy power generation cluster to obtain the power synergy and complementarity relationship between each new energy power generation cluster. The power curve optimization module is used to determine the cluster collaborative feed-in power curve of each new energy power generation cluster based on the initial feed-in power curves corresponding to different types of new energy power plants within each new energy power generation cluster and the power synergy and complementarity relationship between each new energy power generation cluster. The power generation prediction module is used to determine the power generation prediction result of each region to be predicted based on the cluster collaborative feed-in power curve of each new energy power generation cluster and the grid load prediction data of the region within a preset time period.

12. An electronic device, comprising: Memory, used to store computer software programs; A processor for reading and executing computer software programs, characterized in that, when the processor executes the computer software programs, it implements the artificial intelligence-based new energy power generation prediction method as described in any one of claims 1 to 10.

13. A non-transitory computer-readable storage medium, wherein a computer software program is stored therein, characterized in that, When the computer software program is executed by the processor, it implements the artificial intelligence-based new energy power generation prediction method as described in any one of claims 1 to 10.

14. A computer program product, characterized in that, The system includes a computer program, which, when executed by a processor, implements the artificial intelligence-based new energy power generation prediction method as described in any one of claims 1 to 10.

Images