New energy annual generating capacity correction method for evaluating extreme weather influence

Abstract

The invention discloses a new energy annual generating capacity correction method for evaluating extreme weather influence, and belongs to the technical field of electric energy supply and power distribution system management, and the method comprises the steps: dividing a natural stage sequence based on a climate classification rule through multi-source meteorological observation data and climate reanalysis data, and constructing a dynamically growing scene evolution network; generating a path health degree score through a conflict digester, calculating a real-time path value in combination with a value mapping function and a path aging mechanism, and forming a stage anchoring evaluation value at an anchoring time node; and generating a two-dimensional deviation signal by using actually measured power generation data, driving the scoring logic and the growth logic to co-evolve, and synthesizing to obtain an annual power generation amount correction result. According to the method, a dynamically evolved scene network and a two-dimensional deviation feedback mechanism are adopted, and the scoring logic and the growth logic of the evaluation model can be driven to carry out coevolution, so that the self-adaptive accurate correction of annual power generation capacity prediction is realized, and the planning reliability and the operation stability of a power supply system are improved.

Description

Technical Field

[0001] This invention relates to the field of power supply and distribution system management technology, and in particular to a method for adjusting the annual power generation of new energy sources to assess the impact of extreme weather. Background Technology

[0002] New energy power generation, especially wind and solar power, plays an increasingly important role in modern power systems due to its clean and sustainable characteristics. As a crucial component of the power supply and distribution system, the stable integration and efficient utilization of new energy power are essential for ensuring grid security and optimizing the energy structure. However, new energy power generation is intermittent and fluctuating, with its output levels highly dependent on weather conditions, posing challenges to power system planning, dispatching, and stable operation. Therefore, forecasting and revising annual new energy power generation is a key step in ensuring a reliable power supply.

[0003] Among related technologies, Chinese invention patent CN119740786B discloses a method for correcting wind power prediction in power grids under extreme weather conditions. The method includes several steps: initiating a correction process by using extreme weather information contained in regular weather forecasts; optimizing and correcting numerical weather forecasts based on extreme weather event identification; determining the type of extreme weather; if it is a strong wind, then performing a one-way deviation correction for wind speed prediction and fitting the actual power characteristic curve to optimize the wind resource-power conversion model; if it is a cold wave, then directly fitting the actual power characteristic curve to optimize the wind resource-power conversion model; correcting the wind power prediction in power grids under extreme weather conditions by combining environmental protection settings and self-grid connection settings; and finally outputting the correction results.

[0004] Regarding the aforementioned technologies, the inventors believe that they have technical defects in practical applications. Existing wind power forecasting correction methods under extreme weather conditions mainly focus on regular corrections of single or short-term forecast results, making it difficult to uniformly assess the combined impact of multiple and different types of extreme weather events on an annual scale. Furthermore, these methods lack the ability to identify biases in the forecast path structure and do not possess a mechanism for adaptively evolving the correction logic based on actual power generation results, making it difficult to obtain stable and reliable annual power generation correction results under complex extreme climate conditions. Summary of the Invention

[0005] To address the aforementioned issues, this invention provides a method for correcting annual power generation forecasts for new energy sources in response to the impact of extreme weather. Employing a dynamically evolving scenario network and a two-dimensional bias feedback mechanism, this method drives the co-evolution of the scoring and growth logics of the assessment model, thereby achieving adaptive and accurate correction of annual power generation forecasts and improving the planning reliability and operational stability of the power supply system.

[0006] The above objectives can be achieved through the following approach:

[0007] A method for adjusting annual power generation of new energy sources to assess the impact of extreme weather includes: acquiring multi-source meteorological observation data and climate reanalysis data for a target year; dividing the target year into a natural stage sequence based on climate classification rules; acquiring the current natural stage in the natural stage sequence; creating multiple initial assessment paths and applying dynamic growth rules for dynamic management and structural updates to generate a scenario evolution network; acquiring multi-dimensional state information of each assessment path in the scenario evolution network and using a conflict resolver for comprehensive scoring to generate a path health score; based on the path health score, combined with a value mapping function and a path aging mechanism, dynamically calculating and updating the real-time path value of each assessment path in the scenario evolution network; and in the current natural stage sequence, obtaining the current natural stage in the natural stage sequence; acquiring multi-dimensional state information of each assessment path in the scenario evolution network and using a conflict resolver for comprehensive scoring to generate a path health score; and in the current natural stage sequence, acquiring the current natural stage in the natural stage sequence; acquiring multi-dimensional state information of each assessment path in the scenario evolution network and applying a conflict resolution mechanism for comprehensive scoring to generate a path health score; and in the current natural stage sequence, acquiring multi-dimensional state information of each assessment path in the scenario evolution network and applying a dynamic growth rule for dynamic management and structural updates to generate a path health score; and in the current natural stage sequence, acquiring multi-dimensional state information of each assessment path in the scenario evolution network and applying a multi-dimensional state information of each assessment path in the scenario evolution network; acquiring multi-dimensional state information of each assessment path in the scenario evolution network and applying a multi-dimensional state information of each assessment path in the scenario evolution network; acquiring multi-dimensional state information of each assessment path in the scenario evolution network and applying a multi-dimensional state information of each assessment path in the scenario evolution network; and in the current natural stage sequence, acquiring multi-dimensional state information of each assessment path in the scenario evolution network and generating a path health score; and in the current natural stage sequence, acquiring multi-dimensional state information of each assessment path in the scenario evolution network and generating a path health score to generate a path health score; and in the current natural stage sequence, acquiring multi-dimensional state information of each assessment path in the scenario evolution network and generating a path health score to generate a path health score; and At the anchor time node of the stage, the predicted power generation value in the scenario evolution network is weighted based on the real-time path value to generate the stage anchor evaluation value for the current natural stage; the measured power generation data of the current natural stage is obtained and compared with the stage anchor evaluation value to generate a two-dimensional deviation signal; the two-dimensional deviation signal is fed back to the conflict resolver and the dynamic growth rule to drive the scoring logic of the conflict resolver and the growth logic of the dynamic growth rule to co-evolve and generate an updated scoring configuration and growth configuration; based on the scoring configuration and growth configuration, all natural stages in the natural stage sequence are independently evaluated and synthesized with the stage anchor evaluation value to obtain the target annual power generation correction result.

[0008] Optionally, the step of dividing the target year into a natural stage sequence based on climate classification rules includes: acquiring multi-source meteorological observation data and climate reanalysis data for the target year, calculating climate seasonality characteristic indicators; determining climate seasonal boundary information based on the climate seasonality characteristic indicators and in combination with climate classification rules, splitting the target year into stages that are consistent in climate behavior and arranged in chronological order, and generating a natural stage sequence.

[0009] Optionally, the generation of the scenario evolution network includes: creating an initial evaluation path for the current natural stage to obtain an initial evaluation path set; acquiring high-impact weather monitoring data, and when the high-impact weather monitoring data indicates that there is a missing path, performing a replication process on the initial evaluation path set to obtain an updated evaluation path set; performing trend comparison on the updated evaluation path set and performing path merging processing to generate a scenario evolution network.

[0010] Optionally, the path health score generation includes: acquiring the predicted data sequence and actual observed data sequence of the current evaluation path in the near-real-world period in the scenario evolution network to obtain short-term state information; acquiring long-term climate background field data and evaluating the logical consistency with the predicted trend of the current evaluation path to obtain long-term state information; making a reasonable judgment on the physical parameters of the extreme weather events predicted by the current evaluation path to obtain physical state information; combining the short-term state information, long-term state information and physical state information to generate multi-dimensional state information; and inputting the multi-dimensional state information into the conflict resolver for comprehensive scoring to generate a path health score.

[0011] Optionally, the dynamic calculation and updating of the real-time path value of each evaluation path in the scenario evolution network includes: inputting the path health score into a value mapping function to calculate a base value; obtaining the current survival time of the evaluation path and calculating an aging decay coefficient based on an aging factor; multiplying the base value by the aging decay coefficient to obtain the real-time path value; and removing the evaluation path from the scenario evolution network when the real-time path value is lower than an elimination threshold.

[0012] Optionally, generating the stage anchoring evaluation value for the current natural stage includes: at the mid-term point of the current natural stage, performing a weighted calculation on the power generation prediction value based on the real-time path value of each surviving evaluation path in the scenario evolution network to obtain a mid-term evaluation value; performing a consistency analysis on the mid-term evaluation value and the path distribution state of the evaluation paths in the scenario evolution network to obtain an evaluation stability result; at the end of the current natural stage, when the evaluation stability result meets a preset consistency condition, performing a weighted calculation on the power generation prediction value again based on the real-time path value to generate the stage anchoring evaluation value for the current natural stage.

[0013] Optionally, generating the two-dimensional deviation signal includes: acquiring measured power generation data for the current natural stage, calculating the difference between the measured power generation data and the stage anchored assessment value, and generating a numerical deviation component; comparing and analyzing the coverage of the actual extreme weather pattern reflected by the measured power generation data with the survival assessment path in the scenario evolution network, identifying the deviation characteristics of the scenario evolution network at the structural level, and generating key network structural defects; and encapsulating the numerical deviation component and the key network structural defects to generate a two-dimensional deviation signal.

[0014] Optionally, generating the updated scoring configuration and growth configuration includes: parsing the numerical deviation components, adjusting the weight coefficients of each dimension used to calculate the path health score in the conflict resolver, and generating corrected scoring weight parameters; parsing the key network structure defects, adjusting the path bifurcation triggering condition parameters or path merging threshold parameters in the dynamic growth rules, and generating corrected path growth parameters; updating the corrected scoring weight parameters to the conflict resolver, and updating the corrected path growth parameters to the dynamic growth rules, thereby generating the updated scoring configuration and growth configuration.

[0015] Optionally, obtaining the target annual power generation correction result includes: performing evaluation path calculation for each natural stage based on the scoring configuration and growth configuration to obtain stage evaluation results; associating the stage evaluation results with the stage anchored evaluation value of the corresponding stage to generate a stage composite evaluation value; and performing annual-scale synthesis on the stage composite evaluation value to calculate the target annual power generation correction result.

[0016] Based on the same inventive concept, this invention also provides a new energy annual power generation correction system for assessing the impact of extreme weather, including a stage division module for acquiring multi-source meteorological observation data and climate reanalysis data for the target year, and dividing the target year into a natural stage sequence based on climate classification rules; an assessment path construction module for acquiring the current natural stage in the natural stage sequence, creating multiple initial assessment paths, and applying dynamic growth rules for dynamic management and structural updates to generate a scenario evolution network; a path status scoring module for acquiring multi-dimensional status information of each assessment path in the scenario evolution network, and using a conflict resolver for comprehensive scoring to generate a path health score; and a path value update module for dynamically calculating and updating the real-time path value of each assessment path in the scenario evolution network based on the path health score, combined with a value mapping function and a path aging mechanism. The phase anchoring assessment module is used to perform weighted calculations on the predicted power generation values ​​in the scenario evolution network based on the real-time path value at the anchoring time node of the current natural phase, generating a phase anchoring assessment value for the current natural phase. The deviation analysis module is used to acquire the measured power generation data of the current natural phase, compare and analyze it with the phase anchoring assessment value, and generate a two-dimensional deviation signal. The scoring and growth adjustment module is used to feed the two-dimensional deviation signal back to the conflict resolver and the dynamic growth rule, driving the scoring logic of the conflict resolver and the growth logic of the dynamic growth rule to co-evolve, generating updated scoring and growth configurations. The annual power generation correction module is used to independently assess all natural phases in the natural phase sequence based on the scoring and growth configurations, and synthesize them with the phase anchoring assessment value to obtain the target annual power generation correction result.

[0017] Compared with the prior art, the present invention has the following advantages:

[0018] This invention, by constructing a scenario evolution network and implementing dynamic growth rules, can proactively identify and propagate assessment paths representing potential extreme weather events. This overcomes the limitations of traditional ensemble forecasting methods in terms of scenario diversity, thereby enabling a more comprehensive coverage of the possibility space of future weather, improving the ability to capture high-impact extreme weather events, and making the correction results for annual power generation more robust and reliable.

[0019] This invention introduces path health scoring, real-time path value calculation, and a path aging and elimination mechanism, forming a dynamic and multi-dimensional path evaluation and management system. This system can assess the reliability of each path in real time and eliminate inferior paths based on their value, dynamically concentrating computational resources on the most likely weather scenarios. This avoids ineffective resource consumption on a large number of low-probability paths, improving the computational efficiency of the evaluation process and the responsiveness to rapid weather system evolution.

[0020] This invention establishes a closed-loop adaptive evolution mechanism. By generating a two-dimensional deviation signal containing both numerical and structural biases, and feeding this signal back to the conflict resolver and dynamic growth rules, it achieves the co-evolution of scoring logic and growth logic. This enables the evaluation method to learn from prediction errors and continuously optimize itself. The model parameters can automatically adapt to the climate characteristics of specific regions and the response patterns of power plants, thereby continuously improving the evaluation accuracy over time.

[0021] This invention decomposes the annual assessment task into a sequence of natural stages based on climate type and introduces the concept of stage-anchored assessment values. This stage-by-stage and node-by-node assessment method not only makes the complex annual forecasting problem more structured and modular, but also provides clear and verifiable intermediate nodes for the assessment process, enhancing the transparency and controllability of the assessment process. It facilitates stage-by-stage review and adjustment within the annual cycle, improving the scientific rigor and operability of the entire annual correction work.

[0022] Other features and advantages of the invention will be set forth in the description which follows, and will be apparent in part from the description, or may be learned by practicing the invention. The objects and other advantages of the invention may be realized and obtained by means of the structures pointed out in the description, claims and drawings. Attached Figure Description

[0023] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0024] Figure 1 This is a flowchart illustrating a method for adjusting the annual power generation of new energy sources to assess the impact of extreme weather, according to an embodiment of the present invention.

[0025] Figure 2 This is a path similarity and merging determination matrix diagram according to an embodiment of the present invention.

[0026] Figure 3 This is a stage synthesis evaluation and annual correction histogram of an embodiment of the present invention.

[0027] Figure 4 This is a schematic diagram of the structure of a new energy annual power generation correction system for assessing the impact of extreme weather according to an embodiment of the present invention. Detailed Implementation

[0028] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, 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, 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.

[0029] Reference Figure 1 One embodiment of the present invention proposes a method for correcting the annual power generation of new energy sources to assess the impact of extreme weather. It adopts a dynamically evolving scenario network and a two-dimensional deviation feedback mechanism, which can drive the scoring logic and growth logic of the assessment model to evolve in a co-evolution, thereby achieving adaptive and accurate correction of the annual power generation forecast and improving the planning reliability and operational stability of the power supply system.

[0030] The method described in this embodiment specifically includes:

[0031] S1. Obtain multi-source meteorological observation data and climate reanalysis data for the target year, and divide the target year into a natural stage sequence based on climate classification rules;

[0032] Optionally, dividing the target year into a sequence of natural stages based on climate classification rules includes:

[0033] Acquire multi-source meteorological observation data and climate reanalysis data for the target year, and calculate climate seasonality characteristic indicators;

[0034] Based on the climate seasonality characteristic indicators and combined with the climate classification rules, the climate season boundary information is determined, and the target year is divided into stages that are consistent in climate behavior and arranged in chronological order to generate a natural stage sequence.

[0035] Specifically, this involves acquiring multi-source meteorological observation data and climate reanalysis data for the target year. Multi-source meteorological observation data typically refers to hourly or daily meteorological variables collected by ground-based meteorological stations, weather radars, and satellite remote sensing equipment covering the target area, such as surface temperature, air pressure, humidity, wind speed, and solar irradiance. Climate reanalysis data is a spatiotemporally continuous and physically consistent gridded dataset generated by fusing historical observation data with climate models, such as ERA5 or NCEP / NCAR products, providing long-term climate mean state information. To quantify the seasonal evolution of climate, a climate seasonality characteristic index is calculated. This index is calculated using a weighted moving average model.

[0036] ,

[0037] in, These represent seasonal climate characteristics calculated at a specific point in time, comprehensively reflecting the recent climate state. Represents the first time within the sliding time window Key meteorological variable values ​​at a given time point are often selected as the core variables, such as daily average temperature or solar irradiance that is strongly correlated with new energy power generation. This data is extracted from multi-source meteorological observation data. For the first The weighting coefficients at each time point are set as a function that decays exponentially over time. ,in The attenuation constant is This represents the time interval from the current calculation point. The width of the sliding window is typically set between 15 and 30 days to effectively capture signals of seasonal transitions.

[0038] Based on the changing trends of climate seasonality indicators and combined with pre-defined climate classification rules, climate season boundary information is determined. Climate classification rules are a set of logical judgment criteria based on meteorological standards and local climate characteristics. For example, a rule is defined as follows: when climate seasonality indicators... When the temperature consistently exceeds 22 degrees Celsius for five consecutive days, marking the start of summer, the first day of that cycle is designated as the beginning of summer. Conversely, when the temperature consistently falls below a certain threshold, the season ends. These thresholds and the number of consecutive days are key parameters pre-configured by climate experts based on historical data analysis. This is achieved through calculations throughout the year... The sequence performs this type of rule matching, automatically identifying the start and end dates of major climate seasons such as spring, summer, autumn, and winter, i.e., climate season boundary information. Once all seasonal transition points are identified, the target year is divided into chronologically ordered stages that are consistent in climate behavior, thereby generating a natural stage sequence. For example, a target year may be divided into multiple stages such as "early spring," "high-temperature summer," "autumn transition," and "winter cold wave."

[0039] For example, taking a coastal photovoltaic power station with an installed capacity of 300MW as an example, the system divides the annual power generation into natural stages. The system first acquires multi-source meteorological observation data for the target year, including daily average temperature, daily total irradiance, satellite cloud cover data, and ERA5 reanalysis data as the climate background. To quantify the seasonal characteristics of climate, daily average temperature is selected as the core variable, with a sliding window width of 20 days and an exponential decay constant k=0.1. Taking a certain day as an example, the temperature of the most recent 4 days within the window... , , , ,correspond For days 0, 1, 2, and 3, calculate the weights. : , , , .calculate ℃. When When temperatures exceed 22°C for five consecutive days, the system marks that day as the boundary of summer. By calculating climate seasonality indicators, the system can automatically divide the year into natural stages such as spring, summer, autumn, winter, and transitional periods, ensuring consistent climate behavior within each stage. This provides a reliable time benchmark for subsequent extreme weather impact assessments, reduces interference from high-frequency weather fluctuations, and improves the scientific rigor of stage division.

[0040] S2. Obtain the current natural stage in the natural stage sequence, create multiple initial evaluation paths, and apply dynamic growth rules for dynamic management and structural updates to generate a scenario evolution network.

[0041] Optionally, the generative scenario evolution network includes:

[0042] An initial evaluation path is created for the current natural stage, resulting in an initial evaluation path set;

[0043] Acquire high-impact weather monitoring data, and when the high-impact weather monitoring data indicates that there is a missing path, perform a copying process on the initial assessment path set to obtain an updated assessment path set;

[0044] The updated evaluation path set is compared for trends and path merging is performed to generate a scenario evolution network.

[0045] Specifically, an initial assessment path is created for the current natural phase, resulting in a set of initial assessment paths. An assessment path is a complete data sequence containing time-step meteorological variables and corresponding predicted values ​​for renewable energy power generation from the current moment to the end of the natural phase. These initial paths typically originate from ensemble forecast systems of multiple global or regional numerical weather prediction centers, such as ECMWF-EPS or GFS-GEFS. The forecast results of each ensemble member are extracted and converted into corresponding power generation prediction sequences using a calibrated renewable energy power plant power conversion model. Each such sequence constitutes an initial assessment path, collectively forming the initial assessment path set.

[0046] Continuous acquisition of high-impact weather monitoring data is used to detect in real time whether there are potential extreme weather risks not covered by existing assessment paths, i.e., path gaps. High-impact weather monitoring data are physical quantities that indicate extreme weather, calculated based on near-real-time observation data and ultra-short-term forecasting models. They are used to monitor the convective effective potential energy (CAPE) of strong convection or to indicate the rate of change of local vorticity of frontal activity. When this monitoring data indicates a path gap, the initial assessment path set will be replicated. For example, when the CAPE monitoring data... Exceeding the preset trigger threshold If the value is 2000 J / kg, the existing path set is deemed insufficient to represent the thunderstorm risk. In this case, an assessment path that is closest to the current situation is selected as the parent path. A physically reasonable perturbation is applied to "copy" and generate a new assessment path that includes the characteristics of severe thunderstorm weather. This new path is then added to the set to obtain an updated assessment path set.

[0047] To prevent the uncontrolled growth of the number of evaluation paths from exhausting computational resources, a trend comparison and path merging process are performed on the updated evaluation path set. Specifically, the similarity between any two evaluation paths in the updated evaluation path set is calculated periodically. The similarity can be quantified by comparing the root mean square error of the power generation sequences of the two paths during the key prediction period. The core algorithm is as follows:

[0048] ,

[0049] in, Representative evaluation path and evaluation path The degree of difference in power generation sequences between them; and These are two paths at future points in time. The predicted power generation values ​​are obtained from data from each assessment path; The length of the time series used for comparison. When the calculated dissimilarity... Below a preset path merging threshold At this point, the two paths are considered to have a highly consistent trend, representing the same type of weather evolution scenario. This threshold is usually set based on expert experience. After triggering merging, the path with the lower real-time path value is removed from the set, or the two paths are weighted and averaged into a new path. Through this dynamic path bifurcation and merging mechanism, a scenario evolution network with a dynamic structure is ultimately generated, capable of effectively capturing and tracking multiple extreme weather possibilities. Figure 2 As shown, the root mean square error of any two paths is quantified by constructing a path difference matrix. When the calculated difference is lower than a preset merging threshold... At that time, it was determined that the two paths were highly consistent in trend.

[0050] For example, taking the high-temperature period of a coastal photovoltaic power station in midsummer as an example, the system dynamically creates evaluation paths for the current natural phase. The initial set of evaluation paths contains 3 paths, whose predicted cumulative power generation values ​​for the next 5 days are as follows: MWh, MWh, MWh. The system continuously monitors the High Impact Weather Index (CAPE) value. On a certain day... J / kg > Trigger Threshold J / kg, indicating that the existing path set does not adequately cover thunderstorm risk. The system selects a path. As the parent model, a new path is generated by applying a perturbation + 5MWh to the next day. MWh. Subsequently, path similarity was calculated using the root mean square error (RMSE). ,but and Merging. Through dynamic proliferation, perturbation, and merging mechanisms, the system generates assessment paths that can fully cover the probability of extreme weather, while controlling the number of paths to ensure computational efficiency and the comprehensiveness and reliability of power generation forecasts.

[0051] S3. Obtain multi-dimensional state information of each evaluation path in the scenario evolution network, and use a conflict resolver to perform a comprehensive score to generate a path health score.

[0052] Optionally, the generated path health score includes:

[0053] Obtain the predicted data sequence and the actual observed data sequence of the current evaluation path in the near-real period in the scenario evolution network to obtain short-term state information;

[0054] Long-term climate background field data is acquired, and the logical consistency with the predicted trend of the current assessment path is evaluated to obtain long-term state information.

[0055] The physical parameters of the extreme weather events predicted by the current assessment path are judged for reasonableness to obtain physical state information;

[0056] The short-term state information, long-term state information, and physical state information are combined to generate multi-dimensional state information;

[0057] The multi-dimensional status information is input into the conflict resolver for comprehensive scoring, generating a path health score.

[0058] Specifically, the predicted data sequence of the current evaluation path within the near-actual period is extracted from the scenario evolution network. The near-actual period typically refers to a rolling time window of the past 6 to 24 hours. Simultaneously, the actual observation data sequence within the same time period is obtained from the power plant's SCADA or central monitoring system, primarily including power generation. By comparing these two sets of sequences, a short-term fit score is calculated. This score can be calculated using the following formula:

[0059] ,

[0060] in, The score represents short-term state information, and its value range is usually between 0 and 1. The higher the score, the stronger the path's ability to track recent realities. The assessment path is within the near-real-world period. The predicted power generation at any given time is obtained directly from the path data; Is The measured power generation data at any given time comes from the real-time monitoring system; This is the total number of time points in the near-real-time period. This is the rated installed capacity of the power plant, used for normalization to ensure... It is a dimensionless relative index.

[0061] Retrieve long-term climate background data, which refers to gridded data of daily or monthly climate averages and standard deviations for the target area over a period of 30 years or more, such as monthly average total solar radiation and average wind speed. Compare the average meteorological forecast trend for a longer future period in the current assessment path with the statistical characteristics of the corresponding period in the climate background field to assess their logical consistency. For example, if the 7-day average temperature predicted by a path deviates from the local climate average for the same period by more than 3 standard deviations, its climate logical consistency is low. This consistency can be quantified as a long-term state score:

[0062] ,

[0063] in, The score represents long-term state information. This represents the average value of key meteorological variables used in current path forecasting over a future period. and These represent the climate mean and standard deviation of the variable for the same period, obtained from long-term climate background data. This Gaussian function form ensures that the closer the predicted value is to the climate norm, the higher the score.

[0064] The internal physical parameters of the extreme weather events predicted by the current assessment path will be evaluated for reasonableness. This means checking whether the predicted events follow basic physical laws or meteorological statistical regularities. For example, if the path predicts a typhoon, it will simultaneously check whether the predicted minimum central pressure and maximum wind speed satisfy the industry-recognized Dvorak relationship or similar empirical models. If the prediction shows physically inconsistent parameter combinations, its physical reasonableness will be judged as poor. Physical state information score. It is a discrete score based on a set of rules. For example, full compliance with physical constraints is scored as 1.0, slight deviation is scored as 0.5, and serious violation is scored as 0.1.

[0065] short-term status information Long-term status information With physical state information The data are combined to generate a multi-dimensional state information vector. This vector is then input into the conflict resolver. In engineering, the conflict resolver is a configurable weighted scoring engine whose core function is to synthesize scores from different dimensions based on preset weights. The calculation formula is as follows:

[0066] ,

[0067] in, The final output is a path health score. , and These are the weighting coefficients for the three dimensions, summing to 1. These weights are initially set by experts and dynamically adjusted during subsequent feedback learning. For example, when weather systems evolve rapidly, The weights may be dynamically adjusted to favor assessment paths that better track the current situation.

[0068] For example, a health score is calculated for three survival assessment paths of a photovoltaic power plant during the natural high-temperature period of midsummer. The system first obtains the power generation data for the most recent 12 hours from the power plant's SCADA system, including hourly power generation sequences. MWh, and simultaneously extract the corresponding path prediction sequence. MWh. Short-term fit score Substitute MWh, h, MW, get Long-term state score By comparing the average irradiance over the next 7 days along the path kWh / m² and ERA5 long-term climate background value kWh / m² kWh / m² calculation Physical state information Based on the verification of the relationship between temperature, wind speed, and irradiance within the path, it fully complies with the rules. The overall score uses weighted averages. , , Then path health By combining near-real-time data with long-term climate data and physical condition verification, the reliability of the path is quantified, providing a reliable basis for subsequent real-time path value calculation and improving the accuracy of extreme weather scenario assessment.

[0069] S4. Based on the path health score, combined with the value mapping function and the path aging mechanism, dynamically calculate and update the real-time path value of each evaluation path in the scenario evolution network.

[0070] Optionally, the dynamic calculation and updating of the real-time path value for each evaluation path in the scenario evolution network includes:

[0071] The path health score is input into a value mapping function to calculate the basic value;

[0072] Obtain the duration of survival of the current evaluation path and calculate the aging decay coefficient based on the aging factor;

[0073] Multiplying the base value by the aging decay coefficient yields the real-time path value;

[0074] When the real-time path value falls below the elimination threshold, the evaluated path will be removed from the scenario evolution network.

[0075] Specifically, the path health score is input into a preset value mapping function to calculate the base value. This value mapping function is typically a configurable non-linear function, such as an sigmoid function. Its purpose is to non-linearly map the path health scores, distributed within a certain range, to a standardized value interval, thereby amplifying the value difference between high-scoring and low-scoring paths. The core algorithm can be expressed as:

[0076] ,

[0077] in, This represents the calculated basic value; This is the path health score of the current assessment path; It is a slope parameter used to control the steepness of the mapping curve. Its engineering value is usually between 5 and 15, which is used to adjust the sensitivity of the score. It is the center point or reference threshold for the score, usually set to 0.5, which represents the passing grade.

[0078] Obtain the lifetime of the current evaluation path and introduce a path aging mechanism. Maintain a timer for each evaluation path, recording the time step from its creation or last update to obtain the lifetime. Based on this timeframe, an aging decline coefficient is calculated using aging factors. The calculation formula is as follows:

[0079] ,

[0080] in, It is the aging and decay coefficient, and its value is between 0 and 1; It is an aging factor, with a value typically between 0.01 and 0.05, which determines the rate at which value decays over time; The time step in which the current evaluation path has been continuously alive is provided by the system's internal timer.

[0081] Multiplying the base value by the aging decay coefficient yields the real-time path value of the evaluated path at the current moment. The calculation formula is as follows: ,in, This is the final real-time path value, which combines the path's immediate health and historical survivability.

[0082] The path elimination operation is performed based on this real-time path value. When the real-time path value of a given evaluation path... Below the preset elimination threshold If the value is 0.1, for example, the weather scenario represented by the path is determined to no longer have sufficient probability or value. At this point, an elimination mechanism is triggered to completely remove the evaluation path from the data structure of the scenario evolution network.

[0083] For example, taking a certain evaluation path in the scenario evolution network as the object, the calculated path health score is... As input, the system calculates the base path value according to the configured S-shaped mapping function. slope parameter The value is determined to be 10 by model calibration, and the mapping center is... The scoring reference threshold is set to 0.5. Substitute the values ​​and calculate. The system then reads the path's lifespan from the path lifecycle management module. There are 8 time steps, with the time step size consistent with the path update frequency, and each step corresponding to one hour. The system is configured with an aging factor α to control time decay, with an engineering value of 0.03. Therefore, the aging coefficient... Ultimate real-time path value It equals the product of the basic value and the aging coefficient. The system will compare this value with the preset elimination threshold. Comparing the results, since 0.755 > 0.1, the path is selected to proceed to the next round of evaluation. This method combines the immediate reliability and duration of a path into a single comparable real-time value, thereby quantitatively and reproducibly filtering out outdated or unreliable paths. This ensures that computational resources for the scenario evolution network are prioritized for high-reliability scenarios, improving the stability and computational efficiency of subsequent power generation assessments.

[0084] S5. At the anchoring time node of the current natural stage, the predicted power generation value in the scenario evolution network is weighted and calculated based on the real-time path value to generate the stage anchoring evaluation value of the current natural stage.

[0085] Optionally, the generation of the stage anchoring evaluation value for the current natural stage includes:

[0086] At the mid-term point of the current natural stage, the predicted power generation value is weighted and calculated based on the real-time path value of each survival assessment path in the scenario evolution network to obtain the mid-term assessment value.

[0087] Consistency analysis is performed on the interim evaluation values ​​and the path distribution status of the evaluation paths in the scenario evolution network to obtain the evaluation stability results;

[0088] At the end of the current natural stage, when the evaluation stability result meets the preset consistency condition, the power generation prediction value is weighted again based on the real-time path value to generate the stage anchoring evaluation value for the current natural stage.

[0089] Specifically, at a mid-term point in the current natural phase, such as 40% to 60% of the total phase duration, the total power generation predicted by these paths is weighted and calculated based on the real-time path value of each survival assessment path in the scenario evolution network to obtain the mid-term assessment value. The core calculation formula is as follows:

[0090] ,

[0091] in, This represents the interim assessment value, indicating the projected total power generation for the period. The first value obtained from the path value update module Real-time path value of each survival assessment path; It is the first The cumulative power generation predicted by each evaluation path from the current intermediate time point to the end of that natural phase is obtained by integrating or summing the power generation prediction sequence in the path. The summation iterates through all evaluation paths still alive in the scenario evolution network.

[0092] Consistency analysis is performed on the interim assessment value and the path distribution status of the assessment paths in the scenario evolution network to obtain the assessment stability results. The path distribution status is quantified by calculating the standard deviation of the total power generation predicted by all surviving assessment paths. A smaller standard deviation indicates that the prediction results of most high-value paths tend to be consistent, and the network convergence is good. This standard deviation is compared with a dynamic threshold, or it is analyzed whether high-value paths are concentrated in a single prediction cluster, rather than scattered in multiple mutually exclusive scenario clusters. For example, if the difference in the predicted total power generation of the few paths with the highest real-time path value is less than 5% of the overall prediction mean, the assessment stability results are deemed to meet the preset consistency condition.

[0093] At the end of the current natural phase, typically when there is only a short time remaining before the phase ends, such as 85% to 95% of the total phase duration, a final confirmation assessment is performed. The trigger condition is that the stability assessment results meet the consistency conditions set in the forecast. The final anchor value is only generated when the predicted scenario has sufficiently converged and uncertainty has significantly decreased. To avoid drawing premature conclusions when scenarios diverge drastically. If conditions are met, based on the updated real-time path value, the power generation forecasts for each path will be recalculated using the same logic as the interim assessment, generating the stage anchor assessment value for the current natural stage. This stage anchor assessment value will serve as the official assessment conclusion for that natural stage and will be used for subsequent deviation analysis and annual adjustments. If, at the end of the period, the assessment stability results still do not meet the consistency conditions, an alternative strategy will be adopted, such as directly using the forecast of the path with the highest real-time path value, or using the median of all path forecasts as the stage anchor assessment value.

[0094] For example, taking the natural phase of the high-temperature period in midsummer at a coastal wind farm as an example, the system first obtains the real-time path value and corresponding power generation prediction sequence of the surviving paths in the scenario evolution network at the midpoint of this phase—the total phase duration is 20 days, with the midpoint being day 10. There are a total of 3 surviving paths in the network, and their real-time path values ​​are as follows: , , The corresponding cumulative power generation forecasts from day 10 to the end of the phase are as follows: MWh, MWh, MWh. The system is calculated using a weighted average formula. MWh. Subsequently, the system calculates the standard deviation of the path distribution. MWh. Since the 5% total power generation threshold is approximately 22.5MW, and 16.25MWh < 22.5MW, the assessment is deemed stable, and a final anchored assessment value of 452MWh is generated at the end of the phase. By clearly specifying the source of the path value and power generation data, and substituting them into the numerator, denominator, and standard deviation calculations, the system can determine network convergence in the mid-term and generate robust phase anchored values, providing a reliable benchmark for subsequent deviation analysis.

[0095] S6. Obtain the measured power generation data of the current natural stage, compare and analyze it with the stage anchoring evaluation value, and generate a two-dimensional deviation signal.

[0096] Optionally, generating a two-dimensional deviation signal containing numerical and structural deviations includes:

[0097] Obtain the measured power generation data of the current natural stage, calculate the difference between the measured data and the stage anchoring evaluation value, and generate a numerical deviation component.

[0098] By comparing and analyzing the actual extreme weather patterns reflected in the measured power generation data with the coverage of the survival assessment paths in the scenario evolution network, the deviation characteristics of the scenario evolution network at the structural level are identified, and key network structural defects are generated.

[0099] The numerical deviation components are encapsulated with the key network structure defects to generate a two-dimensional deviation signal.

[0100] Specifically, the measured power generation data for the current natural phase is acquired, and the difference between this data and the phase anchoring assessment value is calculated to generate a numerical deviation component. After the natural phase has completely ended, a complete, time-point-by-time measured power generation sequence for that phase is extracted from the power plant's historical database. By integrating or summing this sequence, the total measured power generation for the phase is obtained. Subsequently, the difference between this measured value and the phase anchoring assessment value generated in the previous step is calculated. The calculation formula is as follows:

[0101] ,

[0102] in, This is the numerical deviation component, which directly quantifies the numerical accuracy of the prediction results; It is the total measured power generation of the stage calculated by integrating measured power generation data; This is the stage-anchored assessment value for that natural stage. A positive one... This indicates a systematic underestimation of power generation, while the opposite indicates an overestimation.

[0103] The actual extreme weather patterns reflected in the measured power generation data are identified and classified by analyzing the statistical characteristics of the measured power generation sequence and the associated real-time meteorological data. For example, it may be possible to identify an unexpected "pre-frontal gust" event that caused a sudden surge in power generation. Then, this identified actual weather pattern is compared with the characteristics of all survival assessment paths in the scenario evolution network at the end of the natural phase. If no high-value path in the network can reproduce this "pre-frontal gust" pattern, or if all paths predict a completely different weather scenario, a structural bias is identified, and this "failure to capture the pre-frontal gust event" is recorded as a key network structural defect.

[0104] The numerical bias component and key network structural defects are encapsulated to generate a two-dimensional bias signal. Encapsulation means integrating information from these two different dimensions into a unified data structure for easier parsing and use by subsequent modules. This two-dimensional bias signal is no longer a single error value, but a composite, containing both the numerical bias component indicating "how much the prediction was wrong" and the error value itself. This also explains "why it was wrong".

[0105] For example, the system obtains the hourly power generation for that period from the power plant's SCADA system and sums them up to obtain the total measured power generation for that period. MWh. Phase anchoring assessment value MWh, then numerical deviation MWh. Subsequently, the system identified network structure deviations. By analyzing the power generation sequences of surviving paths in the network and measured extreme events, such as a sudden high wind speed leading to a short-term increase in power generation, it was found that high-value paths in the network failed to capture this increased power generation event, which was recorded as a critical network structure defect: "failure to cover local wind surge leading to a sudden increase in short-term power generation." The system will... A two-dimensional deviation signal was generated by encapsulating the structural defect: {numerical deviation = 13MWh; structural defect = "uncovered local wind surge event"}. Through the methods of acquiring measured data, the source of stage anchor values, difference calculation, and deviation identification, the system quantified the prediction error and its causes, providing actionable feedback information for subsequent adjustment of scoring weights and optimization of dynamic growth rules.

[0106] S7. Feed the dual-dimensional deviation signal back to the conflict resolver and the dynamic growth rule, drive the scoring logic of the conflict resolver and the growth logic of the dynamic growth rule to co-evolve, and generate an updated scoring configuration and growth configuration.

[0107] Optionally, generating the updated scoring configuration and growth configuration includes:

[0108] The numerical deviation components are analyzed, and the weight coefficients of each dimension used to calculate the path health score in the conflict resolver are adjusted to generate the corrected score weight parameters.

[0109] Analyze the key network structure defects, adjust the path bifurcation trigger condition parameters or path merging threshold parameters in the dynamic growth rules, and generate corrected path growth parameters.

[0110] The revised scoring weight parameters are updated to the conflict resolver, and the revised path growth parameters are updated to the dynamic growth rules, generating updated scoring and growth configurations.

[0111] Specifically, analyze numerical deviations. The sign and magnitude of the values ​​are determined and correlated with the characteristics of the assessment paths that led to the bias. If the final result is a significant overestimation of power generation, and the analysis finds that high-value paths surviving at this stage generally deviate significantly from the climate background field, the weights of the long-term state information dimension are inferred. The weights might be too low, leading to inappropriately high values ​​being assigned to predicted paths that were "too extreme" but ultimately did not occur. Based on this inference, a weight adjustment is performed. The core adjustment algorithm is as follows:

[0112] ,

[0113] ,

[0114] ,

[0115] in, This represents the corrected scoring weight parameter. It is a learning rate, typically ranging from 0.01 to 0.1, which controls the step size for each adjustment. It is the input numerical deviation component. It is the average score of those high-value paths that cause bias on the long-term state information dimension. This logic aims to address situations where predictions overestimate... And when the long-term state score of the path is high, increase The weights of the first dimension are reduced if the weights are less than 1, and vice versa. Meanwhile, to ensure the sum of the weights is 1, the weights of other dimensions are reduced accordingly.

[0116] The key network structural defects in the two-dimensional bias signal are analyzed, and the path bifurcation triggering condition parameters or path merging threshold parameters in the dynamic growth rules are adjusted accordingly to generate corrected path growth parameters. If the key network structural defect indicates "failure to capture pre-frontal wind gain events," the key meteorological indicators leading to such events, such as rapid changes in local pressure gradients, are analyzed. Then, the path bifurcation triggering condition parameters related to these indicators in the dynamic growth rules are reduced. For example, previously, a pressure gradient change rate exceeding 10 hPa / (100 km * 3 h) was required to trigger path bifurcation; this is now adjusted to 7 hPa / (100 km * 3 h). This adjustment allows the model to more sensitively generate new evaluation paths to explore such possibilities when encountering similar meteorological precursors in the future. Conversely, if the network generates a large number of redundant paths that are ultimately falsified due to excessive sensitivity, the corresponding path merging threshold parameters are increased. This prompts similar paths to be merged earlier, thus simplifying the network structure.

[0117] The revised scoring weight parameters The conflict resolver is updated, and the corrected path growth parameters are updated to the dynamic growth rules. After this series of updates, an evolved set of model parameter configurations that are better adapted to the local climate and power plant response characteristics is formed, namely the updated scoring configuration and growth configuration.

[0118] For example, the system receives a two-dimensional deviation signal after a certain natural phase has ended, where the numerical deviation component is... MWh, the structural defect points to an "uncovered localized wind surge event". The system analyzes this deviation signal and uses it for adaptive adjustment of the conflict resolver weights and updating of the dynamic growth rule parameters. The initial scoring weight parameters are... , , Learning rate The high-value paths that cause bias within a given phase have an average score of [score missing] on the long-term state information dimension. According to the weighting adjustment formula: , , After adjustment, all weights are non-negative, but the sum is [value missing]. The normalization requirements have been met. When analyzing the structural defects of the two-dimensional bias signal, the system identified "uncovered local wind surge events" and adjusted the dynamic growth rules accordingly. The path bifurcation trigger threshold was lowered to a pressure gradient change rate of 7 hPa / (100 km·3 h) to improve the model's sensitivity to path generation for local wind surge events. Simultaneously, to prevent the generation of too many redundant paths, the path merging threshold was slightly increased. This controls network complexity. Finally, the system updates the conflict resolver with the corrected scoring weights and updates the dynamic growth rules with the corrected path bifurcation and merging thresholds. Through this operation, the model can sensitively capture uncovered extreme events in subsequent natural phases while maintaining a reasonable size for the scenario evolution network, achieving co-evolution of scoring and growth logic.

[0119] S8. Based on the scoring configuration and growth configuration, each natural stage in the natural stage sequence is independently evaluated and synthesized with the stage anchoring evaluation value to obtain the target annual power generation correction result.

[0120] Optionally, obtaining the target annual power generation correction result includes:

[0121] Based on the scoring configuration and growth configuration, the evaluation path is calculated for each natural stage to obtain the stage evaluation results;

[0122] The stage evaluation results are correlated with the stage anchor evaluation values ​​of the corresponding stages to generate stage composite evaluation values;

[0123] The phased composite assessment values ​​are then synthesized on an annual scale to calculate the corrected target annual power generation.

[0124] Specifically, based on the updated scoring and growth configurations, an independent assessment path calculation is performed for all natural stages in the natural stage sequence, excluding the currently completed assessment stage, to obtain the stage assessment results for each stage. For past natural stages, this "assessment calculation" is a "post-mortem review," using historical meteorological data and the evolved model configuration to re-simulate the scenario evolution network at that time, calculating the power generation assessment value that should have been obtained under that configuration. For future natural stages, this is a "forward-looking prediction," using the latest model configuration and future climate prediction data for assessment. Whether it's a review or a prediction, each stage generates an independent stage assessment result. , here The sequence number represents the natural stage.

[0125] The stage evaluation results of each natural stage are correlated with the corresponding stage anchor evaluation value to generate a stage composite evaluation value. This correlation process aims to combine two types of information: the stage anchor evaluation value is an "official" prediction derived from the information available at that time in the real timeline, while the stage evaluation result is a "hindsight" or "updated prediction" derived from the globally optimized model configuration. The correlation process is not a simple replacement but an information fusion. For past stages, the stage composite evaluation value can be a weighted average of these two values, with the stage anchor evaluation value having a higher weight because it represents the output of the actual decision-making process. The calculation formula is as follows:

[0126] ,

[0127] In this formula, It is the first The composite evaluation value of each past stage; It is the stage anchor assessment value generated at that time, obtained directly from the historical record; This is the stage evaluation result calculated based on the updated configuration; This is a weighting coefficient, typically with an engineering value between 0.7 and 0.9, reflecting respect for historical decision-making processes. For future natural stages, since there are no historical stage-anchored assessment values, the composite stage assessment value is directly equal to the stage assessment result predicted based on the latest configuration. For example... Figure 3 As shown, taking the early spring, midsummer, and autumn transition periods as examples, the gray-lined bars represent the final composite evaluation values. It can be seen that the composite values ​​are adjusted by historical weights, preserving the stability of historical anchor values ​​while incorporating correction information from the latest evaluation values.

[0128] The stage-based composite assessment values ​​for all natural stages are synthesized on an annual scale to calculate the corrected power generation for the target year. This is a simple accumulation process, summing the power generation assessment values ​​for all natural stages throughout the target year to form the final prediction of the total annual power generation. The core algorithm is as follows:

[0129] ,

[0130] in, The final output is the target annual power generation correction result, with the summation symbol... This refers to all natural phases within the target year. Phase synthesis evaluation value By accumulating these data, the final result not only includes predictions of future weather scenarios, but more importantly, it systematically corrects for potential annual power generation forecast biases caused by extreme weather events through a complete feedback, evolution, and retrospective evaluation cycle.

[0131] For example, taking a coastal photovoltaic power station in a certain year as an example, the system uses the updated scoring configuration and growth configuration to conduct retrospective evaluation and forward-looking prediction of the natural stages throughout the year. Taking three stages as an example: early spring... MWh, MWh; High summer temperatures MWh, MWh; Autumn Transition Period MWh, MWh, with historical weighting Then the stage synthesis evaluation value MWh, MWh, MWh. Annual revised power generation. MWh. This method achieves year-round phased synthesis and annual cumulative correction, fully considering the impact of extreme weather, providing accurate correction results for the annual power generation of the power plant, and taking into account historical information and optimized forecasts.

[0132] Based on the same inventive concept, such as Figure 4 As shown, the present invention also provides a new energy annual power generation correction system for assessing the impact of extreme weather, the system comprising:

[0133] The phase division module is used to acquire multi-source meteorological observation data and climate reanalysis data for the target year, and divide the target year into a natural phase sequence based on climate classification rules.

[0134] The evaluation path construction module is used to obtain the current natural stage in the natural stage sequence, create multiple initial evaluation paths, and apply dynamic growth rules for dynamic management and structural updates to generate a scenario evolution network.

[0135] The path status scoring module is used to obtain multi-dimensional status information of each evaluation path in the scenario evolution network, and to use a conflict resolver to perform a comprehensive score to generate a path health score.

[0136] The path value update module is used to dynamically calculate and update the real-time path value of each evaluated path in the scenario evolution network based on the path health score, combined with the value mapping function and the path aging mechanism.

[0137] The stage anchoring assessment module is used to perform weighted calculations on the predicted power generation value in the scenario evolution network based on the real-time path value at the anchoring time node of the current natural stage, and generate the stage anchoring assessment value for the current natural stage.

[0138] The deviation analysis module is used to acquire the measured power generation data of the current natural stage, compare and analyze it with the stage anchoring evaluation value, and generate a two-dimensional deviation signal.

[0139] The scoring and growth adjustment module is used to feed back the two-dimensional deviation signal to the conflict resolver and the dynamic growth rule, drive the scoring logic of the conflict resolver and the growth logic of the dynamic growth rule to co-evolve, and generate an updated scoring configuration and growth configuration.

[0140] The annual power generation correction module is used to independently evaluate all natural stages in the natural stage sequence based on the scoring configuration and growth configuration, and synthesize them with the stage anchoring evaluation value to obtain the target annual power generation correction result.

[0141] It should be noted that the electrical connections between the various units described above do not necessarily represent direct or indirect connections. Any indirect connection method can be applied to the embodiments of the present invention as long as it achieves the purpose of the present invention. The above descriptions are merely exemplary embodiments of the present invention and should not be construed as limiting the scope of the present invention.

[0142] All equivalent changes and modifications made in accordance with the teachings of this invention are still within the scope of this invention. Those skilled in the art will readily conceive of other embodiments of this invention upon considering the specification and the disclosure of practical truth. This application is intended to cover any variations, uses, or adaptations of this invention that follow the general principles of this invention and include common knowledge or conventional techniques in the art not described herein.

Claims

1. A method for adjusting annual power generation of new energy sources to assess the impact of extreme weather, characterized in that, The method includes: Acquire multi-source meteorological observation data and climate reanalysis data for the target year, and divide the target year into a natural stage sequence based on climate classification rules; Obtain the current natural stage in the natural stage sequence, create multiple initial evaluation paths, and apply dynamic growth rules for dynamic management and structural updates to generate a scenario evolution network; The multi-dimensional state information of each evaluation path in the scenario evolution network is obtained, and a conflict resolver is used to perform a comprehensive score to generate a path health score. Based on the path health score, combined with the value mapping function and the path aging mechanism, the real-time path value of each evaluation path in the scenario evolution network is dynamically calculated and updated. At the anchoring time node of the current natural stage, the predicted power generation value in the scenario evolution network is weighted and calculated based on the real-time path value to generate the stage anchoring evaluation value of the current natural stage. The measured power generation data of the current natural stage is obtained and compared with the stage anchoring evaluation value to generate a two-dimensional deviation signal. The two-dimensional deviation signal is fed back to the conflict resolver and the dynamic growth rule, driving the scoring logic of the conflict resolver and the growth logic of the dynamic growth rule to co-evolve and generate updated scoring configuration and growth configuration. Based on the scoring configuration and growth configuration, all natural stages in the natural stage sequence are independently evaluated and synthesized with the stage anchoring evaluation value to obtain the target annual power generation correction result.

2. The method for adjusting annual power generation of new energy sources to assess the impact of extreme weather as described in claim 1, characterized in that, The method of dividing the target year into a natural phase sequence based on climate classification rules includes: Acquire multi-source meteorological observation data and climate reanalysis data for the target year, and calculate climate seasonality characteristic indicators; Based on the climate seasonality characteristic indicators and combined with climate classification rules, climate seasonal boundary information is determined, and the target year is divided into stages that are consistent in climate behavior and arranged in chronological order to generate a natural stage sequence.

3. The method for adjusting annual power generation of new energy sources to assess the impact of extreme weather as described in claim 1, characterized in that, The generative scenario evolution network includes: An initial evaluation path is created for the current natural stage, resulting in an initial evaluation path set; Acquire high-impact weather monitoring data, and when the high-impact weather monitoring data indicates that there is a missing path, perform a copying process on the initial assessment path set to obtain an updated assessment path set; The updated evaluation path set is compared for trends and path merging is performed to generate a scenario evolution network.

4. The method for adjusting annual power generation of new energy sources to assess the impact of extreme weather as described in claim 1, characterized in that, The generated path health score includes: Obtain the predicted data sequence and the actual observed data sequence of the current evaluation path in the near-real period in the scenario evolution network to obtain short-term state information; Long-term climate background field data is acquired, and the logical consistency with the predicted trend of the current assessment path is evaluated to obtain long-term state information. The physical parameters of the extreme weather events predicted by the current assessment path are judged for reasonableness to obtain physical state information; The short-term state information, long-term state information, and physical state information are combined to generate multi-dimensional state information; The multi-dimensional status information is input into the conflict resolver for comprehensive scoring, generating a path health score.

5. The method for adjusting annual power generation of new energy sources to assess the impact of extreme weather as described in claim 1, characterized in that, The dynamic calculation and updating of the real-time path value for each evaluation path in the scenario evolution network includes: The path health score is input into a value mapping function to calculate the basic value; Obtain the duration of survival of the current evaluation path and calculate the aging decay coefficient based on the aging factor; Multiplying the base value by the aging decay coefficient yields the real-time path value; When the real-time path value falls below the elimination threshold, the evaluated path will be removed from the scenario evolution network.

6. The method for adjusting annual power generation of new energy sources to assess the impact of extreme weather as described in claim 1, characterized in that, The stage anchoring evaluation value for generating the current natural stage includes: At the mid-term point of the current natural stage, the predicted power generation value is weighted and calculated based on the real-time path value of each survival assessment path in the scenario evolution network to obtain the mid-term assessment value. Consistency analysis is performed on the interim evaluation values ​​and the path distribution status of the evaluation paths in the scenario evolution network to obtain the evaluation stability results; At the end of the current natural stage, when the evaluation stability result meets the preset consistency condition, the power generation prediction value is weighted again based on the real-time path value to generate the stage anchoring evaluation value for the current natural stage.

7. The method for adjusting annual power generation of new energy sources to assess the impact of extreme weather as described in claim 1, characterized in that, The generation of the two-dimensional deviation signal includes: Obtain the measured power generation data of the current natural stage, calculate the difference between the measured data and the stage anchoring evaluation value, and generate a numerical deviation component. By comparing and analyzing the actual extreme weather patterns reflected in the measured power generation data with the coverage of the survival assessment paths in the scenario evolution network, the deviation characteristics of the scenario evolution network at the structural level are identified, and key network structural defects are generated. The numerical deviation components are encapsulated with the key network structure defects to generate a two-dimensional deviation signal.

8. The method for adjusting annual power generation of new energy sources to assess the impact of extreme weather as described in claim 7, characterized in that, The generated updated scoring configuration and growth configuration include: The numerical deviation components are analyzed, and the weight coefficients of each dimension used to calculate the path health score in the conflict resolver are adjusted to generate the corrected score weight parameters. Analyze the key network structure defects, adjust the path bifurcation trigger condition parameters or path merging threshold parameters in the dynamic growth rules, and generate corrected path growth parameters. The revised scoring weight parameters are updated to the conflict resolver, and the revised path growth parameters are updated to the dynamic growth rules, generating updated scoring and growth configurations.

9. The method for adjusting annual power generation of new energy sources to assess the impact of extreme weather as described in claim 1, characterized in that, The revised target annual power generation results include: Based on the scoring configuration and growth configuration, the evaluation path is calculated for each natural stage to obtain the stage evaluation results; The stage evaluation results are correlated with the stage anchor evaluation values ​​of the corresponding stages to generate stage composite evaluation values; The phased composite assessment values ​​are then synthesized on an annual scale to calculate the corrected target annual power generation.

10. A system for correcting annual power generation from renewable energy sources to assess the impact of extreme weather, the system comprising: The phase division module is used to acquire multi-source meteorological observation data and climate reanalysis data for the target year, and divide the target year into a natural phase sequence based on climate classification rules. The evaluation path construction module is used to obtain the current natural stage in the natural stage sequence, create multiple initial evaluation paths, and apply dynamic growth rules for dynamic management and structural updates to generate a scenario evolution network. The path status scoring module is used to obtain multi-dimensional status information of each evaluation path in the scenario evolution network, and to use a conflict resolver to perform a comprehensive score to generate a path health score. The path value update module is used to dynamically calculate and update the real-time path value of each evaluated path in the scenario evolution network based on the path health score, combined with the value mapping function and the path aging mechanism. The stage anchoring assessment module is used to perform weighted calculations on the predicted power generation value in the scenario evolution network based on the real-time path value at the anchoring time node of the current natural stage, and generate the stage anchoring assessment value for the current natural stage. The deviation analysis module is used to acquire the measured power generation data of the current natural stage, compare and analyze it with the stage anchoring evaluation value, and generate a two-dimensional deviation signal. The scoring and growth adjustment module is used to feed back the two-dimensional deviation signal to the conflict resolver and the dynamic growth rule, driving the scoring logic of the conflict resolver and the growth logic of the dynamic growth rule to co-evolve and generate updated scoring configuration and growth configuration. The annual power generation correction module is used to independently evaluate all natural stages in the natural stage sequence based on the scoring configuration and growth configuration, and synthesize them with the stage anchoring evaluation value to obtain the target annual power generation correction result.

Images