Uncertainty analysis method and device for wind power prediction and storage medium

Abstract

The invention discloses an uncertainty analysis method and device for wind power prediction and a storage medium, and relates to the technical field of wind power prediction, and the method comprises the steps: carrying out the data sensing of a target wind power plant region through a plurality of sensors, and obtaining a plurality of operation data sets of a wind power plant; performing wind power similarity analysis based on the plurality of operation data sets, determining a nonlinear coefficient, and training and constructing a Gaussian regression model; inputting variables into the Gaussian regression model, and outputting a regression data set; performing uncertainty analysis on the wind power prediction data according to the data uncertainty interval, and determining a confidence interval of the wind power; and performing control analysis on the target wind power plant, and formulating a scheduling control strategy. According to the method and the device, the technical problem of low operation efficiency and reliability of the wind power plant caused by inaccurate prediction of the wind power in the prior art is solved, and the technical effects of realizing accurate prediction of the wind power and improving the operation efficiency and reliability of the wind power plant are achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of wind power prediction, and in particular to an uncertainty analysis method, device and storage medium for wind power prediction. Background Art

[0002] As the global demand for renewable energy continues to grow, wind power generation, as a clean and sustainable form of energy, has occupied an increasingly important position in the energy field. However, the prediction of wind power has always been one of the key problems in the operation and management of wind farms. Since the operation of wind farms is affected by a variety of complex factors, such as the random changes in meteorological conditions such as wind speed, wind direction, temperature, and air pressure, as well as the performance fluctuations of wind turbines themselves and equipment aging, wind power has strong uncertainty and intermittency. Traditional wind power prediction methods have many limitations. For example, in terms of data collection, there is a lack of systematic and scientific planning, and it is impossible to accurately obtain effective data that fully reflects the operating status of wind farms, resulting in an insufficiently solid data foundation. Some commonly used prediction models fail to fully consider the nonlinear characteristics of wind power and the complex relationship between various factors, resulting in limited prediction capabilities of the models and the inability to accurately capture the changing trend of wind power.

[0003] The existing technology has the technical problem of inaccurate prediction of wind power, which leads to low operating efficiency and reliability of wind farms. Summary of the invention

[0004] The present application provides an uncertainty analysis method, device and storage medium for wind power prediction, which are used to solve the technical problem in the prior art that inaccurate wind power prediction leads to low operating efficiency and reliability of wind farms.

[0005] In view of the above problems, the present application provides an uncertainty analysis method, device and storage medium for wind power prediction.

[0006] In a first aspect of the present application, a method for uncertainty analysis of wind power prediction is provided, the method comprising:

[0007] Data sensing is performed on the target wind farm area through multiple sensors to obtain multiple operating data sets of the wind farm; wind power similarity analysis is performed based on the multiple operating data sets to determine the nonlinear coefficient, and a Gaussian regression model is constructed through training based on the nonlinear coefficient; the multiple operating data sets are input into the Gaussian regression model as variables, and a regression data set is output, wherein the regression data set includes wind power prediction data and data uncertainty intervals; uncertainty analysis is performed on the wind power prediction data according to the data uncertainty interval, simulation calculations are performed based on the analysis results, and a confidence interval of wind power is determined; control analysis is performed on the target wind farm according to the confidence interval of wind power, a dispatching control strategy is formulated, and the dispatching control strategy is executed to dispatch wind power to the target wind farm.

[0008] A second aspect of an embodiment of the present application provides an uncertainty analysis device for wind power prediction, the device comprising:

[0009] An operation data set acquisition module, the operation data set acquisition module is used to perform data sensing on the target wind farm area through multiple sensors to obtain multiple operation data sets of the wind farm; a Gaussian regression model construction module, the Gaussian regression model construction module performs wind power similarity analysis based on the multiple operation data sets, determines the nonlinear coefficient, and constructs a Gaussian regression model based on the training of the nonlinear coefficient; a regression data set output module, the regression data set output module is used to input the multiple operation data sets as variables into the Gaussian regression model, and output a regression data set, the regression data set includes wind power prediction data and data uncertainty intervals; a confidence interval determination module, the confidence interval determination module is used to perform uncertainty analysis on the wind power prediction data according to the data uncertainty interval, perform simulation calculations based on the analysis results, and determine the confidence interval of the wind power; a scheduling control strategy formulation module, the scheduling control strategy formulation module is used to perform control analysis on the target wind farm according to the confidence interval of the wind power, formulate a scheduling control strategy, and execute the scheduling control strategy to dispatch wind power to the target wind farm.

[0010] According to a third aspect of the embodiments of the present application, a computer-readable storage medium is provided, which stores a computer program. When the program is executed by a processor, the uncertainty analysis method for wind power prediction provided by the present application is implemented.

[0011] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0012] The target wind farm area is sensed by multiple sensors to obtain multiple operating data sets of the wind farm; wind power similarity analysis is performed based on the multiple operating data sets to determine the nonlinear coefficient; the multiple operating data sets are input as variables into the Gaussian regression model to output the regression data set; uncertainty analysis is performed on the wind power prediction data according to the data uncertainty interval, simulation calculation is performed based on the analysis results, and the confidence interval of wind power is determined; control analysis is performed on the target wind farm according to the confidence interval of wind power, a dispatching control strategy is formulated, and the dispatching control strategy is executed to dispatch wind power to the target wind farm. The technical effect of realizing accurate prediction of wind power and improving the efficiency and reliability of wind farm operation is achieved. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0014] Figure 1 A schematic diagram of a flow chart of an uncertainty analysis method for wind power prediction provided in an embodiment of the present application;

[0015] Figure 2 A schematic diagram of the structure of an uncertainty analysis device for wind power prediction provided in an embodiment of the present application.

[0016] Explanation of the reference numerals: operating data set acquisition module 10 , Gaussian regression model construction module 20 , regression data set output module 30 , confidence interval determination module 40 , scheduling control strategy formulation module 50 . DETAILED DESCRIPTION

[0017] The present application provides an uncertainty analysis method, device and storage medium for wind power prediction, which is used to solve the technical problem in the prior art that inaccurate wind power prediction leads to low operating efficiency and reliability of wind farms.

[0018] The technical solutions in the embodiments of the present application will be described clearly and completely below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0019] Embodiment 1, as Figure 1As shown, the present application provides an uncertainty analysis method for wind power prediction, the method comprising:

[0020] Step S100: performing data sensing on a target wind farm area through a plurality of sensors to obtain a plurality of operating data sets of the wind farm.

[0021] Specifically, first of all, multiple sensors need to be carefully arranged in the target wind farm area. These sensors are of various types and functions, aiming to fully capture various key information of wind farm operation. They are distributed in different locations to cover every corner and different levels of the wind farm. When the data sensing work is started, the sensors begin to work together. Some sensors focus on monitoring wind speed. By accurately measuring the flow speed of the wind, wind speed sensing data sets are obtained, and wind speed changes at different times and locations are recorded. The wind direction sensor is responsible for determining the direction of the wind and generating a wind direction sensing data set, which provides an important basis for the subsequent analysis of the impact of wind on wind power. The environmental sensor is also actively engaged in work. It will sense and collect data on environmental factors, such as temperature, humidity, air pressure, etc., to form an environmental sensing data set, because these environmental conditions will also have an indirect impact on wind power. At the same time, the power output sensor monitors the power generation of the wind farm in real time, obtains the power sensing data set, and intuitively reflects the actual power generation of the wind farm. Through the comprehensive work of these sensors, the data collected by each is integrated, and finally multiple operation data sets of the wind farm are successfully obtained. These data sets contain rich information, providing a solid data foundation for subsequent wind power forecasting and analysis, enabling a deeper understanding of the operating status of wind farms and taking a critical first step towards achieving accurate wind power forecasting.

[0022] Step S200: performing wind power similarity analysis based on the multiple operating data sets, determining a nonlinear coefficient, and training and constructing a Gaussian regression model based on the nonlinear coefficient.

[0023] Specifically, first, the data is preprocessed and organized, and multiple operating data sets are carefully checked to identify and process outliers and missing values. For outliers, a reasonable data range is set or statistical methods are used to make judgments and corrections. For example, if the wind speed data exceeds the reasonable range of common wind speeds in the area where the wind farm is located, it is considered an outlier and adjusted accordingly. For missing values, appropriate filling methods are used according to the characteristics and relevance of the data, such as using the average value of data in adjacent time periods or interpolation filling based on the trend of historical data to ensure the integrity and continuity of the data. Next, feature engineering technology is used to extract features closely related to wind power from the operating data set. For wind speed data, not only its instantaneous value is considered, but also the mean, variance, maximum, minimum and trend of wind speed are calculated; for wind direction data, the frequency distribution of wind direction, the frequency of occurrence of dominant wind direction and the amplitude of wind direction change are analyzed; for environmental factors, the mean, extreme value and rate of change of temperature, humidity and air pressure are extracted; at the same time, the time series characteristics of power output data itself are analyzed, such as power fluctuation cycle, peak value and valley value. Through these rich feature extractions, more comprehensive information is provided for subsequent similarity analysis. In the process of similarity analysis, a strategy combining multiple methods is adopted. On the one hand, the correlation analysis method is used to calculate the Pearson correlation coefficient to measure the linear correlation between each feature and wind power. In this way, the features with strong correlation with wind power are preliminarily screened out. On the other hand, the mutual information algorithm is introduced to capture the nonlinear correlation between features and wind power. Mutual information can quantify the degree of dependence between two variables without being restricted by linear relationships, thereby more comprehensively revealing the intrinsic connection in the data. For example, it is found that there is a strong nonlinear correlation between the square of wind speed and wind power, which may not be fully reflected in linear correlation analysis. Based on the results of similarity analysis, the nonlinear coefficient is determined. By modeling the complex relationship between relevant features and wind power, a polynomial regression or kernel function-based regression method is used to fit a function model that can describe this nonlinear relationship. In this model, the nonlinear coefficient is the key parameter in the function, which determines the degree and manner of the influence of the characteristic variable on wind power. By continuously adjusting and optimizing the model parameters, the model can best fit the actual data, so as to accurately determine the nonlinear coefficient. After obtaining the nonlinear coefficient, the Gaussian regression model is constructed. Based on the nonlinear coefficient, the initial parameters of the model are set to make it more consistent with the characteristics of wind power data. During the model training process, optimization algorithms such as gradient descent are used to continuously adjust the parameters of the model to minimize the error between the predicted value and the actual value. According to the input features (multiple running data sets) in the training data set and the corresponding actual wind power output, the prediction error of the model is calculated, and then the model parameters are updated in the direction of the error decrease.Through multiple iterations of training, the Gaussian regression model gradually converges and can accurately predict wind power and reasonably estimate the uncertainty of the prediction.

[0024] Step S300: input the multiple operating data sets as variables into the Gaussian regression model, and output a regression data set, wherein the regression data set includes wind power prediction data and data uncertainty intervals.

[0025] Specifically, after the construction and training of the Gaussian regression model is completed, it enters the practical application and output stage, and the multiple operating data sets that have been carefully sorted and processed before are input as variables into this optimized Gaussian regression model. After receiving these operating data sets, the model will conduct in-depth analysis and calculation of the input data based on the complex patterns and relationships that have been learned internally. In this process, the model will first predict the wind power based on various characteristic information such as wind speed, wind direction, environment, etc. in the operating data set, combined with the previously determined model parameters and functional relationships, and obtain the predicted value of wind power. This is the wind power prediction data part in the regression data set. However, the advantage of the Gaussian regression model is not only to provide prediction values, but also to evaluate and quantify the uncertainty of the prediction. The model will calculate a data uncertainty interval based on the characteristics of the input data and the uncertainty estimation mechanism of the model itself. This interval reflects the credibility of the prediction results and the possible fluctuation range. It takes into account a variety of factors, such as the noise that may exist in the operating data set, the adaptability of the model to certain special situations, and the degree of influence of various complex factors on the wind power itself. By calculating the variance, standard deviation and other statistics of the data, and combining the confidence setting of the model, a reasonable data uncertainty interval is determined. In the end, the regression data set output by the model contains two important parts: one is the clear wind power forecast data, which provides the expected value of future wind power for the operation and management of the wind farm; the other is the data uncertainty interval. Such output results enable more comprehensive and scientific planning of wind power production, resource scheduling, and response to various possible situations, thereby improving the overall operational efficiency and stability of the wind farm.

[0026] Step S400: performing uncertainty analysis on the wind power prediction data according to the data uncertainty interval, performing simulation calculations based on the analysis results, and determining a confidence interval of the wind power.

[0027] Specifically, firstly, based on the obtained data uncertainty interval, an in-depth uncertainty analysis is carried out on the wind power forecast data. The forecast data is classified according to the data uncertainty interval and divided into different sub-intervals or categories. For the forecast data in each sub-interval, its distribution characteristics and degree of dispersion are further analyzed. For example, the mean, median, mode and other statistics of the data in each sub-interval are calculated to understand the central trend of the data; at the same time, the standard deviation, variance and other statistics are calculated to measure the degree of dispersion of the data. Through these analyses, the changes and characteristics of the forecast data in different uncertainty intervals are preliminarily grasped. Then, according to the above analysis results, simulation calculations are carried out, and simulation technology is used to construct a variety of possible scenarios to simulate the actual changes in wind power. These scenarios are set based on the uncertainty characteristics of the forecast data, and various factors that may affect wind power and their range of variation are considered. In the simulation process, input parameters that conform to the distribution of the uncertainty interval are randomly generated and substituted into the verified mathematical model or simulation system to simulate a large number of wind power output results. By repeating the simulation many times, a set of simulation results is obtained. Then, the results of these simulation calculations are comprehensively analyzed, the frequency and probability distribution of different simulation results are statistically analyzed, and a reasonable confidence level is determined based on probability statistics theory, such as the common 95% confidence level or 90% confidence level. According to the selected confidence level, the corresponding interval range is found in the probability distribution of the simulation results. This interval range is the confidence interval of wind power. The significance of the confidence interval is that, under a certain confidence level, it is believed that the real wind power has a greater probability of falling within this interval. It provides a quantitative standard for evaluating the reliability of wind power forecasts, allowing wind farm operators to more accurately grasp the possible range of wind power changes, so as to make more reasonable and robust decisions in the formulation of scheduling plans, equipment maintenance arrangements, and interaction with the power grid, effectively reducing the risks caused by wind power uncertainty and improving the safety and economy of wind farm operations.

[0028] Step S500: performing control analysis on a target wind farm according to the confidence interval of the wind power, formulating a dispatch control strategy, and executing the dispatch control strategy to dispatch the wind power of the target wind farm.

[0029] Specifically, based on the confidence interval of wind power, the target wind farm is comprehensively and deeply controlled and analyzed. Taking into account the equipment status, energy storage level, grid access capacity and market demand of the wind farm, the response capacity and potential risks of the wind farm under the different power fluctuation possibilities represented by the confidence interval are analyzed. Based on the analysis results, the dispatching control strategy is carefully formulated. For example, when the confidence interval is narrow, the plan to prioritize the supply of basic loads and reasonably arrange energy storage reserves is determined, and it is clear which wind turbines are prioritized and how to optimize their operating parameters; if the confidence interval is wide, it focuses on flexible allocation of spare resources, such as spare units that are prepared for rapid start and stop, and adjusting the interaction strategy with the grid to cope with large power fluctuations. Subsequently, through advanced monitoring and automation control systems, the dispatching control strategy is strictly implemented, operating instructions are accurately sent to wind turbines, their blade angles and speeds are adjusted, the charging and discharging of energy storage equipment are reasonably controlled, and power output is carried out according to the plan negotiated with the grid. At the same time, the operation data is monitored in real time, compared with the forecast data, and the adjustment strategy is continuously optimized to ensure the stable and efficient operation of the wind farm, and to achieve scientific dispatching of wind power and effective utilization of resources.

[0030] In a possible implementation, step S100 further includes:

[0031] Step S110: Retrieve the historical operation data record log, collect and analyze the target wind farm area based on the historical operation data record log, and determine multiple data collection points.

[0032] Step S120: performing cluster analysis according to the plurality of data collection points to determine a plurality of data categories, and deploying a plurality of sensors according to the plurality of data categories to obtain a plurality of data sensing areas.

[0033] Step S130: traversing the plurality of data sensing areas to perform data sensing on the target wind farm area to obtain an initial operation data set.

[0034] Step S140: performing data denoising on the initial operating data set to obtain the multiple operating data sets.

[0035] Specifically, the historical operation data record logs of the target wind farm are retrieved from the database. These logs contain rich information on the past operation of the wind farm. These logs are analyzed in depth, and factors such as the geographical layout of the wind farm, wind turbine distribution, topography, and the changing patterns of previous data are comprehensively considered to determine multiple reasonable data collection points. The selection of these points is intended to capture the key information of the wind farm operation comprehensively and accurately.

[0036] According to the determined multiple data collection points, cluster analysis is performed, and these points are divided into multiple data categories according to the similarity and relevance of the data, such as wind speed related categories, wind direction related categories, environmental factor related categories, etc. Then, according to these data categories, multiple sensors are deployed at the corresponding locations. Each sensor has its specific monitoring function and accuracy requirements, thus forming multiple data sensing areas. These sensing areas work together to fully cover all areas and different levels of the wind farm, ensuring the integrity and diversity of data collection.

[0037] Traversing the data sensing area, all-round data sensing is performed on the target wind farm area. The sensors work in real time to collect various data, including real-time changes in wind speed, dynamics of wind direction, environmental temperature, humidity, air pressure and other information, as well as wind turbine power output and other data, so as to obtain the initial operation data set. This data set is a preliminary record of the current operating status of the wind farm, but it may contain some noise and interference information.

[0038] The initial operation data set is subjected to data denoising. The operation signal of the initial operation data set is obtained by parsing the wind speed sensor data set, wind direction sensor data set, environmental sensor data set, and power sensor data set. Then, the operation signal is decomposed into wavelet coefficients using wavelet decomposition technology. The threshold is quantified based on these coefficients to determine the appropriate operation signal wavelet selection threshold. The noise signal less than the threshold is set to zero, and the valid signal information greater than the threshold is retained. Finally, the valid signal information is filtered and reconstructed to obtain multiple high-quality operation data sets with noise interference removed. These processed data will provide accurate and reliable basic data support for subsequent wind power prediction and analysis.

[0039] In a possible implementation, step S130 further includes:

[0040] Step S131: traversing the plurality of data sensing areas to perform multi-dimensional sensing analysis on the target wind farm, and generating multi-dimensional sensing information of the wind farm.

[0041] Step S132: performing wind speed sensing on the target wind farm according to the multi-dimensional sensing information of the wind farm and the multiple data sensing areas to obtain a wind speed sensing data set.

[0042] Step S133: performing wind direction sensing on the target wind farm according to the multi-dimensional sensing information of the wind farm and the multiple data sensing areas to obtain a wind direction sensing data set.

[0043] Step S134: performing environmental sensing on the target wind farm according to the multi-dimensional sensing information of the wind farm and the multiple data sensing areas to obtain an environmental sensing data set.

[0044] Step S135: performing power output sensing on the target wind farm according to the multi-dimensional sensing information of the wind farm and the multiple data sensing areas to obtain a power sensing data set.

[0045] Step S136: Correlate and integrate the wind speed sensor data set, the wind direction sensor data set, the environment sensor data set, and the power sensor data set to obtain an initial operation data set.

[0046] Specifically, by traversing multiple data sensing areas, a comprehensive multi-dimensional sensing analysis of the target wind farm is carried out. In this process, the functions of various sensors are comprehensively utilized to collect information from different angles and levels to generate multi-dimensional sensing information of the wind farm. This information covers multiple dimensions of the wind farm, including but not limited to a comprehensive description of the physical state of different locations, meteorological conditions, and equipment operating conditions.

[0047] Targeted wind speed sensing is carried out on the target wind farm according to multiple data sensing areas. Each wind speed sensor accurately measures the wind speed at different positions and heights, and records these data to form a wind speed sensing data set. This data set contains rich wind speed information, such as wind speed changes at different time points and in different areas, providing basic data for subsequent analysis of the impact of wind speed on wind power.

[0048] Similarly, based on the multi-dimensional sensing information of the wind farm, the target wind farm is subjected to wind direction sensing. The wind direction sensor accurately captures the direction of the wind and records the changes in wind direction, thereby obtaining a wind direction sensing data set. This data set is of great significance for understanding the flow direction of the wind and its angle of action on the wind turbine blades, and helps to further analyze the relationship between wind direction and wind power.

[0049] Environmental sensing is performed based on the multi-dimensional sensing information of the wind farm. The environmental sensors monitor the environmental factors of the wind farm, including the measurement of parameters such as temperature, humidity, and air pressure, to obtain the environmental sensing data set.

[0050] According to the multi-dimensional sensor information of the wind farm, the target wind farm is subjected to power output sensing. The power output sensor monitors the power generation of the wind turbine in real time, records the power output of different wind turbines at different times, and obtains a power sensing data set. This data set directly reflects the actual power generation capacity of the wind farm and is one of the key data for evaluating the operation effect of the wind farm and conducting power forecasting.

[0051] The wind speed sensor data set, wind direction sensor data set, environmental sensor data set and power sensor data set obtained earlier are associated and integrated. By establishing the temporal and spatial association relationship between the data, these different types of but interrelated data are fused together to form a complete initial operation data set. This initial operation data set contains various information on the operation of the wind farm, providing comprehensive and rich original data for subsequent data analysis, model construction and prediction, and is an important basis for the entire wind power prediction and analysis work.

[0052] In a possible implementation, step S140 further includes:

[0053] Step S141: performing analysis based on the wind speed sensor data set, the wind direction sensor data set, the environment sensor data set, and the power sensor data set to obtain an operation signal of the initial operation data set.

[0054] Step S142: performing wavelet decomposition on the operation signal of the initial operation data set to obtain the operation signal wavelet coefficients.

[0055] Step S143: performing threshold quantization according to the operating signal wavelet coefficients to determine the operating signal wavelet selection threshold.

[0056] Step S144: the operating signal wavelet coefficients are intercepted according to the operating signal wavelet selection threshold, noise signals smaller than the operating signal wavelet selection threshold are set to zero, and valid signal information greater than the operating signal wavelet selection threshold is obtained.

[0057] Step S145: Filter and reconstruct the effective signal information to obtain the multiple operating data sets.

[0058] Specifically, we first conduct in-depth analysis based on the previously obtained wind speed sensor data set, wind direction sensor data set, environmental sensor data set, and power sensor data set. Through comprehensive analysis of these different types of data sets, we extract the key information of wind farm operation, and thus obtain the operation signal of the initial operation data set. This operation signal contains the dynamic change characteristics and interrelationships of various data, and is the basis for subsequent processing.

[0059] The operation signal of the initial operation data set is subjected to wavelet decomposition. Wavelet decomposition is a powerful signal processing technique that decomposes the operation signal into wavelet coefficients of different frequencies and scales. Through this decomposition, the different components in the signal are clearly separated for further analysis and processing. For example, the high-frequency part corresponds to noise and rapidly changing details, while the low-frequency part contains the main trends and characteristics of the signal.

[0060] Threshold quantization is performed based on the obtained operating signal wavelet coefficients. The purpose of this step is to determine a suitable operating signal wavelet selection threshold. Through statistical analysis of the wavelet coefficients, a threshold that can distinguish between valid signals and noise signals is found. The determination of this threshold requires comprehensive consideration of factors such as signal characteristics, noise levels, and subsequent processing requirements.

[0061] According to the previously determined threshold for wavelet selection of the operating signal, the wavelet coefficients of the operating signal are precisely intercepted, and the relationship between each wavelet coefficient of the operating signal and the threshold is compared one by one. For those coefficients that are less than the threshold for wavelet selection of the operating signal, they are identified as noise signals and set to zero, that is, these small fluctuation components that may interfere with subsequent analysis are completely eliminated. For coefficients greater than the threshold, they are regarded as the part that carries the real and effective information of the operation of the wind farm, and they are extracted as effective signal information. Through such a rigorous screening process, valuable signal components can be separated from the complex original signal, and most of the noise interference can be removed, providing a purer and more accurate basis for effective signal information for subsequent further processing and analysis, so as to more accurately grasp the essential characteristics and laws of wind farm operation.

[0062] The extracted effective signal information is filtered and reconstructed. This process aims to further optimize the effective signal, remove the small amount of noise and abnormal fluctuations that may remain, and make it closer to the actual operation status of the wind farm. By adopting the filtering algorithm, the effective signal is smoothed and the frequency characteristics and amplitude distribution of the signal are adjusted. At the same time, according to the inherent structure and correlation of the wind farm operation data, the effective signal that has been filtered is reconstructed according to certain rules and methods. This reconstruction process not only integrates the effective information obtained by different types of sensors, but also restores the continuity and integrity of the signal in time and space. Finally, multiple operation data sets that remove noise interference and accurately reflect the actual operation of the wind farm are successfully obtained.

[0063] In a possible implementation, step S300 further includes:

[0064] Step S310: defining operation probability distribution information based on the multiple operation data sets, wherein the operation probability distribution information includes wind speed probability distribution information and wind direction probability distribution information.

[0065] Step S320: Perform random sampling based on the wind speed probability distribution information and the wind direction probability distribution information to obtain a sampling data group.

[0066] Step S330: performing environmental simulation analysis on the target wind farm according to the sampling data group, and determining a plurality of simulation input scenarios.

[0067] Step S340: synchronizing the multiple operating data sets to the Gaussian regression model according to the multiple simulation input scenarios to generate the wind power prediction data.

[0068] Step S350: performing critical analysis according to the wind power prediction data to determine a prediction extreme value, wherein the prediction extreme value includes a prediction upper limit value and a prediction lower limit value.

[0069] Step S360: performing mapping calculation according to the prediction upper limit value, the prediction lower limit value and the wind power prediction data to determine the data uncertainty interval.

[0070] Specifically, data related to wind speed and wind direction are extracted from multiple operating data sets, and a detailed statistical analysis is performed on the wind speed data. By calculating the frequency and cumulative frequency, a probability distribution model that best describes the distribution characteristics of the wind speed data is fitted to determine the wind speed probability distribution information. For example, it may be found that the wind speed data conforms to a certain probability distribution function, such as normal distribution, Rayleigh distribution, etc. This distribution information reflects the probability and probability density of different wind speed values. At the same time, similar processing is performed on the wind direction data to analyze the angular distribution law of wind direction and the frequency of occurrence of different wind directions, so as to define the wind direction probability distribution information and clarify the probability distribution of each wind direction interval. The probability distribution information of these wind speeds and wind directions is integrated to form a complete operating probability distribution information, which provides important basic data for subsequent probability-based analysis and sampling, so that the characteristics and laws of wind farm operating data can be understood and described more comprehensively and accurately from a probability perspective.

[0071] According to the determined wind speed probability distribution information and wind direction probability distribution information, a random sampling operation is performed. A certain number of wind speed sample values ​​are randomly selected according to the probability density specified by the wind speed probability distribution function using a random number generation algorithm to ensure that the probability of each sample value being selected is consistent with its probability of occurrence in the actual distribution. Similarly, according to the wind direction probability distribution, the corresponding wind direction sample values ​​are randomly selected, and the extracted wind speed and wind direction sample values ​​are combined and paired to form data points containing wind speed and wind direction information, which together constitute a sampling data group. The sampling data group obtained in this way can well represent the various possible combinations of wind speed and wind direction under their respective probability distributions, providing rich and diverse input data with probability representativeness for subsequent environmental simulation analysis, making the simulation of wind farm operation more comprehensive and accurate, and contributing to a deeper study of the complex relationship between wind power and factors such as wind speed and wind direction, as well as predicting the possible changes in wind power under different environmental conditions.

[0072] The acquired sampling data set is used to conduct a comprehensive environmental simulation analysis of the target wind farm. For each combination of wind speed and wind direction data in the sampling data set, other relevant environmental factors such as temperature, humidity, air pressure, etc. (these factors have also been recorded and analyzed in the previous operating data set) are comprehensively considered to construct a specific environmental simulation scenario. By simulating the flow of wind in the wind farm under different wind speeds, wind directions and environmental conditions, as well as its effect on the wind turbine generator set, including the force and rotation of the blades, the energy conversion efficiency of the generator, etc. According to different sampling data combinations, such a simulation process is repeated to determine multiple simulation input scenarios. Each simulation input scenario restores the operating status of the wind farm under specific environmental conditions as realistically as possible, providing a rich variety of simulation environmental conditions for the subsequent input of the operating data set into the Gaussian regression model for wind power prediction, so that the prediction results can more comprehensively consider various possible actual situations, and improve the accuracy and reliability of the prediction.

[0073] Multiple operating data sets are synchronously input into the Gaussian regression model according to the multiple simulation input scenarios previously determined. The model will process and analyze the corresponding operating data sets for each simulation input scenario, combining the complex mathematical relationships and learned patterns that have been established within it. Through operations such as feature extraction, weight calculation, and function mapping of the input data, the model will output prediction results for each scenario. These results together constitute the wind power prediction data, which reflects the predicted value of wind power generated by the wind farm under different simulation environmental conditions, and provides basic data for further analysis of the uncertainty of the prediction.

[0074] Perform a comprehensive statistical analysis on the forecast data and calculate statistical indicators such as mean, median, standard deviation, etc. to understand the overall distribution characteristics and degree of dispersion of the data. Then, based on the sorting method or quantile calculation, determine the extreme values ​​in the data. In this process, find the boundary conditions of the maximum and minimum values ​​in the forecast data. These boundary values ​​constitute the forecast extreme values. Among them, the forecast upper limit value represents the maximum theoretical value that wind power may reach under the current forecast model and input conditions. It reflects the upper limit of the wind farm's power generation potential under the most ideal conditions; while the forecast lower limit value represents the minimum theoretical value that wind power may appear, taking into account the worst power generation situation under the combined effect of various unfavorable factors. By determining these two forecast extreme values, it is possible to provide key boundary references for the subsequent evaluation of the reliability and uncertainty of wind power forecasts, and better understand the scope and risk of forecast results.

[0075] Based on the obtained upper and lower limits of the forecast, combined with the previous wind power forecast data, mapping calculations are performed to associate and calculate these extreme value information with the entire forecast data set. Taking into account the distribution characteristics of the data, probability factors, and uncertainty characteristics of the model, the data uncertainty interval is determined through a complex calculation process. This interval quantifies the degree of uncertainty in the wind power forecast results, and indicates the range in which the actual wind power may fluctuate within a certain probability range. Based on this uncertainty interval, a more reasonable and robust wind farm operation and management strategy is formulated to balance power generation efficiency and risk control, and ensure the safe and stable operation of the wind farm and maximize economic benefits.

[0076] In a possible implementation, step S400 further includes:

[0077] Step S410: classifying the wind power prediction data based on the data uncertainty interval to generate a plurality of wind power prediction sub-data.

[0078] Step S420: performing calculations based on the plurality of wind power prediction sub-data to obtain a wind power prediction standard deviation.

[0079] Step S430: performing uncertainty assessment on the wind power prediction data based on the wind power prediction standard deviation to obtain a plurality of uncertainty factors, arranging the plurality of uncertainty factors in descending order, and formulating an uncertainty sequence.

[0080] Step S440: Analyze and identify the wind power prediction data according to the uncertainty sequence to generate an analysis priority list.

[0081] Step S450: performing uncertainty analysis on the wind power prediction data based on the analysis priority list to obtain confidence level information, and determining a confidence interval of the wind power based on the confidence level information.

[0082] Specifically, the wind power prediction data are classified according to the data uncertainty interval, and the prediction data are divided according to their relative position to the uncertainty interval or the specific conditions they meet, thereby generating multiple wind power prediction sub-data. Each sub-data set has certain characteristics and ranges, reflecting different degrees of uncertainty or power prediction conditions within a specific interval.

[0083] Calculations are performed based on the multiple wind power prediction sub-data obtained through division. By using statistical methods, the square root of the variance is calculated to obtain the wind power prediction standard deviation. The standard deviation can measure the degree of dispersion of the prediction data, that is, it reflects the fluctuation of the data around the mean value. The larger the standard deviation, the higher the uncertainty of the data.

[0084] Analyze the forecast data from multiple dimensions and consider various factors that affect uncertainty. For example, in terms of data stability, observe the fluctuation of the forecast data in the time series. If the power value changes dramatically in adjacent time periods, then this becomes an uncertainty factor; the error factor of the model, compare the deviation between the forecast value and the actual historical data. A large deviation means that the model may have uncertainty in some aspects; the complexity of environmental factors will also be taken into consideration, such as the variability of wind speed and wind direction and the impact of unstable meteorological conditions on wind power forecasting. Through these analyses, multiple factors that can quantify uncertainty are obtained, and each factor has its corresponding value. The larger the value, the higher the corresponding degree of uncertainty. Then, these multiple uncertainty factors are arranged in descending order according to their numerical values, so that the factors with high uncertainty are ranked first and the factors with low uncertainty are ranked later, thereby formulating a clear uncertainty sequence. This sequence provides a clear order and basis for the subsequent analysis and processing of data according to the degree of uncertainty, which helps to solve the uncertainty problem in wind power forecasting in a more targeted manner and improve the accuracy and reliability of the forecast.

[0085] According to the uncertainty sequence established above, the wind power forecast data is carefully analyzed and identified. Along the uncertainty sequence, those forecast values ​​with higher uncertainty are specially marked and identified as targets that need to be analyzed in detail. For these marked high uncertainty forecast values, further in-depth analysis of their sources involves reviewing the relevant data collection links, such as whether there are problems with the accuracy and stability of the sensor, whether there is interference or error introduced during the data transmission process; re-evaluating the assumptions and parameters of the prediction model to see if some model settings are no longer applicable in specific situations, or whether some parameters are estimated inaccurately. Through such a comprehensive and in-depth analysis, the key factors that cause uncertainty are determined. At the same time, a corresponding priority is assigned to each data point or data subset according to the degree of uncertainty. The higher the uncertainty, the higher the priority, which means that more priority attention and resource investment are required in the subsequent analysis and processing. All these analysis identification results and priority information are integrated to finally generate an analysis priority list.

[0086] Based on the analysis priority list, the uncertainty analysis of wind power forecast data is carried out. Monte Carlo simulation technology is used to randomly generate a large number of simulation samples according to the distribution characteristics and uncertainty factors of the forecast data. For data points with higher priority, the number of simulations is increased to improve the accuracy. Through multiple simulations, the wind power forecast value corresponding to each sample is calculated to obtain a distribution of simulation results. Then, the simulation results are analyzed using parameter estimation methods in statistics, such as maximum likelihood estimation or Bayesian estimation, to estimate the parameters of the distribution, such as the mean and standard deviation. Based on these parameters, the confidence level information is calculated. For example, for a normal distribution, the quantile corresponding to a given confidence level (such as 95%) can be determined based on the properties of the standard normal distribution. Next, based on the calculated confidence level information, the confidence interval of wind power is determined using the interval estimation method. Common methods include confidence interval calculation based on normal distribution, that is, the mean plus or minus a certain multiple of the standard deviation (for a 95% confidence level, it is usually the mean plus or minus 1.96 times the standard deviation). If the data distribution does not conform to the normal distribution, non-parametric methods such as percentile-based interval estimation are used to select the upper and lower percentiles of the corresponding confidence level in the simulation results as the boundaries of the confidence interval. Through such a series of calculations, the uncertainty analysis of wind power forecast data can be accurately performed, reliable confidence level information can be obtained, and a reasonable wind power confidence interval can be finally determined, providing a scientific basis for the operational decision-making of wind farms, effectively dealing with the uncertainty in wind power forecasting, and ensuring the stable operation and economic benefits of wind farms.

[0087] In a possible implementation, step S450 further includes:

[0088] Step S451: establishing a time separation channel according to the data collection time based on the analysis priority list.

[0089] Step S452: extracting power characteristics of the wind power prediction data through the time separation channel to obtain a power characteristic extraction result.

[0090] Step S453: Compensating the confidence level information according to the power feature extraction result.

[0091] Specifically, based on the analysis priority list, a time separation channel is established according to the time order of data collection, and the wind power forecast data collected at different time points are classified and divided into channels according to their time attributes. Each time separation channel corresponds to a specific time period, which helps to more clearly observe and analyze the changing laws and characteristics of the data in the time dimension. In this way, the impact of time factors on the uncertainty of wind power forecast data, as well as the potential connections and differences between data in different time periods, can be better considered.

[0092] By establishing a good time separation channel, power features are extracted from wind power prediction data. Fast Fourier transform is used to extract key features that can reflect the power change characteristics from the original data, such as power fluctuation frequency, amplitude, periodicity, etc. These power feature extraction results not only contain the basic information of the data, but also reveal the inherent pattern and change trend of the data in the time domain.

[0093] According to the power feature extraction results obtained above, the confidence level information is compensated, and the confidence level is adjusted and optimized by analyzing the relationship between power features and uncertainty. If the power features show that the data in a certain time period has high stability or specific rules, then the confidence level corresponding to the time period is appropriately increased; conversely, if the power features show that the data has large fluctuations or anomalies, the corresponding confidence level is lowered to more accurately reflect the actual uncertainty situation. This compensation operation enables the confidence level information to be more in line with the characteristics and changes of the actual data, thereby improving the accuracy and reliability of subsequent decisions based on confidence intervals, and providing more accurate guidance for the operation and management of wind farms.

[0094] Embodiment 2 is based on the same inventive concept as the aquatic product processing and stirring method based on multi-data fusion in the above embodiment. Figure 2 As shown, the present application provides a water product processing stirring device based on multi-data fusion, and the device and method embodiments in the embodiments of the present application are based on the same inventive concept. Among them, the device includes:

[0095] The operating data set acquisition module 10 is used to perform data sensing on a target wind farm area through a plurality of sensors to acquire a plurality of operating data sets of the wind farm.

[0096] The Gaussian regression model construction module 20 performs wind power similarity analysis based on the multiple operating data sets, determines nonlinear coefficients, and constructs a Gaussian regression model through training based on the nonlinear coefficients.

[0097] The regression data set output module 30 is used to input the multiple operating data sets as variables into the Gaussian regression model and output a regression data set, wherein the regression data set includes wind power prediction data and data uncertainty intervals.

[0098] The confidence interval determination module 40 is used to perform uncertainty analysis on the wind power prediction data according to the data uncertainty interval, perform simulation calculation according to the analysis result, and determine the confidence interval of the wind power.

[0099] The scheduling control strategy formulation module 50 is used to perform control analysis on the target wind farm according to the confidence interval of the wind power, formulate a scheduling control strategy, and execute the scheduling control strategy to dispatch the wind power of the target wind farm.

[0100] Furthermore, the operation data set acquisition module 10 further includes:

[0101] The data collection point determination unit is used to retrieve the historical operation data record log, collect and analyze the target wind farm area based on the historical operation data record log, and determine multiple data collection points.

[0102] A data sensing area acquisition unit is used to perform cluster analysis according to the multiple data collection points, determine multiple data categories, arrange multiple sensors according to the multiple data categories, and obtain multiple data sensing areas.

[0103] An initial operation data set acquisition unit is used to traverse the multiple data sensing areas to perform data sensing on the target wind farm area to obtain an initial operation data set.

[0104] An operating data set acquisition unit is used to perform data denoising on the initial operating data set to obtain the multiple operating data sets.

[0105] Furthermore, the initial operation data set acquisition unit further includes:

[0106] A wind farm multi-dimensional sensing information generating unit is configured to traverse the plurality of data sensing areas to perform multi-dimensional sensing analysis on the target wind farm to generate multi-dimensional sensing information of the wind farm.

[0107] A wind speed sensing data set acquisition unit is used to perform wind speed sensing on a target wind farm according to the plurality of data sensing areas based on the multi-dimensional sensing information of the wind farm, so as to obtain a wind speed sensing data set.

[0108] A wind direction sensing data set acquisition unit is used to perform wind direction sensing on a target wind farm according to the plurality of data sensing areas based on the multi-dimensional sensing information of the wind farm, so as to obtain a wind direction sensing data set.

[0109] An environmental sensing data set acquisition unit is used to perform environmental sensing on a target wind farm according to the multi-dimensional sensing information of the wind farm and the multiple data sensing areas to obtain an environmental sensing data set.

[0110] A power sensing data set acquisition unit is used to perform power output sensing on a target wind farm according to the multi-dimensional sensing information of the wind farm and the multiple data sensing areas to obtain a power sensing data set.

[0111] An initial operation data set acquisition unit is used to associate and integrate the wind speed sensor data set, the wind direction sensor data set, the environment sensor data set, and the power sensor data set to obtain an initial operation data set.

[0112] Furthermore, the operation data set acquisition unit further includes:

[0113] An operation signal acquisition unit is configured to obtain an operation signal of the initial operation data set by performing analysis based on the wind speed sensor data set, the wind direction sensor data set, the environment sensor data set, and the power sensor data set.

[0114] An operation signal wavelet coefficient acquisition unit is used to perform wavelet decomposition on the operation signal of the initial operation data set to obtain operation signal wavelet coefficients.

[0115] The operation signal wavelet selection threshold determination unit is used to perform threshold quantization according to the operation signal wavelet coefficients to determine the operation signal wavelet selection threshold.

[0116] An effective signal information acquisition unit is used to intercept the operating signal wavelet coefficients according to the operating signal wavelet selection threshold, set the noise signal smaller than the operating signal wavelet selection threshold to zero, and obtain effective signal information greater than the operating signal wavelet selection threshold.

[0117] A filtering and reconstruction unit is used to perform filtering and reconstruction on the effective signal information to obtain the multiple operating data sets.

[0118] Furthermore, the regression data set output module 30 also includes:

[0119] An operation probability distribution information definition unit is configured to define operation probability distribution information based on the plurality of operation data sets, wherein the operation probability distribution information includes wind speed probability distribution information and wind direction probability distribution information.

[0120] A sampling data group acquisition unit is configured to perform random sampling based on the wind speed probability distribution information and the wind direction probability distribution information to obtain a sampling data group.

[0121] A simulation input scenario determination unit is used to perform environmental simulation analysis on a target wind farm according to the sampling data group to determine a plurality of simulation input scenarios.

[0122] A wind power prediction data generation unit is used to synchronize the multiple operating data sets to the Gaussian regression model according to the multiple simulation input scenarios to generate the wind power prediction data.

[0123] A prediction extreme value acquisition unit, wherein the prediction extreme value acquisition unit is used to perform critical analysis according to the wind power prediction data to determine a prediction extreme value, wherein the prediction extreme value includes a prediction upper limit value and a prediction lower limit value.

[0124] A data uncertainty interval determination unit is used to perform mapping calculation according to the prediction upper limit value, the prediction lower limit value and the wind power prediction data to determine the data uncertainty interval.

[0125] Furthermore, the confidence interval determination module 40 further includes:

[0126] A wind power prediction sub-data generating unit is configured to classify the wind power prediction data based on the data uncertainty interval to generate a plurality of wind power prediction sub-data.

[0127] A wind power prediction standard deviation acquisition unit is used to perform calculations based on the plurality of wind power prediction sub-data to obtain the wind power prediction standard deviation.

[0128] An uncertainty sequence formulation unit is provided, wherein the uncertainty sequence formulation unit performs uncertainty assessment on the wind power prediction data based on the wind power prediction standard deviation, obtains a plurality of uncertainty factors, arranges the plurality of uncertainty factors in descending order, and formulates an uncertainty sequence.

[0129] An analysis priority list generating unit is used to analyze and identify the wind power prediction data according to the uncertainty sequence to generate an analysis priority list.

[0130] A confidence level information acquisition unit is used to perform uncertainty analysis on the wind power prediction data based on the analysis priority list to obtain confidence level information, and to determine a confidence interval of the wind power based on the confidence level information.

[0131] Furthermore, the confidence interval determination module 40 further includes:

[0132] A time separation channel establishment unit is used to establish a time separation channel according to the data acquisition time based on the analysis priority list.

[0133] A power feature extraction result acquisition unit is used to extract the power features of the wind power prediction data through the time separation channel to obtain a power feature extraction result.

[0134] An execution compensation unit is used to perform compensation on the confidence level information according to the power feature extraction result.

[0135] Embodiment three, based on the uncertainty analysis method of wind power prediction in the above embodiment, the same inventive concept, the present application also provides a computer-readable storage medium, the computer-readable storage medium stores a computer program, and the computer program implements the steps of any one of the methods described in the above embodiment one when executed.

[0136] It should be noted that the above-mentioned sequence of the embodiments of the present application is only for description and does not represent the advantages and disadvantages of the embodiments. And the above-mentioned specific embodiments of this specification are described. In addition, the processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0137] The above description is only a preferred embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present application should be included in the protection scope of the present application.

[0138] This specification and the drawings are merely exemplary illustrations of the present application and are deemed to cover any and all modifications, variations, combinations or equivalents within the scope of the present application. Obviously, a person skilled in the art may make various modifications and variations to the present application without departing from the scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the present application and its equivalents, the present application intends to include these modifications and variations.

Claims

1. Uncertainty analysis method for wind power forecasting, characterized in that: The method comprises: Data sensing is performed on the target wind farm area through multiple sensors to obtain multiple operating data sets of the wind farm; Performing wind power similarity analysis based on the multiple operating data sets to determine a nonlinear coefficient, and training and constructing a Gaussian regression model based on the nonlinear coefficient; Input the multiple operating data sets as variables into the Gaussian regression model, and output a regression data set, wherein the regression data set includes wind power prediction data and data uncertainty intervals; Performing uncertainty analysis on the wind power prediction data according to the data uncertainty interval, performing simulation calculations based on the analysis results, and determining a confidence interval for the wind power; A control analysis is performed on the target wind farm according to the confidence interval of the wind power, a dispatch control strategy is formulated, and the dispatch control strategy is executed to dispatch the wind power of the target wind farm.

2. The uncertainty analysis method for wind power prediction according to claim 1, characterized in that: Data sensing is performed on a target wind farm area by using multiple sensors to obtain multiple operating data sets of the wind farm, and the method includes: Retrieving historical operation data record logs, collecting and analyzing the target wind farm area based on the historical operation data record logs, and determining multiple data collection points; Performing cluster analysis according to the multiple data collection points to determine multiple data categories, and deploying multiple sensors according to the multiple data categories to obtain multiple data sensing areas; Traversing the plurality of data sensing areas to perform data sensing on a target wind farm area to obtain an initial operation data set; Data denoising is performed on the initial running data set to obtain the multiple running data sets.

3. The uncertainty analysis method for wind power prediction according to claim 2, characterized in that: The method of traversing the plurality of data sensing areas to perform data sensing on a target wind farm area to obtain an initial operation data set includes: Traversing the plurality of data sensing areas to perform multi-dimensional sensing analysis on the target wind farm to generate multi-dimensional sensing information of the wind farm; According to the multi-dimensional sensing information of the wind farm, wind speed sensing is performed on the target wind farm according to the multiple data sensing areas to obtain a wind speed sensing data set; According to the multi-dimensional sensing information of the wind farm, wind direction sensing is performed on the target wind farm according to the multiple data sensing areas to obtain a wind direction sensing data set; According to the multi-dimensional sensing information of the wind farm, performing environmental sensing on the target wind farm according to the multiple data sensing areas to obtain an environmental sensing data set; According to the multi-dimensional sensing information of the wind farm, power output sensing is performed on the target wind farm according to the multiple data sensing areas to obtain a power sensing data set; The wind speed sensor data set, the wind direction sensor data set, the environment sensor data set, and the power sensor data set are associated and integrated to obtain an initial operation data set.

4. The uncertainty analysis method for wind power prediction according to claim 3, characterized in that: Performing data denoising on the initial running data set to obtain the multiple running data sets, the method comprising: Performing analysis based on the wind speed sensor data set, the wind direction sensor data set, the environment sensor data set, and the power sensor data set to obtain an operation signal of the initial operation data set; Performing wavelet decomposition on the operation signal of the initial operation data set to obtain the operation signal wavelet coefficients; Perform threshold quantization according to the wavelet coefficients of the operating signal to determine the operating signal wavelet selection threshold; The operating signal wavelet coefficients are intercepted according to the operating signal wavelet selection threshold, noise signals less than the operating signal wavelet selection threshold are set to zero, and effective signal information greater than the operating signal wavelet selection threshold is obtained; The effective signal information is filtered and reconstructed to obtain the multiple operating data sets.

5. The uncertainty analysis method for wind power prediction according to claim 1, characterized in that: The multiple running data sets are input as variables into the Gaussian regression model, and the regression data sets are output, the method comprising: Defining operation probability distribution information based on the multiple operation data sets, the operation probability distribution information including wind speed probability distribution information and wind direction probability distribution information; Perform random sampling based on the wind speed probability distribution information and the wind direction probability distribution information to obtain a sampling data group; Performing environmental simulation analysis on the target wind farm according to the sampling data group to determine a plurality of simulation input scenarios; Synchronizing the multiple operating data sets to the Gaussian regression model according to the multiple simulation input scenarios to generate the wind power prediction data; Perform critical analysis according to the wind power prediction data to determine a prediction extreme value, wherein the prediction extreme value includes a prediction upper limit value and a prediction lower limit value; The data uncertainty interval is determined by performing mapping calculation according to the prediction upper limit value, the prediction lower limit value and the wind power prediction data.

6. The uncertainty analysis method for wind power prediction according to claim 1, characterized in that: The uncertainty analysis of the wind power prediction data is performed according to the data uncertainty interval, and simulation calculation is performed according to the analysis result to determine the confidence interval of the wind power. The method includes: Classifying the wind power prediction data based on the data uncertainty interval to generate a plurality of wind power prediction sub-data; Calculating according to the plurality of wind power prediction sub-data to obtain a wind power prediction standard deviation; Perform uncertainty assessment on the wind power prediction data based on the wind power prediction standard deviation to obtain a plurality of uncertainty factors, arrange the plurality of uncertainty factors in descending order, and formulate an uncertainty sequence; Analyze and identify the wind power forecast data according to the uncertainty sequence to generate an analysis priority list; Uncertainty analysis is performed on the wind power prediction data based on the analysis priority list to obtain confidence level information, and a confidence interval of the wind power is determined based on the confidence level information.

7. The uncertainty analysis method for wind power prediction according to claim 6, characterized in that: After obtaining the confidence level information, the method includes: Establishing time separation channels according to data collection time based on the analysis priority list; Extracting power characteristics of the wind power prediction data through the time separation channel to obtain a power characteristics extraction result; Compensation is performed on the confidence level information according to the power feature extraction result.

8. An uncertainty analysis device for wind power prediction, characterized in that: The device comprises: An operation data set acquisition module, the operation data set acquisition module is used to perform data sensing on a target wind farm area through multiple sensors to acquire multiple operation data sets of the wind farm; A Gaussian regression model construction module, wherein the Gaussian regression model construction module performs wind power similarity analysis based on the multiple operating data sets, determines a nonlinear coefficient, and constructs a Gaussian regression model through training based on the nonlinear coefficient; A regression data set output module, the regression data set output module is used to input the multiple operating data sets as variables into the Gaussian regression model, and output a regression data set, the regression data set including wind power prediction data and data uncertainty intervals; A confidence interval determination module, the confidence interval determination module is used to perform uncertainty analysis on the wind power prediction data according to the data uncertainty interval, perform simulation calculation according to the analysis result, and determine the confidence interval of the wind power; A dispatching control strategy formulation module is used to perform control analysis on a target wind farm according to a confidence interval of the wind power, formulate a dispatching control strategy, and execute the dispatching control strategy to dispatch wind power to the target wind farm.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, which implements the steps of the method according to any one of claims 1 to 7 when executed.

Images