A Prediction Method for the Index of New Energy Energy Storage and Absorption Capacity

Abstract

The present application provides a method for predicting the new energy storage and consumption capacity index, which relates to the field of power system operation and includes: for photovoltaic power generation, establishing a non-linear mapping relationship between factors such as solar radiation intensity, photovoltaic module conversion efficiency, temperature coefficient, and aging attenuation rate and power generation efficiency to dynamically correct the predicted value of the potential output of photovoltaic power generation; based on the predicted results of the potential output of photovoltaic and wind power generation, solving the energy storage capacity configuration and operation strategy through an intelligent optimization algorithm to maximize the benefits of the energy storage system, and obtaining the dynamic predicted value of the energy storage and consumption capacity index; and adopting an adaptive weighted fusion strategy to combine the prediction results of a statistical model based on historical data, a kinetic model based on physical mechanisms, and a data-driven model based on machine learning. The present invention realizes the accurate prediction of the potential output of new energy power generation and the dynamic evaluation of the energy storage and consumption capacity, providing effective technical support for the grid connection of new energy and the optimal configuration of the energy storage system.

Description

Technical Field

[0001] The present invention relates to the field of power system operation, and particularly to a method for predicting the new energy storage consumption capacity index. Background Art

[0002] In the prediction of the new energy storage consumption capacity index, a key technical problem is how to accurately predict the potential output of new energy power generation. Taking photovoltaic power generation as an example, the factors affecting its potential power generation include solar radiation intensity, conversion efficiency of photovoltaic modules, temperature coefficient of the modules, aging attenuation rate of the modules, etc. And these factors themselves are also affected by many uncertain factors, such as weather conditions, atmospheric transparency, cleanliness of photovoltaic modules, degradation rate of the modules, etc. Similarly, the potential wind energy resources of wind power generation are also affected by factors such as wind speed, wind direction, air density, and aerodynamic characteristics of wind turbine blades. The uncertainty and volatility of these factors bring great challenges to the prediction of the potential output of new energy power generation.

[0003] How to establish an accurate and robust prediction model for the potential output of new energy power generation considering various uncertain factors, and organically combine it with the evaluation model of the consumption capacity of the energy storage system to form a complete prediction framework for the energy storage consumption capacity index is a key technical problem to be solved urgently. This requires in-depth analysis of the interaction and transmission mechanism between various influencing factors, development of advanced data fusion and mining algorithms, construction of prediction models with multiple time scales and multiple spatial resolutions, and verification and optimization of the models through a large amount of measured data, and finally form a set of practical and reliable prediction technology systems for the energy storage consumption capacity index, providing important decision-making support for the efficient utilization of new energy and the safe and stable operation of the power system. Summary of the Invention

[0004] The present invention provides a prediction of the new energy storage consumption capacity index, mainly including:

[0005] Obtain historical meteorological data and geographical information data, and based on the historical meteorological data and geographical information data, establish a multiple regression model of key factors and weight coefficients affecting the potential output of new energy power generation. Use the Monte Carlo random sampling method to generate a large number of input samples, substitute the input samples into the new energy power generation potential output calculation model to obtain the probability distribution of the potential output, perform wavelet analysis on the probability distribution to obtain the output fluctuation characteristics at different time scales, and conduct abnormal fluctuation early warning according to the output fluctuation characteristics; for photovoltaic power generation, establish a non-linear mapping relationship between factors such as solar radiation intensity, photovoltaic module conversion efficiency, temperature coefficient, and aging attenuation rate and power generation efficiency to dynamically correct the predicted value of the potential output of photovoltaic power generation; for wind power generation, use a computational fluid dynamics model to simulate the wind field distribution, obtain the wind speed and wind direction parameters at the location of the wind turbine, determine the potential output of wind power in combination with the power curve of the wind turbine, and make corrections according to the abnormal fluctuation early warning results; based on the predicted results of the potential output of photovoltaic and wind power generation, solve the energy storage capacity configuration and operation strategy through an intelligent optimization algorithm to maximize the benefits of the energy storage system, and obtain the dynamic predicted value of the energy storage consumption capacity index; and adopt an adaptive weighted fusion strategy to combine the predicted results of the statistical model based on historical data, the dynamic model based on physical mechanism, and the data-driven model based on machine learning to predict the energy storage consumption capacity index.

[0006] Further, the step of using the Monte Carlo random sampling method to generate a large number of input samples includes: according to the uncertainty distribution function of the key factors, use the Monte Carlo random sampling algorithm to generate a large number of random input sample data conforming to the distribution characteristics, and preprocess the random input sample data.

[0007] Further, for the preprocessed data, it includes: using the chi-square test and the Kolmogorov-Smirnov test to judge whether each factor follows the normal distribution or the Weibull distribution. If the factor follows a known distribution type, use the maximum likelihood estimation method to estimate the corresponding distribution parameters, and according to the probability density function obtained by the estimation, generate a large number of random samples following this distribution through the Monte Carlo simulation method for subsequent risk assessment.

[0008] Further, the step of performing wavelet analysis on the probability distribution to obtain the output fluctuation characteristics at different time scales includes: discretize the probability density function and use it as the input signal for wavelet packet decomposition in wavelet transform to obtain wavelet coefficients at multiple time scales. Determine the abnormal fluctuation threshold at each scale according to historical data, judge whether the wavelet coefficients exceed the threshold to obtain abnormal fluctuations, and summarize the abnormal fluctuations to construct a feature vector.

[0009] Further, preprocess the random input sample data, including: removing outliers and missing values, and performing normalization processing using the maximum-minimum normalization method to map the data to the range of [0, 1].

[0010] Within the interval, obtain the preprocessed input sample data. According to the pre-established physical mechanism model of new energy power generation potential output, use the preprocessed input sample data as training data, and minimize the mean square error loss function through the gradient descent algorithm and the backpropagation algorithm to optimize the parameters of the physical mechanism model, obtaining the trained physical mechanism model. Substitute the preprocessed input sample data into the trained physical mechanism model and the data-driven model respectively, calculate the potential output values corresponding to each input sample, and perform weighted averaging on the results predicted by the two models to obtain the final potential output prediction value.

[0011] Further, for photovoltaic power generation, establish a non-linear mapping relationship between solar radiation intensity, photovoltaic module conversion efficiency, temperature coefficient, aging attenuation rate factors and power generation efficiency to dynamically correct the predicted value of photovoltaic power generation potential output, including: using a deep learning algorithm to establish the non-linear mapping relationship between the factors and power generation efficiency, and correcting the predicted value according to the abnormal fluctuation warning result.

[0012] Further, obtain the probability density function of the data, and perform discretization processing on the probability density function using the histogram method to obtain the discretized probability density function, including: using the discretized probability density function as the input signal of wavelet transform, performing multi-scale decomposition using Daubechies wavelets to obtain wavelet coefficients at multiple time scales, determining the abnormal fluctuation threshold of the wavelet coefficients at each time scale according to the preset historical data. If the wavelet coefficient at the current moment exceeds the abnormal fluctuation threshold, it is judged that there is an abnormal fluctuation at this moment, extract the statistical characteristics of the abnormal fluctuation, construct an abnormal fluctuation feature vector, use the random forest algorithm to classify the abnormal fluctuation feature vector to obtain an anomaly detection model. If the abnormal fluctuation feature vector of the newly input data is judged as abnormal by the anomaly detection model, a warning is triggered.

[0013] Further, for wind power generation, use the computational fluid dynamics model to simulate the wind field distribution, obtain the wind speed and wind direction parameters at the location of the wind turbine, determine the wind power potential output in combination with the power curve of the wind turbine, and perform correction according to the abnormal fluctuation warning result, including: according to the three-dimensional wind field model, use the computational fluid dynamics method to simulate the velocity and direction distribution of the wind field, obtain the wind speed and wind direction data at the location of the wind turbine, input the wind speed data into the power curve of the wind turbine to calculate the theoretical power generation power of the wind turbine at this wind speed, and perform correction according to the abnormal fluctuation warning result.

[0014] Further, based on the predicted results of photovoltaic and wind power generation, an intelligent optimization algorithm is used to solve the energy storage capacity configuration and operation strategy to maximize the benefits of the energy storage system, and a dynamic prediction value of the energy storage consumption capacity index is obtained, including: according to the predicted results of the potential output of photovoltaic and wind power generation, an intelligent optimization algorithm is adopted to maximize the benefits of the energy storage system under the conditions of meeting the constraints of new energy consumption and system stability, and a dynamic prediction value of the energy storage consumption capacity index is obtained.

[0015] Further, the adaptive weighted fusion strategy is adopted to combine the prediction results of the statistical model based on historical data, the dynamic model based on physical mechanism, and the data-driven model based on machine learning to predict the energy storage consumption capacity index, including: dynamically adjusting the weight coefficients of each sub-model according to the prediction error to improve the overall accuracy and robustness of the energy storage consumption capacity index prediction.

[0016] The technical solution provided by the embodiment of the present invention may include the following beneficial effects:

[0017] The present invention discloses a method for predicting the energy storage consumption capacity index of new energy. The method first establishes a multiple regression model to determine the key factors affecting new energy power generation and their uncertainty distributions, and generates a potential output probability distribution through Monte Carlo sampling. Wavelet analysis is performed on this distribution to achieve multi-scale decomposition and set thresholds to warn of abnormal fluctuations. For photovoltaic power generation, a deep learning algorithm is used to establish a non-linear mapping relationship to dynamically correct the predicted value. For wind power generation, a computational fluid dynamics model is used to simulate the wind field distribution and combined with the fan power curve to determine the potential output. Based on the above predicted results, an intelligent optimization algorithm is used to solve the energy storage capacity configuration and operation strategy, and a dynamic prediction value of the energy storage consumption capacity index is obtained. Finally, an adaptive weighted fusion strategy is adopted to combine the prediction results of multiple models to improve the overall prediction accuracy. The present invention realizes the accurate prediction of the potential output of new energy power generation and the dynamic evaluation of the energy storage consumption capacity, and provides effective technical support for the grid connection of new energy and the optimal configuration of the energy storage system. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 It is a flowchart of a method for predicting the energy storage consumption capacity index of new energy according to the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0019] Next, the technical solutions in the embodiments of the present invention will be clearly and detailedly described with reference to the drawings in the embodiments of the present invention. The described embodiments are only a part of the embodiments of the present invention.

[0020] As Figure 1 , a method for predicting the energy storage consumption capacity index of new energy in this embodiment may specifically include:

[0021] Step S101: Based on historical meteorological data and geographical information data, a multiple regression model is established using machine learning algorithms to obtain the key factors and weight coefficients affecting the potential output of new energy power generation, and the uncertainty distribution function of each factor is determined. The uncertainty distribution function is used to describe the random variation characteristics of each influencing factor and provide a probability distribution basis for subsequent Monte Carlo sampling.

[0022] Based on historical meteorological data and geographical information data, machine learning algorithms such as random forest are used. Through feature importance evaluation, the key factors affecting the potential output of new energy power generation are screened out. For the selected key factors, their historical data is obtained, and the probability distribution type of each factor is judged by statistical test methods. If it follows a normal distribution, the mean and variance parameters are estimated by the maximum likelihood estimation method; if it follows a Weibull distribution, the scale parameter and shape parameter are estimated to obtain the probability density function of each key factor. According to the estimated probability density function, the Latin hypercube sampling method is used to generate a Monte Carlo sample set, and the sample size can be set to 1000. The generated sample set is input into the subsequent established multiple linear regression model to obtain the predicted values of new energy power generation under each sample. Based on the key factors screened in the first step and the sample set obtained by Monte Carlo simulation, a multiple linear regression model is established to fit the quantitative relationship between new energy power generation and each key factor, and the regression coefficient reflects the influence weight of each factor. Substitute the Monte Carlo sample set into the regression model to obtain the predicted values of new energy power generation under each sample, and draw a probability distribution histogram of the predicted power generation. Calculate the power generation at different quantiles (such as P50, P90, etc.) as an evaluation index of new energy power generation potential to provide reference for new energy project decision-making.

[0023] Exemplarily, there are many factors affecting the potential of new energy power generation, such as meteorological factors like light intensity, wind speed, temperature, humidity, air pressure, etc., and geographical information factors like terrain, altitude, slope, aspect, etc. To accurately evaluate the potential of new energy power generation, key influencing factors need to be screened out. The random forest algorithm can evaluate feature importance. For example, by collecting 10-year historical meteorological data and geographical information data of a certain area, as well as the new energy power generation data for the corresponding time period, a random forest model is constructed. After the model training is completed, the importance ranking of each feature can be obtained. Suppose the ranking result shows that light intensity, wind speed, and temperature are the three most critical factors affecting the potential of new energy power generation in this area. After screening out the key factors, it is necessary to determine the probability distribution types of these factors. Obtain the historical data of these key factors. For example, collect the daily light intensity, wind speed, and temperature data of this area in the past 10 years. Use statistical test methods, such as the Kolmogorov-Smirnov test or the Anderson-Darling test, to judge the probability distribution type of each factor. Suppose the test result shows that the light intensity follows a normal distribution, the wind speed follows a Weibull distribution, and the temperature follows a normal distribution. For the light intensity and temperature that follow a normal distribution, the maximum likelihood estimation method can be used to estimate their mean and variance. For example, according to the historical data, the sample mean of the light intensity is calculated to be 800W / m 2 , and the sample variance is 10000 (W / m 2) 2; The sample mean of the temperature is 20 °C and the sample variance is 9 °C². For the wind speed that follows a Weibull distribution, its scale parameter and shape parameter can be estimated. For example, based on historical data, the scale parameter of the wind speed is calculated to be 6 m / s and the shape parameter is 2. Thus, the probability density functions of the three key factors are obtained. After obtaining the probability density functions, the Latin hypercube sampling method can be used to generate a Monte Carlo sample set. The sample size is set to 1000. For example, according to the probability density functions of light intensity, wind speed, and temperature, 1000 random samples that follow their respective distributions are generated respectively to form a sample set with 1000 rows and 3 columns. Each row represents a possible scenario and contains the values of the three factors: light intensity, wind speed, and temperature. Based on the selected key factors (light intensity, wind speed, temperature) and the 1000 samples generated by Monte Carlo simulation, a multiple linear regression model can be established. This model is used to fit the quantitative relationship between new energy power generation and light intensity, wind speed, and temperature. For example, assume the established regression model is: power generation = a * light intensity + b * wind speed + c * temperature + d, where a, b, c, and d are regression coefficients, reflecting the influence weights of each factor on power generation. By fitting historical data, the specific values of these coefficients can be obtained. Substituting each sample in the Monte Carlo sample set into the established multiple linear regression model, the predicted values of new energy power generation under 1000 samples can be obtained. Plotting these 1000 predicted values as a probability distribution histogram can visually show the distribution of new energy power generation potential. According to the probability distribution histogram, the power generation at different quantiles (such as P50, P90, etc.) can be calculated. P50 means there is a 50% probability that the power generation will reach or exceed this value, and P90 means there is a 90% probability that the power generation will reach or exceed this value. These quantile values can be used as evaluation indicators for new energy power generation potential and provide reference for new energy project decisions. For example, if the power generation corresponding to P90 is lower than the expected value, it may be necessary to re-evaluate the project feasibility. This can quantify the risk and more scientifically guide the investment decisions of new energy projects.

[0024] Step S102: According to historical meteorological data and geographical information data, use a statistical test method to judge the probability distribution type of each factor. If it follows a normal distribution, use the maximum likelihood estimation method to estimate the mean and variance parameters. If it follows a Weibull distribution, estimate the scale parameter and shape parameter to obtain the probability density function of each key factor.

[0025] Obtain historical meteorological data and geographical information data, and preprocess the data, including using interpolation methods to handle missing values, using methods such as box plots to identify and remove outliers, and using methods such as min-max standardization for data standardization. For the preprocessed data, use statistical test methods such as chi-square test and Kolmogorov-Smirnov test to determine whether each factor follows common probability distribution types such as normal distribution and Weibull distribution. For factors that follow known distribution types, use the maximum likelihood estimation method to estimate the corresponding distribution parameters. Taking the normal distribution as an example, construct a log-likelihood function based on the sample data, and obtain the maximum likelihood estimates of the mean and variance by solving the log-likelihood equations. Similarly, for factors that follow the Weibull distribution, estimate the scale parameter and shape parameter by constructing and solving the log-likelihood equations. Substitute the estimated parameters into the corresponding probability density function expression to obtain the probability density functions of each key factor. On this basis, calculate characteristic quantities such as the mean and variance of the probability density function through methods such as numerical integration, and obtain common quantiles by solving the inverse function of the probability distribution function, providing a basis for subsequent analysis. Use the estimated probability density function and adopt the Monte Carlo simulation method to generate a large number of random samples that follow this distribution. Specifically, first obtain the probability distribution function based on the probability density function, then generate uniform random numbers in the interval [0, 1] and substitute them into the inverse function of the probability distribution function to obtain random samples that follow the target distribution. Repeat the above process to obtain a sufficient number of random samples. Based on the generated random samples, evaluate the risk level and uncertainty of the system under different scenarios through methods such as statistical analysis and hypothesis testing. For example, estimate the expected value and volatility of system risk through the mean and variance of the samples; estimate the risk threshold at a given confidence level through sample quantiles; determine whether the difference in risk levels under different scenarios is significant through hypothesis testing, etc. According to the results of risk assessment, optimize the system design and operation strategy and formulate effective risk management measures. For factors that do not follow known distribution types, nonparametric estimation and other methods can be used to directly estimate their probability density functions based on sample data. For example, use methods such as kernel density estimation, and by selecting appropriate kernel functions and bandwidth parameters, fit the empirical distribution of the samples to obtain nonparametric estimates of the probability density function. On this basis, various characteristic quantities of the nonparametric estimated probability density function can be further calculated and used for subsequent risk assessment and decision-making analysis.

[0026] Exemplarily, meteorological data and geographical information data are important bases for evaluating the potential of new energy power generation. After obtaining these data, preprocessing is required first to ensure the quality and reliability of the data. For example, collect 10-year historical meteorological data of a certain area, including daily light intensity, wind speed, temperature, humidity, etc., and geographical information data such as altitude, slope, aspect, etc. During the data collection process, data missing may occur. For example, meteorological data for some days are not recorded due to equipment failures. At this time, interpolation methods can be used to fill in the missing values. For example, linear interpolation, spline interpolation and other methods can be used to estimate the missing values based on the data at adjacent time points. In addition, there may be outliers in the data, such as data deviations caused by measurement errors or extreme weather events. Methods such as box plots can be used to identify and remove outliers. Box plots can visually display the distribution of data and identify data points outside the normal range. To eliminate the differences in the numerical ranges between different factors, the data needs to be standardized. For example, the min-max normalization method can be used to scale the data of each factor to between 0 and 1 for better subsequent analysis. After the data preprocessing is completed, it is necessary to judge the probability distribution types of each factor. For example, to understand the probability distribution of factors such as light intensity, wind speed, and temperature. Statistical test methods such as chi-square test, Kolmogorov-Smirnov test, etc. can be used to judge whether each factor follows common probability distribution types such as normal distribution, Weibull distribution, etc. The Kolmogorov-Smirnov test can compare the difference between the empirical distribution of the sample data and the hypothesized theoretical distribution, so as to judge whether the sample data follows the theoretical distribution. The results of the hypothesis test show that the light intensity follows a normal distribution, the wind speed follows a Weibull distribution, and the temperature follows a gamma distribution. For factors that follow known distribution types, such as the light intensity following a normal distribution, the maximum likelihood estimation method can be used to estimate the corresponding distribution parameters. The basic idea of the maximum likelihood estimation method is to find the parameter values that are most likely to generate the observed data. Specifically, a log-likelihood function is constructed based on the sample data, and by solving the log-likelihood equations, the maximum likelihood estimates of the mean and variance are obtained. Suppose the sample mean of the light intensity calculated based on historical data is 800W / m 2 , and the sample variance is 10000 (W / m 2)2. For wind speeds that follow a Weibull distribution, its scale parameter and shape parameter can be estimated. Suppose the scale parameter of the wind speed calculated based on historical data is 6 m / s and the shape parameter is 2. For temperatures that follow a gamma distribution, its shape parameter and scale parameter can be estimated. Suppose the shape parameter of the temperature calculated based on historical data is 3 and the scale parameter is 2. Substitute the estimated parameters into the corresponding probability density function expressions to obtain the probability density functions of each key factor. For example, the probability density function of light intensity is a normal distribution function, the probability density function of wind speed is a Weibull distribution function, and the probability density function of temperature is a gamma distribution function. After obtaining the probability density function, characteristic quantities such as the mean and variance of the probability density function can be calculated by methods such as numerical integration. The mean reflects the average level of the factor, and the variance reflects the degree of fluctuation of the factor. In addition, common quantiles such as P50 and P90 can be obtained by solving the inverse function of the probability distribution function. Quantiles can be used to describe the value range of the factor at different probability levels. Using the estimated probability density function, a large number of random samples that follow this distribution can be generated by the Monte Carlo simulation method. The Monte Carlo simulation is a numerical simulation method based on random sampling. Specifically, first obtain the probability distribution function based on the probability density function, then generate uniform random numbers in the [0, 1] interval and substitute them into the inverse function of the probability distribution function to obtain random samples that follow the target distribution. For example, to generate 1000 random samples that follow the light intensity distribution, 1000 uniform random numbers in the [0, 1] interval can be generated first, and then these random numbers are substituted into the inverse function of the light intensity probability distribution function to obtain 1000 random samples of light intensity. Repeating the above process can obtain a sufficient number of random samples. Based on the generated random samples, the risk level and uncertainty of the system in different scenarios can be evaluated by methods such as statistical analysis and hypothesis testing. For example, suppose you want to evaluate the risk of new energy power generation in the next 10 years. Through Monte Carlo simulation, 10000 random samples of factors such as light intensity, wind speed, and temperature in the next 10 years are generated. Substituting these samples into the new energy power generation prediction model can obtain 10000 predicted values of new energy power generation in the next 10 years. By analyzing the distribution of these 10000 predicted values, the risk level and uncertainty of new energy power generation can be evaluated. For example, the mean and variance of the predicted values can be calculated to estimate the expected value and volatility of the new energy power generation risk. The quantiles of the predicted values can be calculated to estimate the risk threshold at a given confidence level. Hypothesis testing can also be used to determine whether the difference in risk levels in different scenarios is significant. For factors that do not follow a known distribution type, such as the historical data of some meteorological factors that do not follow any common probability distribution type, non-parametric estimation and other methods can be used to directly estimate its probability density function based on the sample data.For example, the kernel density estimation method can be used. By selecting appropriate kernel functions and bandwidth parameters, the empirical distribution of the samples is fitted to obtain a non-parametric estimation of the probability density function. Kernel density estimation is a non-parametric method for estimating the probability density function based on sample data. It estimates the overall probability density function by weighted averaging the probability densities around each sample point. The choice of the kernel function and the setting of the bandwidth parameter will affect the smoothness of the estimation result.

[0027] In step S103, a large number of input samples are generated by the Monte Carlo random sampling method using the above uncertainty distribution function, and are substituted into the new energy power generation potential output calculation model to obtain the probability distribution of the potential output. This probability distribution reflects the uncertainty characteristics of the new energy power generation potential output.

[0028] According to the uncertainty distribution function, the Monte Carlo random sampling algorithm is used to generate a large number of random input sample data that conform to the distribution characteristics. Necessary preprocessing is performed on the generated random input sample data, including data cleaning, outlier processing, and data normalization, etc., and the characteristic variables related to the new energy power generation potential output are extracted. The preprocessed input sample data is substituted into the pre-established new energy power generation potential output calculation model, which can be a mathematical model based on physical mechanisms or a data-driven model trained by machine learning algorithms (such as support vector machines or neural networks). The potential output value corresponding to each input sample is obtained through model calculation. According to the rated output of the new energy power generation unit and historical operation data, several potential output intervals are reasonably set. The potential output value of each calculated input sample is divided into the corresponding interval. The number of samples in each potential output interval is counted to obtain the frequency distribution of the potential output in different numerical intervals. According to the frequency distribution of the potential output, the empirical probability method is used to calculate the probability of each potential output interval to obtain the probability distribution of the new energy power generation potential output. The probability distribution of the new energy power generation potential output is analyzed, and the uncertainty degree of the potential output is quantitatively evaluated through statistical indicators such as the skewness and kurtosis of the distribution curve. When the absolute value of the skewness is greater than 1 or the kurtosis is greater than 5, it indicates that there is a large uncertainty in the new energy power generation potential output, and measures such as reasonably dispatching reserve capacity and optimizing energy storage configuration need to be taken to cope with its fluctuation risk; when the absolute value of the skewness is less than 5 and the kurtosis is less than 3, it indicates that the uncertainty of the new energy power generation potential output is small, and the grid connection ratio can be appropriately increased to reduce the phenomenon of wind and light abandonment.

[0029] Exemplarily, the core idea of the Monte Carlo random sampling algorithm is to generate random samples from a known probability distribution. In the evaluation of new energy power generation potential, the probability distribution functions of the key influencing factors (such as light intensity, wind speed, temperature) have been determined. Taking light intensity as an example, assume that it follows a normal distribution with a mean of 800W / m 2, with a standard deviation of 100 W / m 2 . Using the Monte Carlo method, a large number of random light intensity samples can be generated, and the distribution of these sample values conforms to the previously determined normal distribution. This is done to simulate the random variation characteristics of light intensity and make it closer to the real situation. The generated sample data usually needs to be preprocessed to improve the accuracy and stability of the model. For example, the data can be cleaned to remove outliers and missing values. Suppose the light intensity sample value for a certain day is -100 W / m 2 , which is obviously an outlier and needs to be removed or replaced. In addition, the data can be normalized to convert data with different dimensions into the same numerical range. For example, the light intensity, wind speed, and temperature can all be converted into values between 0 and 1. This can prevent certain features from having too much impact on the model due to large numerical values. Substituting the preprocessed sample data into the new energy power generation potential output calculation model, the potential output value corresponding to each sample can be obtained. This model can be a mathematical model based on physical mechanisms or a data-driven machine learning model. Suppose a simple linear regression model is adopted: potential output = a * light intensity + b * wind speed + c * temperature, where a, b, and c are model parameters. Substituting the light intensity, wind speed, and temperature values of each sample into this model, the corresponding potential output value can be calculated. To better analyze the distribution of potential output, it can be divided into different intervals. For example, the potential output can be divided into several intervals such as 0 - 10 MW, 10 - 20 MW, 20 - 30 MW, etc. Then, the number of samples in each interval is counted to obtain the frequency distribution of potential output. This can intuitively understand the distribution of potential output within different numerical ranges. Based on the frequency distribution, the probability of each potential output interval can be calculated. For example, suppose among 1000 samples, 200 samples have potential output falling within the 10 - 20 MW interval, then the probability of this interval is 200 / 1000 = 2. By calculating the probability of each interval, the probability distribution of new energy power generation potential output can be obtained. By analyzing statistical indicators such as the skewness and kurtosis of the probability distribution, the uncertainty degree of potential output can be quantitatively evaluated. Skewness measures the skewness of the distribution, and kurtosis measures the peakedness of the distribution. Suppose the calculated skewness of the potential output distribution is 8 and the kurtosis is 3. Since the absolute value of the skewness is less than 1 and the kurtosis is less than 5, it indicates that the uncertainty of potential output is small. In this case, the grid connection ratio of new energy can be appropriately increased to reduce the phenomenon of wind and light abandonment. On the contrary, if the skewness or kurtosis is too large, it indicates that the uncertainty of potential output is high, and measures such as energy storage and reserve capacity need to be taken to cope with its fluctuation risk. For example, energy storage devices can be configured to store the excess electric energy when the power generation is greater than the load demand and release the stored electric energy when the power generation is less than the load demand, thereby smoothing the volatility of new energy power generation and improving the stability of the power grid.

[0030] In step S104, according to the uncertainty distribution function of the potential output of new energy power generation, the Monte Carlo random sampling algorithm is used to generate a large number of random input sample data that conform to this distribution characteristic. By substituting the preprocessed input sample data into the physics-based mathematical model or the data-driven model trained by machine learning algorithms, the potential output value corresponding to each input sample is calculated, and then the probability distribution of the potential output of new energy power generation is obtained. According to statistical indicators such as the skewness and kurtosis of the distribution curve, the uncertainty degree of the potential output is quantitatively judged.

[0031] According to the uncertainty distribution function of the potential output of new energy power generation, the Monte Carlo random sampling algorithm is used to generate at least one thousand random input sample data that conform to this distribution characteristic. The random input sample data generated by the Monte Carlo random sampling algorithm is preprocessed to remove outliers and missing values, and the maximum-minimum normalization method is used for normalization processing to map the data into the interval [0, 1], obtaining the preprocessed input sample data. According to the pre-established physics mechanism model of the potential output of new energy power generation, using the gradient descent algorithm and the backpropagation algorithm, with the preprocessed input sample data as the training data, the model parameters are optimized by minimizing the mean square error loss function to obtain the trained physics mechanism model. According to the pre-established data-driven model of the potential output of new energy power generation, using the random forest regression algorithm, with the preprocessed input sample data as the training data, multiple decision trees are constructed through bootstrap sampling and random feature selection, and the prediction results of all decision trees are averaged to obtain the trained data-driven model. The preprocessed input sample data is respectively substituted into the trained physics mechanism model and the data-driven model, the potential output value corresponding to each input sample is calculated, and the results predicted by the two models are weighted and averaged as the final potential output prediction value. According to the potential output prediction values of all input samples, the probability distribution curve of the potential output of new energy power generation is drawn. Calculate the skewness coefficient and kurtosis coefficient of this distribution curve. When the absolute value of the skewness coefficient is greater than 0 and the kurtosis coefficient is greater than 0, it is judged that the potential output of new energy power generation has high uncertainty; otherwise, it is judged that the uncertainty of the potential output of new energy power generation is low.

[0032] Exemplarily, the Monte Carlo random sampling algorithm is an effective method for generating random sample data from a known probability distribution. In the field of new energy power generation, it can be used to simulate various uncertain factors affecting power output. For example, assume that based on historical data and meteorological forecasts, the probability distribution function of the light intensity for the next 24 hours is obtained, and this function may be a gamma distribution or a beta distribution. Then, the Monte Carlo method can be used to generate 1000 random sample values of light intensity, and these sample values conform to the pre-determined probability distribution characteristics and can more realistically reflect the change of light intensity. There may be outliers or missing values in the generated sample data, which need to be preprocessed. Assume that the light intensity sample value at a certain moment is -500W / m 2 , which is obviously an outlier and needs to be removed. If the light intensity data is missing at a certain moment, the data of adjacent moments can be used for interpolation or filling. In addition, in order to eliminate the influence of the dimension between different features, the data needs to be normalized. For example, the data such as light intensity, wind speed, and temperature are all normalized between 0 and 1. Assume that the light intensity at a certain moment is 600W / m 2 , the minimum value is 0, and the maximum value is 1000W / m 2, the normalized value is (600 - 0) / (1000 - 0) = 6. The potential output calculation model of new energy power generation can be a physics-based model or a data-driven model. Physics-based models usually establish mathematical equations based on energy conversion principles and equipment characteristics. For example, the physics-based model of photovoltaic power generation can consider factors such as light intensity, solar panel area, and conversion efficiency. Data-driven models use machine learning algorithms to learn the relationship between power generation output and various influencing factors from historical data. For example, a data-driven model can be constructed using the random forest regression algorithm. Suppose a physics-based model of the potential output of photovoltaic power generation is established, where the model parameters include the area of the solar panel, conversion efficiency, etc. Using the gradient descent algorithm and the backpropagation algorithm, these parameters can be optimized according to the preprocessed input sample data to minimize the error between the predicted value and the actual value of the model. Suppose a data-driven model is constructed using the random forest regression algorithm. This algorithm constructs multiple decision trees through bootstrap sampling and random feature selection, and then averages the prediction results of all decision trees to obtain the final prediction result. For example, suppose 100 decision trees are constructed, and the prediction results of each tree are slightly different. A more stable and accurate prediction value can be obtained by averaging these results. Substituting the preprocessed input sample data into the trained physics-based model and data-driven model respectively, two prediction results can be obtained. To comprehensively utilize the advantages of the two models, the two prediction results can be weighted and averaged. For example, suppose the weight of the physics-based model is 6 and the weight of the data-driven model is 4, then the final predicted value of the potential output is 6 × predicted value of the physics-based model + 4 × predicted value of the data-driven model. According to the predicted values of the potential output of all input samples, a probability distribution curve can be drawn. This curve can intuitively show the probability of the potential output within different numerical ranges. For example, it can be observed that the probability of the potential output is relatively high within a certain interval and relatively low within other intervals. By calculating the skewness coefficient and kurtosis coefficient of the probability distribution curve, the uncertainty of the potential output can be quantitatively evaluated. The skewness coefficient measures the degree of skewness of the distribution, and the kurtosis coefficient measures the sharpness of the distribution. For example, if the absolute value of the skewness coefficient is 8 and the kurtosis coefficient is 3, it means that the potential output distribution is slightly skewed and the peak is relatively sharp, indicating a high degree of uncertainty. On the contrary, if the skewness coefficient is close to 0 and the kurtosis coefficient is close to 0, it means that the potential output distribution is relatively symmetric and the peak is relatively flat, indicating a low degree of uncertainty. According to the level of uncertainty, corresponding strategies can be formulated, such as configuring energy storage devices or adjusting the grid connection ratio.

[0033] In step S105, wavelet analysis is performed on the obtained probability distribution of the potential output of new energy power generation to achieve multi-scale decomposition and obtain the output fluctuation characteristics at different time scales. When the fluctuation exceeds the preset threshold, it is determined as an abnormal fluctuation and a warning is triggered. The preset threshold is determined according to historical data statistics and the requirements of power system stability. The abnormal fluctuation warning result is used to guide the subsequent correction of power generation prediction and the optimization of energy storage strategies.

[0034] Obtain the probability distribution data of the potential output of new energy power generation, and use non-parametric estimation methods such as kernel density estimation to obtain the probability density function. Discretize the probability density function as the input signal for wavelet transform, and perform wavelet packet decomposition on this input signal to obtain wavelet coefficients at multiple time scales. According to the historical data of new energy power generation, analyze the statistical characteristics of the wavelet coefficients at each time scale, and combine the requirements of power system stability indicators. Use methods such as statistical hypothesis testing to determine the abnormal fluctuation thresholds of the wavelet coefficients at each scale. Judge whether the wavelet coefficients at each time scale exceed the corresponding abnormal fluctuation thresholds. If they exceed, mark the output fluctuation at that time scale as an abnormal fluctuation. Summarize the abnormal fluctuation information at each time scale to construct an abnormal fluctuation feature vector. Use historical data to construct training samples for the power generation power prediction model through feature engineering and data preprocessing. Use the abnormal fluctuation feature vector as an additional input feature, and adopt algorithms such as long short-term memory neural network to predict the new energy power generation power to obtain the corrected power generation prediction result. According to the corrected power generation prediction result, considering the real-time operation state of the power grid and scheduling constraints, with the minimum charging and discharging cost of the energy storage system as the optimization goal, use heuristic algorithms such as particle swarm optimization to optimize the charging and discharging power and time of the energy storage system to obtain the optimal energy storage operation strategy. Convert the optimized energy storage strategy into control instructions for the energy storage system and send them to the energy storage device through the energy management system to achieve real-time scheduling control of the energy storage system. At the same time, according to the requirements of power grid dispatching, coordinate the output of new energy power generation and the charging and discharging of the energy storage system to improve the adaptability of the power grid to the fluctuations of new energy power generation and ensure the safe and stable operation of the power system.

[0035] For example, the power prediction of renewable energy generation is the key to the stable operation of the power system. Since the output of renewable energy generation such as wind power and solar energy is intermittent and volatile, it is crucial to accurately predict its output and formulate corresponding energy storage strategies. Here we will introduce a method for renewable energy power generation power prediction and energy storage optimization based on wavelet packet decomposition and long short-term memory neural network. First, it is necessary to obtain the probability distribution data of the potential output of renewable energy power generation. For example, the potential output at each moment can be calculated by collecting wind speed data of wind farms over a period of time and combining it with the power curve of the wind turbine. Assume that 1000 hours of wind speed data are collected and 1000 potential output data are calculated. Using the kernel density estimation method, the probability density function of the potential output can be obtained. The kernel density estimation method can be understood as fitting the distribution of sample data with small normal distribution curves, and finally obtaining a smooth curve, which is the probability density function. Discretize the probability density function, for example, divide the output range into 10 intervals, each interval corresponds to a probability value, and obtain a discretized probability distribution as the input signal of the wavelet transform. Next, the discretized probability distribution is decomposed by wavelet packets. Wavelet packet decomposition can decompose the signal into components at different time scales, such as decomposing the probability distribution into high-frequency, medium-frequency and low-frequency parts. The high-frequency part reflects the fluctuation of rapid output change, and the low-frequency part reflects the trend of slow output change. Assume that the probability distribution is decomposed into four time scales through wavelet packet decomposition, representing the fluctuation periods of 1 hour, 4 hours, 12 hours and 24 hours respectively. Then, it is necessary to determine the abnormal fluctuation threshold of the wavelet coefficient at each time scale. For example, the mean and standard deviation of the wavelet coefficient at each time scale can be counted based on historical data. Assuming that the mean of the wavelet coefficient at the 1-hour scale is 0 and the standard deviation is 1, the abnormal fluctuation threshold can be set to 2, that is, the fluctuation exceeding 2 standard deviations from the mean is considered to be an abnormal fluctuation. Similarly, the abnormal fluctuation threshold at other time scales can be determined. Next, it is determined whether the wavelet coefficient at each time scale exceeds the corresponding abnormal fluctuation threshold. If it exceeds, the output fluctuation at that time scale is marked as abnormal fluctuation. For example, if the wavelet coefficient at a certain moment at the 1-hour scale is 3, which exceeds the threshold 2, it is marked as an abnormal fluctuation. Abnormal fluctuation information of all time scales is aggregated to construct abnormal fluctuation feature vectors. For example, the abnormal fluctuations of four time scales can be represented by a four-dimensional vector, where 1 represents abnormal fluctuations and 0 represents normal fluctuations. Use historical data to construct training samples for the power generation prediction model. For example, collect power generation data and weather data from the past year, and perform data preprocessing, such as removing outliers and filling missing values. Use the abnormal fluctuation feature vector as an additional input feature, and use algorithms such as long short-term memory neural networks for power prediction. Long short-term memory neural networks are suitable for processing time series data and can capture long-term dependencies in the data.According to the corrected power generation prediction results, considering the real-time operation state of the power grid and dispatching constraints, such as the maximum charge and discharge power, energy storage capacity limit, etc., with the minimum charge and discharge cost of the energy storage system as the optimization goal, heuristic algorithms such as particle swarm optimization are used to optimize the charge and discharge power and time of the energy storage system. For example, when it is predicted that the power generation power will be high in the future, the energy storage system can be controlled to charge and store the excess energy; when it is predicted that the power generation power is low, the energy storage system can be controlled to discharge to supplement the power shortage. Finally, the optimized energy storage strategy is converted into control instructions for the energy storage system and sent to the energy storage device through the energy management system to achieve real-time dispatching control of the energy storage system. For example, information such as the charge power, discharge power, charge time, and discharge time is converted into control signals and sent to devices such as the energy storage converter. By coordinating the output of new energy power generation and the charge and discharge of the energy storage system, the volatility of new energy power generation can be effectively smoothed, the adaptability of the power grid to the fluctuations of new energy power generation can be improved, and the safe and stable operation of the power system can be ensured.

[0036] Step S106: Obtain the probability density function through kernel density estimation, discretize it as the input signal of wavelet transform, use wavelet packet decomposition to obtain multi-time scale wavelet coefficients, determine the abnormal fluctuation thresholds at each scale according to historical data, judge whether the wavelet coefficients exceed the thresholds to obtain abnormal fluctuations, and summarize the abnormal fluctuations to construct a feature vector.

[0037] Obtain the probability density function of the data through kernel density estimation, and then use the histogram method to discretize the continuous probability density function to obtain a discretized probability density function for subsequent wavelet transform processing. Use the discretized probability density function as the input signal of the wavelet transform, and adopt Daubechies wavelets to perform multi-scale decomposition on the input signal to obtain wavelet coefficients at multiple time scales. Calculate the abnormal fluctuation threshold for the wavelet coefficients at each time scale according to historical data. The specific method is as follows: take the absolute value of the wavelet coefficients of the historical data, and then calculate their mean and standard deviation. The abnormal fluctuation threshold is set to the mean plus 3 times the standard deviation. If the wavelet coefficient at the current moment exceeds the abnormal fluctuation threshold at this scale, it is determined that there is an abnormal fluctuation at this moment. Summarize the abnormal fluctuations detected at each time scale, extract the statistical characteristics of the abnormal fluctuations, including the time scale, duration, amplitude, etc. of the abnormal fluctuations, and construct an abnormal fluctuation feature vector. Use the random forest algorithm to classify the abnormal fluctuation feature vector and train an anomaly detection model to determine whether there is an anomaly in the newly input data. If the abnormal fluctuation feature vector of the newly input data is determined to be abnormal by the anomaly detection model, a warning is triggered to indicate the risk of abnormal fluctuations. Classify the abnormal fluctuations according to the time scale and amplitude of the abnormal fluctuations for hierarchical warning. The larger the time scale and amplitude of the abnormal fluctuation, the higher the risk level. According to the preset risk level threshold, the abnormal fluctuations are divided into different levels such as low risk, medium risk, and high risk, and different levels of risks are prompted to provide a basis for business decisions.

[0038] Exemplarily, kernel density estimation is a non-parametric method for estimating probability density functions. It can be understood as using small normal distribution curves (kernel functions) to fit the distribution of sample data, and finally superimposing these small normal distribution curves to obtain a smooth curve, which is the probability density function. For example, 1000 hours of wind speed data of a wind farm are collected, and 1000 potential output data are calculated according to the power curve of the wind turbine. Using the kernel density estimation method, the probability density function of the wind farm's potential output can be obtained, which describes the probability of different output levels occurring. After obtaining the probability density function, it needs to be discretized for subsequent wavelet transform processing. The histogram method is a commonly used discretization method. It divides the value range of the continuous probability density function into several intervals, counts the frequency of data occurrences in each interval, and converts the frequency into a probability value, finally obtaining the discretized probability density function. For example, the value range of the wind farm's potential output can be divided into 10 intervals, representing output levels of 0-10%, 10%-20%... 90%-100% respectively. Then, count the frequency of potential output data occurrences in each interval and convert it into a probability value, finally obtaining a discretized probability density function containing 10 probability values. Take the discretized probability density function as the input signal of the wavelet transform and perform multi-scale decomposition using Daubechies wavelets. Daubechies wavelets are a commonly used wavelet basis function with good time-frequency localization characteristics. Multi-scale decomposition can decompose the signal into components at different time scales, such as high-frequency parts, medium-frequency parts, and low-frequency parts. The high-frequency part reflects the fluctuations of rapid output changes, and the low-frequency part reflects the trend of slow output changes. For example, the discretized probability density function can be decomposed into 4 time scales, representing the fluctuation periods of 1 hour, 4 hours, 12 hours, and 24 hours respectively. To determine whether there are abnormal fluctuations, it is necessary to determine the abnormal fluctuation threshold of the wavelet coefficients at each time scale. A commonly used method is to calculate the mean and standard deviation based on the wavelet coefficients of historical data, and set the value of the mean plus 3 times the standard deviation as the abnormal fluctuation threshold. For example, collect the wavelet coefficient data of the past year, take the absolute value of the wavelet coefficients at the 1-hour time scale, calculate their mean and standard deviation. Suppose the mean is 5 and the standard deviation is 1, then the abnormal fluctuation threshold is 5 + 3*1 = 8. If the wavelet coefficient at the current moment exceeds the abnormal fluctuation threshold at this scale, it is judged that there is an abnormal fluctuation at this moment. For example, the wavelet coefficient at the 1-hour time scale at the current moment is 9, which exceeds the threshold of 8, so it is judged that there is an abnormal fluctuation at this moment. Similarly, it can be judged whether there are abnormal fluctuations at other time scales. Summarize the abnormal fluctuations detected at each time scale, extract the statistical characteristics of the abnormal fluctuations, and construct an abnormal fluctuation feature vector. The abnormal fluctuation feature vector can include information such as the time scale, duration, and amplitude of the abnormal fluctuation.For example, if abnormal fluctuations are detected on both the 1-hour and 4-hour time scales, with durations of 2 hours and 4 hours respectively, and amplitudes of 1 and 2 respectively, an abnormal fluctuation feature vector containing this information can be constructed. The random forest algorithm is used to classify the abnormal fluctuation feature vector and train the anomaly detection model. Random forest is an ensemble learning method that improves the accuracy of classification by constructing multiple decision trees and voting. For example, data from the past year can be used to train a random forest anomaly detection model, with the abnormal fluctuation feature vector as the input and whether there is an abnormal fluctuation as the output. When the abnormal fluctuation feature vector of the newly input data is judged to be abnormal by the anomaly detection model, a warning is triggered. According to the time scale and amplitude of the abnormal fluctuation, a graded warning is issued for the abnormal fluctuation. The greater the time scale and amplitude of the abnormal fluctuation, the higher the risk level. For example, an abnormal fluctuation with an amplitude of 1 on the 1-hour time scale may be classified as a low risk, while an abnormal fluctuation with an amplitude of 5 on the 24-hour time scale may be classified as a high risk. According to the preset risk level threshold, the abnormal fluctuations are classified into different levels such as low risk, medium risk, and high risk, and different levels of risks are prompted to provide a basis for business decisions. This can help better identify and respond to the abnormal fluctuations in the power output of new energy power generation and ensure the safe and stable operation of the power system.

[0039] Step S107: For photovoltaic power generation, use a deep learning algorithm to establish a non-linear mapping relationship between factors such as solar radiation intensity, photovoltaic module conversion efficiency, temperature coefficient, and aging attenuation rate and the power generation efficiency, and dynamically correct the predicted value of the potential power output of photovoltaic power generation. The correction process takes into account the results of the abnormal fluctuation warning to improve the prediction accuracy.

[0040] Obtain historical data and real-time data of multiple factors affecting the power generation efficiency of photovoltaic power generation, including solar radiation intensity, conversion efficiency of photovoltaic modules, temperature coefficient, and aging attenuation rate. Preprocess the obtained historical data using the Pandas library in Python to remove missing values and outliers, and normalize the data to a unified scale of 0-1 using the min-max normalization method. Build a long short-term memory (LSTM) neural network model using the Keras library in Python, with the historical data of influencing factors as input and the actual power generation efficiency in the corresponding time period as output, and train to establish a non-linear mapping relationship between influencing factors and power generation efficiency. Use the ARIMA time series model to predict the values of solar radiation intensity, conversion efficiency of photovoltaic modules, temperature coefficient, and aging attenuation rate in the future period according to the historical data of each influencing factor, and input the predicted influencing factor data into the trained LSTM model to obtain the predicted value of the potential power output of photovoltaic power generation in this time period. Use the isolation forest algorithm in the Scikit-learn library in Python, with the real-time data of influencing factors as input, to perform anomaly detection. Calculate the normal fluctuation range of each influencing factor based on historical data. If the real-time data exceeds this range, it is determined as abnormal fluctuation and a warning message is generated. Conditionally merge the abnormal fluctuation warning message with the predicted value of the potential power output of photovoltaic power generation generated by the LSTM model. If the data of a certain influencing factor is determined to be abnormal, in the prediction result of the LSTM model, the predicted value of the power generation output at the corresponding time is adjusted downward by a fixed ratio to obtain the corrected predicted value. Use the Sklearn library in Python to compare the corrected predicted value of the photovoltaic power generation output with the actual value and calculate the mean absolute percentage error (MAPE). If the MAPE is greater than 10%, then go back to step 3 and retrain the LSTM model using the updated historical data; otherwise, output the final predicted result of the photovoltaic power generation output.

[0041] Exemplarily, the photovoltaic power generation efficiency is affected by various factors, including solar radiation intensity, photovoltaic module conversion efficiency, temperature coefficient, and aging attenuation rate, etc. To accurately predict the photovoltaic power generation output, historical data and real-time data of these influencing factors need to be obtained. Historical data can be obtained through meteorological stations, photovoltaic power plant monitoring systems, etc., and real-time data can be measured in real time by sensors. For example, data such as solar radiation intensity and photovoltaic module temperature can be recorded every 15 minutes. After obtaining the data, the historical data needs to be preprocessed. The Pandas library can be used to conveniently perform data cleaning and transformation. For example, the fillna function of Pandas can be used to fill missing values, and the drop function can be used to remove outliers. In addition, to eliminate the influence of the dimension between different data, the data needs to be normalized. For example, the min-max normalization method can be used to scale the data between 0 and 1. Suppose the highest temperature on a certain day is 35 degrees Celsius and the lowest temperature is 20 degrees Celsius, then 25 degrees Celsius after normalization is (25 - 20) / (35 - 20) = 0.33. Using the preprocessed historical data, a long short-term memory neural network model can be constructed to predict the photovoltaic power generation efficiency. The LSTM neural network is a special recurrent neural network that can learn long-term dependencies in time series data. The historical data of the influencing factors is used as the input, and the actual power generation efficiency of the corresponding time period is used as the output to train the LSTM model to establish a non-linear mapping relationship between the influencing factors and the power generation efficiency. For example, the model can be trained with data from the past year. The input is data such as daily solar radiation intensity and temperature, and the output is the daily photovoltaic power generation efficiency. To predict the photovoltaic power generation output in the future for a period of time, the future values of the influencing factors need to be predicted first. The ARIMA time series model can be used to predict factors such as solar radiation intensity and temperature. The ARIMA model can capture the autocorrelation in time series data and make predictions based on this. For example, based on the solar radiation intensity data of the past month, the solar radiation intensity of the next week can be predicted. The data of the influencing factors predicted by the ARIMA model is input into the trained LSTM model, and the predicted value of the potential photovoltaic power generation output in the future for a period of time can be obtained. For example, by inputting the predicted data such as solar radiation intensity and temperature of the next week into the LSTM model, the predicted value of the photovoltaic power generation output of the next week can be obtained. To improve the prediction accuracy, anomaly detection needs to be performed on the real-time data of the influencing factors. The Isolation Forest algorithm can be used to identify abnormal data. The Isolation Forest algorithm is an algorithm based on anomaly isolation that can effectively identify data points that deviate from the normal range. For example, if the solar radiation intensity suddenly drops to a very low value at a certain moment, the Isolation Forest algorithm can identify it as abnormal data. Based on historical data, the normal fluctuation range of each influencing factor can be calculated.For example, the mean and standard deviation of solar radiation intensity in historical data can be calculated, and the range of the mean plus or minus twice the standard deviation is taken as the normal fluctuation range. If the real-time data exceeds this range, it is determined as an abnormal fluctuation, and a warning message is generated. The warning message of abnormal fluctuation is conditionally merged with the prediction result of the LSTM model. If a certain influencing factor is determined to be abnormal, in the prediction result of the LSTM model, the predicted power generation output value corresponding to the time is adjusted downward by a fixed ratio. This is to avoid the excessive influence of abnormal data on the prediction result. For example, if the real-time solar radiation intensity is determined to be abnormally low, the predicted photovoltaic power generation output value for the corresponding time period is adjusted downward by 10%. Finally, the mean absolute percentage error (MAPE) between the corrected predicted photovoltaic power generation output value and the actual value is calculated using the Sklearn library. MAPE is a commonly used indicator to measure the prediction accuracy. If MAPE is greater than 10%, it indicates that the prediction accuracy of the model is not high enough, and the LSTM model needs to be retrained using the updated historical data. Otherwise, the final predicted photovoltaic power generation output result is output. This can continuously improve the prediction performance of the model and enhance the prediction accuracy.

[0042] Step S108, for wind power generation, use the computational fluid dynamics model to simulate the wind field distribution, obtain the wind speed and wind direction parameters at the location of the wind turbine, and determine the potential wind power output in combination with the power curve of the wind turbine. Also consider the correction of the abnormal fluctuation warning result to improve the accuracy of wind power output prediction.

[0043] According to the three-dimensional wind field model, use the computational fluid dynamics method to simulate the velocity and direction distribution of the wind field, and obtain the wind speed and wind direction data at the location of the wind turbine. Input the obtained wind speed data into the power curve of the wind turbine, and calculate the theoretical power generation power of the wind turbine at this wind speed as the preliminary estimated value of the potential wind power output. Use the support vector machine (SVM) algorithm to train the historical wind speed, wind direction data and the corresponding actual power generation power data to establish a wind power output prediction model. Calculate the mean and standard deviation of the current wind speed and wind direction and the historical data, and determine whether it exceeds the normal range (such as 3 times the standard deviation). If it exceeds, a warning is triggered. The predicted wind power output value is corrected accordingly according to the warning level, such as reducing the weight coefficient of the predicted value. The corrected predicted wind power output value and the preliminary estimated value are weighted and averaged, and the weights can be determined according to the confidence levels of the two to obtain the final predicted wind power output result. Use the cross-validation method to evaluate the wind power output prediction model, calculate evaluation indicators such as the root mean square error (RMSE), and continuously optimize the prediction model by adjusting parameters such as the kernel function type and penalty coefficient of the SVM. Apply the optimized wind power output prediction model to the actual wind farm, combine the real-time monitoring data of the wind turbine (such as wind speed, wind direction, generator speed, etc.), and dynamically adjust the prediction result to provide a decision-making basis for the power generation plan formulation and grid connection scheduling of the wind farm.

[0044] Exemplarily, wind power output prediction is crucial for the effective operation of a wind farm. To predict the wind farm's output more accurately, a more precise prediction model can be established based on the three-dimensional model of the wind farm, combining computational fluid dynamics (CFD) methods and machine learning algorithms. First, using the three-dimensional model of the wind farm, the CFD method is employed to simulate the wind speed and wind direction distribution in the wind farm. For example, professional CFD software such as ANSYS Fluent or OpenFOAM can be used to model and simulate the wind farm, inputting parameters such as terrain data and surface roughness, and simulating the wind speed and wind direction at various positions in the wind farm under different wind speeds and wind directions. Suppose the simulation results show that at a certain moment, the wind speed at the location of wind turbine A is 8 m / s and the wind direction is due south. Then, the obtained wind speed data is input into the power curve of the wind turbine to calculate the theoretical power generation of the wind turbine at this wind speed. Each wind turbine has its unique power curve, which describes the relationship between wind speed and power generation. For example, the power curve of wind turbine A shows that when the wind speed is 8 m / s, its theoretical power generation is 5 MW. This value can be used as a preliminary estimate of the potential wind power output. To improve the prediction accuracy, machine learning algorithms such as support vector machine (SVM) can be used to train historical wind speed, wind direction data, and corresponding actual power generation data to establish a wind power output prediction model. For example, collect wind speed, wind direction, and actual power generation data for the past year, input these data into the SVM model for training, and learn the complex non-linear relationship between wind speed, wind direction, and actual power generation. When making a prediction, calculate the mean and standard deviation of the current wind speed and wind direction from the historical data to determine whether it exceeds the normal range. For example, calculate the mean wind speed for the past year to be 6 m / s and the standard deviation to be 2 m / s. If the current wind speed is 12 m / s, exceeding the mean plus 3 times the standard deviation, a warning is triggered. The warning level can be set according to the multiple of the excess. For example, exceeding 3 times the standard deviation is a level 1 warning, and exceeding 4 times the standard deviation is a level 2 warning. The wind power output prediction value is corrected accordingly according to the warning level. For example, if a level 1 warning is triggered, the weight coefficient of the prediction value is reduced. For example, the weight coefficient is reduced from 8 to 6. This is because when the wind speed exceeds the normal range, the actual power generation of the wind turbine may be affected by other factors. For example, too high a wind speed may cause the wind turbine to operate at a limited power. The output value predicted by the SVM model and the preliminary estimate calculated according to the power curve are weighted and averaged to obtain the final wind power output prediction result. For example, the output value predicted by the SVM model is 4 MW, and the preliminary estimate is 5 MW. Suppose the weight coefficients of the two are 6 and 4 respectively, then the final prediction result is (4×6)+(5×4) = 44 MW. The weight coefficient can be determined according to the confidence levels of the two. For example, if historical data shows that the prediction accuracy of the SVM model is higher, a higher weight is given to it. To evaluate and optimize the prediction model, cross-validation methods are used to calculate evaluation indicators such as root mean square error (RMSE).By adjusting parameters such as the kernel function type and penalty coefficient of the SVM, the prediction model is continuously optimized. For example, different kernel functions can be tried, such as linear kernel, polynomial kernel, and radial basis kernel function, and the value of the penalty coefficient C can be adjusted. Compare the RMSE under different parameter combinations and select the parameter combination with the smallest RMSE. Finally, apply the optimized wind power output prediction model to the actual wind farm. Combining the real-time monitoring data of the wind turbines, such as wind speed, wind direction, and generator speed, dynamically adjust the prediction results to provide a decision-making basis for the power generation plan formulation and grid connection scheduling of the wind farm. For example, if the real-time monitoring data shows that a certain wind turbine fails, the output prediction value of this wind turbine needs to be adjusted according to the actual situation.

[0045] Step S109: Based on the above potential output prediction results of photovoltaic and wind power generation, use an intelligent optimization algorithm to solve the energy storage capacity configuration and operation strategy. Under the conditions of meeting the new energy consumption and system stability constraints, maximize the benefits of the energy storage system to obtain the dynamic prediction value of the energy storage consumption capacity index. The energy storage consumption capacity index is defined as the ratio of the new energy power generation that can be consumed by the energy storage system per unit time to the total power generation, and is used to quantify the contribution of the energy storage system to the new energy consumption.

[0046] Based on the historical data of photovoltaic power generation and wind power generation and weather forecast data, first, data cleaning and preprocessing are carried out to eliminate outliers and missing values, and feature selection and feature engineering are performed to construct a suitable feature set. Then, the support vector machine algorithm is used to establish a potential output prediction model for photovoltaic and wind power generation. The hyperparameters of the model are optimized through methods such as cross-validation and grid search to improve the prediction accuracy of the model, and the potential output prediction values for a future period are obtained. According to the potential output prediction values and the grid load demand, the particle swarm optimization algorithm is used to solve the optimal energy storage capacity configuration and operation strategy under the conditions of considering the energy storage capacity constraint, new energy consumption constraint, and system stability constraint. The objective function is to maximize the benefit of the energy storage system. The particle swarm algorithm searches for the optimal solution through iterative optimization to obtain the energy storage capacity configuration and operation strategy. According to the energy storage capacity configuration and operation strategy obtained from the optimization solution, the charge and discharge power of the energy storage system at different times are obtained through simulation calculations, and then the new energy power generation that can be consumed by the energy storage system per unit time is obtained. The total power generation data of photovoltaic and wind power generation in the same period are obtained from the grid dispatching system, and the ratio of the new energy power generation that can be consumed by the energy storage system to the total power generation is calculated to obtain the energy storage consumption capacity index. The long short-term memory network model is used to dynamically predict the energy storage consumption capacity index. The model is trained with historical data and updated and predicted online in the form of a sliding window to obtain the prediction values for a future period. According to the preset threshold, it is judged whether the prediction value of the energy storage consumption capacity index exceeds the threshold. If it exceeds, it indicates that the energy storage system makes a greater contribution to the consumption of new energy, otherwise it indicates a smaller contribution. According to the judgment result, the model predictive control method is used to dynamically optimize and adjust the energy storage capacity configuration and operation strategy. Through the rolling optimization method, the optimization objectives and constraint conditions are updated in real time to generate a new optimal control strategy, which is applied to the actual operation control of the energy storage system to adapt to the changes in new energy output and load demand, continuously improve the contribution of the energy storage system to the consumption of new energy, and ensure the safe and stable operation of the power grid.

[0047] Exemplarily, data cleaning and preprocessing are the basis of the prediction model. Taking wind power generation data as an example, the original data may include information such as wind speed, wind direction, temperature, humidity, air pressure, etc., as well as the corresponding power generation. First, outliers need to be removed, for example, the instantaneous wind speed suddenly becomes zero or negative due to sensor failures. At the same time, missing values are processed. For example, data missing due to data acquisition system failures during certain periods can be filled using interpolation or mean methods. Feature selection and feature engineering are to construct a more effective feature set. For example, wind speed and wind direction have a greater impact on wind power generation, and these two features can be selected as the main features. In addition, new features can be constructed, such as the daily variation range of wind speed and the stability of wind direction, to improve the prediction accuracy of the model. The support vector machine algorithm can be used to establish a potential output prediction model for photovoltaic and wind power generation. Taking photovoltaic power generation as an example, historical light intensity, temperature and other data can be used as input features, and the actual power generation can be used as the output target to train a support vector machine regression model. Through methods such as cross-validation and grid search, the hyperparameters of the model, such as the kernel function type and penalty coefficient, can be optimized to improve the prediction accuracy of the model. Suppose the predicted potential photovoltaic power generation for the next week is 100 MW, 120 MW, 110 MW, 90 MW, 80 MW, 100 MW, and 130 MW respectively. The particle swarm optimization algorithm can be used to solve the optimal energy storage capacity configuration and operation strategy. Taking a scenario including a wind farm and an energy storage system as an example, the objective function can be set to maximize the benefit of the energy storage system, that is, to maximize the new energy consumption and reduce the curtailment of wind / photovoltaic power. The constraints can include energy storage capacity constraints, such as the maximum charging and discharging rates of the energy storage system, and system stability constraints, such as the grid frequency fluctuation range. Suppose through the particle swarm optimization algorithm, the optimal energy storage capacity obtained is 50 MWh, and the best operation strategy is to store energy in the energy storage system when the wind power generation is sufficient during the day and release the energy to the grid during the peak electricity consumption at night. The energy storage consumption capacity index is an important indicator to evaluate the contribution of the energy storage system to new energy consumption. Suppose the total wind power generation during a certain period is 200 MW, and the wind power generation that the energy storage system can consume is 80 MW, then the energy storage consumption capacity index is 40%. Through the long short-term memory network model, the energy storage consumption capacity index can be dynamically predicted. Suppose the predicted energy storage consumption capacity indexes for the next week are 30%, 40%, 35%, 25%, 20%, 30%, and 45% respectively. The model predictive control method can be used to dynamically optimize the energy storage capacity configuration and operation strategy. Suppose the preset energy storage consumption capacity index threshold is 30%. According to the prediction results, the energy storage consumption capacity indexes on the first, second, third, and last days of the next week all exceed the threshold, indicating that the energy storage system makes a greater contribution to new energy consumption. While the energy storage consumption capacity indexes on the fourth and fifth days are lower than the threshold, indicating that the energy storage system makes a smaller contribution to new energy consumption.At this time, a model predictive control method can be adopted to adjust the operation strategy of the energy storage system, such as increasing the charging rate of the energy storage system or adjusting the dispatching strategy of the power grid to increase the consumption of new energy.

[0048] Step S1010: Use an adaptive weighted fusion strategy to combine the prediction results of multiple models, dynamically adjust the weight coefficients of each sub-model according to the prediction error, and improve the overall accuracy and robustness of the prediction of the energy storage consumption capacity index. The prediction results of multiple models include the prediction results of a statistical model based on historical data, a dynamic model based on physical mechanisms, and a data-driven model based on machine learning.

[0049] Based on the historical data of the energy storage consumption capacity index, establish an ARIMA time series model, and predict the change trend of the future energy storage consumption capacity index by fitting the historical data. Use the laws of physics and the operating mechanism of the power system to establish a differential equation system describing the operating state of the energy storage system, solve the equation system, and obtain the predicted value of the energy storage consumption capacity index by the physical mechanism model. Based on the key factors affecting the energy storage consumption capacity index, such as power load, new energy generation power, etc., construct a BP neural network model, and establish a non-linear mapping relationship between the key factors and the energy storage consumption capacity index through training the network to achieve data-driven prediction. Quantitatively evaluate the prediction errors of each sub-model using the root mean square error and the mean absolute percentage error respectively. According to the magnitude of the prediction error, use the weighted average method to determine the initial weight coefficients of each sub-model. The smaller the error, the larger the weight coefficient. In a new time period, re-obtain the latest predicted values of the ARIMA model, the physical mechanism model, and the BP neural network model for the energy storage consumption capacity index. Calculate the root mean square error and the mean absolute percentage error of each sub-model in the latest period of time. According to the change of the prediction error, adopt an adaptive weighted fusion strategy to dynamically adjust the weight coefficients of each model. Specifically, for the model with a decreasing prediction error, its weight coefficient increases proportionally; for the model with an increasing prediction error, its weight coefficient decreases proportionally. Use the adjusted weight coefficients to perform weighted fusion on the prediction results of the ARIMA model, the physical mechanism model, and the BP neural network model to obtain the final predicted value of the energy storage consumption capacity index. Calculate the prediction error in the current time period and determine whether it exceeds the preset threshold. If it exceeds the threshold, it is considered that the current weight coefficients are no longer applicable and need to be re-adjusted, and return to step 6; if it does not exceed the threshold, it is considered that the current weight coefficients are still valid, maintain the existing weight coefficients, return to step 5, and continue the prediction in the next time period. Repeat steps 5-8 to achieve the adaptive dynamic prediction of the energy storage consumption capacity index. Through continuous error evaluation and weight adjustment, continuously improve the prediction accuracy and enhance the adaptability of the prediction model to the change trend of the energy storage consumption capacity index.

[0050] Exemplarily, the ARIMA model is a commonly used time series prediction method. It analyzes the autocorrelation and partial autocorrelation of time series data to establish a mathematical model to describe the variation law of the data. For example, it is possible to collect the energy storage consumption capacity index data for each day in the past year, use the ARIMA model to fit these historical data, and predict the energy storage consumption capacity index for the next week. Suppose the selected ARIMA model according to the historical data is ARIMA(1, 1, 1), which means that the model contains an autoregressive term, a first-order differencing term, and a moving average term. By fitting the historical data, the parameters of the model can be obtained, and these parameters can be used to predict the energy storage consumption capacity index for the next week. The physical mechanism model is established based on the physical laws of the energy storage system operation and the operation mechanism of the power system. It usually uses a system of differential equations to describe the charging and discharging process, energy conversion process of the energy storage system, and its interaction with the power grid. For example, a power system model including an energy storage system, a photovoltaic power station, and a wind power station can be established. This model can consider the volatility of photovoltaic and wind power generation, the variation of power load, and the charging and discharging characteristics of the energy storage system. By solving the system of differential equations, the operating state of the energy storage system can be obtained, and then the energy storage consumption capacity index can be calculated. Suppose in a specific scenario, the photovoltaic power generation is 100 MW, the wind power generation is 80 MW, the power load is 150 MW, the energy storage system capacity is 50 MWh, and the initial state is 25 MWh. By solving the system of differential equations, the state of the energy storage system after one hour can be obtained. For example, the remaining power of the energy storage system is 35 MWh. Based on this information, the energy storage consumption capacity index for this hour can be calculated. The BP neural network model is a data-driven method that can establish a nonlinear mapping relationship between input variables and output variables. In this example, power load, new energy power generation, energy storage system capacity, etc. can be selected as input variables, and the energy storage consumption capacity index can be used as the output variable. Historical data, such as data for each day in the past year, is collected for training the BP neural network model. Through training, the network can learn the complex relationship between input variables and output variables. For example, when the power load is high and the new energy power generation is also high, the energy storage consumption capacity index may be high; while when the power load is low and the new energy power generation is also low, the energy storage consumption capacity index may be low. The trained BP neural network model can be used to predict the future energy storage consumption capacity index. Root mean square error and mean absolute percentage error are commonly used prediction error evaluation metrics. The root mean square error reflects the deviation degree between the predicted value and the actual value, while the mean absolute percentage error reflects the relative deviation degree between the predicted value and the actual value. For example, the ARIMA model predicts the energy storage consumption capacity index for the next week to be 30%, 40%, 35%, 25%, 20%, 30%, 45% respectively. Suppose the actual values are 32%, 38%, 36%, 27%, 22%, 28%, 42%.The root mean square error and mean absolute percentage error of the ARIMA model can be calculated separately and compared with the errors of other models. According to the magnitude of the prediction errors, the initial weight coefficients of each sub-model can be determined. The smaller the error, the larger the weight coefficient. For example, assuming that the ARIMA model has the smallest root mean square error, it is given the largest weight coefficient; the BP neural network model has the largest root mean square error, so it is given the smallest weight coefficient. In the new time period, the latest prediction values of each sub-model are re-obtained, and the new prediction errors are calculated. According to the changes in the prediction errors, the weight coefficients of each model are dynamically adjusted. For example, if the prediction error of the ARIMA model decreases, its weight coefficient is increased; if the prediction error of the BP neural network model increases, its weight coefficient is decreased. Using the adjusted weight coefficients, the prediction results of each sub-model are weighted and fused to obtain the final prediction value of the energy storage consumption capacity index. For example, assuming that the prediction value of the ARIMA model is 35%, the weight coefficient is 5; the prediction value of the physical mechanism model is 40%, the weight coefficient is 3; the prediction value of the BP neural network model is 30%, and the weight coefficient is 2. Then the final prediction value is 35% * 5 + 40% * 3 + 30% * 2 = 35%. By continuous error evaluation and weight adjustment, the prediction accuracy can be continuously improved, and the adaptability of the prediction model to the changing trend of the energy storage consumption capacity index can be enhanced. This can effectively utilize the advantages of different models, improve the accuracy and reliability of the overall prediction, and provide a more reliable decision-making basis for the optimal operation of the energy storage system.

[0051] As described above, it is only the specific implementation manner of this specification. Those skilled in the art can clearly understand that for the convenience and simplicity of description, the specific working processes of the systems, modules, and units described above can refer to the corresponding processes in the foregoing method embodiments and will not be elaborated here. It should be understood that the protection scope of this specification is not limited thereto. Any person skilled in the art within the technical scope disclosed in this specification can easily think of various equivalent modifications or substitutions, and these modifications or substitutions should be covered within the protection scope of this specification.

Claims

1. A method for predicting the potential output of new energy power generation, characterized in that, Including: Obtain historical meteorological data and geographical information data, and based on the historical meteorological data and geographical information data, establish a multiple regression model of key factors and weight coefficients affecting the potential output of new energy power generation. Use the Monte Carlo random sampling method to generate a large number of input samples, substitute the input samples into the new energy power generation potential output calculation model to obtain the probability distribution of the potential output, perform wavelet analysis on the probability distribution to obtain the output fluctuation characteristics at different time scales, and conduct abnormal fluctuation early warning according to the output fluctuation characteristics; For photovoltaic power generation, establish a non-linear mapping relationship between solar radiation intensity, photovoltaic module conversion efficiency, temperature coefficient, aging attenuation rate factors and power generation efficiency to dynamically correct the predicted value of the potential output of photovoltaic power generation; For wind power generation, use the computational fluid dynamics model to simulate the wind field distribution, obtain the wind speed and wind direction parameters at the location of the wind turbine, determine the potential output of wind power in combination with the power curve of the wind turbine, and correct according to the abnormal fluctuation early warning result; Based on the predicted results of the potential output of photovoltaic and wind power generation, solve the energy storage capacity configuration and operation strategy through an intelligent optimization algorithm to maximize the benefits of the energy storage system, and obtain the dynamic predicted value of the energy storage consumption capacity index; And adopt an adaptive weighted fusion strategy to combine the prediction results of the statistical model based on historical data, the dynamic model based on physical mechanism, and the data-driven model based on machine learning to predict the energy storage consumption capacity index. The statistical model based on historical data refers to establishing an ARIMA time series model according to the historical data of the energy storage consumption capacity index, and predicting the change trend of the future energy storage consumption capacity index by fitting the historical data. The dynamic model based on physical mechanism refers to using physical laws and the operation mechanism of the power system to establish a differential equation group describing the operation state of the energy storage system, and solving the equation group to obtain the predicted value of the physical mechanism model of the energy storage consumption capacity index. The data-driven model based on machine learning refers to constructing a BP neural network model based on the key factors affecting the energy storage consumption capacity index, and establishing a non-linear mapping relationship between the key factors and the energy storage consumption capacity index through training the network to achieve data-driven prediction.

2. The new energy power generation potential output prediction method according to claim 1, wherein The generating a large number of input samples by using the Monte Carlo random sampling method includes: According to the uncertainty distribution function of the key factors, use the Monte Carlo random sampling algorithm to generate a large number of random input sample data conforming to the distribution characteristics, and preprocess the random input sample data.

3. The new energy power generation potential output prediction method according to claim 2, wherein For the preprocessed data, including: Use the chi-square test and the Kolmogorov-Smirnov test to judge whether each factor obeys the normal distribution or the Weibull distribution; If the factor obeys a known distribution type, use the maximum likelihood estimation method to estimate the corresponding distribution parameters; According to the probability density function obtained by the estimation, generate a large number of random samples obeying the distribution through the Monte Carlo simulation method for subsequent risk assessment.

4. The new energy power generation potential output prediction method according to claim 1, characterized in that The performing wavelet analysis on the probability distribution to obtain the output fluctuation characteristics at different time scales includes: Discretize the probability density function and perform wavelet packet decomposition on it as the input signal of wavelet transform to obtain wavelet coefficients at multiple time scales. Determine the abnormal fluctuation thresholds at each scale according to historical data, judge whether the wavelet coefficients exceed the thresholds to obtain abnormal fluctuations, and summarize the abnormal fluctuations to construct a feature vector.

5. The new energy power generation potential output prediction method according to claim 1, wherein Preprocess the random input sample data, including: Remove outliers and missing values, and perform normalization processing using the maximum-minimum normalization method to map the data into the interval [0, 1] to obtain the preprocessed input sample data; According to the pre-established physical mechanism model of the potential power output of new energy power generation, use the preprocessed input sample data as training data, and minimize the mean square error loss function through the gradient descent algorithm and the backpropagation algorithm to optimize the parameters of the physical mechanism model and obtain the trained physical mechanism model; Substitute the preprocessed input sample data into the trained physical mechanism model and the data-driven model respectively, calculate the potential power output values corresponding to each input sample, and perform weighted averaging on the results predicted by the two models to obtain the final potential power output prediction value.

6. The new energy power generation potential output prediction method according to claim 1, wherein For the photovoltaic power generation, establishing a non-linear mapping relationship between factors such as solar radiation intensity, photovoltaic module conversion efficiency, temperature coefficient, and aging attenuation rate and power generation efficiency to dynamically correct the predicted value of the potential power output of photovoltaic power generation, including: Use a deep learning algorithm to establish a non-linear mapping relationship between the factors and power generation efficiency, and correct the predicted value according to the abnormal fluctuation warning result.

7. The method for predicting the potential output of new energy power generation according to claim 6, characterized in that, Obtain the probability density function of the data, and use the histogram method to discretize the probability density function to obtain the discretized probability density function, including: Use the discretized probability density function as the input signal of wavelet transform, and perform multi-scale decomposition using Daubechies wavelet to obtain wavelet coefficients at multiple time scales; Determine the abnormal fluctuation threshold of the wavelet coefficients at each time scale according to the preset historical data; If the wavelet coefficients at the current moment exceed the abnormal fluctuation threshold, it is judged that there is an abnormal fluctuation at this moment; Extract the statistical features of the abnormal fluctuations and construct an abnormal fluctuation feature vector; Use the random forest algorithm to classify the abnormal fluctuation feature vector to obtain an anomaly detection model; If the abnormal fluctuation feature vector of the newly input data is judged to be abnormal by the anomaly detection model, an alarm is triggered.

8. The new energy power generation potential output prediction method according to claim 1, wherein For wind power generation, use a computational fluid dynamics model to simulate the wind field distribution, obtain the wind speed and wind direction parameters at the location of the wind turbine, determine the potential wind power output in combination with the power curve of the wind turbine, and make corrections according to the abnormal fluctuation warning result, including: According to the three-dimensional model of the wind field, use the computational fluid dynamics method to simulate the speed and direction distribution of the wind field, obtain the wind speed and wind direction data at the location of the wind turbine, input the wind speed data into the power curve of the wind turbine to calculate the theoretical power generation power of the wind turbine at this wind speed, and make corrections according to the abnormal fluctuation warning result.

9. The new energy power generation potential output prediction method according to claim 1, wherein Based on the predicted results of photovoltaic and wind power generation, an intelligent optimization algorithm is used to solve the energy storage capacity configuration and operation strategy to maximize the benefits of the energy storage system, and a dynamic predicted value of the energy storage consumption capacity index is obtained, including: According to the predicted results of the photovoltaic and wind power generation, an intelligent optimization algorithm is adopted to maximize the benefits of the energy storage system under the conditions of meeting the constraints of new energy consumption and system stability, and a dynamic predicted value of the energy storage consumption capacity index is obtained.

10. The method for predicting the potential power output of new energy power generation according to claim 1, characterized in that, The adaptive weighted fusion strategy is adopted to combine the prediction results of the statistical model based on historical data, the dynamic model based on physical mechanism, and the data-driven model based on machine learning to predict the energy storage consumption capacity index, including: The weight coefficients of each sub-model are dynamically adjusted according to the prediction error to improve the overall accuracy and robustness of the energy storage consumption capacity index prediction.

Images