A probabilistic power flow calculation method and system based on prediction analysis and interval sampling

Abstract

The application relates to the technical field of power system flow calculation, and particularly discloses a probability power flow calculation method and system based on prediction analysis and interval sampling. The method comprises the following steps: obtaining historical data, wherein the historical data comprises wind speed historical data, solar radiation historical data and distribution network load historical data; obtaining a prediction data series based on the historical data, wherein the prediction data series comprises wind speed prediction data, solar radiation prediction data and distribution network load prediction data; setting a sampling interval for the wind speed prediction data, the solar radiation prediction data and the distribution network load prediction data, and obtaining the probability distribution relationship between wind speed and solar radiation and between wind speed and distribution network load; sampling the wind speed interval, and respectively calculating the power output of a wind power generation system and a solar power generation system; reading in node load data, substituting the power output of the wind power generation system and the solar power generation system, and calculating power flow by using a forward-backward sweep algorithm.

Description

Technical Field

[0001] This invention relates to the field of power flow calculation technology, and in particular to a probabilistic power flow calculation method and system based on predictive analysis and interval sampling. Background Technology

[0002] Electricity load data exhibits significant randomness and periodicity in power systems. Simultaneously, numerous factors, such as public electricity consumption habits, socio-economic activities, weather conditions, and policy guidance, profoundly influence load data. Therefore, a comprehensive and in-depth analysis of the characteristics and related influencing factors of electricity load can more clearly reveal the intrinsic connections between load data and various influencing factors, laying a solid foundation for building accurate load forecasting models. The foundation of electricity load forecasting technology lies in utilizing past load data from user points or individual components, historical and future meteorological information, while combining user living and production habits with other relevant factors such as the macroeconomic environment.

[0003] Distribution networks containing intermittent distributed power supply systems such as wind or solar power will be affected by a double stochastic process: on the one hand, the output of the power generation system is often affected by natural weather conditions and has inherent volatility; on the other hand, the load of the distribution network itself has the characteristics of random volatility. Currently, the methods for studying power flow in distribution networks connected to distributed power supply systems do not consider the characteristics of random volatility in the output of intermittent energy sources.

[0004] Existing technologies also employ the Monte Carlo simulation method to analyze probabilistic power flow in distribution networks containing distributed generation. The Monte Carlo simulation method relies on repeated random sampling to arrive at the final results, but it requires significant computation time for complex systems. Summary of the Invention

[0005] The technical problem this invention aims to solve is to consider the impact of the random fluctuations in intermittent energy output and the random fluctuations in the distribution network load itself on the double-stochastic process of distribution network power flow analysis. It proposes a probabilistic power flow solution method adapted to intermittent distributed power sources using Monte Carlo sampling simulation. The probabilistic power flow analysis results are obtained by analyzing the relationship between the random output fluctuations and the probability distributions of voltage at each node and the power flow distributions of each branch. Power flow tracing analysis is then performed on the distributed power supply system. To address the issue of the long computation time required by the Monte Carlo algorithm for complex systems, data preprocessing is performed to achieve interval sampling optimization.

[0006] To address the aforementioned technical problems, one technical solution adopted by this invention is: a probabilistic power flow calculation method based on predictive analysis and interval sampling, comprising:

[0007] Acquire historical data, including historical wind speed data, historical solar radiation data, and historical power distribution network load data;

[0008] A series of forecast data is obtained by making predictions based on historical data. The series of forecast data includes: wind speed forecast data, solar radiation forecast data, and power distribution network load forecast data.

[0009] Sampling intervals were set for wind speed forecast data, solar radiation forecast data, and power distribution network load forecast data, and the probability distribution relationships between wind speed and solar radiation, as well as between wind speed and power distribution network load, were obtained.

[0010] By sampling the wind speed range, the power output of the wind power generation system and the solar power generation system were calculated respectively.

[0011] Read in the load data of each node, substitute the power output of the wind power generation system and the solar power generation system, and use the forward and backward algorithm to calculate the power flow.

[0012] In a preferred embodiment, the step of obtaining a prediction data series based on historical data includes: inputting historical data into a trained LSTM neural network prediction model to obtain a prediction data sequence.

[0013] In a preferred embodiment, it further includes the step of training an LSTM neural network prediction model, specifically including:

[0014] Use the first 80% of historical data as the training set, 10% as the validation set, and 10% as the test set, and convert them into NumPy arrays.

[0015] Build an LSTM neural network model using Keras;

[0016] The training data is input into the LSTM neural network model for training, resulting in a trained LSTM neural network prediction model.

[0017] In a preferred embodiment, building an LSTM neural network model using Keras includes:

[0018] Create a Sequential model instance and add two LSTM layers with 64 neurons each. Set the return_sequences parameter to True for the first LSTM layer and set the return_sequences parameter to False for the second LSTM layer.

[0019] Add a Dense layer with 10 neurons and an output layer;

[0020] The neural network optimizer is set to Adam optimizer with a learning rate of 0.01, and mean squared error is used as the loss function for model compilation.

[0021] In a preferred embodiment, the step of setting sampling intervals for wind speed prediction data, solar radiation prediction data, and power grid load prediction data, and obtaining the probability distribution relationships between wind speed and solar radiation, and between wind speed and power grid load, includes the following steps before proceeding:

[0022] Using the chi-square test and correlation coefficient, the correlation between wind speed and solar radiation, as well as between wind speed and power distribution network load, is analyzed. If a correlation is found, the steps described above are performed to set sampling intervals for wind speed prediction data, solar radiation prediction data, and power distribution network load prediction data, and to obtain the probability distribution relationship between wind speed and solar radiation, as well as between wind speed and power distribution network load.

[0023] In a preferred embodiment, the step of analyzing the correlation between wind speed and solar radiation and between wind speed and power grid load using the chi-square test and correlation coefficient includes:

[0024] The chi-square independence test is used to test the pairwise independence between wind speed, solar radiation, and user load. The test statistic is:

[0025] (4),

[0026] In equation (4), N is the total amount of data. To determine the number of rows in the split table, For the number of columns, For the first in the partition table Line number The element of the column;

[0027] For pairwise random variables X and Y, the correlation coefficient is calculated as follows:

[0028] (5),

[0029] In equation (5), ,and , , and These correspond to the expected value and standard deviation of random variables X and Y, respectively;

[0030] For two sets of data, {xi} and {yi}, the formula for calculating the correlation coefficient between them is:

[0031] (6).

[0032] In a preferred embodiment, the step of setting sampling intervals for wind speed prediction data, solar radiation prediction data, and power grid load prediction data, and obtaining the probability distribution relationships between wind speed and solar radiation, and between wind speed and power grid load, includes:

[0033] The sampling interval for wind speed prediction data is set: the starting wind speed is determined based on the characteristic curve of wind turbine operation. and cut-off wind speed Wind speed range Divide into several intervals, and at the same time set the wind speed and The wind speeds were grouped into m intervals, denoted as a single level range. ;

[0034] Define the sampling interval for solar radiation prediction data: define the radiation range of the solar radiation prediction data. Divide into k thousand intervals, denoted as ;

[0035] To define sampling intervals for distribution network load forecast data: Based on K-means clustering of the distribution network load forecast data, each load type is further divided into intervals, categorized by maximum and minimum load. Divide into n level intervals, denoted as ;

[0036] The probability distribution relationships between solar radiation and wind speed, and between wind speed and power grid load, were statistically analyzed:

[0037] (7),

[0038] (8),

[0039] Further obtain the load value in the wind speed value The probability distribution under the given conditions is:

[0040] (9),

[0041] The relationship between the probability of solar radiation distribution and wind speed was obtained:

[0042] (10)

[0043] (11).

[0044] In a preferred embodiment, the step of sampling the wind speed range and calculating the power output of the wind power generation system and the solar power generation system respectively includes:

[0045] Wind speed samples are obtained by sampling within the wind speed range;

[0046] The output power of the wind turbine can be calculated using the following formula:

[0047] (1),

[0048] In equation (1), P is the power generation capacity, Prate is the rated power, Vrate is the rated wind speed, Vci is the maximum wind speed at which the wind turbine starts, Vco is the maximum wind speed at which the wind turbine starts, and V is the current wind speed.

[0049] The reactive power output of a wind turbine generator set is calculated using the following formula:

[0050] (2),

[0051] In formula (2): Q is the power factor angle. WTG P WTG These are the reactive and active power of the wind turbine;

[0052] The formula for calculating the output energy of photovoltaic cells is as follows, based on the sampling interval set for solar radiation prediction data:

[0053] (3),

[0054] In equation (3), This refers to the rated capacity of the photovoltaic array; This is the reduction factor; This represents the current actual environmental radiation intensity. is the unit radiation intensity under standard test conditions; is the temperature coefficient of photovoltaic power output; This refers to the current temperature of the photovoltaic array. The temperature is the standard test environment temperature for photovoltaic arrays.

[0055] In a preferred embodiment, it further includes: outputting results when the power flow test meets the convergence condition, and performing probabilistic statistical analysis on the results.

[0056] To address the aforementioned technical problems, another technical solution adopted by this invention is: a probabilistic power flow calculation system based on predictive analysis and interval sampling, configured to implement any of the methods described in the preceding technical solution, the system comprising:

[0057] The acquisition module is used to acquire historical data, including historical wind speed data, historical solar radiation data, and historical power distribution network load data.

[0058] The prediction module is used to make predictions based on historical data to obtain a series of prediction data, which includes: wind speed prediction data, solar radiation prediction data, and power distribution network load prediction data.

[0059] The probability distribution relationship calculation module is used to set the sampling interval for wind speed prediction data, solar radiation prediction data and power distribution network load prediction data, and obtain the probability distribution relationship between wind speed and solar radiation and between wind speed and power distribution network load.

[0060] The power output calculation module is used to sample the wind speed range and calculate the power output of the wind power generation system and the solar power generation system respectively.

[0061] The power flow calculation module is used to read in the load data of each node, substitute the power output of the wind power generation system and the solar power generation system, and calculate the power flow using the forward and backward algorithm.

[0062] The present invention has the following beneficial effects:

[0063] This invention uses an LSTM neural network to build a load forecasting model for predictive analysis. Addressing the randomness of distributed energy resources and loads, this invention optimizes Monte Carlo sampling at the data level by leveraging their interrelationships. Based on the predictive analysis of the data, and considering the randomness and continuous temporal variation of solar radiation, wind energy, and distribution network load, the probabilistic relationships between solar radiation and wind speed, and between wind speed and distribution network load, are statistically derived. Finally, based on the probability distribution relationships between wind speed and solar radiation, and between wind speed and distribution network load, the corresponding solar radiation and load values ​​can be obtained, significantly reducing the number of sampling attempts and computation time. Attached Figure Description

[0064] Figure 1 A flowchart of a probability power flow calculation method based on predictive analysis and interval sampling is provided for an embodiment of the present invention.

[0065] Figure 2 The present invention provides an overall flowchart of a probabilistic power flow calculation method based on predictive analysis and interval sampling.

[0066] Figure 3 The flowchart of the LSTM neural network training model provided in the embodiment of the present invention is shown. Detailed Implementation

[0067] The detailed description and technical content of the present invention are explained below with reference to the accompanying drawings. However, the drawings are provided for reference and illustration only and are not intended to limit the present invention.

[0068] In related technologies, probabilistic power flow modeling and solution methods are divided into two types: analytical methods and simulation methods. Analytical models assume that the load is a normally distributed random variable and that the node-injected power is independent or linearly correlated. Therefore, some scholars have proposed that the probability distribution of the power flow be obtained using the convolution formula from probability theory. Although analytical methods are faster than simulation methods, they suffer from significant errors because the convolution calculations required linearizing the power flow equations near the assumed load points. Furthermore, this method is not adept at handling the correlation of node-injected power.

[0069] This invention employs the Monte Carlo method to analyze probabilistic power flow in distribution networks containing distributed generation. The Monte Carlo method is based on repeated random sampling to statistically analyze the final results. Depending on the random sampling principle, it can be divided into sequential Monte Carlo simulation and non-sequential Monte Carlo simulation. Sequential Monte Carlo simulation can realistically simulate the time-varying conditions of the system operation and reproduce the actual system operation; however, this algorithm is time-consuming, especially for complex systems, often requiring long sampling times. Therefore, this invention, based on the analysis of wind speed, solar power output, and the load fluctuation characteristics of the distribution network itself, studies the digital characteristics and correlations of these random quantities, combines methods for solving typical probabilistic power flow problems, and optimizes the Monte Carlo sampling simulation method to propose a probabilistic power flow solution method suitable for intermittent distributed generation.

[0070] Figure 1 This is a schematic diagram of a probabilistic power flow calculation method based on predictive analysis and interval sampling, provided by an embodiment of the present invention. Figure 2 A flowchart illustrating the overall process of a probabilistic power flow calculation method based on predictive analysis and interval sampling, provided in this embodiment of the invention; Figure 1 , Figure 2 As shown in the figure, the probabilistic power flow calculation method based on predictive analysis and interval sampling provided by this invention includes:

[0071] Step S1: Obtain historical data, including historical wind speed data, historical solar radiation data, and historical power distribution network load data;

[0072] Step S2: Based on historical data, a series of prediction data is obtained, including: wind speed prediction data v(t), solar radiation prediction data r(t), and power distribution network load prediction data l(t).

[0073] Step S3: Set the sampling interval for wind speed prediction data, solar radiation prediction data and power distribution network load prediction data, and obtain the probability distribution relationship between wind speed and solar radiation and between wind speed and power distribution network load.

[0074] Step S4: Sample the wind speed range and calculate the power output of the wind power generation system and the solar power generation system respectively.

[0075] Step S5: Read in the load data of each node, substitute the power output of the wind power generation system and the solar power generation system, and use the forward push-back algorithm to calculate the power flow.

[0076] This invention only requires sampling within a wind speed range, and then, based on the probability distribution relationship between wind speed and solar radiation, as well as between wind speed and power grid load, it can obtain the corresponding power output of wind power generation systems and solar power generation systems, greatly reducing the number of sampling attempts and calculation time.

[0077] In some embodiments, the step of obtaining a series of predicted data based on historical data includes: inputting historical data into a trained LSTM neural network prediction model to obtain a sequence of predicted data.

[0078] This invention uses an LSTM neural network to build a prediction model and implements data preprocessing to achieve interval sampling optimization.

[0079] In some embodiments, the method further includes: a step of training an LSTM neural network prediction model, such as... Figure 3 As shown, it specifically includes:

[0080] Use the first 80% of historical data as the training set, 10% as the validation set, and 10% as the test set, and convert them into NumPy arrays.

[0081] Build an LSTM neural network model using Keras;

[0082] The training set data is input into the LSTM neural network model for training, resulting in a trained LSTM neural network prediction model. During training, 512 data points are fed in each batch for a total of 20 epochs, while the loss value of the validation set is tested simultaneously. After training, the code saves the trained model in HDF5 format and then converts it to TensorFlow.js format for storage.

[0083] In some embodiments, building an LSTM neural network model using Keras includes:

[0084] Create a Sequential model instance and add two LSTM layers with 64 neurons each. Set the return_sequences parameter to True for the first LSTM layer and set the return_sequences parameter to False for the second LSTM layer.

[0085] Add a Dense layer with 10 neurons and an output layer;

[0086] The neural network optimizer is set to Adam optimizer with a learning rate of 0.01, and mean squared error is used as the loss function for model compilation.

[0087] In some embodiments, step S3, which involves setting sampling intervals for wind speed prediction data, solar radiation prediction data, and power grid load prediction data, and obtaining the probability distribution relationships between wind speed and solar radiation, and between wind speed and power grid load, includes the following steps before:

[0088] Step S3a: Use the chi-square test and correlation coefficient to analyze the correlation between wind speed and solar radiation, and between wind speed and power distribution network load. If there is a correlation, proceed to step S3: set the sampling interval for wind speed prediction data, solar radiation prediction data, and power distribution network load prediction data, and obtain the probability distribution relationship between wind speed and solar radiation, and between wind speed and power distribution network load.

[0089] In some embodiments, the steps of analyzing the correlation between wind speed and solar radiation and between wind speed and power grid load using chi-square test and correlation coefficient include:

[0090] The chi-square independence test is used to test the pairwise independence between wind speed, solar radiation, and user load. The test statistic is:

[0091] (4),

[0092] In equation (4), N is the total amount of data. To determine the number of rows in the split table, For the number of columns, For the first in the partition table Line number The element of the column;

[0093] For pairwise random variables X and Y, the correlation coefficient is calculated as follows:

[0094] (5),

[0095] In equation (5), ,and , , and These correspond to the expected value and standard deviation of random variables X and Y, respectively;

[0096] For two sets of data, {xi} and {yi}, the formula for calculating the correlation coefficient between them is:

[0097] (6).

[0098] In some embodiments, step S3, which involves setting sampling intervals for wind speed prediction data, solar radiation prediction data, and power grid load prediction data, and obtaining the probability distribution relationships between wind speed and solar radiation, and between wind speed and power grid load, includes:

[0099] Step S31: Set the sampling interval for the wind speed prediction data: Determine the starting wind speed based on the characteristic curve of the wind turbine operation. and cut-off wind speed Wind speed range Divide into several intervals, and at the same time set the wind speed and The wind speeds were grouped into m intervals, denoted as a single level range. ;

[0100] Step S32, set the sampling interval for the solar radiation prediction data: define the radiation range of the solar radiation prediction data. Divide into k thousand intervals, denoted as ;

[0101] Step S33: Set the sampling interval for the distribution network load forecast data: Based on K-means clustering of the distribution network load forecast data, perform interval processing on each load type, dividing the load intervals according to the maximum load and minimum load. Divide into n level intervals, denoted as ;

[0102] Step S34: Calculate the probability distribution relationships between solar radiation and wind speed, and between wind speed and power grid load.

[0103] (7),

[0104] (8),

[0105] Further obtain the load value in the wind speed value The probability distribution under the given conditions is:

[0106] (9),

[0107] The relationship between the probability of solar radiation distribution and wind speed was obtained:

[0108] (10)

[0109] (11).

[0110] In some embodiments, step S4, which involves sampling the wind speed range and calculating the power output of the wind power generation system and the solar power generation system respectively, includes:

[0111] Step S41: Sampling is performed on the wind speed range to obtain wind speed samples;

[0112] Step S42: There is generally a linear relationship between the output power of a wind turbine and the wind speed, but the curves showing this relationship can have different shapes. Given the wind speed, the output power of the wind turbine can be calculated using the following formula:

[0113] (1),

[0114] In equation (1), P is the power generation capacity, Prate is the rated power, Vrate is the rated wind speed, Vci is the maximum wind speed at which the wind turbine starts, Vco is the maximum wind speed at which the wind turbine starts, and V is the current wind speed.

[0115] When a wind turbine is treated according to the PQ node, its power factor is assumed to remain constant. Therefore, the reactive power output of the wind turbine can be calculated using the following formula:

[0116] (2),

[0117] In formula (2): Q is the power factor angle. WTG P WTG These are the reactive and active power of the wind turbine;

[0118] Step S43: Set the sampling interval for the solar radiation prediction data:

[0119] The output energy of a photovoltaic cell is closely related to the amount of solar radiation received at its geographical location (latitude) and the sun's position relative to the earth. The formula for calculating the output energy of a photovoltaic cell is as follows:

[0120] (3),

[0121] In equation (3), The rated capacity of a photovoltaic array, i.e., its output power under standard test conditions; : This is the reduction coefficient; The actual radiation intensity in the current environment [kW / m2]; : Unit radiation intensity [kW / m2] under standard testing conditions; Temperature coefficient of photovoltaic power output [% / ℃]; The current temperature of the photovoltaic array [°C]; Temperature under standard testing conditions for photovoltaic arrays (typically 25℃).

[0122] In some embodiments, the method further includes: step S6, outputting the result when the power flow calculation meets the convergence condition, and performing probabilistic statistical analysis on the result.

[0123] Based on the same inventive concept, embodiments of the present invention also provide a probabilistic power flow calculation system based on predictive analysis and interval sampling, configured to implement any of the methods described in the above embodiments, the system comprising:

[0124] The acquisition module is used to acquire historical data, including historical wind speed data, historical solar radiation data, and historical power distribution network load data.

[0125] The prediction module is used to make predictions based on historical data to obtain a series of prediction data, which includes: wind speed prediction data, solar radiation prediction data, and power distribution network load prediction data.

[0126] The probability distribution relationship calculation module is used to set the sampling interval for wind speed prediction data, solar radiation prediction data and power distribution network load prediction data, and obtain the probability distribution relationship between wind speed and solar radiation and between wind speed and power distribution network load.

[0127] The power output calculation module is used to sample the wind speed range and calculate the power output of the wind power generation system and the solar power generation system respectively.

[0128] The power flow calculation module is used to read in the load data of each node, substitute the power output of the wind power generation system and the solar power generation system, and calculate the power flow using the forward and backward algorithm.

[0129] The embodiments described above are merely preferred embodiments for fully illustrating the present invention, and the scope of protection of the present invention is not limited thereto. Equivalent substitutions or modifications made by those skilled in the art based on the present invention are all within the scope of protection of the present invention.

Claims

1. A probabilistic power flow calculation method based on predictive analysis and interval sampling, characterized in that, It includes: Acquire historical data, including historical wind speed data, historical solar radiation data, and historical power distribution network load data; A series of forecast data is obtained by making predictions based on historical data. The series of forecast data includes: wind speed forecast data, solar radiation forecast data, and power distribution network load forecast data. Sampling intervals were set for wind speed forecast data, solar radiation forecast data, and power distribution network load forecast data, and the probability distribution relationships between wind speed and solar radiation, as well as between wind speed and power distribution network load, were obtained. By sampling the wind speed range, the power output of the wind power generation system and the solar power generation system were calculated respectively. Read in the load data of each node, substitute the power output of the wind power generation system and the solar power generation system, and use the forward and backward algorithm to calculate the power flow.

2. The method according to claim 1, characterized in that, The steps for obtaining prediction data based on historical data include: inputting historical data into a trained LSTM neural network prediction model to obtain a prediction data sequence.

3. The method according to claim 1, characterized in that, It also includes the steps of training an LSTM neural network prediction model, specifically including: Use the first 80% of historical data as the training set, 10% as the validation set, and 10% as the test set, and convert them into NumPy arrays. Build an LSTM neural network model using Keras; The training data is input into the LSTM neural network model for training, resulting in a trained LSTM neural network prediction model.

4. The method according to claim 3, characterized in that, The method of building an LSTM neural network model using Keras includes: Create a Sequential model instance and add two LSTM layers with 64 neurons each. Set the return_sequences parameter to True for the first LSTM layer and set the return_sequences parameter to False for the second LSTM layer. Add a Dense layer with 10 neurons and an output layer; The neural network optimizer is set to Adam optimizer with a learning rate of 0.01, and mean squared error is used as the loss function for model compilation.

5. The method according to claim 1, characterized in that, Before the step of setting sampling intervals for wind speed prediction data, solar radiation prediction data, and power grid load prediction data, and obtaining the probability distribution relationships between wind speed and solar radiation, and between wind speed and power grid load, the following steps are included: Using the chi-square test and correlation coefficient, the correlation between wind speed and solar radiation, as well as between wind speed and power distribution network load, is analyzed. If a correlation is found, the steps described above are performed to set sampling intervals for wind speed prediction data, solar radiation prediction data, and power distribution network load prediction data, and to obtain the probability distribution relationship between wind speed and solar radiation, as well as between wind speed and power distribution network load.

6. The method according to claim 5, characterized in that, The steps for analyzing the correlation between wind speed and solar radiation, and between wind speed and power grid load, using the chi-square test and correlation coefficient, include: The chi-square independence test is used to test the pairwise independence between wind speed, solar radiation, and user load. The test statistic is: （4）， In equation (4), N is the total amount of data. To determine the number of rows in the split table, For the number of columns, For the first in the partition table Line 1 The element of the column; For pairwise random variables X and Y, the correlation coefficient is calculated as follows: （5）， In equation (5), ,and , , and These correspond to the expected value and standard deviation of random variables X and Y, respectively; For two sets of data, {xi} and {yi}, the formula for calculating the correlation coefficient between them is: （6）。 7. The method according to claim 1, characterized in that, The steps of setting sampling intervals for wind speed prediction data, solar radiation prediction data, and power grid load prediction data, and obtaining the probability distribution relationships between wind speed and solar radiation, and between wind speed and power grid load, include: The sampling interval for wind speed prediction data is set: the starting wind speed is determined based on the characteristic curve of wind turbine operation. and cut-off wind speed Wind speed range Divide into several intervals, and at the same time, wind speed and The wind speeds were grouped into m intervals, denoted as a single level range. ; Define the sampling interval for solar radiation prediction data: define the radiation range of the solar radiation prediction data. Divide into k thousand intervals, denoted as ; To define sampling intervals for distribution network load forecast data: Based on K-means clustering of the distribution network load forecast data, each load type is further divided into intervals, categorized by maximum and minimum load. Divide into n level intervals, denoted as ; The probability distribution relationships between solar radiation and wind speed, and between wind speed and power grid load, were statistically analyzed: （7）， （8）， Further obtain the load value in the wind speed value The probability distribution under the given conditions is: （9）， The relationship between the probability of solar radiation distribution and wind speed was obtained: （10）， （11）。 8. The method according to claim 1, characterized in that, The steps of sampling wind speed ranges and calculating the power output of wind power generation systems and solar power generation systems respectively include: Wind speed samples are obtained by sampling within the wind speed range; The output power of the wind turbine can be calculated using the following formula: （1）， In equation (1), P is the power generation capacity, Prate is the rated power, Vrate is the rated wind speed, Vci is the maximum wind speed at which the wind turbine starts, Vco is the maximum wind speed at which the wind turbine starts, and V is the current wind speed. The reactive power output of a wind turbine generator set is calculated using the following formula: （2）， In formula (2): Q is the power factor angle. WTG P WTG These are the reactive and active power of the wind turbine; The formula for calculating the output energy of photovoltaic cells is as follows, based on the sampling interval set for solar radiation prediction data: （3）， In equation (3), This refers to the rated capacity of the photovoltaic array; This is the reduction factor; This represents the current actual environmental radiation intensity. is the unit radiation intensity under standard test conditions; is the temperature coefficient of photovoltaic power output; This refers to the current temperature of the photovoltaic array. The temperature is the standard test environment temperature for photovoltaic arrays.

9. The method according to claim 1, characterized in that, It also includes: outputting results when the power flow test meets the convergence condition, and performing probabilistic statistical analysis on the results.

10. A probabilistic power flow calculation system based on predictive analysis and interval sampling, characterized in that, The system, configured to implement the method as described in any one of claims 1 to 9, comprises: The acquisition module is used to acquire historical data, including historical wind speed data, historical solar radiation data, and historical power distribution network load data. The prediction module is used to make predictions based on historical data to obtain a series of prediction data, which includes: wind speed prediction data, solar radiation prediction data, and power distribution network load prediction data. The probability distribution relationship calculation module is used to set the sampling interval for wind speed prediction data, solar radiation prediction data and power distribution network load prediction data, and obtain the probability distribution relationship between wind speed and solar radiation and between wind speed and power distribution network load. The power output calculation module is used to sample the wind speed range and calculate the power output of the wind power generation system and the solar power generation system respectively. The power flow calculation module is used to read in the load data of each node, substitute the power output of the wind power generation system and the solar power generation system, and calculate the power flow using the forward and backward algorithm.

Images