A method for simulating and predicting annual scenarios of power of a hydropower station runoff and its associated source and load power

Abstract

A method for simulating and predicting annual scenarios of hydropower station runoff and its associated source load power includes the following steps: Step 1: Collect historical daily runoff data of multiple dispatchable hydropower stations in a certain region, as well as historical daily power data of their externally associated energy and load; Step 2: Use a self-organizing mapping neural network to cluster the above data by ten-day period to form typical ten-day scenarios of runoff and source load power; Step 3: Establish a ten-day state transition probability model for the runoff of the target hydropower station, and simultaneously establish multi-scenario conditional probability models between the runoff of the target hydropower station and associated source load power, and between the runoff of the target hydropower station and the runoff of other dispatchable hydropower stations; Step 4: Randomly simulate the time series scenarios of source load power and dispatchable hydropower station runoff in future years, and aggregate them into typical annual time series simulation scenarios of source load power and dispatchable hydropower station runoff through scenario reduction methods; Step 5: Predict the probability of occurrence of each typical scenario.

Description

Technical Field

[0001] This invention relates to the field of stochastic simulation and prediction technology for hydropower station runoff scenarios, and in particular to a method for stochastic simulation and prediction of annual time-series scenarios of hydropower station runoff and external associated source load power. Background Technology

[0002] As a crucial component of the power system, the correct and rational arrangement of hydropower station operation is vital for the safe, stable, and economical operation of the entire power system. The uncertainty of water inflow to hydropower stations is affected not only by random weather changes but also by the human-induced scheduling behavior of upstream hydropower plants. To effectively reduce peak-shaving water wastage and improve the efficiency of hydropower generation, it is essential to fully consider the volatility and randomness of runoff and source-load power in hydropower scheduling decisions. Starting from the source, it is necessary to analyze the temporal fluctuation characteristics of runoff and to reasonably and accurately simulate the runoff and source-load power sequences.

[0003] Traditional methods, such as causal analysis and statistical methods, have some drawbacks in simulating and predicting scenarios. For example, the causal analysis method was used in the study "An Analysis of the Reasons for the Decline in Runoff in Major River Basins of China from 1960 to 2010" to analyze the combined effects of weather patterns, atmospheric circulation, and underlying physical factors. However, the physical causes affecting long-term runoff series changes are complex, and it is difficult to fully grasp their objective laws. Statistical methods, on the other hand, make predictions based on probability by analyzing and calculating hydrological data. For instance, the study "Long-term runoff study using SARIMA and ARIMA models in the United States" used an autoregressive moving average model estimated from historical data and related derived time series models to maintain the autocorrelation characteristics between runoff periods in the series. However, this method is only suitable for short-term runoff simulations and is difficult to adapt to the probability distribution characteristics of long-term runoff change simulations, thus having certain limitations. The time-series scenario stochastic simulation and prediction method established in this paper has the characteristics of multiple applicability and interpretability. It uses self-organizing map (SOM) to cluster regional runoff and source load power over many years into scenarios, and establishes a multi-scenario conditional probability model. Finally, it uses stochastic simulation and scenario reduction methods to simulate the time-series scenarios of runoff and source load power in future years and make probabilistic predictions of typical time-series scenarios. Summary of the Invention

[0004] The purpose of this invention is to overcome the technical problem that most existing technologies are only applicable to short-term runoff simulation and are difficult to adapt to the probability distribution characteristics of long-term runoff change simulation, thus limiting their applicability. In this invention, a stochastic simulation and prediction method for annual time-series scenarios of hydropower station runoff and external associated source load power is provided.

[0005] A method for simulating and predicting annual scenarios of runoff and associated source load power at a hydropower station, comprising the following steps:

[0006] Step 1: Collect historical daily runoff data of multiple dispatchable hydropower stations in a certain region, as well as historical daily power data of their externally related energy sources and loads;

[0007] Step 2: Use a self-organizing map neural network to cluster the above data by ten-day period to form typical ten-day scenarios of runoff and source-load power.

[0008] Step 3: Establish a ten-day state transition probability model for the runoff of the target hydropower station, and at the same time establish multi-scenario conditional probability models between the runoff of the target hydropower station and the associated source load power, and between the runoff of the target hydropower station and the runoff of other dispatchable hydropower stations.

[0009] Step 4: Randomly simulate the time series scenarios of source load power and dispatchable hydropower station runoff in future years, and aggregate them into typical annual time series simulation scenarios of source load power and dispatchable hydropower station runoff by scenario reduction method;

[0010] Step 5: Predict the probability of each typical scenario occurring.

[0011] In step 2, the self-organizing map neural network clustering algorithm is used to cluster the runoff of each dispatchable hydropower station in step 1 according to the three different periods of non-flood season, general flood season and main flood season, to generate typical scenarios and statistical probabilities of each dispatchable hydropower station in the three different periods; the power data of each source and load are directly clustered by ten days to generate the corresponding typical scenarios and statistical probabilities.

[0012] Step 2 specifically includes the following steps:

[0013] Step 2.1: Using the self-organizing map neural network algorithm, the daily runoff of each dispatchable hydropower station is clustered according to the corresponding ten-day runoff of three different periods: non-flood season, general flood season, and main flood season, to generate typical scenarios for each dispatchable hydropower station in the three different periods. And statistical probability (Q(t) indicates whether the t-th ten-day period corresponds to the non-flood season, general flood season or main flood season), where z∈{1,2,…,Z}, Z is the total number of dispatchable hydropower stations;

[0014] Step 2.2: Use the self-organizing map neural network algorithm to perform cluster analysis on the historical daily power data of external sources and loads associated with the hydropower station in ten-day periods to form typical scenarios of historical power data for each source and load. G represents different source payload types.

[0015] In step 3, a typical ten-day runoff scenario of a benchmark hydropower station is selected from the ten-day scenario set obtained in step 2 as the condition benchmark; the transition probability between typical ten-day scenarios of the hydropower station in history is calculated, and a Markov time series state transition probability model between ten-day periods is established; the condition probabilities of the occurrence of typical scenarios of external associated source load power and typical scenarios of runoff of other dispatchable hydropower stations in historical ten-day periods are statistically analyzed, and a multi-scenario condition probability model between the runoff of the benchmark hydropower station and the power of external associated source load, as well as a multi-scenario condition probability model between the runoff of the benchmark hydropower station and the runoff of other dispatchable hydropower stations are established.

[0016] Step 3 specifically includes the following steps:

[0017] Step 3.1: In the statistically based dispatchable hydropower station, take hydropower station B, which is to be analyzed in the scenario, as the condition benchmark, and form a Markov time series state transition probability matrix P1 based on the typical scenarios of the ten-day runoff corresponding to the three different periods of non-flood season, general flood season and main flood season of the hydropower station.

[0018] Step 3.2: Using the typical scenario of runoff from benchmark hydropower station B as a condition, calculate the conditional probability P2 = P of the typical scenario of external associated source load power occurring in historical ten-day periods. ij (2) Establish a multi-scenario conditional probability model between the runoff of the benchmark hydropower station B and the power of the external associated source load G;

[0019] Step 3.3: Using the typical scenario of runoff from the benchmark hydropower station B as a condition, calculate the conditional probability P3 = P of the typical scenario of runoff from other dispatchable hydropower stations Y in the historical ten-day period. ij (3) A multi-scenario conditional probability model was established between the runoff of hydropower station B and the runoff of other dispatchable hydropower stations Y.

[0020] In step 4, based on the conditional probability matrix and transition probability matrix obtained in step 3, the annual time series scenario of the benchmark hydropower station runoff in future years is randomly simulated. Based on this, and combined with the multi-scenario conditional probability model of external associated source load power and other dispatchable hydropower station runoff, the annual time series scenario of external associated source load power and other dispatchable hydropower station runoff under the benchmark hydropower station runoff scenario is randomly simulated and generated.

[0021] The obtained annual time-series simulation scenario set was further reduced using the K-means method, and aggregated into typical annual time-series simulation scenarios of the benchmark hydropower station runoff, external associated source load power, and runoff of other dispatchable hydropower stations under the flood, normal, and dry water inflow schemes.

[0022] In step 4, obtaining the annual time series scene specifically includes the following steps:

[0023] Step 4.1: Introduce the Markov time series transition probability matrix P1 = [P] for the typical scenario of runoff from the benchmark hydropower station B in Step 3. ij (1) ], the multi-scenario conditional probability matrix P2 = [P ij (2) ] (G) The conditional probability matrix P3 = [P] for the typical scenarios of runoff Y at other dispatchable hydropower stations. ij (3) ] (Y) Typical scenarios for the ten-day runoff of Hydropower Station B Typical scenarios for the ten-day runoff of other dispatchable hydropower stations Where the superscript Q(t) represents the different levels of abundance and scarcity corresponding to the non-flood season, general flood season, and main flood season of the t-th ten-day period, and the set of externally related source load ten-day power scenarios. Where G represents different source payload types;

[0024] Step 4.2: Use stochastic simulation to simulate typical time-series scenarios of runoff and source load power in future years.

[0025] A method for establishing a multi-scenario conditional probability model, wherein the established multi-scenario conditional probability model is a multi-scenario conditional probability model of runoff and source load daily power curves, includes the following steps: based on the historical time series of the occurrence of typical ten-day runoff scenarios of hydropower station B, a Markov time series state transition probability model between ten-day periods is established; taking the typical runoff scenarios of hydropower station B as conditions, the conditional probabilities of the occurrence of externally related source load power and other typical runoff scenarios of hydropower stations in historical ten-day periods are statistically analyzed, and a multi-scenario conditional probability model between the runoff of hydropower station B and the externally related source load power and a conditional probability model between the runoff of hydropower station B and the runoff of other related hydropower stations are established.

[0026] The method specifically includes the following steps:

[0027] Step 3.1: In the statistically based dispatchable hydropower station, take hydropower station B, which is to be analyzed in the scenario, as the condition benchmark, and form a Markov time series state transition probability matrix P1 based on the typical scenarios of the ten-day runoff corresponding to the three different periods of non-flood season, general flood season and main flood season of the hydropower station.

[0028]

[0029] 1≥P ij (1) ≥0,

[0030] In the formula, the number of clusters for the ten-day runoff scenario of the benchmark hydropower station B is k, and the number of transitions N between typical scenario categories i and j in the historical ten-day periods after clustering is calculated. ij The number of times typical scenario category i appears is N. i The transition probability matrix consists of probability values ​​between the interval [0,1] for each element, and the sum of the elements in each row is equal to 1.

[0031] Step 3.2: Using the typical scenario of runoff from benchmark hydropower station B as a condition, calculate the conditional probability P2 = P of the typical scenario of external associated source load power occurring in historical ten-day periods. ij (2) Establish a multi-scenario conditional probability model between the runoff of the benchmark hydropower station B and the power of the external associated source load G;

[0032]

[0033] 1≥P ij (2) ≥0,

[0034] In the formula, each element in the conditional probability matrix is ​​located between [0,1], and the sum of the conditional probabilities of each row or column in the matrix is ​​equal to 1. k is the total number of typical scenario categories for the runoff of hydropower station B, and m is the total number of typical scenario categories for the power of external associated sources and loads. Given that the event of typical scenario i for the runoff of hydropower station B has already occurred, the conditional probability of the event of typical scenario j for the power of associated sources and loads is defined as P. ij (2) Its calculation method is as described above, where M i M represents the number of times typical scenario category i of the runoff of hydropower station B occurs throughout all historical ten-day periods. ij The number of times that the typical scenario category of runoff of hydropower station B in the historical ten-day period is i, and the typical scenario category of external associated source load power in the corresponding ten-day period is j.

[0035] Step 3.3: Using the typical scenario of runoff from the benchmark hydropower station B as a condition, calculate the conditional probability P3 = P of the typical scenario of runoff from other dispatchable hydropower stations Y in the historical ten-day period. ij (3) Establish a multi-scenario conditional probability model between the runoff of hydropower station B and the runoff of other dispatchable hydropower stations Y;

[0036]

[0037] 1≥P ij (3) ≥0,

[0038] In the formula, each element in the conditional probability matrix is ​​located between [0,1], and the sum of the conditional probabilities of each row or column in the matrix is ​​equal to 1. k is the total number of typical scenario categories for the runoff of hydropower station B, and s is the total number of scenario categories for the runoff of other hydropower stations Y. Given that the event of typical scenario i for the runoff of hydropower station B has already occurred, the conditional probability of the event of typical scenario j for the runoff of other hydropower stations Y is defined as P. ij (3) Its calculation method is as described above, where L i L represents the number of times typical scenario category i of runoff from hydropower station B occurs throughout all historical ten-day periods. ij Let i be the number of times the typical scenario category of runoff from hydropower station B in a historical ten-day period is i, and the typical scenario category of runoff from the other hydropower station Y in the corresponding ten-day period is j.

[0039] A method for simulating typical time-series scenarios of runoff and source load power is proposed. This method simulates the annual time-series scenarios of externally associated source load power and other dispatchable hydropower station runoff under the benchmark hydropower station runoff scenario. Based on the conditional probability matrix and transition probability matrix, the method randomly simulates the annual time-series scenarios of the benchmark hydropower station runoff in future years. Using this as a condition, the method combines the multi-scenario conditional probability model of externally associated source load power and other dispatchable hydropower station runoff to randomly simulate and generate the annual time-series scenarios of externally associated source load power and other dispatchable hydropower station runoff under the benchmark hydropower station runoff scenario.

[0040] The method specifically includes the following steps:

[0041] Step 1) Introduce the Markov time series transition probability matrix P1 = [P] for the typical scenario of runoff from the benchmark hydropower station B in Step 3. ij (1) ], the multi-scenario conditional probability matrix P2 = [P ij (2) ] (G) The conditional probability matrix P3 = [P] for the typical scenarios of runoff Y at other dispatchable hydropower stations. ij (3) ] (Y) Typical scenarios for the ten-day runoff of Hydropower Station B Typical scenarios for the ten-day runoff of other dispatchable hydropower stations Where the superscript Q(t) represents the different levels of abundance and scarcity corresponding to the non-flood season, general flood season, and main flood season of the t-th ten-day period, and the set of externally related source load ten-day power scenarios. Where G represents different source payload types;

[0042] Step 2) Use stochastic simulation to simulate typical time-series scenarios of runoff and source load power in future years.

[0043] In step 2), it specifically includes the following steps:

[0044] Step (1) Set the number of random simulations n = 1;

[0045] Step (2) Take the state i of the typical runoff scenario of Hydropower Station B at the end of the previous year's last ten-day period as the initial point, t = 0, and the ten-day scenario number is I(0) = i;

[0046] Step (3) Let t = t + 1, and look up the probability of transferring to the state of the t-th ten-day period from the row where the state I(t - 1) = i in P1 [P i1 (1) , P i2 (1) ,..., P ik (1) , and calculate the cumulative probability in sequence as Then generate a random number a from the uniform distribution If Then the runoff scenario state corresponding to this ten-day period is J, and let I(t) = J;

[0047] Step (4) If t < T (T = 36), then go to step 3; otherwise, end the simulation of the current sample year and output the annual simulation scenario of Hydropower Station B corresponding to {I(t), t = 1,..., T}

[0048] Step (5) Look up the probability of the column corresponding to the source-load G scenario state from the row where the runoff volume state I(t) of Hydropower Station B in P2 Calculate the cumulative probability in sequence Generate a random number b from the uniform distribution If Then the source-load typical scenario state corresponding to this ten-day period is J, and let g(t) = J. Traverse all ten-day periods of the whole year and all types of source-loads to generate the annual scenario of the source-load power under the current annual scenario of the runoff of Hydropower Station B

[0049] Step (6) Look up the probability of the column corresponding to the scenario state of the dispatchable hydropower station Y from the row where the runoff volume state I(t) of Hydropower Station B in P3 [P i1 (3) , P i2 (3) ,..., P is (3) , and calculate the cumulative probability as Generate a random number c from the uniform distribution If Then the runoff typical scenario state of the schedulable hydropower stations corresponding to this ten-day period is J. Let y(t) = J, traverse all ten-day periods of the whole year and all the other schedulable hydropower stations Y, and generate the annual scenarios of the runoff of the other schedulable hydropower stations under the annual scenario of the runoff of Hydropower Station B currently.

[0050] Step (7) If n < N (N is the total number of randomly simulated annual scenarios), then n = n + 1, and go to Step 4.2.2. Otherwise, end the simulation, output the runoff of Hydropower Station B, the annual time-series simulation scenarios of the external source-load power, and the set of annual time-series simulation scenarios of the runoff of the other schedulable hydropower stations.

[0051] Compared with the prior art, the present invention has the following technical effects:

[0052] 1) For the first time, the present invention applies the self-organizing mapping SOM clustering method to the clustering of hydropower station runoff scenarios to generate typical runoff and source-load power scenarios, providing a new clustering method for the analysis of hydropower station runoff scenarios;

[0053] 2) The present invention clusters the runoff of hydropower stations by ten-day periods in three different periods to obtain typical scenarios of each period, and establishes a Markov time-series state transition probability model based on this to simulate the typical time-series scenarios of the runoff in the future year, providing a new idea for the prediction of hydropower station runoff scenarios;

[0054] 3) The present invention considers the external source-load power associated with the hydropower station and the runoff of other associated hydropower stations, establishes a conditional probability matrix of the external source-load power and typical scenarios of the runoff of other associated hydropower stations based on the typical scenarios of the hydropower station, and realizes the prediction of the typical time-series scenarios of the external source-load power and the runoff of other hydropower stations, providing a new method for the analysis of the external associated source-load power and the runoff scenarios of associated hydropower stations;

[0055] 4) The results of actual calculation examples show that this random simulation method can obtain annual time-series prediction scenarios of runoff and external source-load power with high accuracy and rich information. Brief Description of the Drawings

[0056] The following further describes the present invention in conjunction with the drawings and embodiments:

[0057] Figure 1 It is a flow chart of a random simulation and prediction method for annual time-series scenarios of hydropower station runoff and external associated source-load power;

[0058] Figure 2 It is a structure diagram of a SOM neural network;

[0059] Figure 3 It is a flow chart of the SOM algorithm;

[0060] Figure 4 A flowchart for random simulation and scene reduction;

[0061] Figure 5 Here is a flowchart of the K-means algorithm;

[0062] Figure 6 This is a curve diagram of a typical scenario of a hydropower station according to an embodiment of the present invention;

[0063] Figure 7 This is a heatmap of the conditional probability matrix of the runoff of a dispatchable hydropower station Y under the scenario of the runoff of a benchmark hydropower station B, according to an embodiment of the present invention.

[0064] Figure 8 This is a heatmap of the conditional probability matrix of a typical scenario of the active power load of the entire network in an embodiment of the present invention under the scenario of the runoff of the benchmark hydropower station B.

[0065] Figure 9 This is a comparison chart of the predicted and actual annual runoff time series curves of the benchmark hydropower station B under the scenarios of high water year, normal water year and low water year according to an embodiment of the present invention.

[0066] Figure 10 This is a bar chart showing the annual load power prediction in a dry year scenario according to an embodiment of the present invention.

[0067] Figure 11 This is a prediction curve of the annual runoff of a dispatchable hydropower station under a dry year scenario, according to an embodiment of the present invention.

[0068] Figure 12 This is a typical scenario diagram of the Pengshui Hydropower Station's future runoff prediction according to an embodiment of the present invention;

[0069] Figure 13 This is a comparison chart of the low inflow prediction scenario and the empirical prediction results of the Pengshui Hydropower Station according to an embodiment of the present invention;

[0070] Figure 14 This is a typical scenario diagram for wind power prediction under the low inflow scheme of Pengshui Hydropower Station according to an embodiment of the present invention;

[0071] Figure 15 This is a typical scenario diagram for load prediction under the low inflow scheme of Pengshui Hydropower Station, according to an embodiment of the present invention. Detailed Implementation

[0072] like Figure 1 As shown, a method for stochastic simulation and prediction of annual time-series scenarios of hydropower station runoff and external associated source load power includes the following steps:

[0073] Step 1: Collect historical daily runoff data of multiple dispatchable hydropower stations in a certain region, in ten-day periods, as well as historical daily power data of the source loads associated with them.

[0074] Step 2: Use the self-organizing map neural network clustering algorithm to cluster the runoff of each dispatchable hydropower station in Step 1 according to the corresponding ten-day runoff of three different periods: non-flood season, general flood season, and main flood season, to generate typical scenarios and statistical probabilities for each dispatchable hydropower station in the three different periods; directly cluster the power data of each source and load according to ten-day periods to generate the corresponding typical scenarios and statistical probabilities.

[0075] Step 2.1: Using the Self-Organizing Map Neural Network (SOM) algorithm, the daily runoff of each dispatchable hydropower station is clustered according to the corresponding ten-day runoff of three different periods: non-flood season, general flood season, and main flood season, generating typical scenarios for each dispatchable hydropower station under the three different periods. And statistical probability (Q(t) indicates whether the t-th ten-day period corresponds to the non-flood season, general flood season or main flood season), where z∈{1,2,…,Z}, Z is the total number of dispatchable hydropower stations;

[0076] Step 2.2: Use the Self-Organizing Map Neural Network (SOM) algorithm to perform cluster analysis on the historical daily power data of external sources and loads associated with the hydropower station, using ten-day periods as the unit, to form typical scenarios for the historical power data of each source and load. G represents different source payload types.

[0077] according to Figure 3 The process of the Self-Organizing Map (SOM) clustering algorithm based on historical ten-day data is explained:

[0078] 1) Let the iteration number t = 1, and randomly assign weights W to the output nodes. j (j=1,2,…,k) are the smallest initial values, and the learning rate α(t) is initialized.

[0079] 2) Randomly select input sample vector x from n historical scenes. i (i = 1, 2, ..., n), calculate x i With all weight vectors W j The winning neuron is selected based on the Euclidean distance between neurons (j = 1, 2, ..., k) and the distance between them.

[0080]

[0081] In the formula, ||·|| is the distance function, j * The winning unit;

[0082] 3) Adjust the winning neuron j * Connection weights between neurons in the neighborhood and the input neuron:

[0083] w ij (t+1)=w ij (t)+α(t,N)[x i (t)-w ij(t)]

[0084] i = 1, 2, ..., n

[0085] in, For the neighborhood of the winning unit, w ij (t) represents the weights of neurons i to j; α(t,D) represents the training iterations t and the weights of neurons i and j. * A function of the topological distance D between neurons. α(t,D)=α(t)e -D e -D It is a Gaussian function;

[0086] 4) Determine if t has reached the maximum iteration count T. max Otherwise, t = t + 1, proceed to step 2;

[0087] 5) Renumber the categories in ascending order, and output the category number and cluster center C = {C1, C2, ... C} of the samples. k}

[0088] Step 3: Based on the historical time series of typical ten-day runoff scenarios of hydropower station B, establish a Markov time series state transition probability model between ten-day periods; taking the typical runoff scenario of hydropower station B as a condition, statistically analyze the conditional probabilities of the occurrence of external related source load power and other typical runoff scenarios of hydropower stations in historical ten-day periods, and establish a multi-scenario conditional probability model between the runoff of hydropower station B and the external related source load power, and a conditional probability model between the runoff of hydropower station B and the runoff of other related hydropower stations.

[0089] Step 3.1: In the statistically based dispatchable hydropower station, take hydropower station B, which is to be analyzed in the scenario, as the condition benchmark, and form a Markov time series state transition probability matrix P1 based on the typical scenarios of the ten-day runoff corresponding to the three different periods of non-flood season, general flood season and main flood season of the hydropower station.

[0090]

[0091] 1≥P ij (1) ≥0,

[0092] In the formula, the number of clusters for the ten-day runoff scenario of the benchmark hydropower station B is k, and the number of transitions N between typical scenario categories i and j in the historical ten-day periods after clustering is calculated. ij The number of times typical scenario category i appears is N. i The transition probability matrix consists of probability values ​​between the interval [0,1] for each element, and the sum of the elements in each row is equal to 1.

[0093] Step 3.2: Using the typical scenario of runoff from benchmark hydropower station B as a condition, calculate the conditional probability P2 = P of the typical scenario of external associated source load power occurring in historical ten-day periods. ij (2) Establish a multi-scenario conditional probability model between the runoff of the benchmark hydropower station B and the power of the external associated source load G;

[0094]

[0095] 1≥P ij (2) ≥0,

[0096] In the formula, each element in the conditional probability matrix is ​​located between [0,1], and the sum of the conditional probabilities of each row or column in the matrix is ​​equal to 1. k is the total number of typical scenario categories for the runoff of hydropower station B, and m is the total number of typical scenario categories for the power of external associated sources and loads. Given that the event of typical scenario i for the runoff of hydropower station B has already occurred, the conditional probability of the event of typical scenario j for the power of associated sources and loads is defined as P. ij (2) Its calculation method is as described above, where M i M represents the number of times typical scenario category i of the runoff of hydropower station B occurs throughout all historical ten-day periods. ij The number of times that the typical scenario category of runoff of hydropower station B in the historical ten-day period is i, and the typical scenario category of external associated source load power in the corresponding ten-day period is j.

[0097] Step 3.3: Using the typical scenario of runoff from the benchmark hydropower station B as a condition, calculate the conditional probability P3 = P of the typical scenario of runoff from other dispatchable hydropower stations Y in the historical ten-day period. ij (3) A multi-scenario conditional probability model was established between the runoff of hydropower station B and the runoff of other dispatchable hydropower stations Y.

[0098]

[0099] 1≥P ij (3) ≥0,

[0100] In the formula, each element in the conditional probability matrix is ​​located between [0,1], and the sum of the conditional probabilities of each row or column in the matrix is ​​equal to 1. k is the total number of typical scenario categories for the runoff of hydropower station B, and s is the total number of scenario categories for the runoff of other hydropower stations Y. Given that the event of typical scenario i for the runoff of hydropower station B has already occurred, the conditional probability of the event of typical scenario j for the runoff of other hydropower stations Y is defined as P. ij (3)Its calculation method is as described above, where L i L represents the number of times typical scenario category i of runoff from hydropower station B occurs throughout all historical ten-day periods. ij Let i be the number of times the typical scenario category of runoff from hydropower station B in a historical ten-day period is i, and the typical scenario category of runoff from the other hydropower station Y in the corresponding ten-day period is j.

[0101] Step 4: Based on the conditional probability matrix and transition probability matrix obtained in Step 3, randomly simulate the annual time series scenario of the benchmark hydropower station runoff in future years. Using this as a condition, combine the external associated source load power and other dispatchable hydropower station runoff multi-scenario conditional probability model to randomly simulate and generate the annual time series scenario of external associated source load power and other dispatchable hydropower station runoff under the benchmark hydropower station runoff scenario.

[0102] Step 4.1: Introduce the Markov time series transition probability matrix P1 = [P] for the typical scenario of runoff from the benchmark hydropower station B in Step 3. ij (1) ], the multi-scenario conditional probability matrix P2 = [P ij (2) ] (G) The conditional probability matrix P3 = [P] for the typical scenarios of runoff Y at other dispatchable hydropower stations. ij (3) ] (Y) Typical scenarios for the ten-day runoff of Hydropower Station B Typical scenarios for the ten-day runoff of other dispatchable hydropower stations Where the superscript Q(t) represents the different levels of abundance and scarcity corresponding to the non-flood season, general flood season, and main flood season of the t-th ten-day period, and the set of externally related source load ten-day power scenarios. Where G represents different source payload types;

[0103] Step 4.2: Use stochastic simulation methods to simulate typical time-series scenarios of runoff and source load power in future years;

[0104] Step 4.2.1: Let the number of random simulations be n = 1;

[0105] Step 4.2.2: Take the typical scenario state i of the runoff of Hydropower Station B at the end of the last ten days of the year as the initial point, t=0, and the scenario number of the ten-day period is I(0)=i;

[0106] Step 4.2.3: Let t = t + 1, and find the probability of transitioning to the t-th state from the row containing state I(t-1) = i in P1 [P]. i1 (1) ,P i2 (1) ,...,P ik (1), and calculate the cumulative probability in sequence as Then generate a random number a from the uniform distribution If then the runoff scenario state corresponding to this ten-day period is J, and let I(t)=J;

[0107] Step 4.2.4: If t<T (T = 36), then go to Step 3; otherwise, end the simulation of the current sample year and output the annual simulation scenario of the runoff of Hydropower Station B corresponding to {I(t), t = 1,..., T}

[0108] Step 4.2.5: Look up the probability of the corresponding source-load G scenario state column in the row where the runoff state I(t) of Hydropower Station B is located in P2 Calculate the cumulative probability in sequence Generate a random number b from the uniform distribution If then the typical scenario state of the corresponding source-load of this ten-day period is J, and let g(t)=J. Traverse all ten-day periods of the whole year and all types of source-loads to generate the annual scenario of the source-load power under the current annual scenario of the runoff of Hydropower Station B

[0109] Step 4.2.6: Look up the probability [P i1 (3) , P i2 (3) ,..., P is (3) of the corresponding scenario state column of the dispatchable hydropower station Y in the row where the runoff state I(t) of Hydropower Station B is located in P3, and calculate the cumulative probability as Generate a random number c from the uniform distribution If then the typical scenario state of the runoff of the corresponding dispatchable hydropower station of this ten-day period is J, and let y(t)=J. Traverse all ten-day periods of the whole year and the remaining dispatchable hydropower stations Y to generate the annual scenario of the runoff of the remaining dispatchable hydropower stations under the current annual scenario of the runoff of Hydropower Station B

[0110] Step 4.2.7: If n<N (N is the total number of random simulation annual scenarios), then n=n + 1, and go to Step 4.2.2; otherwise, end the simulation and output the set of the annual time series simulation scenarios of the runoff of Hydropower Station B, the external source-load power, and the annual time series simulation scenarios of the runoff of the remaining dispatchable hydropower stations

[0111] Step 5: The annual time series simulation scenario set obtained in Step 4 is reduced to a set of typical annual time series simulation scenarios using the K-means method. These scenarios are then aggregated into typical annual time series simulation scenarios for the baseline hydropower station runoff, external associated source load power, and runoff of other dispatchable hydropower stations under the flood, normal, and dry water inflow schemes. The probability of each typical scenario occurring is then predicted.

[0112] Step 5.1: Use the K-means algorithm to cluster the annual runoff time-series simulation scenarios of Hydropower Station B into K1 classes. Calculate the probability of occurrence of each typical annual scenario based on the ratio of the number of scenarios in each class to the total number of simulation scenarios. Cluster the associated source-load power simulation annual scenarios belonging to the same Hydropower Station B runoff annual scenario category into K2 classes using the K-means algorithm. Then, cluster the remaining dispatchable hydropower station runoff annual time-series simulation scenarios belonging to the same Hydropower Station B runoff annual scenario category into K3 classes using the K-means algorithm. Output the cluster centers of each combined scenario as the typical annual scenario and their statistical probabilities.

[0113]

[0114] Step 5.2: Based on the annual runoff volume of Hydropower Station B, divide the annual typical scenario set into high, medium, or low inflow schemes for output.

[0115] Can be combined Figure 4 The process of the two steps of random simulation and scene reduction in this invention will be explained.

[0116] according to Figure 5 The process of the K-means clustering algorithm is explained below:

[0117] 1) Input the given n sample data;

[0118] 2) Initialize the weights and divide the samples into k data clusters C k Each data cluster C k There is a corresponding cluster center U k ;

[0119] 3) Calculate the distance from the sample data points to the cluster centers U. k The distance is used to select a new cluster center U. k ;

[0120] 4) Iterate again to calculate the distance and select new cluster centers. Repeat the iteration until the cluster centers no longer change.

[0121] This invention also includes a method for establishing a multi-scenario conditional probability model. The established multi-scenario conditional probability model is a multi-scenario conditional probability model of runoff and source load daily power curves, including the following steps: based on the historical time series of typical ten-day runoff scenarios of hydropower station B, establish a Markov time series state transition probability model between ten-day periods; taking the typical runoff scenario of hydropower station B as a condition, statistically analyze the conditional probabilities of the occurrence of externally related source load power and other typical runoff scenarios of hydropower stations in historical ten-day periods, and establish a multi-scenario conditional probability model between the runoff of hydropower station B and the externally related source load power, and a conditional probability model between the runoff of hydropower station B and the runoff of other related hydropower stations.

[0122] The method specifically includes the following steps:

[0123] 1) In the statistically based dispatchable hydropower station, the hydropower station B to be analyzed is used as the condition benchmark, and a Markov time series state transition probability matrix P1 is formed based on the typical scenarios of the ten-day runoff corresponding to the three different periods of non-flood season, general flood season and main flood season of the station.

[0124]

[0125] 1≥P ij (1) ≥0,

[0126] In the formula, the number of clusters for the ten-day runoff scenario of the benchmark hydropower station B is k, and the number of transitions N between typical scenario categories i and j in the historical ten-day periods after clustering is calculated. ij The number of times typical scenario category i appears is N. i The transition probability matrix consists of probability values ​​between the interval [0,1] for each element, and the sum of the elements in each row is equal to 1.

[0127] 2) Using the typical scenario of runoff from benchmark hydropower station B as a condition, calculate the conditional probability P2 = P of the typical scenario of externally related source load power occurring in historical ten-day periods. ij (2) Establish a multi-scenario conditional probability model between the runoff of the benchmark hydropower station B and the power of the external associated source load G;

[0128]

[0129] 1≥P ij (2) ≥0,

[0130] In the formula, each element in the conditional probability matrix is ​​located between [0,1], and the sum of the conditional probabilities of each row or column in the matrix is ​​equal to 1. k is the total number of typical scenario categories for the runoff of hydropower station B, and m is the total number of typical scenario categories for the power of external associated sources and loads. Given that the event of typical scenario i for the runoff of hydropower station B has already occurred, the conditional probability of the event of typical scenario j for the power of associated sources and loads is defined as P. ij (2) Its calculation method is as described above, where M i M represents the number of times typical scenario category i of the runoff of hydropower station B occurs throughout all historical ten-day periods. ij The number of times that the typical scenario category of runoff of hydropower station B in the historical ten-day period is i, and the typical scenario category of external associated source load power in the corresponding ten-day period is j.

[0131] 3) Using the typical scenario of runoff from benchmark hydropower station B as a condition, calculate the conditional probability P3 = P of the typical scenario of runoff from other dispatchable hydropower stations Y during the historical ten-day period. ij (3) Establish a multi-scenario conditional probability model between the runoff of hydropower station B and the runoff of other dispatchable hydropower stations Y;

[0132]

[0133] 1≥P ij (3) ≥0,

[0134] In the formula, each element in the conditional probability matrix is ​​located between [0,1], and the sum of the conditional probabilities of each row or column in the matrix is ​​equal to 1. k is the total number of typical scenario categories for the runoff of hydropower station B, and s is the total number of scenario categories for the runoff of other hydropower stations Y. Given that the event of typical scenario i for the runoff of hydropower station B has already occurred, the conditional probability of the event of typical scenario j for the runoff of other hydropower stations Y is defined as P. ij (3) Its calculation method is as described above, where L i L represents the number of times typical scenario category i of runoff from hydropower station B occurs throughout all historical ten-day periods. ij Let i be the number of times the typical scenario category of runoff from hydropower station B in a historical ten-day period is i, and the typical scenario category of runoff from the other hydropower station Y in the corresponding ten-day period is j.

[0135] The present invention also includes a method for simulating typical time-series scenarios of runoff and source load power. The method simulates the annual time-series scenarios of externally associated source load power and other dispatchable hydropower station runoff under the condition of benchmark hydropower station runoff. Based on the conditional probability matrix and transition probability matrix, the method randomly simulates the annual time-series scenarios of benchmark hydropower station runoff in future years. Based on this, the method combines the multi-scenario conditional probability model of externally associated source load power and other dispatchable hydropower station runoff to randomly simulate and generate the annual time-series scenarios of externally associated source load power and other dispatchable hydropower station runoff under the condition of benchmark hydropower station runoff.

[0136] The method specifically includes the following steps:

[0137] Step 1) Introduce the Markov time series transition probability matrix P1 = [P] for the typical scenario of runoff from the benchmark hydropower station B in Step 3. ij (1) ], the multi-scenario conditional probability matrix P2 = [P ij (2) ] (G) The conditional probability matrix P3 = [P] for the typical scenarios of runoff Y at other dispatchable hydropower stations. ij (3) ] (Y) Typical scenarios for the ten-day runoff of Hydropower Station B Typical scenarios for the ten-day runoff of other dispatchable hydropower stations Where the superscript Q(t) represents the different levels of abundance and scarcity corresponding to the non-flood season, general flood season, and main flood season of the t-th ten-day period, and the set of externally related source load ten-day power scenarios. Where G represents different source payload types;

[0138] Step 2) Use stochastic simulation to simulate typical time-series scenarios of runoff and source load power in future years.

[0139] Step 2) specifically includes the following steps:

[0140] Step (1) Set the number of random simulations n = 1;

[0141] Step (2) takes the typical scenario state i of the runoff of Hydropower Station B at the end of the year as the initial point, t=0, and the scenario number of the ten-day period is I(0)=i;

[0142] Step (3) Let t = t + 1, and find the probability of transitioning to the t-th state from the row containing state I(t-1) = i in P1 [P]. i1 (1) ,P i2 (1) ,...,P ik (1) ], and calculate the cumulative probability in sequence. Then generate a random number a from a uniform distribution If then the runoff scenario state corresponding to this ten-day period is J, and let I(t) = J;

[0143] Step (4) If t < T (T = 36), then go to step 3; otherwise, end the simulation of the current sample year and output the annual simulation scenario of the runoff of Hydropower Station B corresponding to {I(t), t = 1,..., T}

[0144] Step (5) Look up the probability of the corresponding source-load G scenario state column in the row where the runoff state I(t) of Hydropower Station B is located in P2 Calculate the cumulative probability in sequence Generate a random number b from a uniform distribution If then the typical scenario state of this source-load corresponding to this ten-day period is J, and let g(t) = J. Traverse all ten-day periods of the whole year and all types of source-loads to generate the annual scenario of the source-load power under the current annual scenario of the runoff of Hydropower Station B

[0145] Step (6) Look up the probability [P i1 (3) , P i2 (3) ,..., P is (3) of the corresponding scenario state column of the dispatchable hydropower station Y in the row where the runoff state I(t) of Hydropower Station B is located in P3, and calculate the cumulative probability as Generate a random number c from a uniform distribution If then the typical scenario state of the runoff of the dispatchable hydropower station corresponding to this ten-day period is J, and let y(t) = J. Traverse all ten-day periods of the whole year and the remaining dispatchable hydropower stations Y to generate the annual scenario of the runoff of the remaining dispatchable hydropower stations under the current annual scenario of the runoff of Hydropower Station B

[0146] Step (7) If n < N (N is the total number of random simulation annual scenarios), then n = n + 1, and go to step 4.2.2; otherwise, end the simulation and output the set of the annual time-series simulation scenarios of the runoff of Hydropower Station B, the external source-load power, and the annual time-series simulation scenarios of the runoff of the remaining dispatchable hydropower stations

[0147] In this embodiment, historical runoff data of nine dispatchable hydropower stations in a certain region, as well as historical power data of directly dispatchable thermal power, directly dispatchable hydropower, directly dispatchable wind power, directly dispatchable photovoltaic power, non-discrete hydropower, inter-provincial power transmission, and the active power load of the entire grid were collected. When clustering the historical ten-day runoff of each dispatchable hydropower station, clustering was performed according to the ten-day runoff samples corresponding to the non-flood season, general flood season, and main flood season. Each dispatchable hydropower station formed four typical scenarios and statistical probabilities for three different runoff periods. Furthermore, also using ten-day periods as the unit, the historical power of directly dispatchable thermal power, directly dispatchable hydropower, directly dispatchable wind power, directly dispatchable photovoltaic power, non-discrete hydropower, inter-provincial power transmission, and the active power load of the entire grid in this region were clustered using a SOM neural network to form four typical scenarios. Finally, typical runoff and external power source / load scenarios were formed with ten-day periods as the time unit. Using a typical runoff scenario of a hydropower station in period B as a benchmark, the probabilities of other dispatchable hydropower station runoff scenarios (12 scenario types) and source-load power scenarios (4 scenario types) occurring under the 12 typical scenario types of runoff at the hydropower station are statistically analyzed, resulting in a 12×12 (or 12×4) matrix.

[0148] Figure 6 To visualize the typical runoff scenario of a hydropower station in ten-day period B, it can be seen that the SOM clustering algorithm accurately classifies the typical ten-day periods of the hydropower station. Each curve is different in terms of numerical value and curve smoothness, which effectively distinguishes the runoff of each typical ten-day period. Figure 7 and Figure 8 The results of the visualization of the conditional probability matrices of other dispatchable hydropower station Y runoff and the typical active load scenarios of the entire network under the hydropower station B runoff scenario are shown. It can be seen that the runoff of other dispatchable hydropower stations is strongly correlated with the power of external source loads and the typical scenario of hydropower station B runoff. This indicates that it is necessary to consider the typical scenarios of other related hydropower stations and external related source loads when conducting hydropower station scenario analysis. Figure 9 The figure shows a comparison between the predicted and actual annual runoff curves of hydropower station B under the scenarios of abundant water, normal water, and dry water. It can be seen that the predicted runoff curves have significant numerical differences under different levels of abundance and dryness, but the trends of the three predicted curves are very similar to the trends of the actual runoff curves. Figure 10 and Figure 11 It effectively achieves annual time-series scenario prediction of load power and runoff from other dispatchable hydropower stations under dry year conditions. Figure 12-15 Taking the Pengshui Hydropower Station, a dispatchable hydropower station in the embodiment, as an example, the typical scenario predicted for the coming years is presented. Figure 12 For the annual scenario of Pengshui runoff prediction, Figure 13 The superiority of the method of the present invention is verified by comparing the annual runoff of Pengshui River predicted by the method of the present invention with the runoff of Pengshui River predicted by experience and the actual value. Figure 14 and Figure 15 The typical scenarios for wind power prediction and load prediction under the low water inflow scheme in Pengshui were presented respectively.

Claims

1. A method for simulating and predicting annual scenarios of runoff and associated source load power in hydropower stations, characterized in that, It includes the following steps: Step 1: Collect historical daily runoff data of multiple dispatchable hydropower stations in a certain region, as well as historical daily power data of their externally related energy sources and loads; Step 2: Use a self-organizing map neural network to cluster the above data by ten-day period to form typical ten-day scenarios of runoff and source-load power. Step 3: Establish a ten-day state transition probability model for the runoff of the target hydropower station, and at the same time establish multi-scenario conditional probability models between the runoff of the target hydropower station and the associated source load power, and between the runoff of the target hydropower station and the runoff of other dispatchable hydropower stations. Step 4: Randomly simulate the time series scenarios of source load power and dispatchable hydropower station runoff in future years, and aggregate them into typical annual time series simulation scenarios of source load power and dispatchable hydropower station runoff by scenario reduction method; Step 5: Predict the probability of each typical scenario occurring.

2. The method according to claim 1, characterized in that, In step 2, the self-organizing map neural network clustering algorithm is used to cluster the ten-day runoff of each dispatchable hydropower station in step 1 according to the three different periods of non-flood season, general flood season and main flood season, to generate typical scenarios and statistical probabilities of each dispatchable hydropower station in the three different periods. The power data of each source load are directly clustered by ten-day period to generate corresponding typical scenarios and statistical probabilities; Step 2 specifically includes the following steps: Step 2.1: Using the self-organizing map neural network algorithm, the daily runoff of each dispatchable hydropower station is clustered according to the corresponding ten-day runoff of three different periods: non-flood season, general flood season, and main flood season, to generate typical scenarios for each dispatchable hydropower station in the three different periods. And statistical probability, Q ( t ) indicates the first t Ten-day periods correspond to the non-flood season, the general flood season, or the main flood season, among which For the first t The first hydropower station that can be dispatched within ten days k One scenario, z ∈{1,2, …,Z }, Z This represents the total number of dispatchable hydropower stations; Step 2.2: Use the self-organizing map neural network algorithm to perform cluster analysis on the historical daily power data of external sources and loads associated with the hydropower station in ten-day periods to form typical scenarios of historical power data for each source and load. ,in for G Yuanhe's m One scenario, G These represent different source load types.

3. The method according to claim 1, characterized in that, In step 3, a typical ten-day runoff scenario of a benchmark hydropower station is selected from the ten-day scenario set obtained in step 2 as the condition benchmark; the transition probability between typical ten-day scenarios of the hydropower station in history is calculated, and a Markov time series state transition probability model between ten-day periods is established; the condition probabilities of the occurrence of typical scenarios of external associated source load power and typical scenarios of runoff of other dispatchable hydropower stations in historical ten-day periods are statistically analyzed, and a multi-scenario condition probability model between the runoff of the benchmark hydropower station and the power of external associated source load, as well as a multi-scenario condition probability model between the runoff of the benchmark hydropower station and the runoff of other dispatchable hydropower stations are established.

4. The method according to claim 3, characterized in that Step 3 specifically includes the following steps: Step 3.1: In the statistically based dispatchable hydropower stations, select the hydropower station for the scenario analysis. B Using this as a baseline, and based on typical scenarios of ten-day runoff during three different periods—non-flood season, general flood season, and main flood season—the Markov time-series state transition probability matrix is ​​formed. P 1; Step 3.2: Using the benchmark hydroelectric power station B Using typical runoff scenarios as conditions, statistically analyze the conditional probabilities of typical scenarios involving externally related source loads occurring during historical ten-day periods. P 2= P ij (2) Establish benchmark hydroelectric power stations B Runoff and external associated source loads G Multi-scenario conditional probability model of power; Step 3.3: Using the benchmark hydroelectric power station B Using typical runoff scenarios as a condition, statistics were compiled on other dispatchable hydropower stations during historical ten-day periods. Y Conditional probability of typical runoff scenarios P 3= P ij (3) Build a hydroelectric power station B Runoff and other dispatchable hydropower stations Y Multi-scenario conditional probability models for runoff volume.

5. The method according to claim 4, characterized in that, In step 4, based on the conditional probability model and transition probability matrix obtained in step 3, the annual time series scenario of the benchmark hydropower station runoff in future years is randomly simulated. Based on this, and combined with the multi-scenario conditional probability model of external associated source load power and other dispatchable hydropower station runoff, the annual time series scenario of external associated source load power and other dispatchable hydropower station runoff under the benchmark hydropower station runoff scenario is randomly simulated and generated. The obtained annual time-series simulation scenario set was further reduced using the K-means method, and aggregated into typical annual time-series simulation scenarios of the benchmark hydropower station runoff, external associated source load power, and runoff of other dispatchable hydropower stations under the flood, normal, and dry water inflow schemes.

6. The method according to claim 5, characterized in that, In step 4, obtaining the annual time series scene specifically includes the following steps: Step 4.1: Introduce the benchmark hydropower station from Step 3. B Markov time series transition probability matrix for typical runoff scenarios External related source load G Multi-scenario conditional probability matrix of power The remaining dispatchable hydropower stations Y Conditional probability matrix of typical runoff scenarios , B Typical scenarios of ten-day runoff of hydropower stations Typical scenarios for the ten-day runoff of other dispatchable hydropower stations superscript Q ( t ) represents the first t The different levels of abundance and scarcity corresponding to the ten-day period (non-flood season, general flood season, and main flood season) and the set of externally related source load power scenarios for each ten-day period. ,in G Representing different source payload types; Step 4.2: Use stochastic simulation to simulate typical time-series scenarios of runoff and source load power in future years.

7. A method for establishing a multi-scenario conditional probability model, characterized in that, The established multi-scenario conditional probability model is a multi-scenario conditional probability model of runoff and source load daily power curves, including the following steps: based on hydropower stations B Historical time series of typical ten-day runoff scenarios were used to establish Markov time series state transition probability models between ten-day periods; taking hydropower stations as an example. B Using typical runoff scenarios as conditions, the conditional probabilities of occurrence of typical runoff scenarios from external related sources and loads and other hydropower stations during historical ten-day periods are statistically analyzed to establish a hydropower station B Multi-scenario conditional probability model of runoff and external associated source load power and hydropower station B A conditional probability model relating runoff to runoff from other associated hydropower stations.

8. The method for establishing a multi-scenario conditional probability model according to claim 7, characterized in that, The method specifically includes the following steps: Step 3.1: In the statistically based dispatchable hydropower stations, select the hydropower station for the scenario analysis. B Using this as a baseline, and based on typical scenarios of ten-day runoff during three different periods—non-flood season, general flood season, and main flood season—the Markov time-series state transition probability matrix is ​​formed. P 1; （3） In the formula, the benchmark hydropower station B The number of clusters in the ten-day runoff scenario is k After statistical clustering, the typical scene categories in each historical ten-day period were analyzed. i and j Number of transfers between N ij Typical scenario categories i The number of times it appears is N i The transition probability matrix consists of elements that are probability values ​​between the interval [0,1], and the sum of the elements in each row is equal to 1. Step 3.2: Using the benchmark hydroelectric power station B Using typical runoff scenarios as conditions, statistically analyze the conditional probabilities of typical scenarios involving externally related source loads occurring during historical ten-day periods. P 2= P ij (2) Establish benchmark hydroelectric power stations B Runoff and external associated source loads G Multi-scenario conditional probability model of power; （4） In the formula, each element in the conditional probability matrix lies between the interval [0,1], and the sum of the conditional probabilities in each row or column of the matrix is ​​equal to 1. k For hydroelectric power station B Total number of typical scenarios for runoff m This represents the total number of typical scenario categories for externally related source load power in hydropower stations. B Typical runoff scenarios i Given that this event has already occurred, define a typical scenario for the power of associated source and load. j The conditional probability of the event occurring is P ij (2) Its calculation method is as described above, where M i For hydroelectric power station B Typical scenarios for runoff i Number of times it appears in all historical periods M ij For the historical Xunzhong Hydropower Station B Typical scenarios for runoff are: i And the typical scenario category for the external associated source load power of the corresponding ten-day period is: j The number of times; Step 3.3: Using the benchmark hydroelectric power station B Using typical runoff scenarios as a condition, statistics were compiled on other dispatchable hydropower stations during historical ten-day periods. Y Conditional probability of typical runoff scenarios P 3= P ij (3) Build a hydroelectric power station B Runoff and other dispatchable hydropower stations Y Multi-scenario conditional probability models for runoff; （5） In the formula, each element in the conditional probability matrix lies between the interval [0,1], and the sum of the conditional probabilities in each row or column of the matrix is ​​equal to 1. k For hydroelectric power station B Total number of typical scenarios for runoff s For the remaining hydroelectric power stations Y The total number of runoff scenario categories in hydropower stations B Typical runoff scenarios i Given that this event has already occurred, define the remaining hydroelectric power stations. Y Typical scenarios of runoff j The conditional probability of the event occurring is P ij (3) Its calculation method is as described above, where L i For hydroelectric power station B Typical scenarios for runoff i Number of times it appears in all historical periods L ij For the historical Xunzhong Hydropower Station B Typical scenarios for runoff are: i And the remaining hydropower stations corresponding to the ten-day period Y Typical scenarios for runoff are: j The number of times.

Images