Multi-heterogeneous energy power system optimization scheduling method based on time-space correlation scene

Abstract

The invention discloses a space-time correlation scene-based multi-heterogeneous energy power system optimization scheduling method, and relates to the field of power scheduling. The existing method is difficult to ensure time correlation and spatial correlation at the same time. Collecting random variable data in a power system on the same day, and establishing a probability distribution model of random variables; sampling the model by adopting a sampling method, and generating a static scene matrix of a random variable at the initial moment of the next day; calculating the weight of each static scene; and selecting a static scene corresponding to a weight greater than or equal to a threshold value, constructing a dynamic scene model of a full time period of the next day, further establishing a total expected cost minimum probability optimization scheduling model of each time period in the next day, establishing a constraint condition to constrain the model, and solving the model by adopting a solver to obtain a minimum probability optimization scheduling model of the total expected cost of each time period in the next day. And obtaining active power and reactive power of an adjustable generator set in the power system, active power and reactive power of a main transformer network and active power of charging and discharging of an energy storage device. The method is used for optimal scheduling of the power system.

Description

Technical Field

[0001] This invention relates to the field of power dispatching. Background Technology

[0002] Due to various internal random factors or external environmental influences such as transmission line voltage levels, nonlinear loads, and the intermittency, randomness, and volatility of renewable energy generation, there are many parameter values ​​in the power system that cannot be accurately estimated. These parameters are called random variables.

[0003] In power systems, to address the resulting uncertainties and reduce their negative impacts, scientists and engineers have developed numerous techniques to estimate future random variable parameters. Accurate estimation of these parameters is crucial for developing optimal power system scheduling schemes.

[0004] Currently, five main methods are used to estimate the parameters of future random variables: probabilistic methods, likelihood methods, interval analysis methods, robust optimization, and information gap decision theory. Probabilistic methods describe uncertainty through probability distribution functions and are widely used in power system planning and long-term operation, but simulations are time-consuming. Likelihood methods obtain membership functions for uncertain factors and describe them in a fuzzy manner, making them more suitable for practical power systems, but they cannot handle the correlation of random variables. Interval analysis methods assume that the values ​​of uncertain factors are taken from specific intervals to find the boundaries of the output variables, but the results are overly conservative. Robust optimization focuses on the worst-case scenario, seeking solutions that perform well in all realizations of uncertain factors, leading to overly conservative results. Unlike robust optimization, information gap decision theory takes both the worst and best outcomes brought about by uncertainty to make appropriate and reasonable decisions.

[0005] Existing methods struggle to simultaneously guarantee temporal and spatial correlations, resulting in poor accuracy in estimating future random variable parameters. Summary of the Invention

[0006] The purpose of this invention is to address the problem that existing methods cannot simultaneously guarantee temporal and spatial correlation, leading to poor accuracy in estimating future random variable parameters. This invention proposes an optimal scheduling method for multi-heterogeneous energy power systems based on spatiotemporal correlation scenarios.

[0007] A method for optimal scheduling of multi-heterogeneous energy power systems based on spatiotemporal correlation scenarios, the method comprising the following:

[0008] Step 1: Modeling the probability distribution of random variables: Collect random variable data from the power system on the same day and establish a probability distribution model for the random variables;

[0009] Step 2, Static Scene Generation: The probability distribution model of the random variables is sampled using a sampling method to generate a static scene matrix of the random variables at the initial time of the next day.

[0010] Static scene probability: Calculate the weight of each static scene in each static scene matrix;

[0011] Static scene reduction: Select the static scene corresponding to the weight that is greater than or equal to the threshold from all the weights of the static scenes;

[0012] Dynamic scene generation: Construct a dynamic scene model for the entire time period of the next day based on all selected static scenes. The dynamic scene model for the entire time period of the next day includes random variables for the entire time period of the next day.

[0013] Step 3: Based on the dynamic scenario model and the equipment in the power system, establish a minimum probability optimization scheduling model for the total expected cost in each time period of the next day, and establish constraints on the model. Solve the model using a solver to obtain the active and reactive power of the adjustable generator sets, the active and reactive power of the main transformer going offline, and the active power of the energy storage device charging and discharging. Based on this power, schedule the adjustable generator sets, the main transformer going offline, and the energy storage device for the next day.

[0014] Preferably, in step 1, the random variable in the power system refers to the wind speed or solar radiation intensity of multiple wind turbines in the transmission or distribution network.

[0015] Preferably, the sampling method is the Latin hypercube sampling method, the simple random sampling method, or the sampling method based on a low-bias sequence.

[0016] Preferably, in step 1, the probability distribution model of the random variable includes a joint cumulative distribution function and a joint probability density function;

[0017] Joint cumulative distribution function:

[0018] Formula 1,

[0019] In the formula, For the joint cumulative distribution function, For Copula functions, , , For the first The kernel density estimator of the probability density function of a random variable. , , The Gaussian kernel function is used as the kernel smoothing function. Let be the cumulative distribution function of the Gaussian distribution. For data volume, For bandwidth, For the first A random variable, , , For data points, This is the lower bound of the dataset. This is the upper bound of the dataset. For data points Regarding the lower realm The reflection point, For data points Regarding the Upper Realm The reflection point;

[0020] Joint probability density function:

[0021] Formula 2,

[0022] In the formula, Let be the joint probability density function. For Copula density function, , For the first The kernel density estimator of the probability density function of a random variable.

[0023] Preferably, in step 2, the process of generating the static scene matrix is ​​as follows:

[0024] The probability distribution model of the random variable is sampled using a sampling method to obtain a sample matrix, and the static scene matrix is ​​obtained based on the sample matrix.

[0025] Preferably, the sample matrix for:

[0026] Formula 3,

[0027] In the formula, These are the elements in the sample matrix;

[0028] Static scene matrix for:

[0029] Formula 4,

[0030] In the formula, , .

[0031] Preferably, in step 2, the weight of each static scene is:

[0032] Formula 5,

[0033] In the formula, For the first The weight of each static scene, .

[0034] Preferably, the threshold is .

[0035] Preferably, the process of obtaining the dynamic scene model is as follows:

[0036] Establish a multiple linear regression equation:

[0037] Formula 6,

[0038] In the formula, For time The static scene matrix below, For multiple linear regression parameters, For error parameters, This represents the static scene matrix at the initial moment;

[0039] Regarding time The next The dynamic scenario is as follows:

[0040] Formula 7,

[0041] In the formula, For time The nth dynamic scenario with the mth variable as the current variable.

[0042] Regarding time The The dynamic scene transformation is as follows:

[0043] Formula 8,

[0044] In the formula, for Static scene matrix, For one The parameter matrix, For one Error matrix;

[0045] The first at each moment One dynamic scenario:

[0046] Formula 8,

[0047] No. The multiple linear regression model for each scenario is as follows:

[0048] Formula 9,

[0049] In the formula, for The dynamic temporal scene matrix, for The parameter moments, for The error matrix;

[0050] Corresponding time-series dynamic scenarios for:

[0051] Formula 10,

[0052] In the formula, for The estimated value.

[0053] Preferably, the minimum probability optimization scheduling model for the total expected cost of each time period in the next day is as follows:

[0054] Formula 11,

[0055] In the formula, For scenario probabilities, This is the network loss cost coefficient. For the total number of nodes, The penalty coefficient for wind and solar power curtailment. A collection of renewable energy generator sets, including wind power and photovoltaic power. To account for the amount of wind and solar power curtailed, This is the load shedding penalty factor. For node load shedding, For nodes During the period The voltage amplitude, Let be the real part of each element in the nodal admittance matrix. Let be the phase angle difference between nodes ij;

[0056] The constraints are:

[0057] ,

[0058] ,

[0059] ,

[0060] ,

[0061] ,

[0062] ,

[0063] ,

[0064] In the formula, For nodes The wind farm during the time period active power, For nodes During the period Active power load, For nodes During the period voltage amplitude, Let be the imaginary part of the elements in the nodal admittance matrix. For nodes Adjustable generator sets during time periods reactive power, The main change of the net during the time period reactive power, For nodes The wind farm during the time period reactive power, For nodes During the period reactive power load, The maximum downhill climbing speed of the adjustable generator set i. The maximum uphill capacity rate of adjustable generator set i. For nodes Adjustable generator sets during time periods -1 active power, For nodes The lower limit of the active power of the adjustable generator set. For nodes The upper limit of the active power of the adjustable generator set. For nodes The lower limit of reactive power of the adjustable generator set. For nodes The upper limit of reactive power of the adjustable generator set. For nodes During the period voltage amplitude, For nodes During the period The upper limit of voltage amplitude, For nodes During the period The lower limit of the voltage amplitude; the wind farm output is , , The rated active power of the wind farm, For the phase angle of the wind farm, To get to the random variable, For a given random variable, To remove random variables; photovoltaic output is , , The power factor angle, The output power of the photovoltaic module under standard test conditions. Solar irradiance under standard test conditions. The temperature coefficient of photovoltaic power. This refers to the actual operating temperature of the photovoltaic cells. Battery temperature under standard test conditions. The total efficiency of the photovoltaic system;

[0065] Energy storage output constraints:

[0066] ,

[0067] ,

[0068] ,

[0069] ,

[0070] ,

[0071] ,

[0072] ,

[0073] ,

[0074] In the formula, Let t be the state of charge of the stored energy. For energy storage self-discharge rate, The time step for scheduling, , The energy storage charging and discharging power at time t. For energy storage charging efficiency, For energy storage and discharge efficiency, For the rated capacity of the energy storage system, , Rated charging / discharging power, , This represents the lower / upper limit of the energy storage state of charge. For energy storage charging and discharging state variables.

[0075] The beneficial effects of this invention are:

[0076] This invention establishes a probability distribution model of random variables, then uses a sampling method to sample the model to generate static scenarios, calculates the weight of each static scenario to preserve spatial correlation, and then uses multiple linear regression to generate dynamic scenarios, taking into account temporal correlation. Therefore, this invention accurately characterizes the spatiotemporal scenario generation method that depicts the spatial interaction and continuous temporal evolution of uncertain factors, improves the performance of power system optimization scheduling, and provides a more reliable decision-making basis for the economic and safe scheduling of power systems. Attached Figure Description

[0077] Figure 1 A flowchart of an optimized scheduling method for multi-heterogeneous energy power systems based on spatiotemporal correlation scenarios;

[0078] Figure 2 This is a distribution diagram of sample points in two-dimensional space for five sampling methods. Detailed Implementation

[0079] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0080] It should be noted that, unless otherwise specified, the embodiments and features described in the present invention can be combined with each other. The present invention will be further described below with reference to the accompanying drawings and specific embodiments, but this is not intended to limit the scope of the invention.

[0081] Example:

[0082] A method for optimal scheduling of multi-heterogeneous energy power systems based on spatiotemporal correlation scenarios, the method comprising the following:

[0083] Step 1: Modeling the probability distribution of random variables: Collect random variable data from the power system on the same day and establish a probability distribution model for the random variables;

[0084] Step 2, Static Scene Generation: The probability distribution model of the random variables is sampled using a sampling method to generate a static scene matrix of the random variables at the initial time of the next day.

[0085] Static scene probability: Calculate the weight of each static scene in each static scene matrix;

[0086] Static scene reduction: Select the static scene corresponding to the weight that is greater than or equal to the threshold from all the weights of the static scenes;

[0087] Dynamic scene generation: Construct a dynamic scene model for the entire time period of the next day based on all selected static scenes. The dynamic scene model for the entire time period of the next day includes random variables for the entire time period of the next day.

[0088] Step 3: Based on the dynamic scenario model and the equipment in the power system, establish a minimum probability optimization scheduling model for the total expected cost in each time period of the next day, and establish constraints on the model. Solve the model using a solver to obtain the active and reactive power of the adjustable generator sets, the active and reactive power of the main transformer going offline, and the active power of the energy storage device charging and discharging. Based on this power, schedule the adjustable generator sets, the main transformer going offline, and the energy storage device for the next day.

[0089] Specifically, the threshold is Multi-heterogeneous energy refers to energy sources such as wind, solar, energy storage, and thermal power that have significant differences in energy source, control, and output characteristics, and are connected to the power grid together.

[0090] If the random variable is wind speed, this embodiment is equivalent to using today's wind speed data to generate a static scene at 0:00 tomorrow, and then generating a dynamic scene for the next 24 hours.

[0091] To further define, random variable data in a power system refers to the wind speed or solar irradiance of multiple wind turbines in a transmission or distribution network.

[0092] The sampling method is Latin hypercube sampling, simple random sampling, or sampling method based on low-biased sequences.

[0093] Specifically, Figure 2 The figure shows the distribution of sample points in two-dimensional space for different sampling methods. The vertical axis N represents the number of sample points. Figure 2 As can be seen, the sampling points of the five sampling methods are evenly distributed. Therefore, this embodiment can be combined with any one of the methods.

[0094] To further specify, in step 1, the probability distribution model of the random variable includes the joint cumulative distribution function and the joint probability density function;

[0095] Joint cumulative distribution function:

[0096] Formula 1,

[0097] In the formula, For the joint cumulative distribution function, For Copula functions, , , For the first The kernel density estimator of the probability density function of a random variable. , , The Gaussian kernel function is used as the kernel smoothing function. Let be the cumulative distribution function of the Gaussian distribution. For data volume, For bandwidth, For the first A random variable, , , For data points, This is the lower bound of the dataset. This is the upper bound of the dataset. For data points Regarding the lower realm The reflection point, For data points Regarding the Upper Realm The reflection point;

[0098] Joint probability density function:

[0099] Formula 2,

[0100] In the formula, Let be the joint probability density function. For Copula density function, , For the first The kernel density estimator of the probability density function of a random variable.

[0101] Specifically, common Copulas can be divided into elliptical Copulas and Archimedes Copulas, among which there are many families of Copulas to choose from to achieve optimal performance.

[0102] Further specifying step 2, the process of generating the static scene matrix:

[0103] The probability distribution model of the random variable is sampled using a sampling method to obtain a sample matrix, and the static scene matrix is ​​obtained based on the sample matrix.

[0104] Sample matrix for:

[0105] Formula 3,

[0106] In the formula, These are the elements in the sample matrix;

[0107] Static scene matrix for:

[0108] Formula 4,

[0109] In the formula, , .

[0110] Further specifying, in step 2, the weight of each static scene is:

[0111] Formula 5,

[0112] In the formula, For the first The weight of each static scene, .

[0113] Specifically, due to spatial correlation, static scenes cannot be generated simply using the sampling methods described above. To address spatial correlation, importance sampling theory is employed to transform the original distribution into the desired joint distribution of independent multivariates.

[0114] Typically, probabilistic properties are defined by mathematical expectation, such as mean, variance, skewness, and kurtosis. The corresponding equations are described below:

[0115] Formula 12

[0116] Formula 13

[0117] Formula 14

[0118] Formula 15

[0119] At the same time, optimal power flow can be expressed as a multivariable function of random variables:

[0120] Formula 16

[0121] If some probabilistic properties we are interested in are the optimal power flow results The expectation of H(·) of a certain function:

[0122] Formula 17

[0123] in It is the joint probability density function of random variables. Introducing the Copula probability density function, and transforming the integrand into:

[0124] Formula 18

[0125] Therefore, this probability feature It can be regarded as:

[0126] Formula 19

[0127] The expected value, whose probability density function is:

[0128] Formula 20

[0129] Although the original random variable It hasn't changed, but the probability features are being calculated. When they are independent, they can be considered to be independent of each other, because the joint probability density function is the product of the marginal probability density functions of all random variables.

[0130] The estimated value of this integral can be expressed as:

[0131] Formula 21

[0132] Introducing Formula 19:

[0133] Formula 22

[0134] Here and They are approximately equal, therefore It can be replaced with a constant close to 1.

[0135] Formula 23

[0136] In order to calculate value, Assume it is a constant.

[0137] Formula 24

[0138] so, for:

[0139] Formula 25

[0140] Then, combining formulas 23 and 25, the probability characteristics... It can be represented as a weighted sum

[0141] Formula 26

[0142] Based on the structure of Formula 26, the weights This can be viewed as the probability of a scenario occurring, and can be further described as...

[0143] Formula 27

[0144] In practical applications, most probabilities typically occur in a few scenarios. Therefore, we can eliminate scenarios with low probabilities and retain those with high probabilities without significantly affecting the final result. Here, we define valid scenarios that meet the following conditions:

[0145] Formula 28

[0146] Retain all valid scenarios and discard the rest. Use this in Equations 26 and 27. and replace and , representing the number and probability of static scenes after scene reduction, respectively.

[0147] By utilizing the Copula joint probability density function, importance sampling first constructs a function. — This function has the same probabilistic characteristics as the original variable. The probability density function and the marginal probability density function are used, but the random variables are relatively independent. Subsequently, various sampling methods are combined with inverse transforms to generate static scenes, which are then weighted and summed. An estimation is performed. Finally, a simple scenario reduction method based on probability is proposed, which significantly reduces computational complexity while maintaining the highest accuracy. The Copula-importance sampling method cleverly circumvents the sampling challenge of non-independent random variables in high-dimensional spaces.

[0148] To further define the process, the dynamic scene model acquisition process is as follows:

[0149] Establish a multiple linear regression equation:

[0150] Formula 6,

[0151] In the formula, For time The static scene matrix below, For multiple linear regression parameters, For error parameters, This represents the static scene matrix at the initial moment;

[0152] Regarding time The next The dynamic scenario is as follows:

[0153] Formula 7,

[0154] In the formula, For time The nth dynamic scenario with the mth variable as the current variable.

[0155] Regarding time The The dynamic scene transformation is as follows:

[0156] Formula 8,

[0157] In the formula, for Static scene matrix, For one The parameter matrix, For one Error matrix;

[0158] The first at each moment One dynamic scenario:

[0159] Formula 8,

[0160] No. The multiple linear regression model for each scenario is as follows:

[0161] Formula 9,

[0162] In the formula, for The dynamic temporal scene matrix, for The parameter moments, for The error matrix;

[0163] Corresponding time-series dynamic scenarios for:

[0164] Formula 10,

[0165] In the formula, for The estimated value.

[0166] Specifically, The process of obtaining it is as follows:

[0167] Formula 11,

[0168] In the formula, , For the residual vector, For the reason The calculated dynamic scene;

[0169] Differentiating Equation 11, we obtain:

[0170] Formula 12,

[0171] In the formula, For matrix The OK, For time Historical data matrix.

[0172] Multiple linear regression extends static scenarios to dynamic scenarios that include spatiotemporal correlations.

[0173] Static scenes only reflect the spatial correlation of random variables, while dynamic scenes consider both spatial and temporal correlations. For dynamic scene generation that requires simultaneous consideration of both spatial and temporal correlations, sampling methods suffer from the curse of dimensionality and fail to converge. This paper extends the proposed static scene generation method by generalizing static scenes to dynamic scenes to capture temporal correlations.

[0174] Based on the initial random variable data, a spatially correlated static scene is first generated through a sampling method, where the weights can be considered as probabilities. Without loss of generality, it is assumed here that the initial time of the dynamic time-series scene is the same as that of the static scene.

[0175] First, kernel density estimation and the Copula function are used to model the joint distribution of random variables. Then, assuming the random variables are independent, various sampling methods, including simple random sampling, Latin hypercube sampling, and sampling based on low-biased sequences, are used to generate static scenarios. Considering spatial correlation, importance sampling transforms the non-independent joint distribution into an independent distribution by adding weight factors to each static scenario. Multiple linear regression is used to generalize these static scenarios to dynamic time-series scenarios. Finally, the multi-time-period uncertain optimization scheduling problem is formulated as a probabilistic deterministic optimization problem using all dynamic scenarios, where the weight factors represent the corresponding probabilities.

[0176] Further defining the minimum probability optimization scheduling model for the total expected cost of each time period in the next day, we have:

[0177] Formula 11,

[0178] In the formula, For scenario probabilities, This is the network loss cost coefficient. For the total number of nodes, The penalty coefficient for wind and solar power curtailment. A collection of renewable energy generator sets, including wind power and photovoltaic power. To account for the amount of wind and solar power curtailed, This is the load shedding penalty factor. For node load shedding, For nodes During the period The voltage amplitude, Let be the real part of each element in the nodal admittance matrix. Let be the phase angle difference between nodes ij;

[0179] The constraints are:

[0180] ,

[0181] ,

[0182] ,

[0183] ,

[0184] ,

[0185] ,

[0186] ,

[0187] In the formula, For nodes The wind farm during the time period active power, For nodes During the period Active power load, For nodes During the period voltage amplitude, Let be the imaginary part of the elements in the nodal admittance matrix. For nodes Adjustable generator sets during time periods reactive power, The main change of the net during the time period reactive power, For nodes The wind farm during the time period reactive power, For nodes During the period reactive power load, The maximum downhill climbing speed of the adjustable generator set i. The maximum uphill capacity rate of adjustable generator set i. For nodes Adjustable generator sets during time periods -1 active power, For nodes The lower limit of the active power of the adjustable generator set. For nodes The upper limit of the active power of the adjustable generator set. For nodes The lower limit of reactive power of the adjustable generator set. For nodes The upper limit of reactive power of the adjustable generator set. For nodes During the period voltage amplitude, For nodes During the period The upper limit of voltage amplitude, For nodes During the period The lower limit of the voltage amplitude; the wind farm output is , , The rated active power of the wind farm, For the phase angle of the wind farm, To get to the random variable, For a given random variable, To remove random variables; photovoltaic output is , , The power factor angle, The output power of the photovoltaic module under standard test conditions. Solar irradiance under standard test conditions. The temperature coefficient of photovoltaic power. This refers to the actual operating temperature of the photovoltaic cells. Battery temperature under standard test conditions. The total efficiency of the photovoltaic system;

[0188] Energy storage output constraints:

[0189] ,

[0190] ,

[0191] ,

[0192] ,

[0193] ,

[0194] ,

[0195] ,

[0196] ,

[0197] In the formula, Let t be the state of charge of the stored energy. For energy storage self-discharge rate, For the time step of scheduling, , The energy storage charging and discharging power at time t. For energy storage charging efficiency, For energy storage and discharge efficiency, For the rated capacity of the energy storage system, , Rated charging / discharging power, , This represents the lower / upper limit of the energy storage state of charge. For energy storage charging and discharging state variables.

[0198] Specifically, =1 indicates charging. =0 indicates discharge.

[0199] The final scheduling strategy obtained in this embodiment is the active and reactive power of the adjustable generator sets in the power system, the active and reactive power of the main transformer going offline, and the active power of the energy storage device charging and discharging.

[0200] When the random variable data in step 1 is wind speed, This is expressed as wind speed; when the random variable data in step 1 is the actual solar irradiance, This represents the actual solar irradiance.

[0201] The goal of the probability optimization scheduling model for minimizing the total expected cost for each time period of the next day is to minimize the cost, namely, to minimize network loss, wind and solar curtailment penalties, and load shedding penalties.

[0202] While the invention has been described herein with reference to specific embodiments, it should be understood that these embodiments are merely examples of the principles and applications of the invention. Therefore, it should be understood that many modifications can be made to the exemplary embodiments, and other arrangements can be designed without departing from the spirit and scope of the invention as defined by the appended claims. It should be understood that different dependent claims and features described herein can be combined in ways different from those described in the original claims. It is also understood that features described in conjunction with individual embodiments can be used in other described embodiments.

Claims

1. A method for optimal scheduling of multi-heterogeneous energy power systems based on spatiotemporal correlation scenarios, characterized in that, The method includes the following: Step 1: Modeling the probability distribution of random variables: Collect random variable data from the power system on the same day and establish a probability distribution model for the random variables; Step 2, Static Scene Generation: The probability distribution model of the random variables is sampled using a sampling method to generate a static scene matrix of the random variables at the initial time of the next day. Static scene probability: Calculate the weight of each static scene in each static scene matrix; Static scene reduction: Select the static scene corresponding to the weight that is greater than or equal to the threshold from all the weights of the static scenes; Dynamic scene generation: Construct a dynamic scene model for the entire time period of the next day based on all selected static scenes. The dynamic scene model for the entire time period of the next day includes random variables for the entire time period of the next day. Step 3: Based on the dynamic scenario model and the equipment in the power system, establish a minimum probability optimization scheduling model for the total expected cost in each time period of the next day, and establish constraints on the model. Solve the model using a solver to obtain the active and reactive power of the adjustable generator sets, the active and reactive power of the main transformer going offline, and the active power of the energy storage device charging and discharging. Based on this power, schedule the adjustable generator sets, the main transformer going offline, and the energy storage device for the next day.

2. The method for optimal scheduling of multi-heterogeneous energy power systems based on spatiotemporal correlation scenarios according to claim 1, characterized in that, In step 1, the random variables in the power system refer to the wind speed or solar radiation intensity of multiple wind turbines in the transmission or distribution network.

3. The method for optimal scheduling of multi-heterogeneous energy power systems based on spatiotemporal correlation scenarios according to claim 1, characterized in that, The sampling method is Latin hypercube sampling, simple random sampling, or sampling method based on low-biased sequences.

4. The method for optimal scheduling of multi-heterogeneous energy power systems based on spatiotemporal correlation scenarios according to claim 1, characterized in that, In step 1, the probability distribution model of the random variable includes the joint cumulative distribution function and the joint probability density function; Joint cumulative distribution function: Official 1, In the formula, For the joint cumulative distribution function, For Copula functions, , , For the first The kernel density estimator of the probability density function of a random variable. , , The Gaussian kernel function is used as the kernel smoothing function. Let be the cumulative distribution function of the Gaussian distribution. For data volume, For bandwidth, For the first A random variable, , , For data points, This is the lower bound of the dataset. This is the upper bound of the dataset. For data points Regarding the lower realm The reflection point, For data points Regarding the Upper Realm The reflection point; Joint probability density function: Official 2, In the formula, Let be the joint probability density function. For Copula density function, , For the first The kernel density estimator of the probability density function of a random variable.

5. The method for optimal scheduling of multi-heterogeneous energy power systems based on spatiotemporal correlation scenarios according to claim 4, characterized in that, Step 2, the process of generating the static scene matrix: The probability distribution model of the random variable is sampled using a sampling method to obtain a sample matrix, and the static scene matrix is ​​obtained based on the sample matrix.

6. The method for optimal scheduling of multi-heterogeneous energy power systems based on spatiotemporal correlation scenarios according to claim 5, characterized in that, Sample matrix for: Official 3, In the formula, The elements in the sample matrix; Static scene matrix for: Official 4, In the formula, , .

7. The method for optimal scheduling of multi-heterogeneous energy power systems based on spatiotemporal correlation scenarios according to claim 6, characterized in that, In step 2, the weight of each static scene is: Official 5, In the formula, For the first The weight of each static scene, .

8. The method for optimal scheduling of multi-heterogeneous energy power systems based on spatiotemporal correlation scenarios according to claim 1, characterized in that, Threshold is .

9. The method for optimal scheduling of multi-heterogeneous energy power systems based on spatiotemporal correlation scenarios according to claim 7, characterized in that, The process of obtaining the dynamic scene model is as follows: Establish a multiple linear regression equation: Official 6, In the formula, For time The static scene matrix below, For multiple linear regression parameters, For error parameters, This represents the static scene matrix at the initial moment; Regarding time The next The dynamic scenario is as follows: Official 7, In the formula, For time The nth dynamic scenario with the mth variable as the current variable. Regarding time The The dynamic scene transformation is as follows: Official 8, In the formula, for Static scene matrix, For one The parameter matrix, For one Error matrix; The first at each moment One dynamic scenario: Official 8, No. The multiple linear regression model for each scenario is as follows: Official 9, In the formula, for The dynamic temporal scene matrix, for The parameter moments, for The error matrix; Corresponding time-series dynamic scenarios for: Official 10, In the formula, for The estimated value.

10. The method for optimal scheduling of multi-heterogeneous energy power systems based on spatiotemporal correlation scenarios according to claim 9, characterized in that, The minimum probability optimization scheduling model for the total expected cost of each time period in the next day is as follows: Official 11, In the formula, For scenario probabilities, This is the network loss cost coefficient. For the total number of nodes, The penalty coefficient for wind and solar power curtailment. A collection of renewable energy generator sets, including wind power and photovoltaic power. To account for the amount of wind and solar power curtailed, This is the load shedding penalty factor. For node load shedding, For nodes During the period The voltage amplitude, Let be the real part of each element in the nodal admittance matrix. Let be the phase angle difference between nodes ij; The constraints are: ， ， ， ， ， ， ， In the formula, For nodes The wind farm during the time period active power, For nodes During the period Active power load, For nodes During the period voltage amplitude, Let be the imaginary part of the elements in the nodal admittance matrix. For nodes Adjustable generator sets during time periods reactive power, The main change of the net during the time period reactive power, For nodes The wind farm during the time period reactive power, For nodes During the period reactive power load, The maximum downhill climbing speed of the adjustable generator set i. The maximum uphill capacity rate of adjustable generator set i. For nodes Adjustable generator sets during time periods -1 active power, For nodes The lower limit of the active power of the adjustable generator set. For nodes The upper limit of the active power of the adjustable generator set. For nodes The lower limit of reactive power of the adjustable generator set. For nodes The upper limit of reactive power of the adjustable generator set. For nodes During the period voltage amplitude, For nodes During the period The upper limit of voltage amplitude, For nodes During the period The lower limit of the voltage amplitude; the wind farm output is , , The rated active power of the wind farm, For the phase angle of the wind farm, To get to the random variable, For a given random variable, To remove random variables; photovoltaic output is , , The power factor angle, The output power of the photovoltaic module under standard test conditions. Solar irradiance under standard test conditions. The temperature coefficient of photovoltaic power. This refers to the actual operating temperature of the photovoltaic cells. Battery temperature under standard test conditions. The total efficiency of the photovoltaic system; Energy storage output constraints: ， ， ， ， ， ， ， ， In the formula, Let t be the state of charge of the stored energy. For energy storage self-discharge rate, The time step for scheduling, , The energy storage charging and discharging power at time t. For energy storage charging efficiency, For energy storage and discharge efficiency, For the rated capacity of the energy storage system, , Rated charging / discharging power, , This represents the lower / upper limit of the energy storage state of charge. For energy storage charging and discharging state variables.

Images