A multi-temporal and spatial power probability modeling method and system for a new energy station
By introducing the assumptions of predicted power and conditional independence in new energy power plants, and using the copula function to transform multi-temporal probability modeling into a product of single power plant and single time point, the problem of high complexity in high-dimensional modeling is solved, and efficient power probability prediction and scene generation are achieved.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- STATE GRID JILIN ELECTRIC POWER COMPANY LIMITED
- Filing Date
- 2023-01-09
- Publication Date
- 2026-06-12
AI Technical Summary
Existing technologies are difficult to effectively solve the probabilistic modeling of power generation at new energy power plants across multiple time and space, especially when considering spatiotemporal correlations. High-dimensional probabilistic modeling suffers from high computational complexity, operational difficulties, and unreasonable assumptions.
By introducing predicted power as a conditional variable, and using the conditional independence assumption and copula function, the conditional probability of multi-temporal and spatial power generation is expressed as the product of the conditional probabilities of power at a single power station and a single point in time. A multi-temporal and spatial power probability model of new energy power stations is constructed, and sampling is performed using a univariate probability distribution.
It reduces the time and space complexity of high-dimensional probabilistic modeling, simplifies the scene generation process, avoids the complexity of theoretical analysis and unreasonable assumptions, and improves modeling efficiency and accuracy.
Smart Images
Figure CN115983019B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of power technology, and more specifically, to a power probability modeling method and system for multi-temporal and spatial renewable energy power plants. Background Technology
[0002] In the field of power technology, probabilistic power prediction is a probabilistic characterization of power or prediction error to depict the uncertain information of renewable energy power. Probabilistic power prediction can support stochastic optimization of unit combination and spot market clearing. The probabilistic characterization of power mainly includes quantiles, confidence intervals, typical scenarios, and probability distributions of power or prediction error. However, because probability distributions are uniquely determined by random variables, and probability distributions can derive other information such as interval predictions and scenarios, they have received wider research and application. One approach to probabilistic modeling is to construct a parametric probabilistic model, making assumptions about the family of probability distributions beforehand, and using historical data to estimate the parameters of the family of distributions, such as a Gaussian probability model. Another approach is non-parametric methods, which do not require distribution assumptions, such as quantile point regression or kernel density estimation. Scenario generation is essentially based on sampling from probabilistic models. Currently, the difficulty in power probabilistic modeling and scenario generation lies in multi-temporal and spatial probabilistic modeling, which requires considering the spatiotemporal correlation of power, for example, probabilistic modeling of the power of all renewable energy power plants in a certain region over the next 24 hours (96 time points at 15-minute intervals).
[0003] In view of this, the present invention proposes a power probability modeling method and system for multi-temporal and spatiotemporal renewable energy power plants. By introducing predicted power as a conditional variable and reasonable conditional independence assumptions, the spatiotemporal conditional probability is expressed as the product of the power conditional probabilities of a single power plant and a single point in time. This not only reduces the time and space complexity of high-dimensional probability modeling operations, but also provides convenience for scene generation. Moreover, the present invention only involves sampling from the probability distribution of a single variable, avoiding the theoretical analysis difficulties, operational complexity, and unreasonable assumptions that may be involved in sampling from high-dimensional probability distributions. Summary of the Invention
[0004] The purpose of this invention is to provide a power probability modeling method for multi-temporal and spatial renewable energy power plants, characterized by comprising: obtaining a density function and a distribution function based on historical data of the renewable energy power plant; the density function including an actual power density function and a predicted power density function; the distribution function including an actual power distribution function and a predicted power distribution function; the historical data including historical predicted power and historical actual power; determining a copula function relating the historical predicted power and the historical actual power of the renewable energy power plant; and, based on the density function, the distribution function, and the copula function, determining the conditional probability density of the actual power of the renewable energy power plant under the predicted power condition, given the predicted power of the renewable energy power plant in the period to be measured.
[0005] Furthermore, it also includes: traversing and determining the conditional probability density of the actual power of each of the new energy power plants under the predicted power conditions; and obtaining the joint conditional probability density of the power of all the new energy power plants during the period to be measured based on the conditional probability densities of the multiple actual power conditions.
[0006] Furthermore, based on the conditional independence assumption and copula theory, the simplified expression for the conditional probability density of the actual power is as follows:
[0007]
[0008] in, It means A random variable representing actual power. It means The random variable representing the predicted power. It means This represents the value of the random variable representing actual power. It means Let represent the values of the random variable for predicted power, where i is the variable with values from 1 to n, n represents the number of renewable energy power plants, and j is the variable with values from 1 to T, where T represents the number of time periods to be predicted. Represents random variables The copula function, Represents the actual power distribution function. Represents the predicted power distribution function. Represents the actual power density function
[0009] Furthermore, it also includes sampling the conditional probability density of the actual power under a given predicted power condition to obtain the power scenario of the new energy power station, including: sampling the conditional probability density of the actual power under multiple given predicted power conditions to obtain multiple samples of the new energy power station in the test period; splicing the multiple samples to obtain a single sampling of the multiple new energy power stations in the test period; the result of the single sampling is a scenario of the new energy power station; repeating the single sampling to obtain multiple power scenarios.
[0010] Furthermore, the sampling of the conditional probability density of the actual power under multiple given predicted power conditions to obtain samples of multiple new energy power stations during the test period is as follows: randomly select p numbers from U(0,1) to find... Corresponding independent variable value Therefore, we obtain A single sample, i = 1, ..., n; j = 1, ..., T; where n represents the number of new energy power stations, T represents the number of time periods to be predicted, U(0,1) represents a uniform distribution on the interval [0,1], and p represents a randomly selected number. This indicates that the predicted power of the new energy power station is The conditional probability distribution of actual power under the given conditions; This indicates that the predicted power of the new energy power station is The actual power conditional probability density under given conditions.
[0011] Furthermore, the sampled sample is denoted as... By concatenating multiple samples, the expression for the first sampling is:
[0012] Furthermore, the historical data is processed by kernel density estimation to obtain the density function, and the distribution function is obtained by integrating the density function.
[0013] Furthermore, determining the copula function that correlates the historical predicted power and the historical actual power of the new energy power station includes: fitting the historical data to copula functions of various Archimedean families and / or various elliptic families; and using the copula function with the largest log-likelihood value in the fitting results as the copula function that correlates the historical predicted power and the historical actual power of the new energy power station.
[0014] The purpose of this invention is to provide a power probability modeling system for multi-temporal and spatial renewable energy power plants, including a density and distribution function determination module, a copula function determination module, and a conditional probability density determination module. The density and distribution function determination module is used to obtain a density function and a distribution function based on historical data of the renewable energy power plant. The density function includes an actual power density function and a predicted power density function. The distribution function includes an actual power distribution function and a predicted power distribution function. The historical data includes historical predicted power and historical actual power. The copula function determination module is used to determine the copula function associated with the historical predicted power and the historical actual power of the renewable energy power plant. The conditional probability density determination module is used to determine the conditional probability density of the actual power of the renewable energy power plant under the predicted power condition, given the predicted power of the renewable energy power plant during the period to be measured, based on the density function, the distribution function, and the copula function.
[0015] The technical solutions of the embodiments of the present invention have at least the following advantages and beneficial effects:
[0016] This invention uses predicted power and actual power to construct a conditional probability model of power at multiple spatiotemporal scales (i.e., a power prediction model). The introduction of predicted power makes the construction assumptions reasonable and consistent with reality. Thus, the high-dimensional spatiotemporal conditional probability can be transformed into a product of low-dimensional conditional probabilities, specifically a product of the power conditional probabilities of a single power plant and a single point in time, which facilitates the subsequent probabilistic characterization of power.
[0017] This invention represents the conditional probability involving time and space as the product of the power conditional probabilities of a single power station and a single point in time. This not only reduces the time and space complexity of high-dimensional probability modeling operations, but also provides convenience for scene generation.
[0018] This invention only involves sampling from a single-variable probability distribution, avoiding the theoretical analysis difficulties, operational complexity, and unreasonable assumptions that may be involved in sampling from a high-dimensional probability distribution. Attached Figure Description
[0019] Figure 1 An exemplary flowchart of a power probability modeling method for multi-temporal and spatial renewable energy power plants provided in some embodiments of the present invention;
[0020] Figure 2 This is an exemplary module diagram of a power probability modeling system for a multi-temporal and spatial renewable energy power station, provided for some embodiments of the present invention. Detailed Implementation
[0021] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, 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. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.
[0022] Figure 1 This is an exemplary flowchart illustrating a power probability modeling method for multi-temporal and spatial renewable energy power plants, provided in some embodiments of the present invention. The conditional independence assumptions are as follows: given a predicted power, the actual power of each wind farm and photovoltaic power station is independent of each other; given a predicted power for a certain power plant, the actual power of that power plant is independent of the predicted power of other renewable energy power plants; for a certain power plant, given predicted power at several time points, the actual power at each time point is independent of each other; and for a certain power plant, given a predicted power at a certain time point, the actual power at that time point is independent of the predicted power at other time points.
[0023] In some embodiments, process 100 may be executed by system 200. For example... Figure 1As shown, process 100 may include the following:
[0024] Step 110: Based on historical data from the renewable energy power plants, obtain the density function and distribution function. In some embodiments, step 110 can be performed by the density and distribution function determination module 210.
[0025] Historical data can refer to data related to the power output of renewable energy power plants prior to the current time. For example, historical data can include historical predicted power output and historical actual power output. Historical predicted power output can refer to the predicted power output of renewable energy power plants. Historical actual power output can refer to the power output of renewable energy power plants that was actually measured. In some embodiments, historical data can be acquired through various feasible methods, including but not limited to acquisition via a network or extraction from a storage module.
[0026] Density functions include actual power density functions and predicted power density function Distribution functions include actual power distribution functions and predicted power distribution function In some embodiments, historical data can be processed by kernel density estimation to obtain a density function, and the distribution function can be obtained by integrating the density function.
[0027] Step 120: Determine the copula (coupling) function that associates the historical predicted power and historical actual power of the renewable energy power plant. In some embodiments, step 120 can be performed by the copula function determination module 220.
[0028] In some embodiments, historical data can be fitted to copula functions of various Archimedean and / or elliptic families; the copula function with the largest log-likelihood value in the fitting results is used as the copula function for correlating the historical predicted power and historical actual power of renewable energy power plants.
[0029] Step 130: Based on the density function, distribution function, and copula function, predict the power of the new energy power station for the period to be measured. Determine the conditional probability density of the actual power of renewable energy power plants under the predicted power conditions. In some embodiments, step 130 may be performed by the conditional probability density determination module 230.
[0030] The time period to be measured can refer to the time period for which probabilistic modeling is required. For example, the time period to be measured can be time 1-T. Predicted power can refer to the predicted power output of the renewable energy power plant. Actual power can refer to the actual measured power output of the renewable energy power plant.
[0031] In some embodiments, the simplified expression for the conditional probability density of actual power, based on the conditional independence assumption and copula theory, is as follows:
[0032]
[0033] in, It means A random variable representing actual power. It means The random variable representing the predicted power. It means This represents the value of the random variable representing actual power. It means Let represent the values of the random variable for predicted power, where i is the variable with values from 1 to n, n represents the number of renewable energy power plants, and j is the variable with values from 1 to T, where T represents the number of time periods to be predicted. Represents random variables The copula function, The distribution function representing the actual power. The distribution function representing the predicted power. The density function representing actual power.
[0034] In some embodiments, the joint conditional probability density of multiple renewable energy power plants can also be determined based on the conditional probability density, including:
[0035] Step 140: Iterate through and determine the conditional probability density of the actual power of each renewable energy power station under the predicted power conditions. In some embodiments, step 140 may be performed by the joint conditional probability density determination module 240.
[0036] Step 150: Based on the conditional probability densities of multiple actual power sources, obtain the joint conditional probability density of all new energy power plants for the time period to be measured. In some embodiments, step 150 can be performed by the joint conditional probability density determination module 240.
[0037] In some embodiments, the method may further include step 160, sampling the conditional probability density of the actual power under a given predicted power condition to obtain the power scenario of the renewable energy power station. In some embodiments, step 160 may be executed by the power scenario determination module 250, including:
[0038] The conditional probability density of actual power under multiple given predicted power conditions is sampled to obtain samples from multiple new energy power plants during the test period. In some embodiments, a number p can be randomly drawn from U(0,1) to find... Corresponding independent variable value Therefore, we obtain A single sample, i = 1, ..., n; j = 1, ..., T; where n represents the number of new energy power stations, T represents the number of time periods to be predicted, U(0,1) represents a uniform distribution on the interval [0,1], and p represents a randomly selected number. This indicates that the predicted power of the new energy power station is The conditional probability distribution of actual power under the given conditions; This indicates that the predicted power of the new energy power station is The actual power conditional probability density under given conditions.
[0039] Multiple samples are concatenated to obtain a single sample of multiple new energy power stations during the testing period; the result of a single sample constitutes one scenario of the new energy power station. In some embodiments, the sampled sample is denoted as... By concatenating multiple samples, the expression for a single sampling is:
[0040] Repeat the above sampling process to obtain multiple power scenarios.
[0041] Figure 2 This is an exemplary block diagram of a power probability modeling system for multi-temporal and spatial renewable energy power plants, provided for some embodiments of the present invention. For example... Figure 2 As shown, system 200 includes a density and distribution function determination module 210, a copula function determination module 220, and a conditional probability density determination module 230.
[0042] The density and distribution function determination module 210 is used to obtain the density function and distribution function based on historical data from renewable energy power plants. The density function includes the actual power density function and the predicted power density function; the distribution function includes the actual power distribution function and the predicted power distribution function; the historical data includes historical predicted power and historical actual power. For more information on the density and distribution function determination module 210, please refer to [link to relevant documentation]. Figure 1 And its related descriptions.
[0043] The copula function determination module 220 is used to determine the copula function that correlates the historical predicted power and historical actual power of renewable energy power plants. For more information on the copula function determination module 220, please refer to [link to relevant documentation]. Figure 1 And its related descriptions.
[0044] The conditional probability density determination module 230 is used to determine the conditional probability density of the actual power of a renewable energy power plant under the predicted power conditions for the period to be measured, based on the density function, distribution function, and copula function. For more information on the conditional probability density determination module 230, please refer to [link to relevant documentation]. Figure 1 And its related descriptions.
[0045] In some embodiments, the system 200 further includes a joint conditional probability density determination module 240, which is used to iterate and determine the conditional probability density of the actual power of each renewable energy power station under the predicted power condition; based on the conditional probability densities of multiple actual power values, the joint conditional probability density of the actual power of all renewable energy power stations during the test period is obtained. For more information on the joint conditional probability density determination module 240, see [link to documentation]. Figure 1 And its related descriptions.
[0046] In some embodiments, the system 200 further includes a power scenario determination module 250, which is used to sample the conditional probability density of the actual power under a given predicted power condition to obtain the power scenario of the renewable energy power station. For more information on the power scenario determination module 250, see [link to relevant documentation]. Figure 1 And its related descriptions.
[0047] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.
Claims
1. A power probability modeling method for multi-temporal and spatial renewable energy power plants, characterized in that, include: Based on the historical data of the aforementioned renewable energy power plants, a density function and a distribution function are obtained; the density function includes an actual power density function and a predicted power density function; the distribution function includes an actual power distribution function and a predicted power distribution function; the historical data includes historical predicted power and historical actual power. Determine the copula function that associates the historical predicted power and the historical actual power of the new energy power station; Based on the density function, the distribution function, and the copula function, given the predicted power of a renewable energy power station during the measurement period, the conditional probability density of the actual power of the renewable energy power station under the predicted power condition is determined; based on the conditional independence assumption and copula theory, the expression for the conditional probability density of the actual power is simplified as follows: in, It means , representing a random variable indicating actual power. It means , representing a random variable predicting power. It means , representing the value of the random variable of actual power. It means , where i is the variable and takes values from 1 to n, where n represents the number of renewable energy power plants, and j is the variable and takes values from 1 to T, where T represents the number of time periods to be predicted. Represents random variables The copula function, Represents the actual power distribution function. Represents the predicted power distribution function. This represents the actual power density function.
2. The power probability modeling method for multi-temporal and spatial renewable energy power plants according to claim 1, characterized in that, Also includes: The conditional probability density of the actual power of each of the new energy power stations under the predicted power conditions is determined by iterating through the data. Based on the conditional probability densities of multiple actual power sources, the joint conditional probability density of the actual power of all the new energy power plants during the test period is obtained.
3. The power probability modeling method for multi-temporal and spatial renewable energy power plants according to claim 1, characterized in that, The conditional probability density of actual power under multiple given predicted power conditions is sampled to obtain samples of the multiple new energy power plants during the test period: from Random number of samples ,turn up Corresponding independent variable value Thus we obtain One sample, Where n represents the number of renewable energy power stations, and T represents the number of time periods to be predicted. This represents a uniform distribution on the interval [0,1]. This represents the randomly selected number. This indicates that the predicted power of the new energy power station is The conditional probability distribution of actual power under the given conditions; This indicates that the predicted power of the new energy power station is The actual power conditional probability density under given conditions.
4. The power probability modeling method for multi-temporal and spatial renewable energy power plants according to claim 3, characterized in that, The sampled is denoted as By concatenating multiple samples, the expression for a single sampling is obtained as follows: .
5. The power probability modeling method for multi-temporal and spatial renewable energy power plants according to claim 4, characterized in that, It also includes sampling the conditional probability density of the actual power under a given predicted power condition to obtain the power scenario of the new energy power station, including: The conditional probability density of the actual power under multiple given predicted power conditions is sampled to obtain samples of multiple new energy power stations during the test period; Multiple samples are spliced together to obtain a single sample of multiple new energy power stations during the test period; the result of the single sample is a scenario of the new energy power station; Repeat the sampling process to obtain multiple power scenarios.
6. The power probability modeling method for multi-temporal and spatial renewable energy power plants according to claim 1, characterized in that, The historical data is processed by kernel density estimation to obtain the density function, and the distribution function is obtained by integrating the density function.
7. The power probability modeling method for multi-temporal and spatial renewable energy power plants according to claim 1, characterized in that, The copula function used to determine the correlation between the historical predicted power and the historical actual power of the renewable energy power station includes: The historical data is fitted to copula functions of various Archimedean and / or elliptic families; The copula function with the largest log-likelihood value in the fitting results is used as the copula function to correlate the historical predicted power and historical actual power of the new energy power station.
8. A power probability modeling system for multi-temporal and spatial renewable energy power plants, characterized in that, It includes modules for determining density and distribution functions, determining copula functions, and determining conditional probability density; The density and distribution function determination module is used to obtain a density function and a distribution function based on the historical data of the renewable energy power station; the density function includes an actual power density function and a predicted power density function; the distribution function includes an actual power distribution function and a predicted power distribution function; the historical data includes historical predicted power and historical actual power. The copula function determination module is used to determine the copula function associated with the historical predicted power and the historical actual power of the new energy power station. The conditional probability density determination module is used to determine the conditional probability density of the actual power of the new energy power station under the predicted power conditions, given the predicted power of the new energy power station during the period to be measured, based on the density function, the distribution function, and the copula function. The expression for the conditional probability density of the actual power is simplified based on the conditional independence assumption and copula theory as follows: in, It means , representing a random variable indicating actual power. It means , representing a random variable predicting power. It means , representing the value of the random variable of actual power. It means , where i is the variable and takes values from 1 to n, where n represents the number of renewable energy power plants, and j is the variable and takes values from 1 to T, where T represents the number of time periods to be predicted. Represents random variables The copula function, Represents the actual power distribution function. Represents the predicted power distribution function. This represents the actual power density function.