Solar irradiance interval prediction method, apparatus, computer equipment, and storage medium
The solar irradiance interval prediction method addresses the limitations of single-point predictions by using historical data and autonomous sampling to provide confidence intervals, improving the reliability and accuracy of solar power generation forecasting.
Patent Information
- Authority / Receiving Office
- JP · JP
- Patent Type
- Patents
- Current Assignee / Owner
- CHINA THREE GORGES INT CORP
- Filing Date
- 2024-06-13
- Publication Date
- 2026-07-03
Smart Images
Figure 0007884605000003 
Figure 0007884605000004 
Figure 0007884605000005
Abstract
Description
Technical Field
[0004] , , , ,
[0001] This application relates to the technical field of interval prediction, and specifically, to a solar irradiance interval prediction method, apparatus, computer device, and storage medium.
Background Art
[0002] Accurately predicting solar irradiance is the basis for evaluating the power generation of power plants, ensuring power supply plans, and more effectively maintaining the balance of the power grid. At present, the methods used for short-term solar irradiance prediction mainly include physical model methods, statistical methods, artificial neural network methods, deep learning methods, and cluster analysis methods, etc. Among them, the physical model method has a high computing complexity but can provide more accurate prediction results. The statistical method does not require a complex physical model and has high computing efficiency, but its prediction accuracy is relatively low, and it is relatively greatly affected by data quality and feature selection. The artificial neural network method can adapt to complex non-linear relationships, but requires a large amount of data and computing resources, and is sensitive to the selection of network structure and parameters. The deep learning method requires a large amount of data and computing resources, but can provide more accurate prediction results.
[0003] <00,00012>However, all the current short-term prediction methods for solar irradiance are based on single-point prediction. For example, the above methods usually give one determined prediction value instead of the confidence interval of the prediction value, and cannot fully reflect the uncertainty of the prediction result. As a result, the error of the prediction result is large.
Summary of the Invention
Problems to be Solved by the Invention
[0004] In view of this, the present invention provides a solar irradiance interval prediction method, apparatus, computer equipment, and storage medium, thereby solving the problem that conventional short-term solar irradiance prediction methods are all based on single-point predictions and cannot fully reflect the uncertainty of the prediction results, resulting in large errors in the prediction results. [Means for solving the problem]
[0005] In the first aspect, the present application provides a solar irradiance interval prediction method for use in a solar power plant, and the solar irradiance interval prediction method is The method includes the steps of: obtaining a historical environmental parameter dataset and a first environmental parameter dataset for the target period of a solar power plant; determining a first environmental sample dataset based on the historical environmental parameter dataset and the first environmental parameter dataset; establishing a target solar irradiance prediction model based on the first environmental sample dataset; inputting the first environmental parameter dataset into the target solar irradiance prediction model to obtain solar irradiance prediction results for the target period; and performing interval predictions on the solar irradiance prediction results for the target period using an autonomous sampling method based on the first environmental sample dataset to obtain target solar irradiance interval prediction results for the target period.
[0006] The solar irradiance interval prediction method according to this application selects a first environmental parameter dataset within the target period based on a historical environmental parameter dataset, makes predictions using a target solar irradiance prediction model, and considers not only the similarity and regularity of the historical data but also the complexity and accuracy of the model. Furthermore, interval predictions are made to the solar irradiance prediction results using an autonomous sampling method, effectively improving the reliability and trustworthiness of the prediction results.
[0007] In one selectable embodiment, the step of determining a first environmental sample dataset based on a historical environmental parameter dataset and a first environmental parameter dataset is: The method includes the steps of: performing regression analysis based on a historical environmental parameter dataset and constructing a multiple linear regression function, wherein the multiple linear regression function reflects the relationship between environmental parameters and collection time; determining a second environmental parameter dataset based on the multiple linear regression function; and determining a first environmental sample dataset based on the historical environmental parameter dataset and the second environmental parameter dataset.
[0008] This invention determines a second environmental parameter dataset by constructing a multiple linear regression function that reflects the relationship between environmental parameters and collection time, and provides data support for reducing subsequent prediction errors by considering the similarity and regularity of historical data.
[0009] In one selectable embodiment, the step of determining a first environmental sample dataset based on a historical environmental parameter dataset and a second environmental parameter dataset is: The method includes the steps of: calculating the Mahalanobis distance between the historical environmental parameter data corresponding to each pre-set collection time in the historical environmental parameter dataset and the second environmental parameter data corresponding to each pre-set collection time in the second environmental parameter dataset; determining a target period based on each Mahalanobis distance; and determining a first environmental sample dataset in the historical environmental parameter dataset based on the target period.
[0010] This invention can determine the target period and the first environmental sample dataset by combining Mahalanobis distances, consider the correlation between each data feature, and effectively address the problem of different metric scales between different data features.
[0011] In one selectable embodiment, the step of establishing a target solar irradiance prediction model based on a first environmental sample dataset is: The method includes the steps of determining a first environmental sample data subset and a second environmental sample data subset based on a first environmental sample dataset; training a pre-configured neural network using the first environmental sample data subset to obtain an initial solar irradiance prediction model; and validating the initial solar irradiance prediction model using the second environmental sample data subset to obtain a target solar irradiance prediction model.
[0012] This invention establishes a target solar irradiance prediction model using a first environmental sample data subset, verifies it by combining it with a second environmental sample data subset, and improves the accuracy of the model prediction by considering the complexity and accuracy of the model.
[0013] In one selectable embodiment, the step of performing interval predictions on the predicted solar irradiance results within the target period using an autonomous sampling method based on a first environmental sample dataset, and obtaining target solar irradiance interval prediction results within the target period, is as follows: The method includes the steps of: determining at least one second environmental sample dataset in a first environmental sample dataset using an autonomous sampling method; establishing at least one target solar irradiance prediction model based on each second environmental sample dataset; inputting the first environmental parameter datasets into each target solar irradiance prediction model to obtain at least one solar irradiance prediction result; determining at least one confidence interval range and at least one initial solar irradiance interval prediction result within the target period based on each solar irradiance prediction result and each pre-set interval prediction confidence value; determining the accuracy of each initial solar irradiance interval prediction result based on each confidence interval range; and determining a target solar irradiance interval prediction result within the target period based on each accuracy and each initial solar irradiance interval prediction result.
[0014] This invention utilizes an autonomous sampling method to perform interval predictions on solar irradiance prediction results, eliminating the need to make assumptions about the original first environmental sample dataset. It allows for direct estimation of distributions and parameters from the first environmental sample dataset, and quantifies the uncertainty of the estimator by calculating the confidence interval range, thereby effectively improving the reliability and trustworthiness of the prediction results.
[0015] In one selectable embodiment, the step of determining at least one confidence interval range and at least one initial solar irradiance interval prediction result within the prediction period, based on each solar irradiance prediction result and each preset interval prediction confidence value, is: The method includes the steps of: obtaining the number of samples for each second environmental sample dataset; determining the mean solar irradiance and the standard deviation of solar irradiance based on each solar irradiance prediction result; determining at least one interval value based on the standard deviation of solar irradiance, each preset interval prediction confidence value, and each number of samples; determining a confidence interval range corresponding to each preset interval prediction confidence value based on each interval value and the mean solar irradiance; and determining at least one initial solar irradiance interval prediction result within the prediction period based on each confidence interval range.
[0016] This invention quantifies the uncertainty of the estimator by calculating the confidence interval range, thereby effectively improving the reliability and trustworthiness of the prediction results.
[0017] In one selectable embodiment, the step of determining the accuracy of the prediction result for each initial solar irradiance interval based on each confidence interval range is: The method includes the steps of: obtaining the accuracy level of each solar irradiance interval prediction result corresponding to each confidence interval range using a big data method; establishing an evaluation index matrix based on each accuracy level; establishing a solar irradiance matrix based on each solar irradiance prediction result; establishing a degree of belonging matrix based on the evaluation index matrix and the solar irradiance matrix; and determining the accuracy of each initial solar irradiance interval prediction result based on the degree of belonging matrix and each solar irradiance prediction result.
[0018] This invention quantifies the uncertainty of the estimator by calculating the confidence interval range, thereby effectively improving the reliability and trustworthiness of the prediction results.
[0019] In a second aspect, the present invention provides a solar irradiance interval prediction device used in a solar power plant, wherein the solar irradiance interval prediction is The system includes: an acquisition module for obtaining a historical environmental parameter dataset and a first environmental parameter dataset for the target period of a solar power plant; a determination module for determining a first environmental sample dataset based on the historical environmental parameter dataset and the first environmental parameter dataset; an establishment module for establishing a target solar irradiance prediction model based on the first environmental sample dataset; an input module for inputting the first environmental parameter dataset into the target solar irradiance prediction model and obtaining solar irradiance prediction results for the target period; and a prediction module for performing interval predictions on the solar irradiance prediction results for the target period using an autonomous sampling method based on the first environmental sample dataset and obtaining target solar irradiance interval prediction results for the target period.
[0020] In a third aspect, the present invention provides a computer device that includes a memory and a processor that are connected to each other in communication, the memory storing computer instructions, and the processor executing the computer instructions to perform the solar irradiance interval prediction method according to either the first aspect or a corresponding embodiment.
[0021] In a fourth aspect, the present application provides a computer-readable storage medium storing computer instructions for causing a computer to execute the solar irradiance interval prediction method according to any one of the first aspect or the corresponding embodiments thereof.
[0022] To more clearly explain the specific embodiments of the present application or the technical solutions of the prior art, the following briefly explains the drawings necessary for the description of the specific embodiments or the prior art. As is clear, the drawings described below are some embodiments of the present application, and those skilled in the art can obtain other drawings based on these drawings without creative labor.
Brief Description of Drawings
[0023] [Figure 1] It is a flowchart of the solar irradiance interval prediction method according to an embodiment of the present application. [Figure 2] It is a flowchart of another solar irradiance interval prediction method according to an embodiment of the present application. [Figure 3] It is a flowchart of yet another solar irradiance interval prediction method according to an embodiment of the present application. [Figure 4] It is a flowchart of the solar irradiance interval prediction method used in a solar power plant according to an embodiment of the present application. [Figure 5] It is a structural block diagram of the solar irradiance interval prediction device according to an embodiment of the present application. [Figure 6] It is a schematic diagram of the hardware structure of a computer device according to an embodiment of the present application.
Modes for Carrying Out the Invention
[0024] To make the objectives, technical solutions, and advantages of the embodiments of the present application clearer, the following clearly and completely describes the drawings of the embodiments of the present application and the technical solutions of the embodiments of the present application. As is clear, the described embodiments are some embodiments of the present application, not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present application without creative labor belong to the protection scope of the present application.
[0025] Accurately predicting solar irradiance is fundamental to evaluating power generation capacity at power plants, ensuring power supply plans, and more effectively balancing the power grid.
[0026] Therefore, the embodiment of the present application provides a solar irradiance interval prediction method for use in solar power plants. By selecting target period search conditions and making predictions using a target solar irradiance prediction model, the reliability and trustworthiness of the prediction results are improved.
[0027] According to embodiments of the present invention, a method for predicting solar irradiance intervals is provided. The steps shown in the flowchart of the drawings may be executed in a computer system such as a set of computer-executable instructions. Although the flowchart shows a logical order, the steps shown or described may be executed in an order different from that shown herein.
[0028] This embodiment provides a solar irradiance interval prediction method that can be used in a solar power plant. Figure 1 is a flowchart of the solar irradiance interval prediction method according to this embodiment, and as shown in Figure 1, the process includes steps S101 to S105.
[0029] Step S101: Obtain the historical environmental parameter dataset and the first environmental parameter dataset for the predicted period of the solar power plant.
[0030] However, the historical environmental parameter dataset represents a set of environmental parameters, including solar irradiance, temperature, and humidity, collected by a solar power plant within its historical data collection cycle.
[0031] The first environmental parameter dataset represents a collection of environmental parameters, including solar irradiance, temperature, and humidity, collected by a solar power plant within the predicted cycle.
[0032] Specifically, the historical environment parameter dataset includes multiple data sublibraries that store the environment parameters for each category, with each data sublibrary corresponding to one category tag.
[0033] A real-time environmental parameter dataset is constructed based on real-time environmental parameters collected by the solar power plant, and each environmental parameter category in the real-time environmental parameter dataset, such as temperature, humidity, wind speed, and irradiance, is determined.
[0034] Selectively, the real-time environmental parameter dataset is matched with the category tags of each data sublibrary. Environmental parameters for which the matching between the environmental parameter category in the real-time environmental parameter dataset and the category tags of each data sublibrary fails are marked as abnormal data, and the abnormal data in the real-time environmental parameter dataset is deleted.
[0035] Finally, the real-time environmental parameter dataset after removing anomalous data is used as the first environmental parameter dataset within the prediction period.
[0036] Step S102: Based on the historical environment parameter dataset and the first environment parameter dataset, the first environment sample dataset is determined.
[0037] However, the first environmental sample dataset represents a set of environmental parameter data within a collection period similar to the historical collection period corresponding to the historical environmental parameter dataset, within the prediction period.
[0038] Step S103: Establish a target solar irradiance prediction model based on the first environmental sample dataset.
[0039] Specifically, a model can be trained using the first environmental sample dataset to obtain a target solar irradiance prediction model that satisfies the conditions.
[0040] In step S104, the first environmental parameter dataset is input into the target solar irradiance prediction model to obtain the solar irradiance prediction results for the period to be predicted.
[0041] Specifically, a first environmental parameter dataset collected by solar power plants within the target prediction period is input into a constructed target solar irradiance prediction model, and the solar irradiance prediction results for the corresponding target prediction period can be output.
[0042] In step S105, based on the first environmental sample dataset, interval predictions are performed on the predicted solar irradiance results within the target period using an autonomous sampling method to obtain the target solar irradiance interval prediction results within the target period.
[0043] However, autonomous sampling is a method of collecting samples continuously or discontinuously according to a pre-set program by the instrument, without requiring manual intervention in the sampling process.
[0044] In this embodiment, by performing interval predictions on the solar irradiance prediction results within the target period using an autonomous sampling method, the distribution and parameters can be directly estimated from the first environmental sample dataset, effectively improving the reliability and trustworthiness of the prediction results.
[0045] The solar irradiance interval prediction method according to this embodiment selects a first environmental parameter dataset within the target period based on a historical environmental parameter dataset, makes predictions using a target solar irradiance prediction model, and considers not only the similarity and regularity of the historical data but also the complexity and accuracy of the model. Selectively, interval predictions are made to the solar irradiance prediction results using an autonomous sampling method to effectively improve the reliability and trustworthiness of the prediction results.
[0046] This embodiment provides a solar irradiance interval prediction method that can be used in a solar power plant. Figure 2 is a flowchart of the solar irradiance interval prediction method according to this embodiment, and as shown in Figure 2, the process includes steps S201 to S205.
[0047] In step S201, the historical environmental parameter dataset and the first environmental parameter dataset for the predicted period of the solar power plant are obtained. For details, please refer to step S101 of the embodiment shown in Figure 1, and a redundant explanation will be omitted here.
[0048] Step S202: Based on the historical environment parameter dataset and the first environment parameter dataset, the first environment sample dataset is determined.
[0049] Specifically, step S202 above includes steps S2021 to S2023.
[0050] Step S2021: Perform regression analysis based on the historical environmental parameter dataset and construct a multiple linear regression function.
[0051] However, the multiple linear regression function reflects the relationship between environmental parameters and data collection time.
[0052] Specifically, regression analysis is performed using the collection time of each category of environmental parameter in multiple historical collection cycles as the independent variable, and the category of environmental parameter at different collection times in multiple historical collection cycles as the dependent variable. A multiple linear regression function is then constructed to represent the interrelationship between environmental parameters and collection time based on the results of the regression analysis.
[0053] Step S2022: Determine the second environmental parameter dataset based on the multiple linear regression function.
[0054] Specifically, a trend curve for the change of each category environmental parameter is established, which changes over time in the predicted period according to a multiple linear regression function. Each category environmental parameter change trend curve has the collection time on the horizontal axis and each category environmental parameter on the vertical axis, and the category environmental parameters at different collection times in each category environmental parameter change trend curve are used as the second environmental parameter dataset.
[0055] Step S2023: Based on the historical environment parameter dataset and the second environment parameter dataset, the first environment sample dataset is determined.
[0056] Specifically, the second environmental parameter dataset can be used as a condition for determining the first environmental sample dataset, and the first environmental sample dataset can be selected from the historical environmental parameter dataset according to that condition.
[0057] In some selectable embodiments, step S2023 includes steps a1 to a3. Step a1: Calculate the Mahalanobis distance between the historical environmental parameter data corresponding to each pre-set collection time in the historical environmental parameter dataset and the second environmental parameter data corresponding to the pre-set collection time in the second environmental parameter dataset. Step a2: Determine the target period based on each Mahalanobis distance. Step a3: Based on the target cycle, determine the first environment sample dataset in the historical environment parameter dataset.
[0058] Specifically, for each category environmental parameter value in the historical environmental parameter dataset, the Mahalanobis distance is obtained between that value and the corresponding environmental parameter value in the second environmental parameter dataset within the same collection time.
[0059] Subsequently, the Mahalanobis distance values of all categories of environmental parameters in the historical environmental parameter dataset for each historical collection cycle are accumulated to obtain the total Mahalanobis distance, and the historical collection cycle with the minimum total Mahalanobis distance is selected as the target cycle, which is the similar cycle.
[0060] Finally, in the historical environmental parameter dataset, the environmental parameter data corresponding to the target period is selected as the first environmental sample dataset.
[0061] Step S203: Establish a target solar irradiance prediction model based on the first environmental sample dataset.
[0062] Specifically, step S203 above includes steps S2031 to S2033.
[0063] Step S2031: Determine the first environment sample data subset and the second environment sample data subset based on the first environment sample dataset.
[0064] Specifically, the first environment sample dataset is split into a first environment sample data subset and a second environment sample data subset. The first environment sample data subset is used as the model training set, and the second environment sample data subset is used as the model test set.
[0065] In step S2032, a subset of first environment sample data is input into a pre-configured neural network for training to obtain an initial solar irradiance prediction model.
[0066] Specifically, a subset of first environment sample data is input into a pre-configured neural network for training, and an initial solar irradiance prediction model is output until the training of the pre-configured model's loss function stabilizes. The model parameters of this initial solar irradiance prediction model are then saved.
[0067] Step S2033: The initial solar irradiance prediction model is validated using a subset of second environmental sample data to obtain the target solar irradiance prediction model.
[0068] Specifically, using a subset of second environment sample data, a similarity validation is performed on the output data matrix of the initial solar irradiance prediction model that was iteratively trained in step S2032. If the validation is successful, the target solar irradiance prediction model is output.
[0069] If the validation fails, the model parameters are adjusted and the model is retrained using the first environment sample data subset. The target solar irradiance prediction model is output until the output data matrix of the trained model passes the similarity validation.
[0070] In step S204, the first environmental parameter dataset is input into the target solar irradiance prediction model to obtain the solar irradiance prediction results for the period to be predicted. For details, please refer to step S104 of the embodiment shown in Figure 1, and a redundant explanation will be omitted here.
[0071] In step S205, based on the first environmental sample dataset, interval predictions are performed on the predicted solar irradiance results within the target period using an autonomous sampling method to obtain the target solar irradiance interval prediction results within the target period. For details, please refer to step S105 of the embodiment shown in Figure 1, and a redundant explanation will be omitted here.
[0072] The solar irradiance interval prediction method according to this embodiment determines a second environmental parameter dataset by constructing a multiple linear regression function that reflects the relationship between environmental parameters and collection time, determines a target period and a first environmental sample dataset by selectively combining the Mahalanobis distance, taking into account the similarity and regularity of historical data, and can effectively handle the problem of different metric scales between different data features by considering the correlation between each data feature.
[0073] This embodiment provides a solar irradiance interval prediction method that can be used in a solar power plant. Figure 3 is a flowchart of the solar irradiance interval prediction method according to this embodiment, and as shown in Figure 3, the process includes steps S301 to S305.
[0074] In step S301, the historical environmental parameter dataset and the first environmental parameter dataset for the predicted period of the solar power plant are obtained. For details, please refer to step S101 of the embodiment shown in Figure 1, and a redundant explanation will be omitted here.
[0075] In step S302, the first environment sample dataset is determined based on the historical environment parameter dataset and the first environment parameter dataset. For details, please refer to step S202 of the embodiment shown in Figure 2, and a redundant explanation will be omitted here.
[0076] In step S303, a target solar irradiance prediction model is established based on the first environmental sample dataset. For details, please refer to step S203 of the example shown in Figure 3, and a redundant explanation will be omitted here.
[0077] In step S304, the first environmental parameter dataset is input into the target solar irradiance prediction model to obtain the solar irradiance prediction results for the period to be predicted. For details, please refer to step S104 of the embodiment shown in Figure 1, and a redundant explanation will be omitted here.
[0078] In step S305, based on the first environmental sample dataset, interval predictions are performed on the predicted solar irradiance results within the target period using an autonomous sampling method to obtain the target solar irradiance interval prediction results within the target period.
[0079] Specifically, step S305 above includes steps S3051 to S3056. Step S3051: Determine at least one second environment sample dataset from the first environment sample dataset using an autonomous sampling method.
[0080] Specifically, environmental parameters for each category with similar but different collection times are used as sample datasets. The extraction size m and number of extractions k for each sample dataset are determined. One sample of size m is randomly extracted from the sample dataset and repeated k times. Each sampled sample is marked as a Bootstrap sample, and finally k different Bootstrap samples are obtained. Each Bootstrap sample is then used as a second environmental sample dataset.
[0081] Step S3052: Establish at least one target solar irradiance prediction model based on each second environmental sample dataset.
[0082] Specifically, by training each second environmental sample dataset, a single target solar irradiance prediction model corresponding to each second environmental sample dataset can be established. However, the specific training process can be found in the detailed explanation in step S203 above, and a redundant explanation will be omitted here.
[0083] In step S3053, the first environmental parameter dataset is input into each target solar irradiance prediction model to obtain at least one solar irradiance prediction result.
[0084] Specifically, the first environmental parameter dataset is input into each of the constructed target solar irradiance prediction models, and the solar irradiance prediction results output by each target solar irradiance prediction model can be obtained.
[0085] In step S3054, based on each solar irradiance prediction result and each preset interval prediction confidence value, at least one confidence interval range and at least one initial solar irradiance interval prediction result within the prediction period are determined.
[0086] However, interval prediction refers to providing an interval estimate for future or unknown data, based on a certain level of confidence.
[0087] Specifically, by setting pre-configured interval prediction confidence values to different values, different confidence interval ranges can be obtained.
[0088] Selectable, different initial solar irradiance interval prediction results can be obtained depending on different confidence interval ranges.
[0089] Step S3055: The accuracy of the prediction results for each initial solar irradiance interval is determined based on the confidence interval range for each interval.
[0090] Specifically, according to the explanation in step S3054, the initial solar irradiance interval prediction result provides an interval estimation result for the solar irradiance prediction result, based on a certain level of confidence.
[0091] Therefore, the accuracy of the prediction results for each initial solar irradiance interval can be determined according to different confidence interval ranges.
[0092] In step S3056, based on the accuracy and initial solar irradiance interval prediction results, the target solar irradiance interval prediction result within the prediction period is determined.
[0093] Specifically, the initial solar irradiance interval prediction result with the highest accuracy will be used as the target solar irradiance interval prediction result within the target period.
[0094] In some selectable embodiments, step S3054 includes steps b1 to b5. Step b1: Obtain the number of samples for each second environment sample dataset. Step b2: Based on each solar irradiance prediction result, the average solar irradiance and the standard deviation of solar irradiance are determined. Step b3: Determine at least one interval value based on the standard deviation of solar irradiance, each pre-set interval prediction confidence value, and each sample size. Step b4: Determine the confidence interval range corresponding to each preset interval prediction confidence value based on each interval value and the average solar irradiance. Step b5: Based on each confidence interval range, determine the prediction result for at least one initial solar irradiance interval within the target period.
[0095] First, according to the explanation of step S3051, the number of samples in each second environment sample dataset is k, and furthermore, the k solar irradiance prediction results are sorted in descending order, and the mean solar irradiance v and the standard deviation of solar irradiance s are calculated. Next, we calculate the interval values using the following relational equation (1). TIFF0007884605000001.tif13104 where: L is the interval value, NORMSINV is the inverse function of the standard normal distribution, ∂ is the pre-set interval prediction confidence value, and 1-∂ represents the confidence level. Subsequently, the confidence interval range corresponding to each pre-set interval prediction confidence value is determined as (vL, v+L). Finally, the uncertainty of the quantified estimate of the calculated confidence interval range, i.e., each confidence interval range, determines the prediction result for one initial solar irradiance interval within the target period.
[0096] In some selectable embodiments, step S3055 includes steps c1 to c5. Step c1: Obtain the accuracy level of the solar irradiance interval prediction results corresponding to each confidence interval range using a big data method. Step c2: Establish an evaluation metric matrix based on each accuracy level. Step c3: Establish a solar irradiance matrix based on each solar irradiance prediction result. Step c4: Establish the degree of belonging matrix based on the evaluation index matrix and the solar irradiance matrix. Step c5: Determine the accuracy of each initial solar irradiance interval prediction based on the assignment matrix and each solar irradiance prediction result. First, a big data method is used to obtain the accuracy level of the solar irradiance interval prediction results corresponding to different confidence interval range phases, provided that the accuracy level may include 95%, 85%, 80%, and 75%. Next, an evaluation index matrix for prediction interval accuracy is established according to the accuracy level corresponding to different confidence interval ranges, and a solar irradiance matrix is established according to each solar irradiance prediction result predicted by the solar irradiance prediction model. Subsequently, the evaluation index matrix and the solar irradiance matrix are merged using the following relation (2) to obtain a degree of belonging matrix that shows the ambiguous relationship between the solar irradiance prediction result predicted by the solar irradiance prediction model and the prediction interval accuracy. TIFF0007884605000002.tif1299 where: M is the degree of belonging matrix, M1 is the evaluation index matrix, M2 is the solar irradiance matrix, α and β are weight parameters to control the balance between the evaluation index matrix and the solar irradiance matrix in the degree of belonging matrix, and "+" represents the addition of elements at corresponding positions in the evaluation index matrix and the solar irradiance matrix. Finally, based on each solar irradiance prediction result, the corresponding confidence interval range can be determined, and furthermore, the accuracy of each initial solar irradiance interval prediction result in the attribute matrix can be determined based on the confidence interval range.
[0097] The solar irradiance interval prediction method according to this embodiment utilizes an autonomous sampling method to perform interval prediction on the solar irradiance prediction results, eliminating the need to make assumptions about the original first environmental sample dataset. It can directly estimate the distribution and parameters from the first environmental sample dataset, and by calculating the confidence interval range, it can quantify the uncertainty of the estimator, thereby effectively improving the reliability and trustworthiness of the prediction results.
[0098] One example provides a solar irradiance interval prediction method used in a solar power plant, which includes steps S1 to S3, as shown in Figure 4. Step S1: Obtain real-time environmental parameters collected by the solar power plant, mark the collection time, set the collection cycle, and perform data cleansing on the real-time environmental parameters according to the environmental parameter information in the historical data library. Step S2: Set similar cycle search conditions according to the cycle to be predicted, extract sample data that satisfies the similar cycle search conditions from the historical data library, use the environmental parameters of each category of the selected similar cycle as input features, and use solar irradiance as an output tag to construct a solar irradiance prediction model. Step S3: Using the autonomous sampling method, interval prediction is performed on the prediction results, and the accuracy of the prediction results is determined according to the interval range in which the prediction results are located.
[0099] The step of performing data cleansing on real-time environmental parameters according to environmental parameter information in the historical data library, which can be selected, A real-time environmental parameter dataset is constructed based on real-time environmental parameters collected by the solar power plant. Each environmental parameter category in the real-time environmental parameter dataset is determined. The historical data library contains multiple data sublibraries that store each category of environmental parameters. Category tags are generated for each data sublibrary. The real-time environmental parameter dataset is matched with the category tags of each data sublibrary. Environmental parameters for which the matching between the environmental parameter category and the category tag of each data sublibrary fails are marked as abnormal data, and the abnormal data in the real-time environmental parameter dataset is deleted.
[0100] The process of selecting and setting similar period search conditions according to the period to be predicted includes the following: Regression analysis is performed using the collection time of each category environmental parameter in multiple historical data collection cycles of the historical data library as the independent variable, and the category environmental parameter at different collection times in multiple historical data collection cycles of the historical data library as the dependent variable. A multiple linear regression function is constructed that shows the interrelationship between environmental parameters and collection time. A trend curve of change for each category environmental parameter, which changes with the time of the target period according to the multiple linear regression function, is established. In each trend curve of change for each category environmental parameter, the collection time is on the horizontal axis and each category environmental parameter is on the vertical axis, and each category environmental parameter at different collection times in each trend curve of change for each category environmental parameter is used as a similar period search condition.
[0101] The process of extracting sample data from the historical data library that meets similar periodic search criteria, which can be selected, includes the following: The system determines the environmental parameter values for each category at different collection times in the similar period search conditions, retrieves the environmental parameter values for each category at different collection times in multiple historical collection periods in the historical data library, calculates the Mahalanobis distance between each environmental parameter value in each historical collection period and the corresponding environmental parameter value within the same collection time in the similar period search conditions, and obtains the total Mahalanobis distance by summing the Mahalanobis distances for all category environmental parameters in each historical collection period. The historical collection period with the minimum total Mahalanobis distance is then selected as the similar period.
[0102] The steps to construct a solar irradiance prediction model, using the environmental parameters of each category for the selected similar period as input features and solar irradiance as the output tag, include the following:
[0103] In the process of establishing a solar irradiance prediction model, an RBN neural network is constructed, historical datasets are built according to the environmental parameter values for each category at different collection times in similar cycles, the historical datasets are divided into a training set and a test set, the solar irradiance prediction model is trained in real time via the training set, the model parameters are saved until the loss function training of the solar irradiance prediction model stabilizes, and then similarity validation is performed on the output data matrix of the solar irradiance prediction model that has been iteratively trained using the test set. Environmental parameters include parameters such as temperature, humidity, wind speed, and irradiance, and data information on temperature, humidity, wind speed, and irradiance for 1000 sets of historical collection cycles is acquired, of which 950 sets of data are used as the training set and 50 sets of data are used as the test set, and the solar irradiance prediction model is trained until it passes the training.
[0104] The environmental parameters for each category at different collection times within the predicted cycle are input into a solar irradiance prediction model validated by a test set, and solar irradiance at different collection times within the predicted cycle is obtained according to the output layer of the solar irradiance prediction model.
[0105] The process of performing interval prediction on the prediction results using an autonomous sampling method, which can be selected as follows:
[0106] Each category of environmental parameters with similar cycles and different collection times is used as a sample dataset. The extraction size m and extraction order k of the sample dataset are determined. One sample of size m is randomly extracted from the sample dataset and repeated k times. Each sampled sample is marked as one Bootstrap sample, and finally k different Bootstrap samples are obtained. Each Bootstrap sample is used as a training set, and k solar irradiance prediction models are constructed according to the solar irradiance prediction model construction method, for the prediction period. By inputting the environmental parameters for each category at different collection times into k solar irradiance prediction models to obtain k solar irradiance prediction results, sorting the k solar irradiance prediction results in descending order of solar irradiance, calculating the average solar irradiance v and standard deviation s of the k solar irradiance prediction results, determining the interval prediction confidence ∂, determining the interval value L according to the confidence ∂, the solar irradiance standard deviation s and the number of k samples, determining the interval range of confidence ∂ according to the interval value L and the average solar irradiance v, and setting the confidence to different values allows obtaining interval ranges with different confidence levels. However, the specific process for determining the interval range of the interval value L and the confidence level ∂ should be referred to in the explanation in step b5 above.
[0107] The step of determining the accuracy of the prediction result depending on the interval range in which the prediction result is located, as selectable, includes the following:
[0108] Using big data methods, accuracy levels for solar irradiance prediction results corresponding to different confidence intervals are obtained, including accuracy levels of 95%, 85%, 80%, and 75%. An evaluation index matrix for prediction interval accuracy is established according to the accuracy level corresponding to different confidence interval ranges, a solar irradiance matrix is established according to the solar irradiance predicted by the solar irradiance prediction model, and a degree of attribution matrix is established showing the ambiguity relationship between the solar irradiance predicted by the solar irradiance prediction model and the accuracy of the prediction result, according to the evaluation index matrix and the solar irradiance matrix, and the specific process refers to the above relation (2) to obtain the predicted solar irradiance accuracy for the prediction period according to the degree of attribution matrix and the solar irradiance predicted by the solar irradiance prediction model.
[0109] This embodiment further provides a solar irradiance interval prediction device, which is used to implement the above embodiment and optional embodiments, and redundant explanations of parts that have already been described are omitted. As used below, the term “module” can implement a combination of software and / or hardware with a predetermined function, and the devices described in the following embodiments are preferably implemented in software, but can also be implemented in hardware, or a combination of software and hardware, and are conceived.
[0110] This embodiment provides a solar irradiance interval prediction device used in a solar power plant, and as shown in Figure 5, the device is An acquisition module 501 for obtaining a historical environmental parameter dataset and a first environmental parameter dataset within the predicted period of a solar power plant, A decision module 502 for determining the first environment sample dataset based on the historical environment parameter dataset and the first environment parameter dataset, Based on the first environmental sample dataset, establishment module 503 is used to establish a target solar irradiance prediction model, Input module 504 inputs the first environmental parameter dataset into the target solar irradiance prediction model to obtain solar irradiance prediction results within the prediction period, The system includes a prediction module 505 that performs interval predictions on solar irradiance prediction results within a target period using an autonomous sampling method based on a first environmental sample dataset, and obtains target solar irradiance interval prediction results within the target period.
[0111] In some selectable embodiments, the decision module 502 is An analysis construction unit for performing regression analysis based on a historical environmental parameter dataset and constructing a multiple linear regression function, wherein the multiple linear regression function reflects the relationship between environmental parameters and data collection time. A first decision unit for determining a second environmental parameter dataset based on a multiple linear regression function, It includes a second decision unit for determining a first environment sample dataset based on a historical environment parameter dataset and a second environment parameter dataset.
[0112] In some selectable embodiments, the second decision unit is: A computing subunit for calculating the Mahalanobis distance between the historical environmental parameter data corresponding to each pre-set collection time in the historical environmental parameter dataset and the second environmental parameter data corresponding to the pre-set collection time in the second environmental parameter dataset, A first decision subunit for determining the target period based on each Mahalanobis distance, It includes a second decision subunit for determining a first environment sample dataset in a historical environment parameter dataset based on a target period.
[0113] In some selectable embodiments, the establishment module 503 is A third decision unit for determining the first environment sample data subset and the second environment sample data subset based on the first environment sample dataset, A first input unit is used to train a pre-configured neural network by inputting a subset of first environmental sample data to obtain an initial solar irradiance prediction model. This includes a validation unit for verifying an initial solar irradiance prediction model using a subset of second-environment sample data and obtaining a target solar irradiance prediction model.
[0114] In some selectable embodiments, the prediction module 505 is A fourth decision unit for determining at least one second environment sample dataset in the first environment sample dataset using an autonomous sampling method, Based on each second environmental sample dataset, an establishment unit for establishing at least one target solar irradiance prediction model, A second input unit inputs each of the first environmental parameter datasets into a target solar irradiance prediction model to obtain at least one solar irradiance prediction result, A fifth decision unit for determining at least one confidence interval range and at least one initial solar irradiance interval prediction result within the prediction period, based on each solar irradiance prediction result and each pre-set interval prediction confidence value, A sixth decision unit for determining the accuracy of each initial solar irradiance interval prediction result based on each confidence interval range, It includes a seventh decision unit for determining the target solar irradiance interval prediction result within the prediction period, based on each accuracy and each initial solar irradiance interval prediction result.
[0115] In some selectable embodiments, the fifth decision unit is: A first acquisition subunit for obtaining the number of samples in each second environment sample dataset, Based on each solar irradiance prediction result, a third determination subunit is used to determine the average solar irradiance and the standard deviation of solar irradiance, A fourth decision subunit for determining at least one interval value based on the standard deviation of solar irradiance, each pre-set interval prediction confidence value, and each sample size, A fifth decision subunit for determining the confidence interval range corresponding to each preset interval prediction confidence value based on each interval value and the average solar irradiance, It includes a sixth decision subunit for determining the prediction result for at least one initial solar irradiance interval within the target period based on each confidence interval range.
[0116] In some selectable embodiments, the sixth decision unit is: A second acquisition subunit for obtaining the accuracy level of the solar irradiance interval prediction results corresponding to each confidence interval range using big data methods, A first establishment subunit for establishing an evaluation index matrix based on each precision level, A second establishment subunit for establishing a solar irradiance matrix based on each solar irradiance prediction result, A third subunit for establishing a degree of belonging matrix based on the evaluation index matrix and the solar irradiance matrix, It includes a seventh decision subunit for determining the accuracy of each initial solar irradiance interval prediction result based on the degree of assignment matrix and each solar irradiance prediction result.
[0117] Further functional descriptions of each of the above modules and units are the same as those in the corresponding embodiments described above, and therefore, redundant explanations are omitted here.
[0118] The solar irradiance interval prediction device in this embodiment is represented in the form of a functional unit, where the unit refers to an ASIC (Application Specific Integrated Circuit) circuit, a processor and memory that execute one or more software or fixed programs, and / or other devices that can provide the above functions.
[0119] The embodiment of the present invention further provides a computer device comprising a solar irradiance interval prediction device as shown in Figure 5 above.
[0120] As shown in Figure 6, which is a schematic diagram of the structure of a computer device according to an optional embodiment of the present invention, the computer device includes one or more processors 10, memory 20, and interfaces for connecting each component, including high-speed and low-speed interfaces. Each component communicates with one another via different buses and may be mounted on a common motherboard or otherwise mounted as needed. The processors can process instructions executed within the computer device, including instructions in or stored in memory for displaying GUI graphic information on an external input / output device (e.g., a display device coupled to the interface). In some optional embodiments, multiple processors and / or multiple buses may be used together with multiple memories as needed. Similarly, multiple computer devices may be connected, each providing a portion of the required operations (e.g., functioning as a server array, a set of blade servers, or a multiprocessor system). In Figure 6, one processor 10 is used as an example.
[0121] The processor 10 may be a central processing unit, a network processor, or a combination thereof. The processor 10 may further include hardware chips. The hardware chips may be application-specific integrated circuits, programmable logic devices, or a combination thereof. The programmable logic devices may be complex programmable logic devices, field programmable logic gate arrays, generic array logic, or any combination thereof.
[0122] However, the memory 20 stores instructions that can be executed by at least one processor 10, thereby enabling at least one processor 10 to execute and implement the method shown in the above embodiment.
[0123] Memory 20 may include an operating system, a program storage area for storing application programs required for at least one function, and a data storage area for storing data created in accordance with the use of the computer equipment. Memory 20 may also include high-speed random-access memory and may further include non-temporary memory such as at least one disk storage device, flash memory device, or other non-temporary solid-state storage device. In some selectable embodiments, memory 20 may optionally include memory remotely installed relative to the processor 10, and these remote memories may be connected to the computer equipment via a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof.
[0124] Memory 20 may include volatile memory such as random access memory, or non-volatile memory such as flash memory, hard disk, or solid-state drive, and memory 20 may include a combination of the above types of memory.
[0125] The computer equipment further includes a communication interface 30 for the computer equipment to communicate with other equipment or a communication network.
[0126] Embodiments of the present application further provide a computer-readable storage medium, and the methods according to the embodiments of the present application may be implemented in hardware, firmware, or in a recordable manner on a storage medium, or as computer code downloaded over a network, originally stored on a remote storage medium or a non-temporary machine-readable storage medium but stored on a local storage medium, thereby enabling the methods described herein to be processed by a general-purpose computer, a dedicated processor, or software stored on a storage medium using programmable or dedicated hardware. The storage medium may be a magnetic disk, an optical disk, read-only memory, random access memory, flash memory, a hard disk, or a solid-state drive, and optionally, the storage medium may include a combination of the above types of memory. As understood, the computer, processor, microprocessor controller, or programmable hardware includes a storage component capable of storing or receiving software or computer code, and when the software or computer code is accessed and executed by the computer, processor, or hardware, the methods shown in the embodiments are implemented.
[0127] While embodiments of the present application have been described with reference to the drawings, those skilled in the art can make various changes and modifications without departing from the spirit and scope of the present application, and such changes and modifications all fall within the scope defined in the attached claims.
Claims
1. A method for predicting solar irradiance intervals used in solar power plants, The steps include obtaining a historical environmental parameter dataset and a first environmental parameter dataset within the predicted period of the solar power plant, The steps include determining a first environment sample dataset based on the aforementioned historical environment parameter dataset, The steps include establishing a target solar irradiance prediction model based on the aforementioned first environmental sample dataset, The steps include inputting the first environmental parameter dataset into the target solar irradiance prediction model and obtaining the solar irradiance prediction result within the prediction target period, The step includes performing interval predictions on the solar irradiance prediction results within the target prediction period using an autonomous sampling method based on the first environmental sample dataset, and obtaining target solar irradiance interval prediction results within the target prediction period. Based on the first environmental sample dataset, the step of performing interval predictions on the solar irradiance prediction results within the target prediction period using an autonomous sampling method, and obtaining target solar irradiance interval prediction results within the target prediction period, is as follows: A step of determining at least one second environment sample dataset in the first environment sample dataset using an autonomous sampling method, The steps include establishing at least one target solar irradiance prediction model based on each of the second environmental sample datasets, The steps include inputting the first environmental parameter dataset into each of the target solar irradiance prediction models to obtain at least one solar irradiance prediction result, The steps include determining at least one confidence interval range and at least one initial solar irradiance interval prediction result within the prediction target period, based on each of the aforementioned solar irradiance prediction results and each of the predetermined interval prediction confidence values, A step of determining the accuracy of each initial solar irradiance interval prediction result based on each confidence interval range, A solar irradiance interval prediction method characterized by comprising the step of determining the target solar irradiance interval prediction result within the prediction target period based on each of the accuracy and each of the initial solar irradiance interval prediction results.
2. The step of determining a first environment sample dataset based on the aforementioned historical environment parameter dataset is as follows: A step of performing regression analysis based on the historical environmental parameter dataset and constructing a multiple linear regression function, wherein the multiple linear regression function reflects the relationship between environmental parameters and collection time. The steps include determining a second environmental parameter dataset based on the aforementioned multiple linear regression function, The method according to claim 1, comprising the step of determining the first environmental sample dataset based on the historical environmental parameter dataset and the second environmental parameter dataset.
3. The step of determining the first environmental sample dataset based on the historical environmental parameter dataset and the second environmental parameter dataset is as follows: A step of calculating the Mahalanobis distance between the historical environmental parameter data corresponding to each preset collection time in the historical environmental parameter dataset and the second environmental parameter data corresponding to the preset collection time in the second environmental parameter dataset, A step of determining a target period based on each of the aforementioned Mahalanobis distances, The method according to the 2nd, further comprising the step of determining the first environmental sample dataset in the historical environmental parameter dataset based on the target period.
4. The step of establishing a target solar irradiance prediction model based on the first environmental sample dataset is as follows: The steps include determining a first environment sample data subset and a second environment sample data subset based on the first environment sample dataset, The steps include: inputting the aforementioned first environmental sample data subset into a pre-configured neural network for training to obtain an initial solar irradiance prediction model; The method according to claim 1, further comprising the step of verifying the initial solar irradiance prediction model using the second environmental sample data subset to obtain the target solar irradiance prediction model.
5. The step of determining at least one confidence interval range and at least one initial solar irradiance interval prediction result within the prediction target period, based on each of the aforementioned solar irradiance prediction results and each of the predetermined interval prediction confidence values, is: A step of obtaining the number of samples for each of the aforementioned second environment sample datasets, The steps include determining the average solar irradiance and the standard deviation of solar irradiance based on each of the aforementioned solar irradiance prediction results, A step of determining at least one interval value based on the standard deviation of solar irradiance, each of the predetermined interval prediction confidence values, and each of the sample sizes, The steps include determining the confidence interval range corresponding to each of the preset interval prediction confidence values based on each of the interval values and the average solar irradiance value, The method according to claim 1, comprising the step of determining at least one initial solar irradiance interval prediction result within the prediction target period based on each of the aforementioned confidence interval ranges.
6. The step of determining the accuracy of each initial solar irradiance interval prediction result based on each confidence interval range is as follows: A step of obtaining the accuracy level of the prediction results for each initial solar irradiance interval corresponding to each confidence interval range using a big data method, The steps include establishing an evaluation index matrix based on each of the aforementioned precision levels, The steps include establishing a solar irradiance matrix based on each of the aforementioned solar irradiance prediction results, The steps include establishing a degree of belonging matrix based on the aforementioned evaluation index matrix and the aforementioned solar irradiance matrix, The method according to claim 1, further comprising the step of determining the accuracy of each initial solar irradiance interval prediction result based on the degree of assignment matrix and each of the solar irradiance prediction results.
7. A solar irradiance interval prediction device used in solar power plants, An acquisition module for acquiring a historical environmental parameter dataset and a first environmental parameter dataset within the predicted period of the solar power plant, A decision module for determining a first environmental sample dataset based on the historical environmental parameter dataset and the first environmental parameter dataset, Based on the aforementioned first environmental sample dataset, an establishment module for establishing a target solar irradiance prediction model, An input module for inputting the first environmental parameter dataset into the target solar irradiance prediction model and obtaining solar irradiance prediction results within the prediction target period, The prediction module includes, based on the first environmental sample dataset, an autonomous sampling method to perform interval predictions on the solar irradiance prediction results within the prediction target period, and to obtain target solar irradiance interval prediction results within the prediction target period, The prediction module, A fourth decision unit for determining at least one second environment sample dataset in the first environment sample dataset using an autonomous sampling method, Based on each second environmental sample dataset, an establishment unit for establishing at least one target solar irradiance prediction model, A second input unit inputs the first environmental parameter dataset into each target solar irradiance prediction model to obtain at least one solar irradiance prediction result, A fifth decision unit for determining at least one confidence interval range and at least one initial solar irradiance interval prediction result within the prediction period, based on each solar irradiance prediction result and each pre-set interval prediction confidence value, A sixth decision unit for determining the accuracy of each initial solar irradiance interval prediction result based on each confidence interval range, A solar irradiance interval prediction device characterized by including a seventh determination unit for determining a target solar irradiance interval prediction result within the prediction period based on each accuracy and each initial solar irradiance interval prediction result.
8. Computer equipment, A computer device comprising a memory and a processor that are connected to each other in communication, wherein the memory stores computer instructions, and the processor executes the solar irradiance interval prediction method according to any one of claims 1 to 6 by executing the computer instructions.
9. A computer-readable storage medium characterized in that it stores computer instructions for causing a computer to execute the solar irradiance interval prediction method described in any one of claims 1 to 6.