Method for predicting generated power of fan and computing equipment

Abstract

The invention provides a fan generation power prediction method and computing equipment. The method comprises the following steps: dividing historical data of a station into a pre-processing stage and a post-processing stage; calculating the data in the pre-processing stage to obtain a first quantile; according to the first quantile, performing calculation to obtain feature data of the post-processing stage; and according to the feature data, calculating to obtain a probability prediction interval of the post-processing stage. According to the technical scheme of the invention, the accuracy and robustness of power probability prediction can be improved, the requirement on a prediction sample is low, the data acquisition cost is reduced, and the error of probability prediction is reduced.

Description

Technical Field

[0001] The present invention relates to the field of new energy, and in particular to a method and computing equipment for predicting wind turbine power generation. Background Art

[0002] Probabilistic forecasting is a method of predicting the probability of future events based on historical data and statistical models. Probabilistic forecasting is widely used in many fields, such as finance, insurance, healthcare, and natural disaster prevention. Probabilistic forecasting is a method of predicting the probability of future events based on historical data and statistical models. Probabilistic forecasting analyzes historical data and uses statistical techniques to build mathematical models to predict future events or outcomes. This method can discover patterns and trends in data and use these patterns and trends to make predictive inferences. Probabilistic forecasting helps us understand and predict future uncertain events by analyzing historical data and building statistical models. This method has a wide range of applications in many fields, from financial market analysis to medical and health predictions, all of which are inseparable from the support of probabilistic forecasting.

[0003] The challenges faced by probabilistic forecasting mainly include the quality and completeness of data, the rationality of model assumptions, the accuracy of parameter selection, insufficient generalization ability due to overfitting historical data, and the complexity and uncertainty of the real world.

[0004] Therefore, a technical solution is needed that can improve the accuracy and robustness of power probability prediction, have lower requirements on prediction samples, reduce data collection costs, and reduce the error of probability prediction. Summary of the invention

[0005] The present invention aims to provide a method and computing equipment for wind turbine power generation prediction, which can improve the accuracy and robustness of power probability prediction, have lower requirements for prediction samples, reduce data collection costs, and reduce the error of probability prediction.

[0006] According to one aspect of the present invention, a method for predicting wind turbine power generation is provided, the method comprising:

[0007] Divide the station historical data into pre-processing stage and post-processing stage;

[0008] Calculate the data in the pre-processing stage to obtain a first quantile;

[0009] According to the first quantile, characteristic data of the post-processing stage is calculated;

[0010] The probability prediction interval of the post-processing stage is calculated based on the feature data.

[0011] According to some embodiments, the station data is divided into a pre-processing stage and a post-processing stage, including:

[0012] The pre-processing stage and the post-processing stage are divided according to the timestamp;

[0013] The data in the pre-processing stage is training set data;

[0014] The data in the post-processing stage is the test set data.

[0015] According to some embodiments, the training set data is divided into processing unit data and control unit data.

[0016] According to some embodiments, the station historical data needs to be processed as outliers;

[0017] The abnormal values ​​include null values, dead values ​​and out-of-limit values.

[0018] According to some embodiments, the station historical data needs to be processed for outliers, including:

[0019] Among the abnormal values, if the proportion of null values ​​and dead values ​​in the total data volume is less than one tenth, the average value of the abnormal values ​​is used to fill it.

[0020] According to some embodiments, the station historical data needs to be processed for outliers, further comprising:

[0021] Among the abnormal values, if the number of consecutive null values ​​is equal to and / or greater than three, the consecutive null values ​​are directly eliminated.

[0022] According to some embodiments, the station historical data needs to be processed for outliers, further comprising:

[0023] The over-limit values ​​are directly eliminated, and the over-limit values ​​are overflow values ​​and negative values ​​less than 0.

[0024] According to some embodiments, the pre-processing stage is interactive with the algorithm parameters during calculation;

[0025] During calculation in the post-processing stage, only one-way prediction is performed with respect to the algorithm parameters.

[0026] According to another aspect of the present invention, a computer program product is provided, comprising a computer program, wherein when the computer program is executed by a processor, the method as described above is implemented.

[0027] According to another aspect of the present invention, there is provided a computing device, comprising:

[0028] Processor; and

[0029] A memory stores a computer program, and when the computer program is executed by the processor, the method described in any one of the above items is implemented.

[0030] According to an embodiment of the present invention, the historical data of the station is divided into a pre-processing stage and a post-processing stage, the data of the pre-processing stage is calculated to obtain the first quantile, and then the characteristic data of the post-processing stage is calculated based on the first quantile, and then the probability prediction interval of the post-processing stage is calculated based on the characteristic data to achieve probability prediction. Through the present invention, the accuracy and robustness of probability prediction can be improved, the requirements for prediction samples are relatively low, the data collection cost is reduced, and the error of probability prediction is reduced.

[0031] According to some embodiments, a large amount of data can be predicted through a neural network, but for scenarios with a smaller amount of data, the method of the present invention can achieve better prediction results.

[0032] According to some embodiments, quantile random forests are more robust to outliers and noise because quantile regression is based on ranking statistics rather than relying on the average of all data points like mean regression.

[0033] According to some embodiments, the present invention does not need to assume that the data follows a specific probability distribution, and the random forest itself can well capture the complex nonlinear relationship between features, which is very useful for many real-world data sets.

[0034] According to some embodiments, the present invention can provide information about the importance of input features, which is helpful for feature selection and understanding models. By estimating different quantiles, prediction intervals can be constructed, which is very helpful for risk assessment and uncertainty quantification. Multiple quantiles can be predicted simultaneously, thereby providing a more comprehensive risk profile or data distribution. Building trees in random forests can be easily parallelized, which helps improve computational efficiency, especially on large data sets.

[0035] It is to be understood that the foregoing general description and the following detailed description are exemplary only and are not restrictive of the invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings required for describing the embodiments are briefly introduced below.

[0037] Figure 1 A flow chart of a function prediction method according to an example embodiment is shown.

[0038] Figure 2 A schematic diagram of a probability interval prediction process according to an example embodiment is shown.

[0039] Figure 3 A block diagram of a computing device is shown according to an exemplary embodiment. DETAILED DESCRIPTION

[0040] Example embodiments will now be described more fully with reference to the accompanying drawings. However, example embodiments can be implemented in many forms and should not be construed as limited to the embodiments set forth herein; rather, these embodiments are provided so that the present invention will be comprehensive and complete and fully convey the concepts of the example embodiments to those skilled in the art. The same reference numerals in the figures represent the same or similar parts, and thus their repeated description will be omitted.

[0041] In addition, the described features, structures or characteristics may be combined in one or more embodiments in any suitable manner. In the following description, many specific details are provided to provide a full understanding of the embodiments of the present invention. However, those skilled in the art will appreciate that the technical solution of the present invention can be practiced without one or more of the specific details, or other methods, components, devices, steps, etc. may be adopted. In other cases, known methods, devices, implementations or operations are not shown or described in detail to avoid blurring various aspects of the present invention.

[0042] The block diagrams shown in the accompanying drawings are merely functional entities and do not necessarily correspond to physically independent entities. That is, these functional entities may be implemented in software form, or in one or more hardware modules or integrated circuits, or in different networks and / or processor devices and / or microcontroller devices.

[0043] The flowcharts shown in the accompanying drawings are only exemplary and do not necessarily include all the contents and operations / steps, nor must they be executed in the order described. For example, some operations / steps can be decomposed, and some operations / steps can be combined or partially combined, so the actual execution order may change according to actual conditions.

[0044] It should be understood that although the terms first, second, third, etc. may be used herein to describe various components, these components should not be limited by these terms. These terms are used to distinguish one component from another component. Therefore, the first component discussed below can be referred to as the second component without departing from the teachings of the present inventive concept. As used herein, the term "and / or" includes any one of the associated listed items and all combinations of one or more.

[0045] The user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in the present invention are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with the relevant laws, regulations and standards of relevant countries and regions, and provide corresponding operation entrances for users to choose to authorize or refuse.

[0046] Those skilled in the art will appreciate that the drawings are merely schematic diagrams of example embodiments, and the modules or processes in the drawings are not necessarily necessary for implementing the present invention, and therefore cannot be used to limit the protection scope of the present invention.

[0047] Prediction models are widely used in finance, healthcare, electricity, transportation and other industries, becoming an important means to improve operational efficiency and reduce risks. In the financial field, probabilistic prediction models can help analyze stock prices, market trends, etc., and provide a scientific basis for investment decisions. In the medical field, these models can be used for disease prediction, drug effect evaluation, etc., to improve the quality and efficiency of medical services. In addition, probabilistic prediction also involves in-depth research on random processes, which is a mathematical method for extracting information from uncertainty, used to describe and predict movements or changes full of complex and unknown factors.

[0048] The mainstream probability prediction methods now include regression algorithm and Bayesian algorithm. In practical applications, it is necessary to select appropriate methods for probability prediction according to specific problems and data characteristics. In addition, the current methods have high sample requirements and are sensitive to the selection of initial values. The traditional linear quantile regression converges slowly and may have deviations from function missetting. In order to improve the accuracy and reliability of predictions, it is necessary to continuously optimize and improve the model structure and parameter settings, as well as perform cross-validation and error analysis.

[0049] To this end, the present invention proposes a method for predicting wind turbine power generation, which can improve the accuracy and robustness of power probability prediction, has low requirements for prediction samples, reduces data collection costs, and reduces the error of probability prediction.

[0050] Exemplary embodiments of the present invention are described below with reference to the accompanying drawings.

[0051] Figure 1 A flow chart of a power prediction method according to an example embodiment is shown.

[0052] See also Figure 1 , in S101, the station historical data is divided into a pre-processing stage and a post-processing stage.

[0053] According to some embodiments, the pre-processing stage and the post-processing stage are divided according to timestamps; the data in the pre-processing stage is training set data; the data in the post-processing stage is test set data; the training set data is divided into processing unit data and control unit data.

[0054] According to some embodiments, historical data of wind turbine power generation is collected from various sites, and the data set is divided into a pre-processing stage, i.e., a training set, and a post-processing stage, i.e., a test set, according to timestamps.

[0055] The station historical data needs to be processed as abnormal values; the abnormal values ​​include null values, dead values ​​and over-limit values. Among the abnormal values, if the proportion of null values ​​and dead values ​​in the total data volume is less than one tenth, they are filled with the mean value of the abnormal values; among the abnormal values, if the number of consecutive null values ​​is equal to and / or greater than three, the consecutive null values ​​are directly eliminated; and over-limit values ​​are directly eliminated, and the over-limit values ​​are overflow values ​​and negative values ​​less than 0.

[0056] Abnormal values ​​include null values, dead values, and over-limit values. The processing of abnormal values ​​includes: when the proportion of null values ​​and dead values ​​in the feature is less than 10% of the total data volume, fill it with the average of the upper and lower values, and directly remove the null values ​​with more than 3 consecutive null values; directly delete the data of over-limit values ​​(that is, greater than the installed capacity and less than 0).

[0057] According to some embodiments, the pre-processing stage interacts with the algorithm parameters during calculation; and the post-processing stage only performs one-way prediction with the algorithm parameters during calculation.

[0058] In S103, the data in the pre-processing stage is calculated to obtain a first quantile.

[0059] Calculate the first quantile using the quantile random forest calculation formula

[0060] The meanings of the variables involved in the quantile random forest calculation formula are as follows:

[0061] Variable Y it , i=1,……,n+1, t=1,……T0,T0+1,……,T0+T1, the time dimension T is divided into T0 and T1, which are called the pre-processing stage and the post-processing stage respectively.

[0062] Data {Y it}, where i=1 is the processing unit, i=2, ... n+1 is the control unit, where t is the time variable, T 2,t . is a specific time period.

[0063] Y t is the dependent variable of treatment; X t =(T 2,t ,……,T n+1,t ,Z 1,t ,……,Z p,t ), where X t is the characteristic data, Z 1,t ,……Z p,t It is a covariate. The covariate is calculated based on the characteristic data, mainly meteorological element data and its derivative data. and Y t The α and 1-α quantiles of .

[0064] Using pre-processed data, the quantile random forest (QRF) is used to calculate exist Quantile regression at α / 2 and 1-α / 2 on the graph. The formula for calculating the quantile random forest is as follows:

[0065]

[0066] Among them, M is the number of trees in the random forest, θ m is the hyperparameter of the mth tree.

[0067] Quantile random forest improves prediction performance and model stability by building multiple decision trees. To achieve optimal performance, a series of hyperparameters usually need to be adjusted. The following are common hyperparameters in random forests and their functions:

[0068] n_estimators, the number of decision trees in the random forest. Increasing the number of trees can usually improve the performance of the model; max_depth, the maximum depth of the decision tree. A larger depth may lead to overfitting, while a smaller depth may lead to underfitting. By default, the tree will continue to split until all leaf nodes contain samples that are less than the minimum number of samples required for leaf nodes; min_samples_split, the minimum number of samples required for a node to be further split. This hyperparameter can help control the growth of the tree and prevent overfitting; min_samples_leaf, the minimum number of samples required for a leaf node. Increasing this value can reduce the variance of the model; max_features, the maximum number of features considered when finding the best split point, can be set to an integer, floating point number, or string. Common options are "sqrt" (default, representing square root) and "log2" (representing logarithm); bootstrap, whether to use bootstrap sampling (bootstrap sampling) to extract training samples. If set to True, each tree will extract samples with replacement from the training set. If set to False, each tree will extract samples without replacement from the training set. oob_score, whether to use out-of-bag samples to evaluate the generalization ability of the model. If set to True, the out-of-bag score will be calculated as one of the evaluation indicators of the model. criterion, a criterion for measuring the quality of segmentation. random_state, controls the seed of the random number generator to ensure that the results of each run are the same. class_weight, for unbalanced data sets, the importance of each category can be adjusted by setting the category weight.

[0069] The goal of tuning these hyperparameters is to maximize the predictive performance of the model while avoiding overfitting. Typically, methods for tuning hyperparameters include Grid Search or Random Search, which systematically try different hyperparameter combinations and select the best performing one.

[0070] In S105, characteristic data of the post-processing stage is calculated based on the first quantile.

[0071] For the post-processing stage t=T0+1, ..., T0+T1, calculate the of and Characteristic data in the post-processing stage

[0072] In S107, a probability prediction interval of the post-processing stage is calculated based on the feature data.

[0073] Calculate feature data For the post-processing stage t = T0+1, ..., T0+T1, calculate the predicted interval 1-α of the processing unit in the post-processing stage This interval is the probability prediction interval of the predicted value of wind turbine power generation.

[0074] Random forest is an ensemble learning method that makes predictions by building multiple decision trees and taking their average. It can handle classification and regression problems with high accuracy and robustness. In probability prediction, random forest can be used to train a probability model to predict the probability of an event by inputting features. The advantage of random forest is that it can automatically select the most important features and is robust to noise and outliers in the data. However, the conventional random forest method is sensitive to small samples, and the optimized algorithm performs better in small samples.

[0075] This paper uses the machine learning method of "quantile random forest" to perform quantile regression to construct the confidence interval of the treatment effect. This method is also called quantile control method. As a non-parametric ensemble learning method, quantile control method is still robust in the case of heteroskedasticity, autocorrelation or function misspecification, and performs relatively well in small samples.

[0076] Figure 2 A schematic diagram of a probability interval prediction process according to an example embodiment is shown.

[0077] See also Figure 2Before data processing, outlier processing and feature processing are performed first. Outlier processing includes identifying and processing data points that significantly deviate from the normal range. The processed data enters the pre-processing stage. In the pre-processing stage, the pre-processed data will be calculated by the quantile control algorithm (QCM) to obtain the quantile. After obtaining the quantile, the post-processing stage begins, and the probability prediction interval is calculated by the quantile control algorithm.

[0078] Figure 3 A block diagram of a computing device is shown according to an exemplary embodiment.

[0079] like Figure 3 As shown, computing device 30 includes processor 12 and memory 14. Computing device 30 may also include bus 22, network interface 16, and I / O interface 18. Processor 12, memory 14, network interface 16, and I / O interface 18 may communicate with each other via bus 22.

[0080] The processor 12 may include one or more general-purpose CPUs (Central Processing Units, processors), microprocessors, or application-specific integrated circuits, etc., for executing relevant program instructions. According to some embodiments, the computing device 30 may also include a high-performance graphics card (GPU) 20 for accelerating the processor 12.

[0081] The memory 14 may include a machine system readable medium in the form of a volatile memory, such as a random access memory (RAM), a read-only memory (ROM) and / or a cache memory. The memory 14 is used to store one or more programs including instructions and data. The processor 12 can read the instructions stored in the memory 14 to execute the above-mentioned method according to the embodiment of the present invention.

[0082] The computing device 30 may also communicate with one or more networks via the network interface 16. The network interface 16 may be a wireless network interface.

[0083] The bus 22 may include an address bus, a data bus, a control bus, etc. The bus 22 provides a path for exchanging information between components.

[0084] It should be noted that, in the specific implementation process, the computing device 30 may also include other components necessary for normal operation. In addition, those skilled in the art may understand that the above device may only include components necessary for implementing the embodiments of this specification, and need not include all components shown in the figure.

[0085] The present invention also provides a computer-readable storage medium having a computer program stored thereon, which implements the steps of the above method when executed by a processor. The computer-readable storage medium may include, but is not limited to, any type of disk, including a floppy disk, an optical disk, a DVD, a CD-ROM, a microdrive, and a magneto-optical disk, a ROM, a RAM, an EPROM, an EEPROM, a DRAM, a VRAM, a flash memory device, a magnetic card or an optical card, a nanosystem (including a molecular memory IC), a network storage device, a cloud storage device, or any type of medium or device suitable for storing instructions and / or data.

[0086] An embodiment of the present invention further provides a computer program product, which includes a computer program. The computer program is operable to enable a computer to execute part or all of the steps of any one of the methods described in the above method embodiments.

[0087] Those skilled in the art can clearly understand that the technical solution of the present invention can be implemented with the help of software and / or hardware. "Unit" and "module" in this specification refer to software and / or hardware that can independently complete or cooperate with other components to complete specific functions, where the hardware can be, for example, a field programmable gate array, an integrated circuit, etc.

[0088] It should be noted that, for the above-mentioned method embodiments, for the sake of simplicity, they are all described as a series of action combinations, but those skilled in the art should know that the present invention is not limited by the described action sequence, because according to the present invention, certain steps can be performed in other sequences or simultaneously. Secondly, those skilled in the art should also know that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily required by the present invention.

[0089] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0090] In the several embodiments provided by the present invention, it should be understood that the disclosed devices can be implemented in other ways. For example, the device embodiments described above are only schematic, such as the division of units, which is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some service interfaces, and the indirect coupling or communication connection of devices or units can be electrical or other forms.

[0091] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0092] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0093] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable memory. Based on this understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art or the whole or part of the technical solution can be embodied in the form of a software product, which is stored in a memory and includes several instructions for a computer device (which can be a personal computer, server or network device, etc.) to perform all or part of the steps of the methods of various embodiments of the present invention.

[0094] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0095] The exemplary embodiments of the present invention are specifically shown and described above. It should be understood that the present invention is not limited to the detailed structures, configurations or implementations described herein; on the contrary, the present invention is intended to cover various modifications and equivalent configurations included in the spirit and scope of the appended clauses.

Claims

1. A method for predicting wind turbine power generation, the method comprising: Divide the station historical data into pre-processing stage and post-processing stage; Calculate the data in the pre-processing stage to obtain a first quantile; According to the first quantile, characteristic data of the post-processing stage is calculated; The probability prediction interval of the post-processing stage is calculated based on the feature data.

2. The method according to claim 1, characterized in that: The station data is divided into pre-processing and post-processing stages, including: The pre-processing stage and the post-processing stage are divided according to the timestamp; The data in the pre-processing stage is training set data; The data in the post-processing stage is the test set data.

3. The method according to claim 2, characterized in that include: The training set data is divided into processing unit data and control unit data.

4. The method according to claim 1, characterized in that: include: The historical data of the station needs to be processed as outliers; The abnormal values ​​include null values, dead values ​​and out-of-limit values.

5. The method according to claim 4, characterized in that The historical data of the station needs to be processed for outliers, including: Among the abnormal values, if the proportion of null values ​​and dead values ​​in the total data volume is less than one tenth, the average value of the abnormal values ​​is used to fill it.

6. The method according to claim 4, characterized in that The site historical data needs to be processed for outliers, including: Among the abnormal values, if the number of consecutive null values ​​is equal to and / or greater than three, the consecutive null values ​​are directly eliminated.

7. The method according to claim 4, characterized in that The site historical data needs to be processed for outliers, including: The over-limit values ​​are directly eliminated, and the over-limit values ​​are overflow values ​​and negative values ​​less than 0.

8. The method according to claim 1, characterized in that include: The pre-processing stage interacts with the algorithm parameters during calculation; During calculation in the post-processing stage, only one-way prediction is performed with respect to the algorithm parameters.

9. A computer program product, characterized in that The method comprises a computer program, which implements the method according to any one of claims 1 to 8 when being executed by a processor.

10. A computing device, characterized in that include: processor; as well as A memory storing a computer program, wherein when the computer program is executed by the processor, the method according to any one of claims 1 to 8 is implemented.

Images