Wind energy prediction method which combines spatial-temporal features and error processing

A spatio-temporal feature and error processing technology, applied in feature engineering, wind energy forecasting, and data mining, can solve problems that are not universal and applicable to specific scenarios, and achieve better consistency, improved prediction accuracy, and strong stability Effect

Active Publication Date: 2018-11-06
TIANJIN UNIV
View PDF3 Cites 5 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0006] The present invention provides a wind energy prediction method that combines spatio-temporal features and error processing. The present invention can extract more effective features combined with integrated learni

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Wind energy prediction method which combines spatial-temporal features and error processing
  • Wind energy prediction method which combines spatial-temporal features and error processing
  • Wind energy prediction method which combines spatial-temporal features and error processing

Examples

Experimental program
Comparison scheme
Effect test

Example Embodiment

[0034] Example 1

[0035] In order to achieve the above objective, the embodiment of the present invention proposes a wind energy prediction method that combines temporal and spatial characteristics and error processing, see figure 1 , The method includes the following steps:

[0036] 101: Extract time features from wind energy time series, extract information between wind turbines that are closer together through a multiple-input-single-output model, and introduce spatial features;

[0037] On the one hand, the wind energy time series reflects the changes in the output power of the wind turbine over time, so feature data is extracted from the wind energy time series for training prediction models.

[0038] On the other hand, the embodiments of the present invention refer to wind turbines that are relatively close to each other as "neighbors". Extract the information between wind turbines that are closer together, and introduce spatial features.

[0039] The embodiment of the present ...

Example Embodiment

[0056] Example 2

[0057] The following combines specific calculation formulas, examples, and figure 2 The solution in Example 1 is further introduced, as detailed in the following description:

[0058] 201: In the process of training the VFMLEs model, the first step is to Group according to characteristic variance;

[0059] 202: Extract spatio-temporal features in corresponding groups. The embodiment of the present invention adopts a multiple-input-single-output mode, namely , Where X is the vector, the instantaneous feature, and y is the output. The spatiotemporal feature extraction methods are as figure 2 Shown.

[0060] Such as figure 2 As shown, for a single observation object nt i , The time feature is obtained from the data of a single wind turbine, and the tb sequence is obtained. At any moment, select several recent measurements in the past as features, and use the measured value of a specific time distance in the future as the output corresponding to the time feature. ...

Example Embodiment

[0085] Example 3

[0086] The feasibility verification of the schemes in Examples 1 and 2 is carried out below in conjunction with specific experimental data, as detailed in the following description:

[0087] The wind power forecasting problem is essentially a numerical forecasting problem, and there are general standards for evaluating numerical forecasting problems, such as: mean absolute error MAE, mean square error MSE, and root mean square error RMSE. Usually, the "error rate (percentage of error and actual value)" is used to judge the model. This method has certain defects. For example, the value of the error rate depends on the actual value. When the actual value is very small, even if the prediction error is small, the error The rate may also be large. Conversely, when the actual value is large, even if the model performs poorly, the error rate may be small. The embodiments of the present invention mainly use MSE to evaluate and compare experimental results. The calculat...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

PUM

No PUM Login to view more

Abstract

The invention discloses a wind energy prediction method which combines spatial-temporal features and error processing. The method comprises the following steps: extracting time features from a wind energy time sequence; extracting the information between the wind motors close to each other via a multi-input and single-output mode and introducing spatial features; preprocessing the spatial-temporalfeatures by a k-nearest neighbor based noise data detection method; analyzing the variance attributes of the preprocessed spatial-temporal features, and training multiple predictor models based on the result of the analysis; integrating the multiple predictor models by a weighted average method to generate an integrated learning model based on the spatial-temporal feature variance, which is usedto forecast the error of the integrated predictor model; using the integrated learning model to get a predicted value y, inputting the error features corresponding to the spatial-temporal features into an auxiliary model to get the result y ', then getting the final predicted value y + y'; and combining the integrated learning model with the auxiliary model to generate the final model.

Description

technical field [0001] The invention relates to the fields of data mining, feature engineering and wind energy forecasting, in particular to a wind energy forecasting method combining spatiotemporal features and error processing. Background technique [0002] Currently, machine learning algorithms for wind energy forecasting mainly include artificial neural networks, decision trees, and support vector machine regression. Since the problem of wind energy prediction itself is a problem of "predicting values ​​based on features", it has good compatibility with general machine learning methods, so that most of the commonly used machine learning methods include: random forest, neural network, and various regression Algorithms, etc. can be easily migrated and applied to this field. [0003] At present, it is impossible to prove that one model is better than another model: first, there is no generally accepted evaluation standard to judge each model, and second, to compare the eff...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

Application Information

Patent Timeline
no application Login to view more
IPC IPC(8): G06Q10/04
CPCG06Q10/04
Inventor 于瑞国喻梅于健赵满坤刘志强安永利
Owner TIANJIN UNIV
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap