Wind power random missing value imputation method and system

By using the multi-round ESN fitting-replacement reconstruction iterative processing of the RDESN network, the problem of balancing accuracy and efficiency in wind power random missing value interpolation is solved, achieving efficient and accurate data interpolation, which is suitable for wind power dispatch and power prediction.

CN122241032APending Publication Date: 2026-06-19NANCHANG UNIV
View PDF 4 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
NANCHANG UNIV
Filing Date
2026-05-20
Publication Date
2026-06-19

AI Technical Summary

Technical Problem

Existing technologies struggle to simultaneously balance interpolation accuracy and execution efficiency in interpolating random missing values ​​of wind power, thus failing to meet practical application requirements.

Method used

The RDESN network is adopted to process the raw wind power data through segmentation and location encoding. The ESN module and the replacement module are combined to perform multiple rounds of ESN fitting-replacement reconstruction iteration. The model error is evaluated using preset evaluation indicators to generate the target interpolation model.

Benefits of technology

This achieves a two-way improvement in interpolation efficiency and accuracy, ensuring the model's generalization performance and engineering applicability, and meeting the actual needs of wind power dispatch and power prediction.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122241032A_ABST
    Figure CN122241032A_ABST
Patent Text Reader

Abstract

This invention provides a method and system for imputing random missing values ​​in wind power data. The method includes: encoding each set of original datasets to construct corresponding location codes; reconstructing the original wind power data based on the location codes to generate a corresponding initial validation set and activating an adapted RDESN network; performing multiple rounds of ESN fitting-replacement reconstruction iterative processing on the initial validation set through an ESN module and a replacement module to output a corresponding target validation set; training the RDESN network using the original wind power data and the target validation set to generate a corresponding RDESN model; evaluating the prediction error of the RDESN model using preset evaluation metrics; and generating a corresponding target imputation model after the prediction error meets preset requirements, and using the target imputation model to complete the imputation of missing values ​​in the original wind power data. This invention can accurately complete the imputation of missing wind power data.
Need to check novelty before this filing date? Find Prior Art