A fan nonlinear load adaptive control method and device and storage medium

By constructing a dynamic neural network model with a hybrid structure of long short-term memory network and attention mechanism, and an adaptive model predictive controller, the problems of low prediction accuracy and slow response in wind turbine load control are solved, achieving efficient and stable control of nonlinear loads of wind turbines and improving the reliability and economy of wind power generation systems.

CN120949556BActive Publication Date: 2026-07-21GUIZHOU YAGUANG ELECTRONICS TECH +1
View PDF 2 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
GUIZHOU YAGUANG ELECTRONICS TECH
Filing Date
2025-07-17
Publication Date
2026-07-21

AI Technical Summary

Technical Problem

Existing wind turbine load control technologies are insufficient in processing complex dynamic data, resulting in low accuracy of wind turbine load prediction and slow control response, which cannot meet the requirements of modern wind power systems for efficient and stable operation.

Method used

A dynamic neural network model with a hybrid structure of long short-term memory network and attention mechanism is adopted. It combines an adaptive model prediction controller and an incremental learning algorithm with gradient constraints to update the model parameters in real time. The parameters are then normalized using the Sigmoid function to generate safety control quantities.

Benefits of technology

It improves the prediction accuracy and control response speed of wind turbine nonlinear loads, enhances the reliability and economy of wind power generation systems, and meets the requirements of modern wind power systems for efficient and stable operation.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN120949556B_ABST
    Figure CN120949556B_ABST
Patent Text Reader

Abstract

The application discloses a fan nonlinear load adaptive control method and device and a storage medium, and is used for efficiently realizing dynamic tracking control on a fan nonlinear load. The application comprises the following steps: collecting running state data of a fan system; constructing a dynamic neural network model comprising a long short-term memory network-attention mechanism hybrid structure; inputting the running state data into the dynamic neural network model; outputting a dynamic prediction value of the fan nonlinear load; constructing an adaptive model predictive controller based on the dynamic prediction value; generating a control sequence; calculating a prediction residual of the dynamic neural network model; when the prediction residual exceeds a standard threshold, triggering a model updating mechanism; using an incremental learning algorithm with gradient constraint to update parameters of the dynamic neural network model, and obtaining updated parameters; and outputting a safety control amount to a fan actuator, so as to realize dynamic tracking control on the fan nonlinear load.
Need to check novelty before this filing date? Find Prior Art