A two-stage attention MCNN-BiLSTM short-term photovoltaic power prediction method and system

By using a two-stage attention MCNN-BiLSTM model, combined with multi-scale dilated convolution and BiLSTM temporal modeling, the problem of feature fusion and hyperparameter optimization under complex meteorological conditions in photovoltaic power prediction is solved, achieving higher prediction accuracy and stability.

CN122393924APending Publication Date: 2026-07-14LANZHOU UNIVERSITY OF TECHNOLOGY
View PDF 0 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
LANZHOU UNIVERSITY OF TECHNOLOGY
Filing Date
2026-06-17
Publication Date
2026-07-14

AI Technical Summary

Technical Problem

Existing photovoltaic power prediction methods struggle to coordinate local and global temporal features and integrate information from multiple time scales under complex weather conditions. Furthermore, hyperparameter configuration relies on human experience, resulting in insufficient prediction accuracy and stability.

Method used

A dual-stage attention MCNN-BiLSTM model is adopted, which combines a multi-scale dilated convolution feature extraction module, a BiLSTM temporal modeling module, and a sparrow search algorithm optimization module. Multi-scale dilated convolution is used to extract multi-scale fluctuation features of photovoltaic power, BiLSTM enhances the global context information modeling capability, the dual-stage attention mechanism improves the key feature recognition capability, and the hyperparameters are optimized by the sparrow search algorithm.

Benefits of technology

It improves the accuracy and stability of photovoltaic power prediction. By fusing multi-scale feature extraction and bidirectional time-series dependence, it reduces the subjectivity of the model and improves the prediction performance and the rationality of parameter configuration.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122393924A_ABST
    Figure CN122393924A_ABST
Patent Text Reader

Abstract

The application discloses a kind of two-stage attention MCNN-BiLSTM short-term photovoltaic power prediction method and system, and relates to photovoltaic power generation technical field.The method comprises: obtaining photovoltaic power station dataset;Build two-stage attention MCNN-BiLSTM model framework, use the historical time series dataset after pre-processing to the model framework training, obtain the training completed optimal photovoltaic power prediction model;The meteorological characteristic data to be predicted of target photovoltaic power station is input into the training completed optimal photovoltaic power prediction model, and the future time photovoltaic power prediction value is output.The prediction framework constructed in the application has achieved significant improvement in prediction accuracy and stability, and can provide technical support for renewable energy and fine management of renewable energy under complex weather conditions.
Need to check novelty before this filing date? Find Prior Art