A REVIEW OF DEEP LEARNING FORECASTING BY USING WIND ENERGY

A REVIEW OF DEEP LEARNING FORECASTING BY USING WIND ENERGY

Padma Pandranki 1, Saipriya Laveti2, D. Rajesh Babu3*

1,2B. Tech Student, Department of EEE, GMR Institute of Technology, Rajam-532127, Andhra Pradesh, India

3*Assistant Professor, Department of EEE, GMR Institute of Technology, Rajam-532127, Andhra Pradesh, India

Email: pandrankipadma2005@gmail.com 1

---------------------------------------------------------------------***---------------------------------------------------------------------

Abstract - The "Convolution-based Spatial-Temporal Wind Power Predictor" (CSTWPP) is a new deep learning method designed to forecast wind power and support energy management. It combines data about location and time to improve prediction accuracy. CSTWPP uses convolutional neural networks (CNNs) to study past wind data from various places. These CNNs help detect patterns related to wind direction and landscape effects that traditional methods might miss. The model also uses recurrent neural networks (RNNs) to capture changes in wind conditions over time. By merging location-based and time-based data, CSTWPP forecasts wind power output more precisely. This accuracy is essential for managing energy effectively. Utility companies and grid operators can use these predictions to plan energy distribution, decrease dependence on fossil fuels, and make the best use of wind energy. Better wind power forecasting supports grid stability and helps prepare for power fluctuations. In short, CSTWPP combines deep learning, spatial-temporal data analysis, and CNNs with RNNs to make accurate wind power predictions. These forecasts improve energy management, encouraging sustainable and reliable wind energy use.

 Keywords: Convolution neural network, deep learning, incremental learning, short term wind power forecast, Spatial-Temporal correlation.