Wind Power Forecasting Using K-Means Clustering and Convolutional Neural Network

EasyChair Preprint 2732

Abstract

Wind energy is a renewable energy that is free, clean and readily available. They also cause less impact to the environment. However, the random nature of the wind and its associated uncertainty create challenges in dealing with the generation of wind power effectively, which can result in unnecessary restrictions to the potential of a wind farm. Wind power forecasting is essential for energy trading and can also be used to decide on economic dispatch based on the forecast. This system makes use of data mining techniques and deep learning to predict the wind power by the combined approach of K-Means Clustering and Convolutional Neural Network. At first, the dataset is preprocessed in MATLAB using Self Organizing Map for dimensionality reduction. Then K-Means is used for clustering the datasets based on meteorological conditions, historical power data and wind turbine parameters. Finally, the cluster is used for training and testing the Convolutional Neural Network to generate the result.

Keyphrases: Convolutional Neural Networks, Data Mining, K-means clustering, MATLAB, Self-Organizing Map, deep learning

@booklet{EasyChair:2732,
  author    = {M. Bhavani and S. Mohith Vasan and S. Muthu Kumar and N. Padma Gokul},
  title     = {Wind Power Forecasting Using K-Means Clustering and Convolutional Neural Network},
  howpublished = {EasyChair Preprint 2732},
  year      = {EasyChair, 2020}}