• Login
    JavaScript is disabled for your browser. Some features of this site may not work without it.

    Browse

    All of DSpaceCommunities & CollectionsBy Issue DateAuthorsTitlesSubjectsThis CollectionBy Issue DateAuthorsTitlesSubjects

    My Account

    LoginRegister

    Statistics

    View Usage StatisticsView Google Analytics Statistics

    Wind energy forecasting using recurrent neural networks

    Thumbnail
    View/Open
    201611057 (1.193Mb)
    Date
    2018
    Author
    Rani, Neha
    Metadata
    Show full item record
    Abstract
    Wind energy has the potential to meet all our electricity demands and is a cost effective source of energy. Power operators dealing with the electricity generation from wind energy often face many problems due to the fluctuations in the wind energy and the uncertainty associated with it . Therefore, system operators need to have enough resources which can ensure the proper functioning of the system in the situation of fluctuating wind energy generation. Proper wind energy forecasting is a crucial part of the smart energy grid. The new era of machine learning has become very popular due to its fast training and good forecasting performance. This capability of machine learning techniques can be applied to predict the wind energy. In this thesis work, an efficient recurrent neural network based forecasting of wind energy is being proposed. ELMAN network is developed for short term forecast specifically, we consider 24 hour ahead forecast. An ELMAN neural network is a kind of recurrent neural network that incorporates the dynamic dependency in the data due to the presence of the feedback path present in the network. It has three main layers i.e, an input layer, hidden layers and the context layer that captures the dynamic behaviour in the time series. Supervised learning method is used for the forecasting purpose using themeteorological data as the training features. Three ELMAN networks having different input feature vectors are developed i.e. two weather sensitive networks having input vector of size (4 1) and (7 1) respectively and a non- weather sensitive network having input vector of size (4 1) and these networks are compared with other conventional methods. Experimental results show that there is a significant reduction in the evaluation criteria i.e. Root Mean Square Error (RMSE) and Mean Absolute Percentage Error (MAPE) for the proposed method when compared to other approaches.
    URI
    http://drsr.daiict.ac.in//handle/123456789/773
    Collections
    • M Tech Dissertations [820]

    Resource Centre copyright © 2006-2017 
    Contact Us | Send Feedback
    Theme by 
    Atmire NV
     

     


    Resource Centre copyright © 2006-2017 
    Contact Us | Send Feedback
    Theme by 
    Atmire NV