Variable-structure neural network with real-coded genetic algorithm and its application on short-term load forecasting

S.H. Ling, F.H.F. Leung, H.K. Lam, Herbert Iu

    Research output: Contribution to journalArticle

    18 Citations (Scopus)


    This paper presents a novel neural network with a variable structure, which is trained by a real-coded genetic algorithm (RCGA), and its application on short-term load forecasting. The proposed variable-structure neural network (VSNN) consists of a Neural Network with Link Switches (NNLS) and a Network Switch Controller (NSC). In the NNLS, switches are introduced in the links between the hidden and output layers. By using the NSC to control the on-off states of the switches in the NNLS, the proposed neural network can model different input patterns with variable network structures. It gives better results and learning ability than the fixed-structure network with link switches (FSNLS) [3], wavelet neural network (WNN) [25] and feed-forward fully-connected neural network (FFCNN) [9]. In this paper, an improved RCGA [2] is used to train the parameters of the VSNN. An industrial application on short-term load forecasting in Hong Kong is given to illustrate the merits of the proposed network.
    Original languageEnglish
    Pages (from-to)23-40
    JournalInternational Journal of Information and Systems Sciences
    Issue number1
    Publication statusPublished - 2009


    Cite this