dimension corresponds to the number of input variables. of the deterministic multiplayer perceptron (MLP) model of the reconstructed time series [61, [7].A new approach to short-term load forecasting (STLF) in power systems is described in this paper. The methodThe MLP has been trained and then has been used for uses a chaotic time series and artificial neural network.load forecasting in Greek Electric Power System. The paper describes chaos time series analysis of daily The paper is organized as follows: in the next section power system peak loads. Nonlinear mapping of restoring the chaotic time series is considered. Then, in deterministic chaos is identified by multiplayer Section 111, it is explained how to use the so-called perceptron (MLP). Using embedding dimension and correlation dimension in order to determine the delay time, an amactor in pseudo phase plane and an embedding dimension of the time series. In Section IV is ANN model trained by this amactor are constructed. The shown how to use the Lyapunov s p e c " analysis and proposed approach is demonstrated by an example. more specially the largest Lyapunov exponent for determining the embedded dimension. In Section V, following the data for Greek Electric Power System, some experiment results are given. Some conclusions are made in the lasf section'

No abstract

No abstract

The multicompanion controllable form of a multivariablo system (A, B, 0) is used to derive an algebraic description of the (A, B).invariant and controllability eubepecee of the system. The largest such aubspacea in Ker (0) arc also determined.

No abstract