Forecasting of Iran Electricity Market Indices with Least Squares Support Vector Machines

Accurate forecasting of electricity load and price have been the most important subjects in the deregulated electricitymarkets. Modern data mining methods have played an effectiverole. Support vector machines (SVMs) have been successfully applied to solve nonlinear regression and time series problems.Least squares support vector machines (LS-SVM), a new type of machine learning technique based on statistical learning theory,can be used for time series prediction. We find the chaotic characteristics of the load and price series by analyzing. Mutual Information (MI) method is used to find the optimal time lag. With the optimal time lag and embedding dimension, LS-SVM is used to predict future load series of the load and price of Iran. Animportant issue for the performance of these models is the choice of the kernel parameters and the hyper-parameters which definethe optimal function which to be minimized. In this paper a new method for setting both the σ parameter of the RBF kernel and theregularization hyper-parameter is used.

کلیدواژه ها:

Least Squares Support Vector Machines ، Time Series Prediction ، Chaos Theory ، Hyper-parameters Optimization