Mid-term load forecasting based on dynamic least squares SVMS

In this study, a dynamic model based on least squares support vector machines is proposed to forecast the daily peak loads of a month. The model function is got from the training data set using least squares support vector machines. In the time series prediction process, new data points are included into training data set and some of the old ones are deleted, so as to track the dynamics of the nonlinear time-varying feature of load demand. The electricity load data from European Network on Intelligent Technologies (EUNITE) network competition are used to illustrate the performance of the proposed dynamic least squares support vector machines. The experimental results reveal that the proposed model outperforms the least squares support vector machines, which outperforms the support vector machine. Consequently, the dynamic least squares support vector machines provides a promising alternative for forecasting mid-term electricity load in power industry.