Combination of Text Mining and Corrective Neural Network in Short-term Load Forecasting

Short-term load forecasting refers to short period load prediction of utility ranging from one hour to several days ahead. It is meaningful in planning and dispatching the load to meet the electricity system demand. The inaccuracy load forecasting can increase the electricity operating costs. In this paper, a novel method is presented and discussed which combines text mining and corrective neural network (TM-CNN) methods. Subsequently, a numeric example of daily maximum load forecasting is used to illustrate the performance of TM-CNN method, and the experiment results also reveal that TM-CNN method outperforms the autoregressive moving average(ARMA) and BP Artificial Neural Network(BPNN) approaches.