Parameters Estimated Subsection Predicting Model for Electric Power Consumption in the Spring Festival

Because of the Spring Festival effect, it´s hard for traditional methods to predict monthly electricity consumption in January and February. This paper presents a specialized subsection-model which can better reflect the situation to solve this problem. In terms of the holiday condition, the model divides the days of the two months into two parts: the usual electricity days and holiday electricity days. In addition, an algorithm is proposed to estimate the parameters of the subsection-model according to data distribution. The data of similar situation are rare because the date of Spring Festival is different every year, the algorithm overcomes this difficulty. Experiments conducted on real dataset prove the proposed model is effective and accurate.