Abstract—Short-term electric load forecasting is significant for safe and economic operation in power system. In order to improve the accuracy of predicted results, this paper proposes a PSO-SVM model, which is based on cluster analysis techniques and data accumulation pretreatment. Specifically, K-medoids cluster analysis is employed to extract similar days, then original data is divided into k category clusters. In addition, this paper carries on the accumulation of the original data to weaken the effect of the irregular disturbance and enhance the regularity of the sequence. And we use PSO to optimize the parameters of SVM system. In case analysis, the results show that the method proposed in this paper can effectively promote the forecasting performance.
Index Terms—Short-term power load forecasting, data mining, K-medoids (KM), partial swam algorithm (PSO), support vector machine (SVM).
The authors are with the North China Electric Power University, Beijing, China (e-mail: wuhan930108@163.com, niudx@126.com, 961056925@qq.com).
[PDF]
Cite: Han Wu, Dongxiao Niu, and Zongyun Song, "Short-Term Electric Load Forecasting Based on Data Mining," International Journal of Engineering and Technology vol. 9, no. 3, pp. 250-253, 2017.
