摘要
针对电力公司在电力物资供应过程中可能会产生供需关系不平衡,进而造成社会或者电力公司经济损失的情况,以及由于电力需求具有周期性、季节性的特点,文中提出一种基于SARIMA与ANFIS的组合预测模型,即SARIMA-ANFIS模型。该模型首先单独使用SARIMA与ANFIS这两种预测方法对电力物资需求分别进行预测,然后使用BP神经网络将这两种预测得到的结果进行整合,得到最终的预测结果。最后将该方法运用到某电网公司中,结果验证了该方法的可靠性及准确性。文中提出的预测模型可以较好地获取电力需求数据中的线性特征、非线性特征和季节性特征,对于具有此类特征的数据具有较高的预测精度。
Since the imbalance between supply and demand of power materials may cause economic loss to the society or power companies, and the power demand has cyclical and seasonal characteristics, in this article a combined prediction model based on SARIMA and ANFIS, that is the SARIMA-ANFIS model, is proposed. Firstly, SARIMA and ANFIS are used separately to predict the demand of power materials. Then, the BP neural network is adopted to integrate the prediction results for final calculation. Finally, the method is applied to a power-grid company, and the results have verified that this method is reliable and accurate. Thanks to the prediction model, the power-demand data’s linear, non-linear and seasonal characteristics can be identified in a desirable manner, with higher accuracy in prediction.
作者
宋鑫磊
黎莫林
詹勤辉
林东明
杨雅茗
李慧
SONG Xin-lei;LI Mo-lin;ZHAN Qin-hui;LIN Dong-ming;YANG Ya-ming;LI Hui(Shenzhen Power Supply Bureau Co.,Ltd.,Senzhen 518000)
出处
《机械设计》
CSCD
北大核心
2022年第6期66-72,共7页
Journal of Machine Design
基金
基于数字画像与智能分析技术的物资仓储配送管理方法研究(2019GZX-0025)。
作者简介
宋鑫磊(1985-),男,工程师,学士,研究方向:供应链管理。E-mail:songxinlei@sz.csg.cn。
