摘要
常见的电价预测模型一般建立在时间序列法和神经网络法的基础上。本文将小波变换与自回归积分滑动平均模型结合起来得到小波ARIMA模型并使用该模型进行预测,相对其他时间序列方法,自回归积分滑动平均模型在处理电价这类非平稳时间序列时有更好的表现。经过预测误差的对比分析可以得知小波ARIMA模型的预测效果要优于传统的ARIMA模型。
Common electricity price prediction models are generally based on time series and neural network methods.In this paper,the wavelet transform and the autoregressive integral moving average model are combined to obtain the wavelet ARIMA model and used to predict.Compared with other time series methods,the autoregressive integral moving average model is better when dealing with non-stationary time series such as electricity price.The comparative analysis of the prediction error shows that the prediction effect of the wavelet ARIMA model is better than the traditional ARIMA model.
作者
张一泓
朱国荣
蔡永自
朱瑶琪
ZHANG Yi-hong;ZHU Guo-rong;CAI Yong-zi;ZHU Yao-qi(Economic Research Institute of Zhejiang Electric Power Company,Hangzhou 310000 China)
出处
《自动化技术与应用》
2020年第1期125-129,139,共6页
Techniques of Automation and Applications
基金
电力体制改革背景下的浙江交叉补贴机制研究（编号5211JY16000V）
