本文已被：浏览 811次 下载 510次
投稿时间：1996-09-19
投稿时间：1996-09-19
中文摘要: 本文基于在线辨识理论,用AR模型作为即时模型,通过最小二乘法的参数识别、Stoica方法对模型适用性检验和方差最小选优等措施使模型完善化。在预报过程中不断采用新的数据信息修正模型并将新信启、数据纳入模型进行预报,以每15分钟为间隔的短期负荷进行了模拟预报,其相对于均误差在1.2％左右,单点负荷预报值相对误差小于3％的精度概率为95％。
Abstract:Based on the on - line recognization theory, this paper ases the AR model as real - time model and least square method as model recongnization method. The Stoica method is used for model adjustment purpose. The minimum variance is used for model optimation. During the forecast, new data are continuously input into the model for updating and forcasting. The method in this paper has made forecasts on the load in 15 minutes intermission. The relative average error is about 1.2%. The relative error of single-point load forecast is less than 3% with the probability of 95%.
keywords: AR model on-line forecast short一term load
文章编号：19960410 中图分类号： 文献标志码：
基金项目:
| Author Name | Affiliation |
| Ding Hui-kai | Shanghai Institute of Electric Power |
| Liang Jian-ming | Guangxi Hechi Bureau of Power Supply |
引用文本：
丁会凯,梁建敏.用于短期负荷在线预报的AR模型建立与实例分析[J].上海电力大学学报,1996,12(4):58-63.
Ding Hui-kai,Liang Jian-ming.AR Model's Creation and Example Analysis on On-line Forecast for Short - term Load[J].Journal of Shanghai University of Electric Power,1996,12(4):58-63.
丁会凯,梁建敏.用于短期负荷在线预报的AR模型建立与实例分析[J].上海电力大学学报,1996,12(4):58-63.
Ding Hui-kai,Liang Jian-ming.AR Model's Creation and Example Analysis on On-line Forecast for Short - term Load[J].Journal of Shanghai University of Electric Power,1996,12(4):58-63.