摘要
针对电力用户数量众多、数据量大、计算量大等特点,提出了基于Hadoop处理框架的大数据技术解决方法;针对用户与系统负荷存在的差异,在用户负荷分析和影响因素分析方面,引入了在处理大数据量和知识学习等方面具有独特优势的数据挖掘技术,为大幅提升预测模型的使用效率奠定了基础,从而大幅提高了短期负荷预测精度。与传统的系统负荷预测方式相比,实例证明该方法具有明显优势。
With the wide installation of smart meters and large-scale application of Customer Power Consumption Information Collection System,power user information has achieved ' full coverage and full collection',and provided the data foundation for the user-based short-term load forecasting. Considering the large quantity of power consumers,data and calculation,this paper proposes customer load forecasting solution ideas based on big data technical architecture. On the problem of customer load analysis and factors impact on load,studies have been done to deal with large volumes of data,knowledge learning and data mining techniques to enhance the efficient use of forecast models to improve prediction accuracy. The application has verified the superiority of the proposed method to the traditional load forecasting method.
出处
《华东电力》
北大核心
2014年第10期2007-2010,共4页
East China Electric Power
