基于数据中台的电力数据报表模型研究与应用

Research and application of power data report model based on data center

针对风、光等可再生能源发电不断并入电力系统各个地域,使配电网分散协调的主动性日益增强,由此使传统报表系统存在数据源单一、无法互联和动态扩展的问题,在电力系统全时空量测大数据环境下,基于数据中台,利用深度学习理论中的双向长短时记忆网络方法,提出了电力系统动态数据报表的设计方法以解决数据源互联、动态扩展问题。对此,给出了大数据报表系统的四种传统结构,并基于电力系统各个子系统大数据分散、分布的结构,以目前报表系统的框架为基础,搭建了动态报表系统的结构模型;在此基础上,依次设计了动态报表系统中的数据贴源层、数据共享层、数据分析层、动态数据格式层,总结了各层的特点、实现方式和优点,使报表能够实现多数据源之间协同、扩展和智能推测功能,通过实际测试验证了这些功能。

Wind,light and other renewable energy power generation continue to be integrated into various regions of power system,which makes the distribution network more and more active in decentralized coordination.As a result,the traditional report system has the problems of single data source,no interconnection and dynamic expansion.In the environment of power system full time and space measurement big data,based on the data center,the bidirectional long short time memory network in deep learning theory is used methods,the design method of power system dynamic data report is proposed to solve the problems of data source interconnection and dynamic expansion.In this paper,four traditional structures of big data report system are given,and based on the decentralized and distributed structure of big data in each subsystem of power system,the structure model of dynamic report system is built based on the framework of current report system.On this basis,the data pasting layer,data sharing layer,data analysis layer and dynamic data grid of dynamic report system are designed in turn.This paper summarizes the characteristics,implementation methods and advantages of each layer,so that the report can realize the functions of collaboration,expansion and intelligent speculation among multiple data sources.The actual test shows these functions.