基于粗糙集的电池储能电站海量数据处理方法

锂离子电池的储能电站在运行过程中上传的数据量庞大，数据采样率高，增大了实时在线评估工作的难度。如何提取并压缩有效数据，同时保证压缩后的数据具有保真性，成为数据挖掘预处理所面对的主要问题。针对上述问题，选取储能电站特定工况下一个电池簇的数据，利用粗糙集方法，对数据的属性进行约简，根据正态分布的$ 2sigma $原则对属性值进行二元逻辑划分，将单体电池划分为频繁检测对象和普通检测对象，压缩了在线处理的数据量。最后，对同一电池簇20天内不同工况数据进行验证，证明了该处理方法的有效性。

Massive Data Processing Method for Battery Energy Storage Power Stations Based on Rough Sets

During the operation of energy storage power stations with lithium-ion batteries,huge amounts of uploaded data and highfrequency data sampling increase the difficulty of online real-time evaluation.How to extract and compress effective data and ensure the fidelity of the compressed data has become the main content of preprocessing in data mining.Given the above problems,the attributes of the data from a battery cluster are first reduced by the method using rough set theory.The battery cluster is selected from the energy storage power station under a specific working condition.Then,according to the 2σprinciple of normal distribution,the attribute values are divided on the basis of binary logic,and the battery cells are classified into frequent detection objects and infrequent detection objects,which reduces the amount of data processed online.Finally,the data of the same battery cluster under different working conditions in 20 days are taken to verify the effectiveness of the processing method.