Research and Implementation of Data Feature Extraction Technology for Multisource Heterogeneous Data in Electric Distribution Network

With the development of information technology, the degree of information of electric power enterprises is getting higher and higher, and the data generated by business systems is also growing rapidly. Data has become the new power of enterprise production and plays a vital role in the business growth of power companies. State Grid Corporation of China has brought the benefits of rapid business growth by using data, but at the same time, it is also faced with data redundancy, difficult data discovery, low efficiency, resource consumption, and other issues, as is the case in the power industry. In the power industry, the data preparation stage of the distribution network will also occupy most of the time and energy of data scientists. The high degree of specialization of data extraction and feature engineering work will result in a high threshold, which will increase obstacles for business personnel to participate in data value exploration. As long as traditional automatic data extraction and automatic feature engineering are based on business experience, data analysis and data extraction must be carried out from top to bottom, which is far from enough in the era of artificial intelligence. In order to reduce the technical threshold of data utilization, this paper fully integrates data science with power grid business, carries out research on multisource heterogeneous data extraction and automatic feature engineering technology, and realizes intelligent data management, improves data availability, improves data management level, and reduces the threshold of data utilization through automatic feature engineering methods, Finally, it supports typical business scenarios such as Internet of data management or power consumption information acquisition data management of State Grid Corporation of China.