Research on Deep Learning-based AI Information Extraction Methods of Substation Engineering Design

Substation engineering construction has accumulated a large amount of documents and a wealth of explicit and implicit knowledge in the long-term work practice. Using artificial intelligence technology to extract knowledge from unstructured documents, drawings and photos into structured data, is a prerequisite for providing guidance for new engineering design. Through the research on application of artificial intelligence in the design of substation engineering, two functions are realized in the structuring of data: 1) using deep learning-based image recognition technology (SSD_mobilenet), the elements intelligent identification and quantity statistic in engineering drawings is realized. It provides a possibility of automatic generation for the structuring of design data. 2) through computer vision and machine learning technology (CTPN and CRNN), an electrical information recognition system is established to errata automatically, which saves manpower and time and provides auxiliary functions for data collection operations of previous engineering projects. The implementation results verify the effectiveness of the deep learning methods using in substation engineering design.