Design and Implementation of Artificial Intelligence Platform Dedicated to Power System
Jingpo Bai,Yu Yin,Jing Liu,Yu Qin,Yingbao Cui
DOI: https://doi.org/10.1109/icsai53574.2021.9664153
2021-11-13
Abstract:With the rapid development of the digital age and the tremendous changes in traditional industries, global artificial intelligence (AI) has become a general trend. In the power system, the combination of AI technology and traditional power domain has greatly promoted the technological progress of the digital transformation of the power industry. However, the difficulty of AI algorithm research and development, the lack of unified management of data and models, and other issues restrict the wide application of AI technology in power systems. Establishing a unified AI platform for the power system and providing a unified AI model operation service support has become the key to power system AI technology progress. In this work, we propose a power system AI platform design and implementation plan, which provides the full-chain functions from model production to service formation, including sample processing, model training, and shared release. Specifically, the platform includes three modules—data set, model set, and learning environment—which have functions such as model deployment, data management, service release, and service management, etc. The data set includes functions such as data preprocessing, tag management, and metadata management, and has multi-person collaborative service interfaces. The model set includes model evaluation, model compression, model resources and catalog service interfaces. The learning environment includes functions such as custom algorithm management, automated model training, resource scheduling, etc. The results of applying AI technology in typical business scenarios such as power transmission, power transformation, and safety supervision verify the platform's full-chain support capability in the model formation process.