Aiming at improving the safety management level and operational efficiency of power plant, an intelligent video analysis platform based on edge artificial intelligence technology is proposed. By integrated applications of artificial intelligence, machine learning and edge computing, the platform is structured of three components, named as end device, edge computing and cloud. We use front-end sensing devices to collect scene video and other data, by using traditional computer vision, neural networks, deep learning and other algorithms to analyze the collected data, to discover as far as possible the irregular behavior of personnel, equipment and environmental abnormalities. The platform forms corresponding event and alarm information and actively pushing alarm results, improving intrinsic safety management and control capabilities. Moreover, the container technology is used to expand the platform flexibility. The cloud supports remote update and replacement of edge-side functions. Based on cloud-edge collaboration, intelligent release can be directly realized, which guarantees the reliability, availability and scalability of the platform.

暂无