- The "Infrared Brain" of State Grid Zhejiang Electric Power Research Institute: The independently developed "Intelligent Recognition Algorithm for Heating Defects in Infrared Images of Suspension Point Fittings of Transmission Lines" was launched on the UAV Autonomous Inspection Platform of State Grid Zhejiang Electric Power Company. This marks the full implementation of the AI intelligent recognition module for heating defects of the five major key components of the transmission lines of the Zhejiang power grid. This algorithm can automatically identify heating defects in infrared images in complex environments. Verified by actual combat, the average detection rate of the system exceeds 81%, and the false detection rate is lower than 1.7%. The analysis efficiency is increased by a hundred times compared with manual work. It only takes 1 hour to complete the intelligent diagnosis of 12,000 images per day, effectively ensuring the safety of the power grid. At present, the "Infrared Brain" has been applied in actual combat on the extra-high voltage transmission lines across Zhejiang Province, discovering 37 cases of heating defects and supporting the operation and maintenance units to complete the defect elimination.
The AI Power Supply Service Robot of State Grid Xingtai Power Supply Company: Based on the comprehensive integration of the power supply service grid and the community grid, an AI power supply service robot was innovatively developed and promoted and applied in nearly a thousand WeChat groups in 4 counties and districts including Nanhe District. It can monitor group messages 24 hours a day, automatically reply to customers' inquiries about electricity-related business with a response speed at the "second level", and automatically match the power supply stations to dispatch repair work orders according to the repair locations reported by customers. Since the robot "took up the post", the standardization, intelligence level and response speed of the company's power supply service have been comprehensively improved. The response rate of grid workers within 30 minutes has reached 99.26%, and the number of online work orders has decreased by 29% year-on-year.
