Actively research and apply artificial intelligence in power grid management and operation

Over the years, Quang Tri Power Company (PC Quang Tri) has actively researched, built and applied science and technology in the fields of automation of operation, inspection, assessment of power grid quality, construction investment management and customer service business, contributing to improving the company's production and business activities.

Some images of unsafe power grids detected by drones/UAVs - Photo: TN

PC Quang Tri is a pioneer in the Vietnam Electricity Group (EVN) to participate in research and development and be recognized for its initiative with an integrated version of field information software and grid engineering management software. In particular, the company has conducted research in the field of artificial intelligence (AI) in automation to detect abnormal phenomena of the power line system and transformer stations on the operating grid using images.

Some of these programs include: Automatically checking objects of interest in images captured and stored on the construction investment management system (EVN-IMIS). This program has helped automate the inspection and analysis of images captured annually at investment projects; or the application program of artificial intelligence, automatically detecting abnormal heat through images captured from energized devices on the grid. The program automatically analyzes and issues warnings to help technical staff have appropriate solutions to handle those abnormalities to prevent possible electrical incidents.

In 2022, PC Quang Tri researched and applied AI to detect grid safety risks from images/videos collected by flying drones. Although the electricity industry has applied many programs to serve the management and operation of the grid such as grid management software (PMIS), medium voltage field inspection (KTHT) with the purpose of digitizing the inspection of lines and transformer stations, however, the detection of existence according to images from PMIS and KTHT programs is still done by the naked eye.

This method takes a lot of time to detect from images and videos. Therefore, images and videos collected from flycam/drone devices will be synchronized into the PMIS-AI program and automatically analyzed and detected risks of grid safety instead of workers performing visual inspections or binoculars. Therefore, the application of AI models in detecting grid safety risks from images/videos collected from flying drones has brought positive effects in the management and operation of the grid.

In order for the system to operate with high accuracy, in addition to building models, standardizing data, labeling objects and training the object recognition program, the company has applied the Yolov5 model solution to the PMIS-AI program.

With this model, the processing time of a 4MB image takes only 1/10 of a second. Therefore, Quang Tri PC is a unit that has taken a step forward in participating in research in this field, especially with many solutions proposed for widespread implementation. Typically, the program of applying artificial intelligence in image recognition in the construction steps of the field of construction investment management, automatic recognition of thermal cameras for units under the Central Power Corporation is highly appreciated and effectively applied in practice.

In 2024, the topic "Research and application of artificial intelligence to detect risks of power grid insecurity from images/videos collected by drones/UAVs from flight missions" by the group of authors: Masters Phan Van Vinh, Nguyen Van Tai, Le Cong Hieu, Le Van Minh, Nguyen Xuan Thuy of PC Quang Tri won the Second Prize at the 17th National Technical Innovation Contest (2022-2023) organized by the Vietnam Union of Science and Technology Associations, Vietnam Fund for Supporting Technical Innovation (VIFOTEC) in the fields of: information technology, electronics, telecommunications.

With the AI ​​application solution to detect risks of power grid insecurity from images/videos collected by drones/UAVs, programming automatic flight paths according to flight tasks of PC Quang Tri belongs to the category of AI recognition software combined with data analysis to give warnings and detect risks of power grid insecurity from images/videos collected by flying drones.

Applying Yolov8 artificial intelligence model, other supporting tools (LabelMe for labeling, Google Colab for training) to detect existence/abnormalities of 110kV, 22kV lines through images and videos collected from flycam/drones, specifically focusing on detecting objects of frayed bare conductors, loose porcelain neck ties, dirty, broken, cracked insulation and other abnormal objects on the grid.

Automated flight path programming for drones flying over the power grid is an advanced technology in the field of safety and efficiency of power grid monitoring. The system is designed to monitor the power grid automatically and continuously, while providing full information to detect risks of power grid safety. The solution helps increase the efficiency of power grid safety monitoring; save costs; reduce monitoring costs; increase accuracy; increase operational efficiency; reduce time and manpower.

With the aim of maximizing the power of digital technology to improve the efficiency of technical management and ensure the safety of grid operations, the research and application of AI in technical management is an inevitable trend. Because this will greatly contribute to improving labor productivity and the efficiency of power quality management. Thereby, providing a stable and safe power source to serve the socio-economic development of the locality.

Tan Nguyen

Source: https://baoquangtri.vn/tich-cuc-nghien-cuu-ung-dung-tri-tue-nhan-tao-trong-quan-ly-van-hanh-luoi-dien-189890.htm