DEVELOPMENT OF POWER LINE DETECTION ALGORITHM AND BACKEND FOR AI SYSTEM TO DETECT TREES WITH POTENTIAL DAMAGE TO POWER LINE

• Main Author: Pambudi Adhiluhung, Gelar
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/55327
• Institution: Institut Teknologi Bandung

The high demand for electricity for daily activities requires the electricity to always be available. One of the causes of electrical disruption is the presence of trees or vegetation growing on a Right of Way of power line. Good vegetation management is needed to ensure that the power grid is not disturbed. Some of the problems that often arise when carrying out the vegetation management process are difficult access to locations and the mapping process sometimes takes a long time. The AI system for detecting trees that interfere with power lines, named Druma, was developed as a solution for managing power line vegetation. The system is developed as a web application so that user can access the system easily with internet connection. Users don’t need PC with specific requirements to run the system. The system will detect the power line and trees on a satellite or UAV imagery image and calculate the distance from each tree to a power line. Trees within dangerous area will be marked with a certain color by the system. The main parts of the system to detect tree and power line will be created as a backend. The development of the power line detection algorithm is based on Hough Transform. Backend testing is carried out to ensure the backend is working properly. Testing the power line detection algorithm is performed by calculating and comparing the precision and recall of the results for each algorithm created. Based on the test results, the system’s backend can satisfy its functionality specification. The power line detection algorithm test results shows that algorithm II that used in this system has highest precision and recall.