Sensors Fusion and Multidimensional Point Cloud Analysis for Electrical Power System Inspection

Area of Science:

• Engineering
• Computer Science
• Materials Science

Background:

• Thermal inspection is vital for early component diagnosis.
• Infrared reflections from external sources can distort thermal readings.
• Manual camera positioning is time-consuming and costly.

Purpose of the Study:

• To develop an autonomous system for identifying and filtering infrared reflections.
• To enhance the accuracy of thermal inspection data.
• To reduce the operational costs and time associated with thermal inspections.

Main Methods:

• Acquiring data from multiple observation points (OPs) using stereo and thermal cameras.
• Fusing 3D point cloud and thermal image data at each OP.
• Generating a temperature profile for each spatial point by combining data from multiple poses.
• Filtering erroneous readings caused by reflections and other interference.

Main Results:

• A dense point cloud with accurate spatial position and temperature for each point.
• Successful identification and filtering of infrared reflections.
• Demonstrated improved performance in real-life conditions using a Directional Robotic System (DRS).

Conclusions:

• The proposed autonomous system effectively mitigates issues caused by infrared reflections in thermal inspections.
• Data fusion from stereo and thermal cameras enables robust temperature profiling.
• The Directional Robotic System (DRS) facilitates autonomous tracking and inspection, enhancing efficiency.