What question did this study set out to answer?

The study aims to optimize laser paint stripping processes using hyperspectral imaging and machine learning techniques.

March 23, 2026Procedia Computer Science0 citationsOpen Access

Achieving selectivity of laser paint stripping by using a convolutional neural network in combination with Hyperspectral imaging

HBHauke BrüningFraunhofer Institute for Manufacturing Technology and Advanced Materials JAJannes AdamFraunhofer Institute for Manufacturing Technology and Advanced Materials RSRüdiger SachsFraunhofer Institute for Manufacturing Technology and Advanced Materials

Key Points

The study aims to optimize laser paint stripping processes using hyperspectral imaging and machine learning techniques.
Utilized hyperspectral imaging for surface analysis of aircraft paint systems.
Implemented convolutional neural networks (CNNs) for processing hyperspectral data.
Compared the performance of CNNs to Random Forests in paint identification and removal tasks.
CNNs demonstrated improved accuracy in identifying paint materials compared to Random Forests.
Laser treatments were performed with enhanced speed and efficiency.
The integration of hyperspectral imaging into laser processes reduced the reliance on hazardous chemicals.

Abstract

Laser processes hold significant potential in the aircraft industry, especially for paint removal during Maintenance, Repair, and Operation (MRO), which is typically manual and costly. This study explores automated laser-based paint stripping, a promising alternative that can enhance efficiency and reduce the use of hazardous chemicals and labor costs. To optimize laser parameters for effective paint removal, hyperspectral imaging is proposed as a surface analysis tool, providing vital material properties that inform subsequent processing steps. Hyperspectral imaging has shown efficacy across various applications, including paint identification, through machine learning models like Random Forests and Convolutional Neural Networks (CNNs). This paper compares these models’ performance, highlighting the adaptability of CNNs for real-time hyperspectral data processing, enabling precise laser treatments of aircraft paint systems. The results demonstrate improved accuracy and speed, underscoring hyperspectral imagings potential in automated paint removal processes.

AI से पूछें

Bookmark

View Full Paper