14th International Conference on Shot Peening

Application of Deep Learning to Predict Shot Peening Coverage

Aldrich Chuaa, Wang Zhenbiaoa, Ang Huichena, Muhammad Azrul Shukri Azmi

Advanced Remanufacturing and Technology Centre, 3 Cleantech Loop, #01/01 CleanTech Two, Singapore 637143

Corresponding author. Tel.: +65 9852 6166. Email address: aldrich-chua@artc.a-star.edu.sg

Introduction

Coverage is one of the key output parameters for shot peening, and is described as the percentage of a peened area over the total surface area of the component. Coverage is typically represented as a percentage value and it is a requirement to obtain a minimum of 98% coverage for the shot peening process according to AMS specifications [1].

Coverage is typically determined through operators’ visual assessments with the aid of magnifiers of minimum 10x magnification, but this method can be subjective and inconsistent between operators, especially while determining high coverage percentages on complex features. Deep learning has been widely used in many applications to detect patterns, especially in image analysis.

This study intends to investigate how deep learning can be applied to predict shot peening coverage, and propose how it can be applied to industry.

Objectives

The objective of this paper is to investigate the effectiveness and limitations of using deep learning to predict different coverage ranges based different peening intensities and materials.