A Reinforcement Learning Based Approach for Welding Sequence Optimization

Abstract

We develop and implement a Q-learning based Reinforcement Learning (RL) algorithm for Welding Sequence Optimization (WSO) where structural deformation is used to compute reward function. We utilize a thermomechanical Finite Element Analysis (FEA) method to predict deformation. We run welding simulation experiment using well-known Simufact® software on a typical widely used mounting bracket which contains eight welding beads. RL based welding optimization technique allows the reduction of structural deformation up to ~66%. RL based approach substantially speeds up the computational time over exhaustive search.