Developing machine learning based approaches to weld residual stress problems

Determining weld residual stress from welding simulations is computational expensive and sensitive to key process parameters, such as material properties. Conversely, when experimental coupons are required to undertake tests which consider the influence of weld residual on properties it is not feasible to estimate coupon properties on a case by case basis. Tis project bridges this gap by utilising the rapidly developing field of machine learning and artificial intelligence to use simplified procedures to generate and inform suitable test specimens representative of real welding processes. Different algorithms, different validation measurements and different processes and case studies will all be considered to produce a candidate methodology.