Physics of Degradation based Surrogate Model using Deep Learning for Lifetime Prognostics of Solder Joints in Microelectronics

At SAE L3 level and higher, there are significant requirements for fault-tolerance of the electronic system. Due to the automotive constraints in terms of cost, space, weight, energy consumption etc., the use of hardware redundancy for meeting these requirements is not suitable. Health management based on predicting the remaining useful life (RUL) of individual electronic components offers a promising alternative. A measure of component degradation forms the basis of the required prognostics and health monitoring of a system [1].