Evolutionary algorithms for compact thermal modelling of microsystems: application to a micro-pyrotechnic actuator

Abstract

In this work, we approach the problem of extracting a dynamic multiport thermal compact model from thermal impedance transients of microsystems using genetic algorithms. The model takes the form of a unique RC network, using a thermal-electrical analogy. The model topology is codified in a binary chromosoma and nonlinear least squares is used for sizing their components. The compact model topology evolution is genetically controlled to obtain the RC network that minimizes the reconstruction error of the thermal impedance transients. As an example, the proposed methodology is applied to an innovative silicon microthruster and compared with random search and sequential forward selection.