Causality assignment using multi-objective evolutionary algorithms

Causality assignment is an important task in physical modeling by bond graphs. Traditional causality assignment procedures have specific aims and particular purposes. The traditional SCAP and MSCAP are algorithms that may fail if the bond graph has loops or contains junction causality violations. The RCAP algorithm focuses on the generation of differential algebraic equations to take into account junction violations caused by nonlinear multi-port devices and is not suitable for general bond graphs. We present a formulation of the causality assignment problem as a constrained multiobjective optimization problem. We then solve the resulting constrained system by a simple yet effective multi-objective evolutionary technique.