Topology optimization of truss subjected to static and dynamic constraints by integrating simulated annealing into passing vehicle search algorithms

Abstract

Three modified versions of passing vehicle search (PVS) are proposed and tested on truss topology optimization with static and dynamic constraints. PVS works on the mechanism of passing a vehicle on a two-lane highway. The heuristic nature of PVS allows the search to jump into non-visited regions (exploration) and also permits a local search of visited regions (exploitation). First, the original PVS algorithm is improved to avoid a local optima trap using a novel parallel run mechanism. Then, population diversity is improved by incorporating the selection of simulated annealing. The various versions of PVS are verified on the truss design problems. Comparative results show that the parallel run concept improves the original PVS algorithm. The selection using the Boltzmann probability as used in simulated annealing further improves the algorithm.