Model Updating for Large-Scale Railway Bridge Using Grey Wolf Algorithm and Genetic Alghorithms

Abstract

This paper proposes a novel hybrid algorithm to deal with an inverse problem of a large-scale truss bridge. Grey Wolf Optimization (GWO) Algorithm is a well-known and widely applied metaheuristic algorithm. Nevertheless, GWO has two major drawbacks. First, this algorithm depends crucially on the positions of the leading Wolf. If the position of the leaderis far from the best solution, the obtained results are poor. On the other hand, GWO does not own capacities to improve the quality of new generations if elements are trapped into local minima. To remedy the shortcomings of GWO, we propose a hybrid algorithm combining GWO with Genetic Algorithm (GA), termed HGWO-GA. This proposed method contains two key features (1) based on crossover and mutation capacities, GA is first utilized to generate the high-quality elements (2) after that, the optimization capacity of GWO is employed to seek the optimal solutions. To assess the effectiveness of the proposed approach, a large-scale truss bridge is employed for model updating. The obtained results show that HGWO-GA not only provides a good agreement between numerical and experimental results but also outperforms traditional GWO in terms of accuracy.

Keywords

• model updating
• grey wolf algorithm
• large-scale bridges
• stiffness of truss joints