Evolutionary based approach to modal parameter identification

Kurzfassung

An important task in modal analysis is the optimal selection of physical poles of a system. In this paper an Artificial Intelligence (AI) evolutionary algorithm is used to optimize the modal parameters from Output Only Modal Analysis (OMA). The evolutionary algorithm begins by generating a population selected randomly from mode clusters. The fitness of each individual is then evaluated based on the difference between the measured and synthesized spectra. The best individuals are then reproduced using a crossover strategy. A random mutation allows the exploration of the larger solution space in order to avoid convergence in local minima. The algorithm is tested on simulated data as well as the DLR AIRMOD laboratory structure, and is found to provide accurate and robust heuristic solutions to large and complex solution spaces.