A novel experience-based learning algorithm for structural damage identification: simulation and experimental verification

• Published in: Engineering optimization 2020-10, Vol.52 (10), p.1658-1681
• Main Authors: Zheng, Tongyi, Luo, Weili, Hou, Rongrong, Lu, Zhongrong, Cui, Jie
• Format: Article
• Language: English
• Subjects: Cantilever beams; Computer simulation; Damage detection; Damage identification; experience-based learning algorithm; Functionally gradient materials; Machine learning; mode shape; natural frequency; Optimization algorithms; Parameters; Particle swarm optimization; Resonant frequencies; Robustness (mathematics); Search algorithms; Structural damage; structural health monitoring
• ISSN: 0305-215X; 1029-0273
• DOI: 10.1080/0305215X.2019.1668935

A simple yet powerful optimization algorithm, named the experience-based learning (EBL) algorithm, is proposed in this article for structural damage identification based on vibration data. This algorithm is free from any algorithm-specific control parameters and requires only common control parameters. The natural frequencies and/or mode shapes are utilized in establishing an objective function. The efficiency and robustness of the proposed method are demonstrated by two numerical examples, namely a television tower and a functionally graded material beam. A set of experimental work on a cantilever beam is studied for further verification. Both numerical and experimental results confirm the superiority of the proposed EBL algorithm in terms of convergence and accuracy for structural damage identification, in comparison with particle swarm optimization, the cloud model-based fruit fly optimization algorithm, squirrel search algorithm and teaching-learning-based optimization.