Perpustakaan Digital ITB

The active control system as an alternate solution in structural dynamics proposes many merits since it could increase the energy dissipation capacity of a structure provided by control devices. Despite the benefits it proposes, the active structural control system still has limited acceptance in the industry. One of the main reasons for this is to deliver the optimum control actuator configuration from performance and cost perspectives. For a complex problem, such as a high-rise building, developing an optimized active control configuration becomes challenging due to the large search space. This study proposes an efficient design framework to obtain the optimum active control actuator configuration on the controlled multistory steel buildings via a combined predictive and metaheuristic method. The active control device tested in this study is the active tendon system analyzed via the non-linear time history analysis. The optimization objectives are to minimize the structural response, the number of active control actuators, and the required force capacity of the actuators. The optimum actuator configuration is searched via multi-objective evolutionary optimization based on the Weighted Sum Genetic Algorithm (WSGA), Non-Dominated Sorting Genetic Algorithm II (NSGA-II), and the newly proposed Moving Average Fitness Genetic Algorithm (MAFGA), and the novel Population Guidance and Modified Reference-Point Based Non-Dominated Sorting Genetic Algorithm II (PMR-NSGA-II). The PMR-NSGA-II utilizes initial population guidance and a modified reference-point-based technique. The preliminary prediction methods via modified modal controllability and input energy distribution approach are used to guide the metaheuristic method by providing a prospective initial population. A modified reference point based on a normalized Euclidean distance is introduced to focus the search towards a preferred region in the search space. The results show that the proposed PMR-NSGA-II could expedite the computation speed up to two times faster than the widely applied NSGA-II while successfully providing non-dominated solutions. The WSGA could only be applied where the value between objectives could be translated into a commensurate unit. The MAFGA could negate the need to use weighting factors, thus making it suitable for cases with non-commensurate optimization objectives. Conclusively, the PMR-NSGA-II is superior to the rest of the studied algorithms in terms of quality of solutions and computational efficiency.