عنوان مقاله [English]
One of the main concerns in the application of active and semi-active structural control in large scale structures, as a seismic protection strategy, is the consistency and reliability of real-time data transmission. Coaxial cables, as the primary communication link, provide reliable links for transferring data. However, their installation is time consuming and costly, and it is a tiresome task to change system architecture and reroute wires, especially in large-scale structures. Wireless sensors and communication have the potential to significantly reduce the cost and increase the architecture flexibility of a feedback structural control system. However, replacing coaxial cables with wireless communication channels has drawbacks, such as the coordination of sensing and control nodes, communication range, feedback time delay and potential data loss. Feedback time delay, due to wireless communication, will cause degradation to the performance of a structural control system. A decentralized control strategy is proposed as a remedy to improve the performance of the system due to time delay. In decentralized control strategies, a large control system is divided into distributed sub-systems. Each sub-system has its own controller. These controllers make control decisions through acquiring data from local and neighbouring sensors. On the other hand, in large scale structures, large numbers of degrees of freedom cause numerical difficulties in dynamic analysis and may increase requirements of storages in computation. The reduced-order system solves this problem if it acquires the essential properties of the full-order model, such as passivity and stability. A useful method of the reduced order model is Singular Value Decomposition (SVD). In SVD, the system is transformed to a basis where the states are that are difficult to reach or observe. Then, the reduced model is obtained simply by truncating the states which have this property. This study presents the feasibility of developing a decentralized controller using H2/LQG as a robust-optimum algorithm in a cable-stayed bridge. Due to the large-scale of the structure, the order of the model is reduced using two different reduction methods in finite time and frequency. The controller is developed in centralized and decentralized solutions. The decentralized controller, through developing a gradient-based method, is designed to efficiently solve the system closed-loop H2 optimization problem. The structural responses in centralized and decentralized solutions are compared each other with time delay in data transmission. Results indicate that for large-scale bridges with a time delayed data communication issue, the H2/LQG controller in the decentralized solution gives the maximum reduction of response indices in comparison with other strategies.