Parametric identification of nonlinear devices for seismic protection using soft computing techniques

Passive devices for vibration control are widely adopted in earthquake engineering for mitigation of seismic effects obtaining an efficient, robust and not expensive structural protection. They are largely used in the seismic protection of industrial machines, technical equipment, buildings, bridges and others more as reliable and affordable solutions. Moreover their performances are extremely sensitive to their dynamic mechanical behavior; a reliable identification of their mechanical behavior is therefore of key importance, despite the current lack of accurate and simple standard procedures to identify parameters and models for those devices. In this work, a new procedure for the dynamic identification of passive devices is described, through standard laboratory dynamic tests and the use of evolutionary algorithms. This procedure allows to find proper mechanical law and parameters to use for an accurate structural analysis and earthquake-resistant structure design. The procedure uses standard pre-qualification and quality-control tests, and consists in the minimization of the integral measure of the difference between mathematic and experimental applied force to the device under an imposed displacement time history. Due to the amount of corruption source of the experimental data and to the deep non linear nature of the problem, the use of evolutive algorithms is the main way to solve hard numerical task in an efficient way. The proposed procedure is applicable to a wide range of mathematical expressions because of its inherent stability and low computational cost, and allows comparing different mechanical laws by ranking their agreement with experimental data. Results are obtained for different experimentally tested devices, that are viscous dampers and seismic isolators, and are reported in order to demonstrate the efficiency of the proposed strategy