SENSITIVITY ANALYSIS OF A GENETIC ALGORITHM-BASED ACTIVE/HYBRID CONTROL SYSTEM UNDER NEAR SOURCE STRONG GROUND MOTIONS

摘要

After Northridge (1994) and Kobe (1995) earthquakes, development of robust and reliable control systems to encounter large velocity and displacement ground pulses has attracted considerable attention. Genetic Algorithm (GA), with its evolutionary optimization power, has also become an attractive artificial intelligence tool in the field of earthquake engineering. We developed a computer program for simulating a GA-based open-closed loop control system for application to structures under seismic excitations. The developed program is capable of evaluating the optimum modal control gain for an active or hybrid system for a predefined allowable structural response. As a proof of concept, behavior of the control system is simulated using mathematical model of a benchmark structure. Minimum active mass damper acceleration time histories are obtained for different limitations on the response quantities under near source strong ground motions. The simulation results shows that the presence of large input pulses could strongly affect the behavior of the open-closed loop control system in terms of the gain values. Sensitivity analyses of the GA parameters are also performed and the results are discussed.