Application of advanced control algorithms in seismic isolation of structures

Algorithms used in the active control systems are the most determinative characteristics. The control forces generated by the means of the used algorithms are sent to actuators. In this study, the advanced control algorithms have been first designed then applied to the seismic isolation after that the performances of the designed control algorithms have been compared with each other.LQR control, one of the active control algorithms, has commonly been used because of its stability and robust features. The limited control forces can be applied since all actuators have the control bounds in practical applications in the active control systems. Because of the bounds on the control variables, the performance of LQR control algorithm has decreased and especially been insufficient against structural uncertainities. On the other hand, the bang-bang control, one of the optimum controls used successfully in many areas, has not sufficiently been used in earthquake-resistant structure design. The success of this control algorithm is similar to that of LQR control algorithm in the simulations.Sliding-mode controller, a special case of a variable structure control algorithm, has been one of the most popular and the most widely accepted control algorithms since the most important feature of this control method is not to be sensitive against structural uncertainities and disturbances. In addition, this method has become more attractive by rapid technological developments to allow the implementation of these kinds of controllers. However, chattering phenomena, too many switches in the control bounds, is the most disadvantage of this control method. For this reason, the obtained time histories of control forces can not be realized by the real controllers.The successes of artificial intelligence methods among the alternative solutions to eliminate the mentioned disadvantage without degrading control performance have extremely been good. Hence, fuzzy sliding-mode control algorithm was designed and the successful results were obtained. The chattering phenomena was especially eliminated and the control algorithm was able to applicable in practical sense. However, obtaining to the optimum result is very difficult because some parameters of fuzzy logic (FL) have been determined by trial-error technique. The studies using Artificial Neural Network (ANN) as a controller have achieved a great successful. The simulation results of the controller designed by using ANN has verified a good achievement however it has not reached to the success of the sliding-mode controller if considered against structural uncertainities. Combining FL having properties to eliminate disadvantages of parametric and structural uncertainities and ANN having properties training capabilities has created a successful control algorithm. However, it has not reached to the success of the sliding-mode controller if considered against structural uncertainities. Therefore, a neural based fuzzy sliding-mode control algorithm has been designed by combining ANN, FL and sliding-mode control. The excellent results were obtained. In order to obtain better performance of the controllers, sliding-mode controller with moving sliding surfaces has been developed by rotating the sliding surfaces in the phase plane in such a direction that the seismic isolation can be improved. Thus, fuzzy sliding-mode control, neural based sliding-mode control and neural based fuzzy sliding-mode control methods have designed and applied by moving sliding surfaces for the seismic isolation of structures. The simulation results showed that the applied control algorithms had great achievements.The selection of the optimum control parameters directly affects the performance of controllers. For this reason, genetic algorithm technique was used to obtain the optimum control parameters.In order to apply the designed control algorithms and analyze the performances of these controllers, 3, 8 and 16-floor building models have been considered. In addition to that, two different practical control systems have been applied for each building model. These active control systems called active tendon and active mass systems have most commonly been used in practical applications. The mathematical models of the considered systems were obtained then they were transformed in the state-space forms. Six different seismic events have been considered as ground accelerations acting to the systems. Besides, 30% parametric uncertainities and 20 ms time delay were added to the system to see the effects of parametric uncertainities and time delay. For simulations, the MATLAB code was prepared and Genetic Algorithm Toolbox in MATLAB was used for the optimizations. The obtained results were presented in graphical and table forms and examined. As a result, it has been seen that the performances of the fuzzy, neural based and neural based sliding-mode controller with moving sliding surfaces are much better and applicable if the displacement, acceleration responses, parametric uncertainities and time delay have been considered.Keywords : Seismic Isolation, Control Algorithms, Artificial Neural Network, Fuzzy Logic, Genetic Algorithms.