Abstract
Parameter specification of a fuzzy inference system (HS) with the aid of artificial neural networks allows the creation of complex, multi-dimensional models that are computationally efficient and numerically robust. An adaptive neuro-fuzzy inference system (ANFIS) model is designed to emulate the dynamic behavior of an SD-1000 magnetorheological (MR) damper. The fuzzy damper model enables design and testing of semi-active control strategies within a numerical environment. Modern neuro-fuzzy design techniques are again used with the introduction of a new form of expert control. With this new method, fuzzy logic controllers (FLC) are developed to diminish seismically induced vibrations of single and multiple DOF buildings. The controllers specify a time varying voltage to be applied to MR dampers installed in the buildings. Determination of the voltage signal is based solely on acceleration feedback from the structure being controlled. Effectiveness of the RC is tested through a series of numerical simulations during which the building models are subjected to five different earthquake signals. Signal noise and time delays are found to have only a small effect on operation of the fuzzy controllers. For both structures, performance of the semi-active control scheme is compared with that of similarly scaled passive damping strategies. With regards to the SDOF building, it is determined that a passive control system is superior to a semi-active strategy of similar strength for reduction of lateral displacement and acceleration. For a majority of the tests involving the MDOF building, a semi-active scheme is found to be more beneficial than passive damping approachs for the reduction of lateral acceleration. Thus, more research in the use of MR dampers with tall buildings is advocated.
Schurter, Kyle Christopher (2000). Neuro-fuzzy control of an MDOF building with a magnetorheological damper using acceleration feedback. Master's thesis, Texas A&M University. Available electronically from
https : / /hdl .handle .net /1969 .1 /ETD -TAMU -2000 -THESIS -S38.