Dynamic System Modeling
Dr. Bekey has worked extensively in dynamic system modeling, including:
Mathematical models of human operators
Building on his Ph.D. thesis, Dr. Bekey and his students mathematically modeled human operator behavior (papers 195, 209, 213, 215 and 229) including tactile inputs (paper 161) and car drivers' actions ( papers 171, 174 and 183), predating current interest by several decades.
Optimization by random search
Dr. Bekey developed several algorithms for optimizing dynamic systems by random search, including an important paper published in 1980 (papers 155 and 223). He and USC civil engineering professor Dr. Sami Masri applied these methods to structural problems such as building design to minimize the effects of earthquake-induced oscillations (papers 147, 152, 157 and 182).
System identification and other control system contributions
For the first half of his academic career, Dr. Bekey worked primarily in estimation of system parameters from input-output measurements. He introduced the notions of parameter identifiability and near-identifiability (papers 123 and 184), identified non-linear systems with hysteresis (paper 137) and applied binary input signals to enhance identification (paper 88). Among his other contributions: studying sensitivity of discrete systems to sampling interval variation (papers 224 and 225), developing a new signal flow graph for mixed discrete-continuous systems (paper 220) and developing means to speed neural network learning times (paper 72)