This paper describes multiple layered neural netowrk to simulate the hysteretic behavior of high damping rubber bearing. The units of input layer and output layer were selected appropriately, and the effect of hidden unit and temperature constant were confirmed under many cases of analysis. Sinusoidal wave loading test was conducted to acquire the data for learning of the netowkr. And random wave loaidng test was conducted to examine the applicability of learned network to unlearned data. Function approximation ability of the network without any mathematicla assumption was evaluated through comparing with analytical data and experimental data. unlearned hysteresis behavior was simulated by learned network using the data of sinusoidal wave loading test.
展开▼