抄録
A new analytical technique is presented for the assesment of stationary or non-stationary random earthquake responses of a single-degree-of-freedom structural system with poly-linear hysteretic characteristics by extending and making use of the previous analytical procedure developped in Part 1. The availability of this new technique is ascertained by investigating the agreement between the analytical and simulated non-stationary r.m.s. displacement and velocity response processes of a nonlinear system with tri-linear hysteretic characteristics. The following two interesting problems in relation to the substitution of an approximate bi- or tri-linear hysteretic characteristics for an actual poly-linear one are studyed ; i.e., (1) what qualitative and quantitative influences the difference between an actual and approximate hysteretic models have on the random response characteristics and (2) how the slopes in the approximate elasto-plastic hysteretic characteristics should be determined to get the closer random responses of the approximate model to those of the acturl one.
