Automated analysis of building seismic records

System identification has been used before for post-earthquake evaluation of building behavior. It invariably involves the matching of recorded acceleration-time histories to calculated time histories based on an assumed mathematical model. The matching is achieved by minimizing some measure of the difference between the two sets of time histories through the adjustment of the parameters of the assumed model. Frequently the minimization solution may not be physically sensible. Thus, intervention of an experienced user is needed to get a sensible solution by repeated trials. The work described herein is an automated method by which this iterative solution process is carried out without the intervention of a user. This is achieved by using an expert system that contains the knowledge and experience of the user. Current advancements in hardware technology allow the automatic recording and digitizing of seismic records using a personal computer (PC). With this automatic system identification method implemented on a PC-based recording system, it is possible to have the results of system identification minutes after the end of an earthquake event. This in turn lends itself to an automated damage assessment which is valuable in the absence of detailed on-site inspection by experienced engineers. A methodology of automated damage assessment is developed utilizing the results of system identification. Prior to an earthquake event, expert opinion on the various damage indicators or indices and their meaning pertaining to the particular building is collected and is input to a fuzzy logic inference machine. The machine in turn outputs a fuzzy language assessment of the degree of damage ranging from no damage to severe damage. The state-of-the-art of damage assessment based on recorded measurements does not yet render itself to a precise assessment. The output of the inference machine, however, serves a useful purpose as a preliminary assessment.