This study aims to introduce a new application of random decrement (RD) technique to struc-tural health monitoring for improving the results of statistical process control (SPC). The RD technique transforms random time series data into free vibration forms by averaging them at a specific time. Autoregressive (AR) model is utilized to fit a mathematical model to the av-eraged time series data in order to extract residual errors of AR model as a damage-sensitive feature. After that, Shewhart X-bar control chart is applied to identify the damage by sub-grouping the residual errors. Acceleration time series data from a laboratory experiment con-ducted on a three-story frame are exploited to validate the proposed algorithms for the pro-cess of damage identification. Based on these data and mentioned procedures, it is possible to demonstrate the superior performance of random decrement technique in the statistical pro-cess control as an applicable and useful method for structural health monitoring. Results show that the RD technique improves the process of time series modelling. This technique one can assist to attain better uncorrelated residual errors that should be used in the statistical process control. Furthermore, RD technique reduces the influences of operational and envi-ronmental conditions for damage identification due to having the averaging and normaliza-tion processes. In addition, the quality control with aid of the X-bar control chart is per-formed precisely in association with the random decrement technique that clearly discrimi-nates the damaged from undamaged states.

Keywords