Transient conditions and overlapping sub-records: An excellent pair for FRF measurements

A detailed understanding of a measured system requires some knowledge on the system´s behavior at different frequencies. The linear behavior of a system at different frequencies is estimated by the Frequency Response Function (FRF). If the system is excited by a periodic multi-frequency signal, the FRF can be obtained by computing the ratio of the discrete Fourier coefficients of the response of the system and of the excitation signal at the excited frequencies. This is a widely applied practical method to measure the system´s FRF. However, this non-parametric method may return incorrect results if the system is not operating under steady-state conditions. One way to overcome this problem is by only considering the measurements taken when the system operates under steady state conditions. This significantly increases the measurement time. For expensive measurement problems like biochemical, biomedical and geochemical processes increasing the measurement time is not allowed or simply not possible. This paper proposes a fully non-parametric effective reduction of the system transients based on overlapping sub-records.