- This paper aims at improving the robustness and accuracy of the least square fit technique utilized in blade tip timing (BTT) measurements of blade vibrations by proposing two modifications. The
first proposal is to replace the lines of the original least square problem by differences of consecutive lines. Thereby, the static deflection as well as the circumferential blade positioning error cancels out and the robustness is improved by removing these uncertainties inherently. The second proposal is to replace the fit of piecewise constant vibration amplitudes within the chosen block length by a linear or cubic spline in the frequency (integral) or time (non-integral) regime. This does not only suppress overshoots due to distorted acceleration or deceleration manoeuvres but also allows for a ”coarser analysis grid” (i.e. larger block length) without loosing amplitude accuracy. Thereby it smooths out random errors more efficiently and increases the orthogonality of the relevant EO or vibration frequency to unwanted signal components.