A parametric approach to data cleaning of ion current measurements
Fredrik Andréen Blomdell · Lund University Publications Student Papers (Lund University) · 2017
In this thesis, a few possible methods to clean nuisance signals from ion current measurements are explored. Ion current measurements can be used to optimise a combustion engine, but interference from the ignition system coupled with clipping caused by limitations in the measurement equipment limit the amount of data that can be extracted. This thesis develops a model which is used for simulation of the ion current measurements, that is then used to construct an algorithm to remove the interference. The algorithm developed uses a parametric approach as the interference has a distinct structure that can be exploited. A corner stone of the developed algorithm is the HTLS-algorithm, which estimates frequency and damping coefficients of uniformly sampled data. The thesis implements a recursive HTLS to exploit the structure of the interference efficiently. This method performs well on simulated data, but testing on real data shows that it needs further refinement if it is to be used commercially.