Conceptualisation of a Parameterizable Low-Pass Filter for Resolving Measurement Data
Michael Schmid, Bastian Eisenmann, Osman Aksu, Mišel Radosavac, Florian Bierwirth, Hans–Georg Herzog · 2025
The data acquisition process through measurements translates continuous physical quantities, such as temperature or magnetic flux, into discrete digital data sets. Different noise sources, such as quantization or Gaussian noise, distort the sampled points and inhibit subsequent processing during this process. Filtering the measured data is, thereby, a necessary step. In contrast to conventional filter design, this paper describes a novel discrete low-pass filter with a variable shape. It is configured by two independent parameters, allowing it to be sensitive to time-critical peaks at high levels while providing adequate noise suppression at low measured quantities. The theoretical background of this work's filter approach is explained, and guidelines for parameter selection are suggested. In the utilized test data set, the novel approach results in a 19 % better peak preservation while providing the same noise suppression as a conventional low-pass filter.