Multidimensional complex signal processing and feature extraction for fatigue testing data

Abstract

This study focuses on implementing a multiaxial racetrack amplitude filter designed for processing fatigue testing data. The filter utilizes the concept of a multiaxial racetrack to map a multidimensional signal into a multidimensional space and employs amplitude filtering to eliminate noise while preserving structural outliers and sequential characteristics inherent in the signal. This research evaluates the effectiveness of this method in processing multidimensional data and compares it to conventional data preprocessing methods in the context of fatigue test data. Furthermore, the effectiveness of two optimization strategies proposed to address the limitations related to the filter's “filtering radius” and “filtering direction” has been validated. The findings reveal that the proposed filter is a versatile multidimensional signal processing technique suitable for diverse domains requiring signal order and shape preservation, particularly in fatigue analysis.