Comparative Study on Data Editing Techniques for Fatigue Time Series Signals

This paper presents the comparative study on three types of fatigue data editing technique for summarising long records of fatigue data. Two of techniques were developed based on time-frequency domain (continous wavelet and discrete wavelet) and another one technique was developed based on time domain. These three techniques are used to extract fatigue damaging events in the record that cause the majority of fatigue damage, whilst preserving the load cycle sequence in the data. The objective of this study is to observe the capability of each technique in summarising long records of fatigue data. For the purpose of this study, two set of nonstationary data that exhibits random behaviour was used. This random data was measured in microstrain unit on the SAE1045 material that were used as a lower suspension arm of a car. Experimentally, the data was collected for 60 seconds at sampling rate of 500Hz, which gave 30, 000 discrete data points. The result of the study indicates that all techniques are applicable in detecting and extracts fatigue damaging events that exist in the fatigue data. However, it was found that continous wavelet can extract the data better than the other technique based the shorten signals, retention of damage and statistical parameter.