A Probability Box Modeling Method of Dimensionless Mechanical Fault Features in Time Domain
Yi Du, Jiaman Ding, An Liu
Available Online March 2014.
- https://doi.org/10.2991/mce-14.2014.112How to use a DOI?
- Probability box theory; modeling; Uncertainty; Mechanical Fault; fault feature
- The simply multi-segment averages of mechanical fault features lead to the loss of signal uncertainty probability statistics which may contribute to fault diagnosis. To solve the problem, a probability box modeling method of mechanical fault features is proposed. Three similar dimensionless factors (peak factor form factor margin factor) in time domain were used to obtain probability boxes. The algorithm of modeling was then proposed. Firstly, the mechanical fault raw data was divided into several segments according to the sampling frequency. Secondly, the dimensionless features in time domain were obtained for every segment. After verifying the normal distribution probability distribution type, the row vector of features was divided into several sets. From these sets, the range of the mean and variance were obtained. Finally, the DSS of mean and variance were integrated into the probability modeling algorithm of normal distribution to get the probability box. The rolling bearing’s faults raw data was used to verify this modeling method. The results show that the probability box modeling method can effectively reflect the uncertainties of fault state, which can be used for fault diagnosis.
- Open Access
- This is an open access article distributed under the CC BY-NC license.
Cite this article
TY - CONF AU - Yi Du AU - Jiaman Ding AU - An Liu PY - 2014/03 DA - 2014/03 TI - A Probability Box Modeling Method of Dimensionless Mechanical Fault Features in Time Domain BT - 2014 International Conference on Mechatronics, Control and Electronic Engineering (MCE-14) PB - Atlantis Press SP - 502 EP - 505 SN - 1951-6851 UR - https://doi.org/10.2991/mce-14.2014.112 DO - https://doi.org/10.2991/mce-14.2014.112 ID - Du2014/03 ER -