A Data-Driven Fault Tree for a Time Causality Analysis in an Aging System

Abstract

This paper develops a data-driven fault tree methodology that addresses the problem of the fault prognosis of an aging system based on an interpretable time causality analysis model. The model merges the concepts of knowledge discovery in the dataset and fault tree to interpret the effect of aging on the fault causality structure over time. At periodic intervals, the model captures the cause–effect relations in the form of interpretable logic trees, then represents them in one fault tree model that reflects the changes in the fault causality structure over time due to the system aging. The proposed model provides a prognosis of the probability for fault occurrence using a set of extracted causality rules that combine the discovered root causes over time in a bottom-up manner. The well-known NASA turbofan engine dataset is used as an illustrative example of the proposed methodology.

Recent Advances in Intelligent Algorithms for Fault Detection and Diagnosis

“…The failure detection and diagnosis process is carried out in two stages: the first stage consists of a first online verification [ 27 , 28 , 29 ]. Fault detection is carried out by measuring the error values.…”

“…For a change in the set-point or reference value, two possible solutions are considered: Eliminate the change effect, where, for this purpose, the transfer function representative of the set-point variation is designed. This transfer function represents the relationship between the control signal and the change in the set-point and is subtracted from the value of the control signal when the change occurs [ 27 , 28 , 35 , 42 , 43 ]. On the other hand, the design of a second bank of calculated models, including the variation in the set point, could be considered: A variation of 10% of the original reference value is taken as an example, and the models are calculated in the same way as the previous case [ 22 , 44 , 45 , 46 ].…”

“…On the other hand, the design of a second bank of calculated models, including the variation in the set point, could be considered: A variation of 10% of the original reference value is taken as an example, and the models are calculated in the same way as the previous case [ 22 , 44 , 45 , 46 ]. A new alternative for fault detection and diagnosis was proposed based on and taking advantage of the characteristics of a method originally used for the control of processes with different operating regimes [ 23 , 27 , 47 , 48 , 49 ]. One of the advantages of the proposed method is the use of only the input signal (control signal) and the output of the process to obtain the models.…”

Explainable Predictive Maintenance: A Survey of Current Methods, Challenges and Opportunities

“…However, domain experts weighed in on the explanations provided in a positive manner. Waghen et al [79] utilized fault trees to perform interpretable time causality analysis. Their methodology consisted of building multiple logic trees for each subset of data.…”