Abstract:Timed failure propagation graphs (TFPGs) perform fault diagnosis in a transparent way. However, accurate TFPGs depend on experts’ knowledge or accurate model of the system, which is hard to obtain for complex systems. This paper presents a data-driven TFPG construction approach for fault diagnosis, which finds spectral-timed failure propagation graphs (sTFPG) directly from data. The sTFPG construction problem is transformed into a spectral-temporal logic inference problem and solved with a tree-structured long short-term memory (LSTM) network. Therefore, no expert or accurate model is needed to construct the TFPG. Moreover, the training process is guided and sped up by the supremacy property of the spectral-temporal logic, which focuses on the structure information of the signals and incorporates the physical meanings in the learning process. Experimental results on real rolling element bearing data sets illustrate that the performance of the proposed fault diagnosis method is comparable with state-of-the-art machine learning methods in fault diagnosis accuracy, and outperforms the logic-based method in computational efficiency. Additionally, fault diagnosis with TFPG can be understood by humans and reveal the fault mechanism.

Fault Diagnosis of Timed Event Graph

Fault Diagnosis for Discrete Event Systems Modeled By Bounded Petri Nets: Fault Diagnosis of Petri Nets

Fault Diagnosis for Discrete Event Systems Using Partially Observed Petri Nets

Fault Diagnosis Based on the Structure Information of Partially Observed Petri Nets.

Fault Diagnosis of Bounded Petri Nets Using Path Marking Graphs.

Fault Diagnosis System For Electric Vehicle Charging Devices Based On Fault Tree Analysis

Technology and Development of Fault Diagnosis Based on Graph Theory

FAULT DIAGNOSIS FOR POWER ELECTRONIC CIRCUITS BASED ON HYBRID SYSTEM THEORYAND EVENT IDENTIFICATION

A New Fault Diagnosis Approach in Power System Based on Element-Oriented Petri Nets with Time-Stamp

Diagnosis of Structural and Temporal Faults for $k$ -Bounded Non-Markovian Stochastic Petri Nets

A Fault Prognosis Strategy Based on Time-Delayed Digraph Model and Principal Component Analysis

An Electric Power System Fault Diagnosis Method Based On Time Series Petri Net

Temporal Constrained Dynamic Uncertain Causality Graph for Root Cause Analysis of Intermittent Faults

Timed failure propagation graph construction with supremal language guided Tree-LSTM and its application to interpretable fault diagnosis

Fuzzy Timing Petri Net for Fault Diagnosis in Power System

An algebraic model for performance evaluation of timed event multigraphs

Algebraic approach to continuous event graphs without input

Approach To Fault Diagnosis Using Sdg Based On Fault Revealing Time

Power System Fault Diagnosis with an Enhanced Fuzzy Petri Net Accommodating Temporal Constraints

A new approach to power system fault diagnosis based on fuzzy temporal order Petri nets

FAULT DIAGNOSIS OF POWER-PLANT ALARM SYSTEM