Fault diagnosis for complex systems based on reliability analysis and sensors data considering epistemic uncertainty

Abstract

Fault diagnosis For complex systems based on reliability analysisand sensors data considering epistemic uncertainty diagnozowanie błędów w systemach złożonych na podstawie analizy niezawodności oraz danych z czujników z uwzględnieniem niepewności epistemicznejThis paper presents an information fusion method to diagnose system fault based on dynamic fault tree (DFT) analysis and dynamic evidential network (DEN). In the proposed method, firstly, it uses a DFT to describe the dynamic fault characteristics and evaluates the failure rate of components using interval numbers to deal with the epistemic uncertainty. Secondly, qualitative analysis of a DFT is to generate the characteristic function via a traditional zero-suppressed binary decision diagram, while quantitative analysis is to calculate some importance measures by mapping a DFT into a DEN. Thirdly, these reliability results are updated according to sensors data and used to design a novel diagnostic algorithm to optimize system diagnosis. Furthermore, a diagnostic decision tree (DDT) is obtained to guide the maintenance workers to recover the system. Finally, the performance of the proposed method is evaluated by applying it to a train-ground wireless communication system. The results of simulation analysis show the feasibility and effectiveness of this methodology.

Fault Diagnosis Strategy for Complex Systems Based on Multi-Source Heterogeneous Information Under Epistemic Uncertainty

“…One part is a static evidence network, and the other part is a time attribute. DFT can be converted into a static evidence network first, and then dynamic evidence network can be obtained by adding time attribute [43]. After establishing the DFT model of the system, the DFT is mapped into a corresponding dynamic evidence network according to the above method and inference algorithm can be used to calculate some importance factors.…”

“…As for diagnostic algorithms, many researchers have proposed many effective methods. Reference [24] incorporates the sensor information into diagnosis process and considers DIF of components and the minimum cut set to diagnose the system. However, it is impossible to make decisions on the case where DIF of the minimal cut set is large and DIF of the component is small.…”

Untitled

“…This outcome is interesting but again, vibration assessment causes some noise in the final system state determination. Inspirational examples on specific diagnostic data forms analysis can be found in [9,13,14,25,28,32].…”

Modified convolutional neural network with global average pooling for intelligent fault diagnosis of industrial gearbox

“…High transmission ratio, strong load-bearing and high efficiency makes modern gearboxes are always considered to be critical components in various industrial applications, such as wind turbine generator system, helicopter main speed reducer, aerospace engineering and etc [27,38]. In real practice, however, gearboxes will inevitably be subjected with dynamic heavy-duty loads under complex operating conditions, making the breakdown or even accidents of the engineering system [3,4,8,32,39]. Therefore, it is of great significance to Przekładnie stanowią kluczowe elementy układów napędowych i jako takie znajdują szerokie zastosowane w przemyśle.…”