Neural networks for model-based prognostics

Jaw, Link C.

doi:10.1109/aero.1999.789761

Cited by 17 publications

(10 citation statements)

References 5 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…It is not as extensively applied as direct RUL estimation; however, some examples have been published. Jaw demonstrated a 3 layer MLP to model the tip clearance of a turbine, which was known to be related to blade life [159]. Gebraeel et al tested direct versus indirect estimation of RUL and using various MLP and GRNN approaches; parameter estimation of an exponential degradation function was found to be superior to direct time forecasting, especially real-time sensory parameters were included as network inputs [136].…”

Section: Rul Via Parameter Estimationmentioning

confidence: 99%

Prognostic modelling options for remaining useful life estimation by industry

Sikorska¹,

Hodkiewicz

2011

Mechanical Systems and Signal Processing

781

513

View full text Add to dashboard Cite

Section: Rul Via Parameter Estimationmentioning

confidence: 99%

Prognostic modelling options for remaining useful life estimation by industry

Sikorska¹,

Hodkiewicz

2011

Mechanical Systems and Signal Processing

781

513

View full text Add to dashboard Cite

“…Blade tip clearance [Ja99], [Fl00], [Do99]: In turbine engines, the clearance of the blade tips must be kept low for the engine to be efficient, but not so low that there is rubbing against the race; this is made difficult by thermal expansion. Sensors can be used to monitor the clearance of the blades.…”

Section: Prior Workmentioning

confidence: 99%

“…and "If a failure has occurred, which one?" Techniques for diagnostic information fusion and decision might be physical models/expert knowledge/rule-based [Ro00], [Ga01]; neural networks trained by real data [Br00], [Ga01], [Br99], [Ja99]; fuzzy logic [Br00], [Ga01], [Br99]; or statistical (Bayesian, Dempster-Shafer) [Ro01].…”

Section: Prior Workmentioning

confidence: 99%

Predicting Time to Failure Using the IMM and Excitable Tests

Phelps

Willett

Kirubarajan

et al. 2007

IEEE Trans. Syst., Man, Cybern. A

View full text Add to dashboard Cite

Prognostics, which refers to the inference of an expected time-to-failure for a system, is made difficult by the need to track and predict the trajectories of real-valued system parameters over essentially unbounded domains, and by the need to prescribe a subset of these domains in which an alarm should be raised. In this paper we propose an idea, one whereby these problems are avoided: instead of physical system or sensor parameters, a vector corresponding to the failure probabilities of the system's sensors (which of course are bounded within the unit hypercube) is tracked. With the help of a system diagnosis model, the corresponding the fault signatures can be identified as terminal states for these probability vectors. To perform the tracking, Kalman filters and interacting multiple model (IMM) estimators are implemented for each sensor. The work that has been completed thus far shows promising results in both large and small-scale systems, with the impending failures being detected quickly and the prediction of the time until this failure occurs being determined accurately.

show abstract

“…An effective means to assure the system reliability and availability is to fast and reliably detect and identify the system performance degradation caused by the sensors, actuator or system component failures, so that appropriate remedies can be undertaken. During the last two decades, significant research efforts have been drawn in the field of fault detection and identification [1,2,3,16,17,20,18,15]. All these FDI technologies utilize parameter and state estimation, detection filters, statistical pattern recognition, multiple model estimator, maximum likelihood techniques and Bayes' theorem.…”

Section: Introductionmentioning

confidence: 99%

Design of Dynamic System Fault-Tolerant Control using IMM Estimation and RBF Neural Network

Wang

Syrmos

2006

2006 14th Mediterranean Conference on Control and Automation

View full text Add to dashboard Cite

In this paper, a strategy of failure detection, identification and reconfigurable scheme for a dynamic system is proposed. The proposed scheme provides detection and identification of sensor, actuator and/or system component failures, dynamic system state estimation and system performance recovery. Fault detection and identification is carried out using radial basis function (RBF) neural network and interacting multiple model (IMM) estimation. The RBF-NN is used to form a statistical model of nominal or faulty data and estimate the mode-conditional probability densities as the choice of likelihood function. The IMM mechanism carries out the interaction among mode-based filters, update the mode probability and provide the overall state estimate as the control input. Eigenstructure assignment (EA) technique is used for the reconfigurable controller design. The proposed approach is evaluated using an aircraft example, and the results obtained show that it can reliably and accurately detect, identify the faults and recover the impaired dynamic performance to the desired one.

show abstract

Neural networks for model-based prognostics

Cited by 17 publications

References 5 publications

Prognostic modelling options for remaining useful life estimation by industry

Prognostic modelling options for remaining useful life estimation by industry

Predicting Time to Failure Using the IMM and Excitable Tests

Design of Dynamic System Fault-Tolerant Control using IMM Estimation and RBF Neural Network

Contact Info

Product

Resources

About