Condition Monitoring Using Empirical Models: Technical Review and Prospects for Nuclear Applications

Heo, Gyun

doi:10.5516/net.2008.40.1.049

Cited by 31 publications

(14 citation statements)

References 13 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…These models can be very sophisticated and often require the use of complex computer algorithms and long-term calculations. However, they can simulate the tracks behaviour in different conditions, even before its exploitation, and help to define tracks most important properties needed for improving its exploitation behaviour [36][37][38][39].…”

Section: Choosing the Modelling Approachmentioning

confidence: 99%

Track Gauge Degradation Modelling on Small Urban Rail Networks: Zagreb Tram System Case Study

Ahac¹,

Lakušić²

2017

Urban Transport Systems

View full text Add to dashboard Cite

Track gauge is the most significant tram track geometry parameter. Its degradation, which manifests as gradual increase of gauge deviation from prescribed values during track exploitation, causes poor ride quality, reduces safety and triggers most of the maintenance activities. To optimize tram track maintenance procedures, it is necessary to increase the proportion of preventive maintenance at the expense of corrective maintenance. This requires creation of physical model of track degradation. Conducted survey of conventional track degradation models showed that, in order to quantify the influence of track design, construction and exploitation characteristics on gauge degradation, it is most favourable to adopt the mechanistic-empirical modelling approach. Zagreb high-capacity tram network presents an optimal testing ground for exploration of the possibilities for tram track gauge degradation model development. Analysis of modelling results gave new, practical insights about the effects of tram track design and construction elements and exploitation characteristics on gauge degradation. These models represent the first step towards predictive maintenance system establishment on Zagreb tram tracks.

show abstract

Section: Choosing the Modelling Approachmentioning

confidence: 99%

Track Gauge Degradation Modelling on Small Urban Rail Networks: Zagreb Tram System Case Study

Ahac¹,

Lakušić²

2017

Urban Transport Systems

View full text Add to dashboard Cite

show abstract

“…Anticlockwise flow in Figure 4 shows architecture and greedy layer by layer unsupervised pretraining procedure for all hidden layers in DAASM stack. For each hidden layer ℎ , a DAE block is shown, in which an encoder function (⋅) and a decoder function (⋅) are learnt by minimizing the loss function corresponding to fault free reconstruction of the inputs as in relation (6). For the case of first hidden layer ℎ 1 , the corresponding DAE-1 is trained directly on sensor data using ( ,̂, ) loss function in (6).…”

Section: Daasm Architecture and Regularizationmentioning

confidence: 99%

“…For each hidden layer ℎ , a DAE block is shown, in which an encoder function (⋅) and a decoder function (⋅) are learnt by minimizing the loss function corresponding to fault free reconstruction of the inputs as in relation (6). For the case of first hidden layer ℎ 1 , the corresponding DAE-1 is trained directly on sensor data using ( ,̂, ) loss function in (6). However, hidden layers ℎ 2 through ℎ 3 are learnt on data from preceding hidden layer activations using recursive relation in (7).…”

Section: Daasm Architecture and Regularizationmentioning

confidence: 99%

See 1 more Smart Citation

A Denoising Based Autoassociative Model for Robust Sensor Monitoring in Nuclear Power Plants

Shaheryar

Yin

Hao

et al. 2016

Science and Technology of Nuclear Installations

View full text Add to dashboard Cite

Sensors health monitoring is essentially important for reliable functioning of safety-critical chemical and nuclear power plants. Autoassociative neural network (AANN) based empirical sensor models have widely been reported for sensor calibration monitoring. However, such ill-posed data driven models may result in poor generalization and robustness. To address above-mentioned issues, several regularization heuristics such as training with jitter, weight decay, and cross-validation are suggested in literature. Apart from these regularization heuristics, traditional error gradient based supervised learning algorithms for multilayered AANN models are highly susceptible of being trapped in local optimum. In order to address poor regularization and robust learning issues, here, we propose a denoised autoassociative sensor model (DAASM) based on deep learning framework. Proposed DAASM model comprises multiple hidden layers which are pretrained greedily in an unsupervised fashion under denoising autoencoder architecture. In order to improve robustness, dropout heuristic and domain specific data corruption processes are exercised during unsupervised pretraining phase. The proposed sensor model is trained and tested on sensor data from a PWR type nuclear power plant. Accuracy, autosensitivity, spillover, and sequential probability ratio test (SPRT) based fault detectability metrics are used for performance assessment and comparison with extensively reported five-layer AANN model by Kramer.

show abstract

“…Detailed reviews of state estimation and fault detection methods are given in Hashemian (1995), Hines and Seibert (2006), Heo (2008), and Ramachandran et al (2010). State estimation and fault detection will be briefly described in the following sections.…”

Section: Monitoring Methodsmentioning

confidence: 99%

Prognostics and Health Management in Nuclear Power Plants: A Review of Technologies and Applications

Coble¹,

Ramuhalli²,

Bond³

et al. 2012

View full text Add to dashboard Cite

Recent years have seen a resurgence of nuclear power worldwide, with interest in extending the operating life of the approximately 436 reactors currently in service (as of March, 2012), 61 new reactors being constructed, and as many as 162 under consideration. Renewed worldwide interest in nuclear power has been somewhat tempered by the March 2011 incident at Fukushima Dai-ichi in Japan. However, nuclear power is still considered a key element in meeting future worldwide sustainable energy, energy security, and emissions goals. Currently, three separate thrusts to safe and economical nuclear power development for energy security are being pursued in the United States: (i) longer term operation for the legacy fleet, from 40-60 and possibly 60-80 years; (ii) near-term new nuclear plants with a 60-year design life; and (iii) small modular reactors, which are expected to employ light water reactor technology at least in the medium term. Within these activities, attention is turning to enhanced methods for plant component and structural health management.The operating U.S. fleet includes 104 light water reactors. In addition, there are now (as of May 2012) four new nuclear power plants (AP-1000 plants) under construction in the United States, and two delayed plants are being completed by the Tennessee Valley Authority. There is also interest in the United States in small modular reactors (SMRs), which could be easier to match to existing grid infrastructure and which could replace aging coal fired plants. The current low price for natural gas presents a challenge to the economics of nuclear power, at least in the short term; however, some recent studies have demonstrated that nuclear generation will be competitive in the longer term (at least in some markets) when anticipated escalation in gas prices and the cost of building, operating, and maintaining gas-fired plants are considered over those same time periods.This report reviews the current state of the art of prognostics and health management (PHM) for nuclear power systems and related technology currently applied in field or under development in other technological application areas, as well as key research needs and technical gaps for increased use of PHM in nuclear power systems. The historical approach to monitoring and maintenance in nuclear power plants (NPPs), including the Maintenance Rule for active components and Aging Management Plans for passive components, are reviewed. An outline is given for the technical and economic challenges that make PHM attractive for both legacy plants through Light Water Reactor Sustainability (LWRS) and new plant designs. There is a general introduction to PHM systems for monitoring, fault detection and diagnostics, and prognostics in other, non-nuclear fields. The state of the art for health monitoring in nuclear power systems is reviewed. A discussion of related technologies that support the application of PHM systems in NPPs, including digital instrumentation and control systems, wired and wireless sensor techno...

show abstract

Condition Monitoring Using Empirical Models: Technical Review and Prospects for Nuclear Applications

Cited by 31 publications

References 13 publications

Track Gauge Degradation Modelling on Small Urban Rail Networks: Zagreb Tram System Case Study

Track Gauge Degradation Modelling on Small Urban Rail Networks: Zagreb Tram System Case Study

A Denoising Based Autoassociative Model for Robust Sensor Monitoring in Nuclear Power Plants

Prognostics and Health Management in Nuclear Power Plants: A Review of Technologies and Applications

Contact Info

Product

Resources

About