Comparison of Different Classification Algorithms for Fault Detection and Fault Isolation in Complex Systems

Jung, Marcel; Niculita, Octavian; Skaf, Zakwan

doi:10.1016/j.promfg.2018.01.016

Cited by 38 publications

(27 citation statements)

References 9 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…For example, Jung et al . [24] compared the performance of three classifiers SVM, weighted k‐NN, and DT, for fault detection and fault isolation in complex systems. According to Nowaczy et al .…”

Section: Approachmentioning

confidence: 99%

“…Jung et al . [24] compared the performance of three classifiers: SVM, k‐NN, and DT – using data from the optimised and non‐optimised sensor set solutions for fault detection in complex systems on a fuel system. The fuel delivery system comprises standard components such as a filter, pump, valve, nozzle, pipes, and two tanks, and their model detects any component‐related fault in the system.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Classification model for predictive maintenance of small steam sterilisers

Musabayli¹,

Osman

Dirix³

2020

IET Collaborative Intelligent Manufacturing

View full text Add to dashboard Cite

With 35,000 small steam sterilisers in the German market, after‐sales service and maintenance are critical issues for manufacturers and distributors. At present, preventive maintenance is one of the most commonly‐implemented maintenance strategies. However, with an average failure probability of 10%, ∼3500 autoclaves require unplanned repair per year, causing customers’ business interruptions and increased maintenance costs. From the authors’ observation, a predictive failure detection mechanism is needed to prevent failures and reduce the significant safety risk. Hence, this study proposes a predictive maintenance mechanism for small steam sterilisers. The predictive maintenance mechanism is constructed from classification models that categorised the health condition of two critical components in small steam sterilisers, i.e. a vacuum pump and a steam generator. The classification models were built from multisensory data, obtained from 1000 protocol records of CertoClav Vacuum Pro steam sterilisers. They perform exploratory experiments to find a suitable classification model. This study found that the random forest algorithm performed best in terms of accuracy for both the vacuum pump and steam generator data sets (83.5 and 82.0%, respectively). They also found that the features related to the pre‐vacuum stage profoundly influence the condition of the vacuum pump and the steam generator.

show abstract

Section: Approachmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Classification model for predictive maintenance of small steam sterilisers

Musabayli¹,

Osman

Dirix³

2020

IET Collaborative Intelligent Manufacturing

View full text Add to dashboard Cite

show abstract

“…The samples that are closest to the decision boundary, thus defining the hyper-planes, are called support vectors. In [62] the authors compare several classifiers for fault detection, including distance-weighted K-nearest neighbors and support vector machines among others.…”

Section: Classificationmentioning

confidence: 99%

On the Relevance of Preprocessing in Predictive Maintenance for Dynamic Systems

Cernuda

2019

Predictive Maintenance in Dynamic Systems

View full text Add to dashboard Cite

The complexity involved in the process of real-time data-driven monitoring dynamic systems for predicted maintenance is usually huge. With more or less in-depth any data-driven approach is sensitive to data preprocessing, understood as any data treatment prior to the application of the monitoring model, being sometimes crucial for the final development of the employed monitoring technique. The aim of this work is to quantify the sensitiveness of data-driven predictive maintenance models in dynamic systems in an exhaustive way.We consider a couple of predictive maintenance scenarios, each of them defined by some public available data. For each scenario, we consider its properties and apply several techniques for each of the successive preprocessing steps, e.g. data cleaning, missing values treatment, outlier detection, feature selection, or imbalance compensation. The pretreatment configurations, i.e. sequential combinations of techniques from different preprocessing steps, are considered together with different monitoring approaches, in order to determine the relevance of data preprocessing for predictive maintenance in dynamical systems.

show abstract

“…Para o diagnóstico de falhasé possível utilizar classificadores como Redes Neurais Artificiais (ANN), K-vizinhos Próximos (KNN),Árvore de Decisão (DT), Máquina de Vetores Suporte (SVM), ou outros, treinados a partir de dados. Jung et al (2018) comparou o desempenho dos classificadores SVM, KNN ponderado e DT, testados em um protótipo construído em laboratório para representar um sistema de combustível de veículos aéreos não tripulados (UAV). Tambémé possível utilizar inferência Bayesiana para diagnosticar falhas, como em Ge e Song (2009) e Md.…”

Section: Introductionunclassified

Diagnóstico de Falhas em um Sistema de Tanques Acoplados Operando em Múltiplas Regiões

Ribeiro

Bissoli

Munaro

2019

Anais Do 14º Simpósio Brasileiro De Automação Inteligente

View full text Add to dashboard Cite

In this work, a methodology for fault detection and diagnosis in systems whose variables are non-stationary but vary continuously in an operation region is evaluated. The external analysis method was evaluated estimating different models to generate residues used for detection and diagnosis. Fault detection was performed using principal component analysis followed by monitoring with Hotteling's T 2 statistics, quadratic prediction error and combined. Different classifiers were used for diagnosis using different hyperparameters and features. The methodology was applied to a four coupled tanks system subjected to 9 faults manifested in 4 different ways. The good results demonstrated that the methodology has good generalization capacity for treating failures in large operation regions without having been trained for all of them. Resumo: Neste trabalhoé avaliada uma metodologia para detecção e diagnóstico de falhas em sistemas cujas variáveis são não estacionárias, mas variam continuamente em uma região de operação. O método de análise por variáveis externas foi avaliado estimando diferentes modelos para geração dos resíduos usados para detecção e diagnóstico. A detecção de falhas foi feita usando análise de componentes principais seguida do uso de monitoramento com as estatísticas T 2 de Hotteling, erro de predição quadrático e combinada. Para o diagnóstico foram usados diferentes classificadores, avaliados com diferentes hiperparâmetros e features. A metodologia foi aplicada a um sistema de quatro tanques acoplados sujeito a 9 falhas que se manifestam de 4 formas diferentes. Os bons resultados demonstraram que a metodologia tem boa capacidade de generalização por tratar falhas em grandes regiões de operação sem ter sido treinada para todas elas.

show abstract

Comparison of Different Classification Algorithms for Fault Detection and Fault Isolation in Complex Systems

Cited by 38 publications

References 9 publications

Classification model for predictive maintenance of small steam sterilisers

Classification model for predictive maintenance of small steam sterilisers

On the Relevance of Preprocessing in Predictive Maintenance for Dynamic Systems

Diagnóstico de Falhas em um Sistema de Tanques Acoplados Operando em Múltiplas Regiões

Contact Info

Product

Resources

About