Search

In this paper, we propose an offline and online machine health assessment (MHA) methodology composed of feature extraction and selection, segmentation-based fault severity evaluation, and classification steps. In the offline phase, the best representative feature of degradation is selected by a new filter-based feature selection approach. The selected feature is further segmented by utilizing the bottom-up time series segmentation to discriminate machine health states, ie, degradation levels. Then, the health state fault severity is extracted by a proposed segment evaluation approach based on within segment rate-of-change (RoC) and coefficient of variation (CV) statistics. To train supervised classifiers, a priori knowledge about the availability of the labeled data set is needed. To overcome this limitation, the health state fault-severity information is used to label (eg, healthy, minor, medium, and severe) unlabeled raw condition monitoring (CM) data. In the online phase, the fault-severity classification is carried out by kernel-based support vector machine (SVM) classifier. Next to SVM, the k-nearest neighbor (KNN) is also used in comparative analysis on the fault severity classification problem. Supervised classifiers are trained in the offline phase and tested in the online phase. Unlike to traditional supervised approaches, this proposed method does not require any a priori knowledge about the availability of the labeled data set. The proposed methodology is validated on infield point machine sliding-chair degradation data to illustrate its effectiveness and applicability. The results show that the time series segmentation-based failure severity detection and SVM-based classification are promising.

International audience; This paper proposes a new adaptive prognostics approach consisting of hybrid feature selection and remaining-useful-life (RUL) estimation steps for railway point machines. In step-1, different time-domain based features are extracted and the best ones are selected by the hybrid feature selection method. Then, a degradation model is fitted to each of the selected features and the parameters are estimated. In step-2, the RUL of the component is predicted by using the proposed adaptive prognostics approach. The adaptive prognostics is based on the weighted likelihood combination of the estimated model parameters. The model parameters each of which estimated by curve fitting are used in the calculation of the likelihood probability weights. Then, an adaptive degradation model is built by using the weighted combination of the model parameter estimates and the component RUL is estimated. The proposed approach is validated on in-field point machine sliding-chair degradation and the results are discussed.

International audience; Detection of unexpected events (e.g. anomalies and faults) from monitoring data is very challenging in machine health assessment. Hence, abrupt or incipient fault detection from the monitoring data is very crucial to increase asset safety, availability and reliability. This paper presents a generic methodology for abrupt and incipient fault detection and feature fusion for health assessment of complex systems. Proposed methodology consists of feature extraction, feature fusion, segmentation and fault detection steps. First of all, different features are extracted using descriptive statistics. Secondly, based on linearly weighted data fusion algorithm, extracted features are combined to get the generic and representative feature. Afterward, combined feature is divided into homogeneous segments by sliding window segmentation algorithm. Finally, each segment is further evaluated by coefficient of variability which is used in inferential statistics, to evaluate health state changes that indicate asset faults. To illustrate its effectiveness, the methodology is implemented on point machine and Li-ion battery monitoring data to detect abrupt and incipient faults. The results show that proposed methodology can be effectively used in fault detection for asset monitoring.

Tüm dünyada tren, metro ve tramvay sistemlerini içeren demiryollarına olan talebin gün geçtikçe arttığı görülmektedir. Bu artışa cevap verilebilmesi için demiryolları sistemlerinin güvenilirliği, hazır olma oranı ve yolcu emniyetinin arttırılması gerekmektedir. Bu ise demiryollarına uygulanan bakım sayısının arttırılmasıyla gerçekleştirilebilse de artan bakım sayısı maliyeti yükseltmektedir. Örneğin, İngiltere'de 1000 km'lik demiryolunda bulunan makas sistemlerinin bakımı için her sene 3,4 milyon sterlin harcanmaktadır. Bu yüzden bakım maliyetini en aza indiren erken hata teşhis metodlarının geliştirilmesi önem arz etmektedir. Bu metodlar oluşmakta olan hataları önceden haber vererek bakım maliyetini azaltmakta ve güvenliği arttırmaktadır. Bu tezde zaman serisi analiz yöntemleri kullanılarak demiryollarının en önemli bileşeni olan makas sistemlerinde arıza tespit metodları üzerine çalışilmıştır.Bu çalışma kapsamında arızasız bir makas sisteminden çeşitli sensörler kullanılarak sınırlı miktarda veri toplanmıştır. Bu gerçek veriler simüle edilerek arızasız ve iki arıza türünde çok miktarda veri elde edilmiştir. Arızasız ve hatalı olarak simüle edilen veriler öğrenme ve test olmak üzere iki gruba ayrılmıştır. Bu veriler eksponansiyel düzleştirme algoritması kullanılarak gürültüden arındırılmıştır. Daha sonra uzman sistem ve Self Organizing Map (SOM) kullanılarak hata tespiti gerçekleştirilmiştir. Geliştirilen uzman sistemde arıza benzerlik seviyesi Dinamik Zaman Eğilimi (Dynamic Time Warping), ve Öklid Uzaklık (Euclidean Distance) metodları ile belirlenmiştir. The importance of railway systems that include trains, metro, and tramway has been increasing in the world. It is now an obligation to improve reliability, availability, and safety of railway systems in order to accommodate increasing passenger and cargo transportation with higher train speeds, greater axle loads, and increased service frequency. Increasing reliability, availability and safety can be achieved by increasing the number of maintenance in the railway system. However, this will increase the operating and support cost. For example in England, 3.4 million pound is spent every year for maintenance of point machines in 1000 km of railways. Hence, it is now a great need to develop diagnostics methods that will help to identify the failures early enough. These failure diagnostics methods not only reduce the cost, but also increase the safety. In this thesis, failure identification methods in railway turnouts using time series analysis are studied.Limited sensory data is collected from a healthy turnout system. Large numbers of data are simulated for healthy and faulty system using the limited real data. Simulated data in both healthy and faulty system are divided into two groups: training and testing. Noise is removed from the data using exponential smoothing algorithm. Self Organizing Map (SOM) and expert system are used for failure identification. Dynamic Time Warping and Euclidean Distance are employed for identification of similarity level between the real healthy system and signal under observation. Similarity level is used in expert system for failure identification. 82