Browsing by Author "Namoano, Bernadin"

Now showing 1 - 19 of 19
  • Thumbnail Image
    ItemOpen Access
    Automation of knowledge extraction for degradation analysis
    (Elsevier, 2023-07-13) Addepalli, Sri; Weyde, Tillman; Namoano, Bernadin; Oyedeji, Oluseyi Ayodeji; Wang, Tiancheng; Erkoyuncu, John Ahmet; Roy, Rajkumar
    Degradation analysis relies heavily on capturing degradation data manually and its interpretation using knowledge to deduce an assessment of the health of a component. Health monitoring requires automation of knowledge extraction to improve the analysis, quality and effectiveness over manual degradation analysis. This paper proposes a novel approach to achieve automation by combining natural language processing methods, ontology and a knowledge graph to represent the extracted degradation causality and a rule based decision-making system to enable a continuous learning process. The effectiveness of this approach is demonstrated by using an aero-engine component as a use-case.
  • Thumbnail Image
    ItemOpen Access
    Change detection in streaming data analytics: a comparison of Bayesian online and martingale approaches
    (Elsevier, 2020-12-18) Namoano, Bernadin; Emmanouilidis, Christos; Ruiz-Carcel, Cristobal; Starr, Andrew G.
    On line change detection is a key activity in streaming analytics, which aims to determine whether the current observation in a time series marks a change point in some important characteristic of the data, given the sequence of data observed so far. It can be a challenging task when monitoring complex systems, which are generating streaming data of significant volume and velocity. While applicable to diverse problem domains, it is highly relevant to monitoring high value and critical engineering assets. This paper presents an empirical evaluation of two algorithmic approaches for streaming data change detection. These are a modified martingale and a Bayesian online detection algorithm. Results obtained with both synthetic and real world data sets are presented and relevant advantages and limitations are discussed.
  • Thumbnail Image
    ItemOpen Access
    Cognitive data imputation: case study in maintenance cost estimation
    (Elsevier, 2023-07-13) Erkoyuncu, John Ahmet; Namoano, Bernadin; Kozjek, Dominik; Vrabič, Rok
    Cost estimation is critical for effective decision making in engineering projects. However, it is often hampered by a lack of sufficient data. For this, data imputation techniques can be used to estimate missing costs based on statistical estimates or analogies with historical data. However, these techniques are often limited because they do not consider the existing knowledge of experts. In this paper, a novel cognitive data imputation technique is proposed for cost estimation that uses explanatory interactive machine learning to integrate and improve human knowledge. Through a case study in maintenance cost estimation the effectiveness of the approach is demonstrated.
  • Thumbnail Image
    ItemOpen Access
    Data-driven wheel slip diagnostics for improved railway operations
    (Elsevier, 2022-09-27) Namoano, Bernadin; Ruiz-Carcel, Cristobal; Emmanouilidis, Christos; Starr, Andrew G.
    Wheel slip activity detection is crucial in railway maintenance, as it can contribute to avoiding wheel damage but also track deteriorations leading to significant maintenance costs, trains delays, as well as the risk of accidents. Wheel slip activity is characterised by lower adhesion between track and wheel, especially in braking conditions, locking the wheels. It is complex to model or predict, being influenced by a multitude of factors including ambient conditions, global vehicle load, track and axle quality, leaves and objects present on the rail, steep incline, oxidation of the rails, and braking forces applied to the wheels. This paper presents a combined wavelet and tuned Long-Short Term Memory (LSTM) approach for the detection of wheel slip from time series data collected from real-world trains. Results provide evidence of superior performance over methods such as decision trees and random forests, naïve Bayes, k-nearest neighbours, logistic regression, and support vector machines.
  • Thumbnail Image
    ItemOpen Access
    Designing a semantic based common taxonomy of mechanical component degradation to enable maintenance digitalisation
    (Elsevier, 2023-07-08) Addepalli, Sri; Namoano, Bernadin; Oyedeji, Oluseyi Ayodeji; Farsi, Maryam; Erkoyuncu, John Ahmet
    Digital data management and enterprise systems have become key to support the digitalisation of maintenance activities. With traditional maintenance activities still striving for efficiencies, platforms such as the natural language processing (NLP) are supporting industries to mine textural data, not just extracting degradation terminologies but providing the maintainer with holistic insights on the degradation process. Traditionally, the degradation analysis, the first step in maintenance, is a manual process for defect characterisation, followed by failure investigation and a remaining useful life estimation. To enable digitalisation, transfer of human cognitive decision making from the physical world to the digital world is key. This paper enables this cognitive knowledge transfer through the design of a common degradation taxonomy and extracting terminology relationships to produce degradation causality with an NLP knowledge extraction approach. Further, this paper proposes and demonstrates a framework to present the data in the form of a knowledge graph populated using an application-level ontology. Use cases in the aerospace context have been used to show the power of the NLP and conceptual journey into the digitalisation of maintenance.
  • Thumbnail Image
    ItemOpen Access
    Detecting wheel slip from railway operational data through a combined wavelet, long short-term memory and neural network classification method
    (Elsevier, 2024-11) Namoano, Bernadin; Emmanouilidis, Christos; Starr, Andrew
    — Detecting wheel slip is important in railway operations, to prevent damage to wheels and tracks, reduce maintenance costs, improve safety and enhance passenger comfort. Slip activity is characterised by reduced adhesion between the wheel and the rail and limits effective braking or acceleration, causing also operational risks. It is influenced by environmental conditions, vehicle load, track and axle quality, contaminants, inclines, rail oxidation, and braking forces. This paper introduces an innovative method for wheel slip detection in operational trains, utilizing wavelet analysis combined with Long-Short Term Memory (LSTM) modelling. This method analyzes operational data to effectively identify wheel slip, showing promising results when compared to traditional classification-based machine learning methods such as decision trees, forests, logistic regression, naïve Bayes, and support vector machines. This novel approach addresses the complexities of wheel slip detection and is capable of identifying the conditions leading to slip several seconds prior to the commencement of the slip event, offering a practical solution for real-world railway systems.
  • Thumbnail Image
    ItemOpen Access
    Digital twin architecture for a sustainable control system in aircraft engines
    (Springer , 2024-08-08) Farsi, Maryam; Namoano, Bernadin; Latsou, Christina; Subhadu, Vaishnav Venkata; Deng, Haoxuan; Sun, Zhen; Zheng, Bohao; D’Amico, Davide; Erkoyuncu, John Ahmet; Karakoc, T. Hikmet; Colpan, Can Ozgur; Dalkiran, Alper
    Over the past decades, climate change has remained one of the major global challenges in the world. In the aviation and aerospace industry, the environmental sustainable development strategies towards carbon-neutral mainly focus on efficiency and demand measures, sustainable fuels, renewable energies, and removal and carbon offsetting. The carbon dioxide equivalent (CO2e) emissions footprint of an aircraft is primarily determined by energy and fuel efficiency. The advanced engine control systems of an aircraft can optimise the engine performance to achieve energy efficiency, fuel optimal consumption, and emission reduction. This paper proposed a digital twin architecture of a sustainable aircraft control system that allows the system to collect, analyse, and optimise sustainability-related data and to provide insight to operators, engineers, maintainers, and designers. The required information, knowledge and insight databases across flight environment, engine specification, and gas emissions are identified. The research argued that the proposed architecture could enhance engine energy efficiency, fuel consumption, and CO2e footprint reduction and enable (near) real-time data monitoring, proactive anomaly detection, forecasting, and intelligent decision-making within an automated sustainability control system. This research suggests ontology-based digital twin as an effective approach to further develop a cognitive twin that facilitates automated decision-making within the aircraft control system.
  • No Thumbnail Available
    ItemOpen Access
    Doors datasets
    (Cranfield University, 2020-08-18 11:55) Namoano, Bernadin; Starr, Andrew; Emmanouilidis, Christos; Ruiz Carcel, Cristobal
    Data represents urban train condition monitoring raw data. Data contains, current and door motions during opening and closing periods.
  • No Thumbnail Available
    ItemOpen Access
    Engines datasets
    (Cranfield University, 2020-08-18 11:56) Namoano, Bernadin; Starr, Andrew; Emmanouilidis, Christos; Ruiz Carcel, Cristobal
    Data represents urban DMU trains engines condition monitoring data.
  • Thumbnail Image
    ItemOpen Access
    A fault detection technique based on deep transfer learning from experimental linear actuator to real-world railway door systems
    (PHM Society, 2022-10-28) Shimizu, Minoru; Perinpanayagam, Suresh; Namoano, Bernadin
    Fault detection for railway door systems based on data-driven approaches has been investigated in recent years due to the massive amount of available monitoring data. Despite much attention to its application, the major challenge is the lack of available faulty datasets to build a reliable model since railway maintenance is usually conducted regularly to avoid significant defects from economic and safety points of view. We aimed to tackle the issue by employing transfer learning. Firstly, we built a long-short term memory-based deep learning model using linear actuator experimental datasets. Then, we employed a transfer learning technique to adjust the deep learning model to be available to real-world railway door systems using a small amount of faulty data. As a result, high fault detection accuracy can be obtained at 0.979 as F1 score. The result reveals that an accurate fault detection model can be built even though a large amount of labelled datasets is unavailable. In addition, the proposed method is applicable to other door systems or electro-mechanical actuators since the method is unspecific to physical mechanisms and fault modes, and the only motor current signal is used in this research. The signal is primarily available from the controller or motor drive without additional sensors.
  • Thumbnail Image
    ItemOpen Access
    A full 3D reconstruction of rail tracks using a camera array
    (Elsevier, 2023-12-14) Wang, Yizhong; Liu, Haochen; Yang, Lichao; Durazo-Cardenas, Isidro; Namoano, Bernadin; Zhong, Cheng; Zhao, Yifan
    This research addresses limitations found in existing 3D track reconstruction studies, which often focus solely on specific rail sections or encounter deployment challenges with rolling stock. To address this challenge, we propose an innovative solution: a rolling-stock embedded arch camera array scanning system. The system includes a semi-circumferential focusing vision array, an arch camera holder, and a Computer Numerical Control machine to simulate track traverse. We propose an optimal configuration that balances accuracy, full rail coverage, and modelling efficiency. Sensitivity analysis demonstrates a reconstruction accuracy within 0.4 mm when compared to Lidar-generated ground truth models. Two real-world experiments validate the system's effectiveness following essential data preprocessing. This integrated technique, when combined with rail rolling stocks and robotic maintenance platforms, facilitates swift, unmanned, and highly accurate track reconstruction and surveying.
  • Thumbnail Image
    ItemOpen Access
    Multi-channel anomaly detection using graphical models
    (Springer, 2024-12-31) Namoano, Bernadin; Latsou, Christina; Erkoyuncu, John Ahmet
    Anomaly detection in multivariate time-series data is critical for monitoring asset conditions, enabling prompt fault detection and diagnosis to mitigate damage, reduce downtime and enhance safety. Existing literature predominately emphasises temporal dependencies in single-channel data, often overlooking interrelations between features in multivariate time-series data and across multiple channels. This paper introduces G-BOCPD, a novel graphical model-based annotation method designed to automatically detect anomalies in multi-channel multivariate time-series data. To address internal and external dependencies, G-BOCPD proposes a hybridisation of the graphical lasso and expectation maximisation algorithms. This approach detects anomalies in multi-channel multivariate time-series by identifying segments with diverse behaviours and patterns, which are then annotated to highlight variations. The method alternates between estimating the concentration matrix, which represents dependencies between variables, using the graphical lasso algorithm, and annotating segments through a minimal path clustering method for a comprehensive understanding of variations. To demonstrate its effectiveness, G-BOCPD is applied to multichannel time-series obtained from: (i) Diesel Multiple Unit train engines exhibiting faulty behaviours; and (ii) a group of train doors at various degradation stages. Empirical evidence highlights G-BOCPD's superior performance compared to previous approaches in terms of precision, recall and F1-score.
  • Thumbnail Image
    ItemOpen Access
    Named entity recognition in aviation products domain based on BERT
    (Institute of Electrical and Electronics Engineers (IEEE), 2024-12-12) Yang, Mingye; Namoano, Bernadin; Farsi, Maryam; Ahmet Erkoyuncu, John
    The aviation products' manufacturing industry is undergoing a profound transformation towards intelligence, among which the construction of a knowledge graph specifically for the aviation field has become the core link in achieving cognitive intelligence. In the process of knowledge graph construction, named entity recognition (NER) is a key step and one of the main tasks of knowledge extraction. Given the high degree of specialisation of aviation product text data and the wide span of contextual information, existing models often perform poorly in entity extraction. This paper proposes a new Named Entity Recognition (NER) method specifically tailored for the aviation product field (BBC-Ap), introducing an innovative approach that leverages domain-specific ontologies and advanced deep learning algorithms to significantly enhance the accuracy and efficiency of entity extraction from complex technical documents. The first step of this method is to establish an ontology model of aviation products and annotate the relevant text data to form a dataset for training the named entity model. Next, it adopts a multi-level model structure based on BERT, in which BERT is used to generate word vector representations, a bidirectional long short-term memory network (BiLSTM) is used as an encoder to extract semantic features, and a conditional random field (CRF) is used as a decoder to achieve optimal label assignment. Through experiments on the constructed aviation product dataset, the model achieved a Precision value of 91.74%, a Recall value of 92.46%, and an F1 score of 92.1%, Compared with other baseline models, the F1-score is improved by 0.9% to 1.5%. At the same time, the model also performs well on standard datasets such as CoNLLpp, with a Precision value of 92.87%, a Recall value of 92.54%, and an F1-Score of 92.70%. Finally, the model was used to successfully construct a knowledge graph reflecting the relationships between aviation products in Neo4j, further demonstrating the effectiveness and practicality of the method.
  • Thumbnail Image
    ItemOpen Access
    Online change detection techniques in time series: an overview
    (IEEE, 2019-08-29) Namoano, Bernadin; Starr, Andrew; Emmanouilidis, Christos; Cristobal, Ruiz Carcel
    Time-series change detection has been studied in several fields. From sensor data, engineering systems, medical diagnosis, and financial markets to user actions on a network, huge amounts of temporal data are generated. There is a need for a clear separation between normal and abnormal behaviour of the system in order to investigate causes or forecast change. Characteristics include irregularities, deviations, anomalies, outliers, novelties or surprising patterns. The efficient detection of such patterns is challenging, especially when constraints need to be taken into account, such as the data velocity, volume, limited time for reacting to events, and the details of the temporal sequence.This paper reviews the main techniques for time series change point detection, focusing on online methods. Performance criteria including complexity, time granularity, and robustness is used to compare techniques, followed by a discussion about current challenges and open issues
  • Thumbnail Image
    ItemOpen Access
    A real-time fault detection framework based on unsupervised deep learning for prognostics and health management of railway assets
    (IEEE, 2022-09-08) Shimizu, Minoru; Perinpanayagam, Suresh; Namoano, Bernadin
    Fault detection based on deep learning has been intensively investigated in the recent decade due to increasing availability of data and its ability to engineer features with deep neural network architectures. Despite much attention to its application, the major challenge is the lack of available labelled datasets to build the models since maintenance is usually conducted regularly to avoid significant defects. This paper aims to propose a successful real-time fault detection framework based on unsupervised deep learning using only healthy normal data. The approach is based on autoencoder architecture and a one-class support vector machine as a classifier. As a case study, large real-world datasets acquired from railway door systems have been employed. The five different types of deep learning models and a one-class classifier are trained and comprehensively validated based on performance metrics and sensitivity analysis. In addition, two experiments have been carried out to verify the model’s adaptability and robustness to variational time-series data. The result shows a typical autoencoder is the least sensitive to a decision boundary set by the one-class classifier. However, the two experiments show that the fault detection accuracy for a bidirectional long short-term memory-based autoencoder is considerably higher than other autoencoder-based models at 0.970 and 0.966 as F1 score, meaning only this model is adaptable and robust to variational data. The experimental result allows us to obtain the understandability of the deep learning models. Furthermore, the regions of anomalies are localised with unsupervised models, which enables diagnosing the cause of failure.
  • Thumbnail Image
    ItemOpen Access
    Real-time prognostics and health management without run-to-failure data on railway assets
    (IEEE, 2023-03-20) Shimizu, Minoru; Perinpanayagam, Suresh; Namoano, Bernadin; Starr, Andrew
    Prognosis is a challenging technology that aims to accurately predict and estimate the remaining useful life of a component or system in order to enhance its reliability and performance. Although prognosis research for predictive maintenance is a well-researched topic, practical examples of successful prognostic applications remain scarce. This is due to the lack of available run-to-failure data to build the prediction model as maintenance is usually conducted regularly to avoid significant defects. This paper proposes a novel prognosis method that can be applied to real-world railway maintenance planning without employing runto-failure data. The key idea is that the fault severity assessment and approximate remaining time prediction are often all that is needed in order to plan maintenance. Firstly, using motor current signals, a degradation indicator on railway door systems is generated based on the dynamic time warping method to measure similarity between typical normal and faulty behaviour. Then, the K-means algorithm is applied to assess fault severity, followed by the representative time estimation for each level of fault severity. This estimation thus allows the remaining time prediction until reaching the critical fault severity level without using runto-failure data. As a result, the proposed method enables predictive maintenance planning for railway door systems. In addition, the fault severity threshold can be updated by additional operational data, enabling the remaining time prediction to be more reliable. Furthermore, the proposed method can be applied to conventional railway assets and other electro-mechanical actuators as motor current signals are primarily available from the controller or motor drive without additional sensors.
  • Thumbnail Image
    ItemOpen Access
    Real-time techniques for fault detection on railway door systems
    (IEEE, 2022-08-10) Shimizu, Minoru; Perinpanayagam, Suresh; Namoano, Bernadin
    This paper focuses on real-time techniques for fault detection in railway assets through large real-world datasets. It aims to investigate data mining methods to detect faulty behaviour in time series data. A fault detection on railway door systems is carried out using motor current and encoder signal. The door data highlighted start-stop characteristics, with discontinuities in the data. This paper presents a successful fault detection technique, which is a feature-based machine learning method that requires several steps for time-series data processing, such as signal segmentation and the extraction of features. Principal Component Analysis (PCA) is applied to reduce the dimensionality of the extracted feature set and generate condition indicators. Then, the k-means algorithm is employed to separate normal and abnormal behaviour. This is followed by an evaluation of the proposed method and discussion about current challenges and prognosis possibility.
  • Thumbnail Image
    ItemOpen Access
    A robust design for lifecycle cost with reliability analysis integration
    (Elsevier, 2023-07-08) Farsi, Maryam; Namoano, Bernadin; Sonmez, Ayse Nur; Addepalli, Pavan; Erkoyuncu, John Ahmet
    Maintenance, repair, and overhaul (MRO) is the most significant cost driver over a complex engineering asset lifecycle. Therefore, high-value manufacturers are required to plan MRO occurrences to optimize the overhaul cost while achieving the desired performance. This trade-off imposes a shift towards a proactive maintenance strategy. However, creating a long-term proactive maintenance plan is challenging due to uncertainties in the performance of the asset and its critical components. Hence, this paper presents a robust design framework for the lifecycle cost estimation process by integrating reliability life data analysis. The level of data availability across the lifecycle is considered. The framework is proposed based on a literature review and the Delphi method. This study highlights that the level of robustness in the lifecycle cost estimates can be achieved by continuous feedback to the design phase and to the body of knowledge over the asset lifecycle. Moreover, this study suggests that the optimization model for the trade-off between cost and reliability should fulfil safety and environmental sustainability requirements when providing a cost-effective reliability solution.
  • Thumbnail Image
    ItemOpen Access
    Track geometry deterioration modelling for asset management: a visual analytics approach
    (Cranfield University, 08/11/2022) Alotaibi, Abdulaziz; Durazo-Cardenas, Isidro; Namoano, Bernadin; Starr, Andrew
    To maintain safe operations and cost-effective maintenance, British railway tracks must be monitored. Track recording assets which include trains and cars, regularly monitor key components of the track in order to detect and diagnose early incipient faults. The measurements accumulate over time, providing time series data that can be used to model track geometry deterioration process. However, the modelling results are often too sophisticated to be used to their full potential in track asset management. As a result, the goal of this research is to use visualisation approaches to display the results of track geometry deterioration, which would simplify and enhance track asset management. Two visual techniques have been used. The first visual includes two dimensional plots enabling visual fault detection and localisation and the second is a 3D plot which gives a better sight for the decision makers to act. These visual analytics allowed a better understanding of fault occurrence, enable a vast amount of data integration, flexible and simple for stakeholders to use. The limitations of such approaches include the inability to visualise more than 5 dimensions and human interpretation.