Dr Samir Khan

Dr Samir Khan

Associate Professor at the University of Tokyo

Publications

On the requirements of digital twin-driven autonomous maintenance

Published in Annual Reviews in Control, 2020

Autonomy has become a focal point for research and development in many industries. Whilst this was traditionally achieved by modelling self-engineering behaviours at the component-level, efforts are now being focused on the sub-system and system-level through advancements in artificial intelligence. Exploiting its benefits requires some innovative thinking to integrate overarching concepts from big data analysis, digitisation, sensing, optimisation, information technology, and systems engineering. With recent developments in Industry 4.0, machine learning and digital twin, there has been a growing interest in adapting these concepts to achieve autonomous maintenance; the automation of predictive maintenance scheduling directly from operational data and for in-built repair at the systems-level. However, there is still ambiguity whether state-of-the-art developments are truly autonomous or they simply automate a process. In light of this, it is important to present the current perspectives about where the technology stands today and indicate possible routes for the future. As a result, this effort focuses on recent trends in autonomous maintenance before moving on to discuss digital twin as a vehicle for decision making from the viewpoint of requirements, whilst the role of AI in assisting with this process is also explored. A suggested framework for integrating digital twin strategies within maintenance models is also discussed. Finally, the article looks towards future directions on the likely evolution and implications for its development as a sustainable technology.

Recommended citation: Khan, et al (2020). "On the requirements of digital twin-driven autonomous maintenance." Annual Reviews in Control. 1(1).

Olfactory-based augmented reality support for industrial maintenance

Published in IEEE Access, 2020

Augmented reality (AR) applications have opened innovative ways for performance improvement in the IoT industry. It can enhance user perception of the real-world by providing valuable information. This article presents how odors and scents in an industrial environment can be used to provide visual virtual information onto a head-mounted device (HMD). Such information is important for maintainers to quickly detect abnormalities. Since odors are made up of volatile compounds at low concentration, they can be identified by using sensors whose outputs reach specific values depending on the number of volatile compounds. An electronic nose, composed of three gas sensors is developed for the purposes of air analysis. The device is integrated with AR smart glasses (Microsoft HoloLens). Results are then sent through a local network to the HMD and displayed with other relevant information concerning the diagnosed odor. To validate the device, four odors including engine oil, sun lotion, medical alcohol and perfume have been analyzed. The experiments recorded miscellaneous behaviors of each sample to demonstrate the repeatability of the process. The presented technology incorporates sampling methods, cleaning processes and statistical analysis that can be further scrutinized to allow correct smell identification.

Recommended citation: Khan, et al (2018). "Olfactory-based augmented reality support for industrial maintenance." IEEE Access. 1(1).

A review on the application of deep learning in system health management

Published in Mechanical Systems and Signal Processing, 2018

Given the advancements in modern technological capabilities, having an integrated health management and diagnostic strategy becomes an important part of a system’s operational life-cycle. This is because it can be used to detect anomalies, analyse failures and predict the future state based on up-to-date information. By utilising condition data and on-site feedback, data models can be trained using machine learning and statistical concepts. Once trained, the logic for data processing can be embedded on on-board controllers whilst enabling real-time health assessment and analysis. However, this integration inevitably faces several difficulties and challenges for the community; indicating the need for novel approaches to address this vexing issue. Deep learning has gained increasing attention due to its potential advantages with data classification and feature extraction problems. It is an evolving research area with diverse application domains and hence its use for system health management applications must been researched if it can be used to increase overall system resilience or potential cost benefits for maintenance, repair, and overhaul activities. This article presents a systematic review of artificial intelligence based system health management with an emphasis on recent trends of deep learning within the field. Various architectures and related theories are discussed to clarify its potential. Based on the reviewed work, deep learning demonstrates plausible benefits for fault diagnosis and prognostics. However, there are a number of limitations that hinder its widespread adoption and require further development. Attention is paid to overcoming these challenges, with future opportunities being enumerated.

Recommended citation: Khan, S., & Yairi, T. (2018). A review on the application of deep learning in system health management. Mechanical Systems and Signal Processing, 107, 241-265. https://doi.org/10.1016/j.ymssp.2017.11.024

Unsupervised anomaly detection in unmanned aerial vehicles

Published in Applied Softcomputing, 2017

A real-time anomaly detection solution indicates a continuous stream of operational and labelled data that must satisfy several resources and latency requirements. Traditional solutions to the problem rely heavily on well-defined features and prior supervised knowledge, where most techniques refer to hand-crafted rules derived from known conditions. While successful in controlled situations, these rules assume that good data is available for them to detect anomalies; indicating that these rules will fail to generalise beyond known scenarios. To investigate these issues, current literature is examined for solutions that can be used to detect known and unknown anomalous instances whilst functioning as an out-of-the-box approach for efficient decision-making. The applicability of the isolation forest is discussed for engineering applications using the Aero-Propulsion System Simulation dataset as a benchmark where it is shown to outperform other unsupervised distance-based approaches.

Recommended citation: Khan, et al (2019). "Unsupervised anomaly detection in unmanned aerial vehicles." CIRP Applied Soft Computing. 83, 105650.

Evaluating diagnostic failures during System Design

Published in CIRP Journal of Manufacturing Science and Technology, 2017

Modern complex mechanical systems, such as a fuel system, often face a high number of No Fault Found (NFF) events due to design limitations associated with testability. Therefore, this paper investigates how the diagnostic analysis of a system that can be improved to help recognise and reduce failure ambiguity groups that lead to NFF events. The simulation-based evaluation investigates the replacement of failed components to demonstrate a method for estimating cumulative replacement costs due to false avionic unit removals and hence, the overall system life-cycle costs during the design stage. The analysis may be used not only as a baseline for the prediction of NFF rates, but also as a measure of how maintenance requirements might change over the intended lifetime of a system.ture work.

Recommended citation: Khan, et al (2018). "Evaluating No Fault Found problems during System Design." CIRP Journal of Manufacturing Science and Technology. 1(1).

Perspectives on trading cost and availability for corrective maintenance at the equipment type level

Published in Reliability Engineering & System Safety, 2017

This article investigates a process that has been developed to estimate performance based support contract costs attributed to corrective maintenance. These can play a dominant role in the through-life support of high values assets. The case context for the paper is the UK Ministry of Defence. The developed approach allows benchmarking support contract solutions, and enabling efficient planning decisions. Emphasis is placed on learning from feedback, testing and validating current methodologies for estimating corrective maintenance costs and availability at the Equipment Type level. These are interacting sub-equipments that have unique availability requirements and hence have a much larger impact on the capital maintenance expenditure. The presented case studies demonstrate the applicability of the approach towards adequate savings and improved availability estimates.

Recommended citation: Erkoyuncu, et al (2017). "Perspectives on trading cost and availability for corrective maintenance at the equipment type level." Reliability Engineering & System Safety. 1(1). https://doi.org/10.1016/j.ress.2017.05.041

A novel approach for No Fault Found decision making

Published in CIRP Journal of Manufacturing Science and Technology, 2017

This paper focuses on outlining current industrial attitudes regarding the No Fault Found (NFF) phenomena and identifies the drivers that influence the NFF decision-making process. It articulates the contents of tacit knowledge and addresses a knowledge gap by developing NFF management policies. The paper further classifies the NFF phenomenon into five key processes that must be controlled by using the developed policies. In addition to the theoretical developments, a Petri net model is also outlined and discussed based on the captured information regarding NFF decision-making in organisations. Since NFF decision-making is influenced by several factors, Petri nets are sought as a powerful tool to realise a meta-model capability to understand the complexity of situations. Its potential managerial implications can help describe decision problems under conditions of uncertainty. Finally, the conclusions indicate that engineering processes, which allow decision-making at various maintenance echelons, can often obfuscate problems that then require a systems approach to illustrate the impact of the issue.

Recommended citation: Khan et al (2017). "A novel approach for No Fault Found decision making." CIRP Journal of Manufacturing Science and Technology. 17, 18–31. http://dx.doi.org/10.1016/j.cirpj.2016.05.011

Modelling the positional and orientation sensitivity of inductively coupled sensors for industrial IoT applications

Published in International journal of simulation systems, science and technology, 2016

As the Internet of Things (IoT) sector continually expands there is a growing abstraction between physical objects and the data associated with them. At the same time, emerging Industrial-IoT applications rely upon diverse and robust hardware sensing interfaces in order to deliver high quality data. In this paper, the fundamental limitations associated with inductive proximity sensing interfaces are considered in terms of positional and orientation sensitivity and a triaxial approach is proposed that enables arbitrary source-sensor positioning. A matrix transformation model based on the field coupling equations is applied to a number of candidate configurations assessed according their relative source-sensor coverage and graphical visualization of coupling quality. Particular attention is paid to the recombination of tri-sensor outputs involving direct-summation, rectifysummation, best-coil and root-mean-square methods. Of these, the rectify-summation method was observed to provide favorable performance, exceeding 70% coverage for practical cases, thus far exceeding that of traditional co-planar arrangements.

Recommended citation: McWilliam et al (2016). "Modelling the positional and orientation sensitivity of inductively coupled sensors for industrial IoT applications." International journal of simulation systems, science and technology. 17(35), 23-1. http://ijssst.info/Vol-17/No-35/paper23.pdf

Research Study from Industry-University Collaboration on No Fault Found Events

Published in Journal of Quality in Maintenance Engineering, 2015

This paper emphasises the importance of the breath of interaction channels and demonstrates the opportunities for effective knowledge exchange by using the activities at Cranfield University to demonstrate their usefulness. The arguments clearly lead to the necessity of academia in this type of industrial problem. However, the presence of a university in this case is not as the sole problem solver, but the rather to act as a collaborative medium between various other outlets. Further ideas proposed, such as constructing guidelines for industries in handling NFF problems and benchmarking tools, can serve as real products that can be benefit industries. The study also aims to promote best practice in the field of maintenance management and outlines the foundations for NFF training material. The originality of the paper is that it presents a structured methodology for engaging with industry. It also outlines a curriculum for NFF training. It essentially serves as a road-map for research and offers a detailed account of areas that need to be taken into account in order to reduce the likely event of NFF.

Recommended citation: Khan (2015). "Research Study from Industry-University Collaboration on No Fault Found Events." Journal of Quality in Maintenance Engineering. 21(2). http://dx.doi.org/10.1108/JQME-01-2014-0004

Studying the impact of intermittent variations using sensitivity analysis

Published in International Journal of Condition Monitoring, 2015

Fault diagnostics focuses on the detection, identification and isolation of failures. However, this becomes challenging when investigating fault alarms that cannot be verified, diagnosed or even duplicated under standard manual inspection regimes. To improve system effectiveness, it is essential to investigate these instances, along with the effects of design parameters on system dynamic characteristics. Recent research has identified intermittent fault behaviour within components as one of the primary focuses for false alarms, and hence a direct consequence to the phenomenon of no fault found. This paper examines the performance characteristics of an electronic system under intermittent component variations. Understanding occurrences in parameter deviations (and their impact) can help with understanding the requirements for improving system fault tolerance. It is shown that, in many cases of practical importance, components do not have the same sensitivity to intermittent variations and hence can be better suited for monitoring. The analysis provides extra information and guidance for the maintenance decision-making process in organisations on resource requirements.

Recommended citation: Khan, et al (2015). "Studying the impact of intermittent variations using sensitivity analysis." International Journal of Condition Monitoring. 5(3), 6-11(6). http://dx.doi.org/10.1784/204764215816195883

A framework to estimate the cost of No-Fault Found events

Published in International Journal of Production Economics, 2015

The article investigates a generic framework to estimate maintenance costs attributed to the No Fault Found (NFF) phenomenon. Such overhead costs are particularly difficult to quantify due to potentially serviceable equipment being returned for repair. Other factors, such as a reduction in the availability of the system, compromising reliability of high value assets, and logistical factors, can all contribute to the cost of resolving an unknown fault. Here we apply the soft systems methodology to capture the critical cost drivers of NFF across the supply chain and build a framework to estimate the cost of NFF. We use a multi-method design including an online survey, workshops and semi-structured interviews to study NFF related cost practices based on information from 12 key participants across 7 UK organisations. The study identifies the major NFF cost drivers across the supply chain (e.g. transportation), the OEM (e.g. inventory) and the customer (e.g. lost man hours). An agent based model is used to evaluate the impact of these cost drivers on the overall NFF cost. The analysis shows how the most appropriate drivers can be selected to represent the cumulative costs due to NFF events and their impacts across the supply network. From the academic perspective, the generic framework for NFF cost estimation demonstrates how qualitative and quantitative information can be used together to achieve maintenance objectives. From a practical perspective, by applying the framework on one component, an organisation has the liberty to analyse the cost of NFF for that particular unit only.

Recommended citation: Erkoyuncu et al (2015). "A framework to estimate the cost of No-Fault Found events." International Journal of Production Economics. 173(1). http://dx.doi.org/10.1016/j.ijpe.2015.12.013

No Fault Found events in maintenance engineering Part 2: Root causes, technical developments and future research

Published in Reliability Engineering & System Safety, 2014

This is the second half of a two paper series covering aspects of the no fault found (NFF) phenomenon, which is highly challenging and is becoming even more important due to increasing complexity and criticality of technical systems. Part 1 introduced the fundamental concept of unknown failures from an organizational, behavioral and cultural stand point. It also reported an industrial outlook to the problem, recent procedural standards, whilst discussing the financial implications and safety concerns. In this issue, the authors examine the technical aspects, reviewing the common causes of NFF failures in electronic, software and mechanical systems. This is followed by a survey on technological techniques actively being used to reduce the consequence of such instances. After discussing improvements in testability, the article identifies gaps in literature and points out the core areas that should be focused in the future. Special attention is paid to the recent trends on knowledge sharing and troubleshooting tools; with potential research on technical diagnosis being enumerated.

Recommended citation: Khan, et al (2014). "No Fault Found events in maintenance engineering Part 2: Root causes, technical developments and future research." Reliability Engineering & System Safety. 123, 196-208. http://dx.doi.org/10.1016/j.ress.2013.10.013

No Fault Found events in maintenance engineering Part 1: Current trends, implications and organizational practices

Published in Reliability Engineering & System Safety, 2014

This paper presents the first part of a state of the art review on the No Fault Found (NFF) phenomenon. The aim has been to compile a systematic reference point for burgeoning NFF literature, and to provide a comprehensive overview for gaining an understanding of NFF knowledge and concepts. Increasing systems complexities have seen a rise in the number of unknown failures that are being reported during operational service. Units tagged as NFF are evidence that a serviceable component was removed, and attempts to troubleshoot the root cause have been unsuccessful. There are many reasons on how these failures manifest themselves and these papers describe the prominent issues that have persisted across a variety of industrial applications and processes for decades. This article, in particular, deals with the impact of NFF from an organizational culture and human factors point of view. It also highlights recent developments in NFF standards, its financial implications and safety concerns.

Recommended citation: Khan et al (2014). "No Fault Found events in maintenance engineering Part 1: Current trends, implications and organizational practices." Reliability Engineering & System Safety. 123, 183-195. http://dx.doi.org/10.1016/j.ress.2013.11.003