Keynote Speech 1: Helicopter main gearbox planetary bearing fault diagnosis
Professor David Mba, Pro Vice-Chancellor and Dean of Technology at De Montfort University, UK.
The condition monitoring of helicopter main gearbox (MGB) is crucial for operation safety, flight airworthiness and maintenance scheduling. Currently, the helicopter health and usage monitoring system, HUMS, is installed on helicopters to monitor the health state of their transmission systems and predict remaining useful life of key helicopter components. However, recent helicopter accidents related to MGB failures indicate that HUMS is not sensitive and accurate enough to diagnose MGB planetary bearing defects. The keynote paper presents the necessity for improvements in current MGB diagnostic systems and presents some recent studies in enhancing planetary bearing fault diagnosis on helicopter transmission systems.
Professor David Mba is a leading authority in machine condition monitoring, machine diagnosis and prognosis. He has contributed to the development and publication of international standards in the subject area of Acoustic Emission and Vibration diagnosis. David’s research has been funded via numerous commercial, EU and EPSRC grants; he has published over 250 journal and conference papers and in 2010, he was awarded the Ludwig Mond prize in recognition of the best contribution to the progress of Mechanical engineering of interest to the chemical industry (UK). David’s current research is focused on machine fault diagnosis, model based prognostics and machine performance prediction.
Professor Mba is currently Pro Vice-Chancellor and Dean of Technology at De Montfort University, UK.
Keynote Speech 2: Uncovering the Value of PHM in Industrial Applications
Dr. Nicholas Williard, Data Scientist, Schlumberger, USA.
Prognostics and health management has been propelled by a larger, overarching trend to collect and extract meaning from data. Internet of things (IoT) architectures are becoming commoditized so that sensor and operational data can be easily collected and streamed in real time. Machine learning algorithms have been packaged in open source frameworks so they can be used to rapidly develop and deploy models for a wide variety of different applications. And there is a new generation of data scientists with cross functional skillsets including algorithm development, computer science, statistics and general engineering, that are equipped to solve problems in predictive analytics. These factors have helped drive revenue for tech companies looking to profit from selective advertisement and e-commerce. However, the value proposition of predicting failures in industrial applications is a little less clear. Being able to predict a failure doesn’t necessarily mean that you can prevent it. Providing false alarms could potentially be costlier than any benefits gained from true failure prediction. There are several factors from model accuracy to the nature of the equipment being monitored, that effects the overall value of a PHM system. This discussion will touch on lessons learned on how to maximize the value of a PHM system to gain the most benefit in an industrial setting.
Keynote Speech 3: Enabling Modern Day Automotive Electronics using Physics of Failure
Dr. Preeti Chauhan, Quality and Reliability Program Manager, Intel Corporation, USA.
Automotive industry is going through a major innovation phase with the introduction of several groundbreaking technological advances such as Automated Driver Assistance System (ADAS), and autonomous driving. While these advances will radically change the face of automotive electronics, they come with novel electronic package architectures, increased design and manufacturing complexity and higher uncertainties in use conditions. Given these boundary conditions, the historical approach of using a database of known reliability qualification tests will no longer work.
Physics of Failure (PoF) approach utilizes the key technology attributes of the electronic packages and uses fail-mode specific reliability models and qualification tests. The approach starts from the design phase wherein the package architecture is selected based on the application requirement as well as customer needs. Once the initial package architecture is frozen, the key fail modes identified during Failure Mode and Effect Analysis are assessed on the prototypes and the final product.
The talk will present the application of the PoF approach for enabling some of the advanced packaging architectures used in ADAS and autonomous driving applications in order to enable development and qualification for accelerated time to market. The approach can then be further augmented by the prognostics and health management (PHM) techniques to help tackle the uncertainty in the use conditions, which is one of the biggest challenges for automotive electronics.
Dr. Preeti S. Chauhan is currently the Quality and Reliability Program Manager in Assembly Test and Technology Development division at Intel Corporation. Her job role comprises of providing technical leadership to a team of over 15 engineers in the area of semiconductor packaging to drive quality and reliability certification of Intel’s server microprocessors. Dr Chauhan received Ph.D and M.S. in Mechanical Engineering from CALCE at University of Maryland. Her research focused on the reliability evaluation of lead-free solder interconnects, copper wire bonding and prognostics and health management of electronics. Dr. Chauhan authored a book on the challenges and technology enablers for Copper Wire Bonding in 2013, and has authored several book chapters in the area of reliability of electronic packages and PHM. She has published more than twenty refereed articles in high impact journals and conferences. She has also been involved in peer review of journal articles for Microelectronics Reliability and Transactions on Material Device Reliability. Dr. Chauhan was awarded the 2017 Early Career Award by her Alma Mater - James Clark School of Engineering at University of Maryland, College Park in recognition of her professional achievements at an early stage of career. Preeti is an IEEE member and has recently been elected to the IEEE Reliability Society Administrative Committee.
Keynote Speech 4: Self-Certification of Robotics – The Route to Persistent Autonomy
Dr. David Flynn, Eminent Overseas Professor of Nagasaki University and Associate Professor (Reader) at Heriot-Watt University, UK.
At present, robots can undertake constrained semi-autonomous inspections, using predetermined tasks (missions) with minimum supervision. In most applications, the state of the world changes with time, and sensors are employed to measure strategic priorities of this evolving picture from the ambient world state. However, sensors often fail during operation, feeding decision-making with wrong information about the world. Moreover, hardware degradation may alter dynamic behaviour and subsequently alter the capabilities of an autonomous inspection system, rendering the original mission infeasible.
In this talk, we propose an initial architecture to the safe verification and validation of robot inspections for offshore assets. Our first contribution relates to the verification and validation architecture, which takes into account risks associated with asset inspection, safety protocols, evolving ambient changes, as well as the inherent state of health of the robot.
The second part of our paper looks to how machine learning (ML) can be used to create prognostic algorithms that account for skewed data sets. ML techniques reviewed in this paper include: using incremental kernel regression is used for dynamic modeling: whenever the environment of operation changes, or the use of Bayesian reasoning to provide a rigorous framework for addressing uncertainty. In addition, Bayesian Networks are also used to provide complex inference regarding hardware degradation. Finally, Relevance Vector Machines can be used to provide accurate predictions of the remaining useful lifetime (RUL) of battery assets.