Use this URL to cite or link to this record in EThOS:
Title: Prognostics and health management for multi-component systems
Author: Assaf, Roy
ISNI:       0000 0004 7430 9836
Awarding Body: University of Salford
Current Institution: University of Salford
Date of Award: 2018
Availability of Full Text:
Access from EThOS:
Access from Institution:
The ever increasing number of manufacturing requirements is pushing original equipment manufacturers (OEM) to design more complex systems to meet industrial needs. These systems are being fitted with more components which bear stochastic and economic dependencies. Therefore maintaining such systems is becoming more and more of a challenge, especially due to their degradation processes becoming highly stochastic in nature. This thesis is concerned with the prognostics and health management (PHM) of such complex multi-component systems, whereby signal processing and health indicator extraction, diagnostics, prognostics and maintenance decision making in light of present stochastic and economic dependencies are considered. We introduce several novel approaches for dealing with systems that have multiple components. We first introduce a gearbox accelerated life testing platform that was designed with the objective of gathering experimental data for multi-component degradation models, for the reason that multi-component systems with inter-dependencies follow a highly stochastic degradation process which depends to an extent on their complex mechanical design. We then present our methodology for extracting accurate health indicators from multi-component systems by means of a time-frequency domain analysis. This sets the stage for degradation modelling, and so we show the development of a generic degradation model in which the degradation process of a component may be dependent on the operating conditions, the component's own state, and the state of the other components. We then show how to fit the models to data using particle filter. This method is then used for the data generated by the gearbox. Afterwards a diagnostic procedure is presented and uses Gaussian mixture models. This is used to uncover accelerated wear processes that take place when old worn out components are coupled with new healthy components. Finally economic dependency is considered where combining multiple maintenance activities has lower cost than performing maintenance on components separately. To select a component or components to be preventively maintained, adaptive preventive maintenance and opportunistic maintenance rules are proposed. A cost model is developed to find the optimal values of decision variables. In our work, we find that stochastic dependencies between components lead to accelerated degradation which causes unexpected faults and failures, and consequent economic losses. Although this work deals with stochastic dependence between components, it involves some engineering knowledge of the systems under study, and this makes application of the models on a large scale challenging to automate. Therefore, we make recommendation for future research that includes the development of end-to-end learning techniques such as deep learning. In doing so we can potentially use the time wave data and automatically extract the most relevant features for doing accurate prognostics, and therefore health management, of such systems. The research work in this thesis was motivated by the problems faced by industrial partners such as the world leading food system manufacturing company Marel in the Netherlands, which were part of the sustainable manufacturing and advanced robotics training network in Europe (SMART-e).
Supervisor: Not available Sponsor: SMART-e
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID:  DOI: Not available