This repository contains all of the R, Stan, and LaTex code to reproduce the PhD Thesis Bayesian Hierarchical Modelling of Equipment Reliability in Mining: A Pragmatic Approach by Ryan K. Leadbetter. The PhD was done at the Center for Transforming Maintenance Through Data Science (CTMTDS), Curtin University: a partnership between Curtin University, University of WA, and CSIRO and the mining companies Alcoa, BHP, and Roy Hill. The centre is part funded by the Australian Research Council and part funded by the industry partners. The over arching theme of the thesis is useful (and novel) examples of Bayesian model building for reliability modelling in industrial settings.
My PhD sat under Theme 2 of the CTMTDS, who's goal is to support the engineer. Therefore, the foundational goal of my PhD Thesis was to assist engineers in maintenance decision making by building statistical models for their industry data. Throughout my PhD I have had to undertake a number of placements within the industry partners to identify and work on statistical problems which are relevant to them and the wider mining industry. The purpose of embedding these placements within the PhD was to direct the final body of work to be more applied and industry relevant. Initially I proposed large sophisticated models for summarising the reliability of entire complex assets. However, I quickly found that there was more fundamental problems that had to be tackled first due to the messiness and noise in observational industry data sets. Many of the assets where maintenance bottlenecks were present, and hence statistical modelling would have the biggest impact, either had small amounts of data with small signal to noise ratios or large datasets with very messy/incomplete data collection. In these cases, the main reason that maintenance decision making was so difficult was the maintainer's uncertainty resulting from them trying to understand a complex processes through flawed data. This conclusion naturally focussed my PhD on filling the gaps necessary to translate robust statistical modelling methods in the reliability literature over to these messy industrial settings so that the maintainer's have a formal framework to quantify the uncertainty in these difficult maintenance decisions. To do this, I used the Bayesian framework.
This repository can be used to fully reproduce the Curtin Thesis. The Thesis is written using LaTex and a Curtin astronomy template. All of the code to produce the analysis was written in R and is documented using Quarto. The Bayesian models are defined and fit using Stan and the RStan package.
The directory structure of the repository is laid out bellow.
.
├── thesis.tex # The main Tex file for the thesis
├── references.bib # The bibTex references
├── chapters/ # The child Tex files called by thesis.tex
│ ├── abstract.tex
│ ├── publications.tex
│ ├── contributors.tex
│ ├── copyright.tex
│ ├── acknowledgement.tex
│ ├── acknowledgement-of-country.tex
│ ├── chapter1.tex
│ ├── chapter2.tex
│ ├── chapter3.tex
│ ├── chapter4.tex
│ ├── chapter5.tex
│ ├── chapter6.tex
│ ├── chapter7.tex
│ └── appendixA.tex
├── r-code/ # All of the R code to reproduce figures for the main body chapters
│ ├── ch-1/
│ ├── ch-2/
│ ├── ch-3/
│ ├── ch-4/
│ ├── ch-5/
│ ├── ch-6/
│ ├── DSTM-Logo-RGB.png # CTMTDS logo for Quarto docs
│ └── styles.css # Style file for the Quarto docs
├── data/ # The datasets used in the thesis
├── figures/ # The figures generated by the R code, organised by chapters
├── tables/ # The Tex tables generated by the R code, organised by chapters
├── outputs/ # The intermediate outputs of the code
└── ... # Aux git files Curtin Thesis LaTeX template
To generate any of the .html files to reproduce the analysis using quarto simply install quarto and run
quarto render file_name.qmd
-
R. Leadbetter, A. Phatak, A. Polpo and M. Hodkiewicz, Informative Bayesian Survival Analysis to Handle Heavy Censoring in Lifetime Data, 2021 International Conference on Maintenance and Intelligent Asset Management (ICMIAM), Ballarat, Australia, 2021, pp. 1-6, doi: 10.1109/ICMIAM54662.2021.9715184.
-
Leadbetter, R., G. Gonzalez Caceres, & A. Phatak, Bayesian hierarchical modelling of noisy gamma processes: Model formulation, identifiability, model fitting, and extensions to unit-to-unit variability, 2024.
-
Leadbetter, R. & A. Phatak, Functional degradation modelling of the wearing surface of conveyor belting using Bayesian hierarchical modelling and Gamma processes, 2024.