This newsletter focuses on the activities of the Centre during May and June; a busy time for preparing and participating in the 26th World Mining Congress in Brisbane
TEAM NEWS #
BHP MECoE and CTMDS researchers from Perth that travelled to the World Mining Congress took the opportunity to catch up in person. This catch up was a great opportunity to get to know the MECoE team based in Brisbane to and build relationships to further research collaborations.
EVENTS #
World Mining Congress #
The World Mining Congress (WMC) is the leading international forum for the global mining and resource industries. Starting in 1958, the World Mining Congress has served as the pre-eminent gathering of the worldwide mining and resources industry. In June this year, CSIRO hosted the first Australian WMC in Brisbane. Curtin University was a Platinum sponsor, and the Centre exhibited our research through participation in the pre-conference activities, presentations and exhibition. CI Prof Melinda Hodkiewicz was the chair and organiser of the Artificial Intelligence Stream for the Congress
The congress allowed our researchers to listen to world leaders across the full spectrum of mining and share their research with people from a broad range of the mining sector, including international investors, operators, regulators, and suppliers.
Pre Congress activities #
At the Work Mining congress in Brisbane the Training Centre for Transforming Maintenance through Data Science hosted two workshops as part of the Pre-Symposia program.
Workshop 1 - Technical Language Processing and Knowledge Graphs Demystified
This full day workshop provided an overview of state of the art in technical language processing. It describes the motivation and explains key concepts relating to natural language processing of technical texts, knowledge graphs and ontologies. While the examples used will relate to asset maintenance data, the concepts are applicable to safety records, geology and other technical texts.
Theme Leader A Prof Wei Liu, CI Professor Melinda Hodkiewicz , and PDF Dr Sirui Li and PDF Michael Stewart facilitated a workshop on Technical Language Processing.
Workshop 2 - The Data Fit Organisation: A new approach to building data capability and getting value from data
This half-day training, provided by CORE Skills in collaboration with the Future of Work Institute, detailed an approach called 'lean for data' for delivering value through data. The workshop outlined a methodology for mapping data capability in an organisation. Participants mapped industrial data workflow for their organisation; they identified requisite roles and capabilities and worked through how to overcome the challenges of enabling and embedding data workflows in their business. PDF Dr Eden Li and Partner Zane Prickett (CORE Innovation) ran a half day pre-conference workshop on the Data Fit Organisation.
Congress activities #
CI Prof Melinda Hodkiewicz was the chair and organiser of the Artificial Intelligence(AI) Stream for the World Mining Congress. As part of the AI Stream, Prof Andrew Rohl presented Transforming Maintenance Through Data Science ; in this presentation, Andrew focussed on how the Centres research solutions are driven by maintenance problems from the mining industry and the barriers to implementing our solutions in the industry, both from a technical and sociological perspective. PhD Ryan Leadbetter presented how to get the most out of condition monitoring data in his presentation focused on improving conveyor belt wear forecasts to make better maintenance decisions.
PhD Gabriel Gonzalez gave a dynamic poster presentation on his research focused on a practical approach to risk-based inspection interval estimations.
Prof Andrew Rohl and Ryan Leadbetter participated in a roundtable panel discussion chaired by Karin Breitman.
The Centre exhibited our research along side the WA school of Mines and the Trail Blazer. PhD students Yingying Yang, Srimali Gunasekara, Ponpot Jartnillaphand, Gabriel Gonzalez, Ryan Leadbetter and Research Fellows Shuixiu Lu, Sirui Li and Eden Li supported this exhibit discussing their research and the broader research efforts involved in their research themes.
The conference was an excellent forum for the Centre's research, where it could be shared industry personnel interested in how we are using data science to improve mining and maintenance.
PhD Ryan Leadbetter, PDF Sirui Li and PDF Shuixiu Lu network with international visitors interested in their research
PhD candidate Srimali Gunasekara joins exhibitors from WA School of Mines and the Curtin TrailBlazer team.
PhD candidate Yingying Yang takes the time to network with other exhibitors
Workshop on standard for Coding and Unit Testing Guidelines #
CTMTDS PhD students participated in a workshops facilitated by Professor Andrew Rohl, John Hille and PDF Michael Stewart
Researchers taken through the importance of unit reminding them that there are a lot of different types of software testing; unit testing is the lowest level and the one most closely associated with developing code.
It is important for the following reasons
- Quality Assurance. For the researcher and and anyone that may use their code
- Confidence that code changes will not have unexpected side effects
- Acts as additional documentation on how the code is expected to work.
- Bugs that are found quickly are also much quicker and easier to fix
Participants in the workshop went through the common testing frameworks and how to write a unit test. The workshop included code coverage and the importance of well designed and robust tests that exercise the "critical" parts of code and validate their results.
The Centre has a draft set of guidelines that include the following requirements
For code that is part of a translation project (i.e. may be deployed in industry), the requirement is 70% test coverage.
- For all other code, researchers should aim for 50% coverage.
- The Centre recommends that our tests are written using pytest, with coverage reports generated using pytest-cov.
- Researchers should aim to write tests that give them a strong level of confidence that their code works as expected, and not just for the purpose of reaching a certain level of coverage.
PRESENTATIONS #
Researcher Catch-up Series for May and June 2023 #
May presentations were hosted by Dr Eden Li
Dr Sirui Li presented on Question Answering over Temporal Knowledge Graphs #
The use of knowledge graphs for natural language processing tasks, including question answering, has been gaining traction in recent years. However, existing approaches often need help to capture temporal information, which is critical in industry data as it is naturally time variant. Industry data presents a unique opportunity to explore temporal knowledge graphs, where the head entity represents an event, and the tail entity is a time-variant state. For example, Engine X (head entity) was observed leaking oil (tail) with a start-time (temporal point) and an optional end-time (temporal point).
Sirui introduced existing question-answering systems over temporal knowledge graphs, and she will present an approach under development. Her proposed approach enables accurate retrieval of time-sensitive information from large-scale knowledge repositories. She will leverage a neural network architecture that integrates temporal information with question-and-answer embeddings, allowing the system to reason over the temporal structure of the graph to generate accurate answers.
Sirui's proposed approach will offer a promising solution for enhancing knowledge management and decision-making in industries that rely on historical data.
Dr Eden Li presented on Unlocking the Potential of Data Science: Developing an Assessment Tool for Organisational Capability #
Eden asked her audience the following questions
- Have you ever considered how your organisation can fully unlock the potential of data science and stay ahead of the competition?
- Have you ever questioned the effectiveness of cross-team and multi-disciplinary collaborations in sharing and utilising data in your organisation?
- Have you ever reflected on the skills your employees will need to work alongside these new data science solutions?
Translation Theme 4 aims to support and organisation to ensure that they have the capability and workplace practices to fully utilise Data Science innovations . Eden Li is developing an assessment tool that will inform the capabilities of an organisation to recognise, adapt, and implement data science innovations.
Eden's presentation covered the assessment tool's development, its various capability components, and how implementing it can provide valuable insights and benefits for your organisation.
June presentations were hosted by Dr Michael Stewart
Ryan Leadbetter presented on how to get the most out of condition monitoring data: Improving conveyor belt wear forecasts to make better maintenance decisions #
A reliable conveying system is critical to the productivity of an iron ore mine. If a conveyor in the system fails unexpectedly during a planned period of operation, production in that part of the mine stops. Therefore, engineers collect condition monitoring data to decide when the component should be preventatively replaced before its degraded state causes unplanned downtime and loss of production.
One critical component is the conveyor belt. Reliability engineers take ultrasonic thickness (UT) measurements at multiple locations across the width of the belt surface to monitor the wear of the protective rubber topcoat and manage the risk of the structural part of the belt being damaged by the ore. Engineers try to predict when the wear will exceed an unacceptable threshold, beyond which the protective topcoat no longer provides sufficient protection for the structural part of the belt to plan when the belt should be replaced.
Ryan presented a new method for modelling belt wear. He demonstrated how to forecast the evolution of the wear profile through time to get a predictive distribution of the wear at a particular time horizon, illustrating how predictive distributions can be used to derive remaining useful life and inform maintenance decisions.
Srimali Gunasekara delivered her first presentation for the Centre, she presented on Risk-Based Maintenance Scheduling Optimisation #
Srimali will study the problem of scheduling preventive maintenance activities in maintenance plans, which requires the plant for each maintenance plan to be shut down to carry out the maintenance activities. She will consider two consecutive maintenance plans separated by a long interval.
Srimali presented a case in which the maintenance of components can be postponed from the first maintenance plan to the second. Such a postponement may be required due to constraints on budget or resource availability. She will focus on the risk of failure of components that comes with the postponement decision and formulate the problem as a two-stage stochastic programming problem.
Srimali presented how to minimise the total cost of scheduled maintenance activities in both maintenance plans and the expected cost of corrective maintenance. She plans to extend the existing deterministic solution methodologies for maintenance scheduling to stochastic programming problem.
PUBLICATIONS #
- Hodkiewicz, M., Bikaun, T., & Stewart, M. (2023). RelOps – A Whole-of-Organisation Approach for Reliability Analytics. In A. Crespo Márquez, J. F. Gómez Fernández, V. González-Prida Díaz, & J. Amadi-Echendu, 16th WCEAM Proceedings Cham.
- Lu, S., & Oberst, S. (2023). Recurrence-based reconstruction of dynamic pricing attractors. Nonlinear Dynamics, 111(16), 15263-15278. https://doi.org/10.1007/s11071-023-08629-x
- Nguyen, C., French, T., Liu, W., & Stewart, M. (2023 . CylE: Cylinder Embeddings for Multi-hop Reasoning over Knowledge Graphs. Proceedings of the 17th Conference of the European Chapter of the Association for Computational Linguistics,
- Pesch, T., Polpo, A., Cripps, E., & Cramer, E. (2023). Reliability inference with extended sequential order statistics. Applied Stochastic Models in Business and Industry, n/a(n/a). https://doi.org/https://doi.org/10.1002/asmb.2764
- Spiers, S., Bui, H. T., & Loxton, R. (2023). An exact cutting plane method for the Euclidean max-sum diversity problem. European Journal of Operational Research. https://doi.org/https://doi.org/10.1016/j.ejor.2023.05.014
Stay tuned for our next issue in August where we will cover:
- WADSIH Data and AI conference for Business
- Research presentations by Braden Thorne, Melinda Hodkiewicz, Shuixiu Lu and Tim Pesch
- New publications for 2023
- Research updates