Explainable Maintenance Support
In this project, I offered consultancy services to the Lumada Data Science Lab of Hitachi in Japan. The project focused on the exploration of Explainable AI solutions for maintenance support, drawing from actual IoT data from Hitachi manufacturing sites.
Maintenance operations in large manufacturing machines are complex and require expert knowledge and experience. The goal of the project was to explore novel AI solutions that incorporate IoT data, knowledge and domain expertise. For the solutions to be effective, they have to go beyond the black-box inference and diagnosis provided by traditional Machine Learning. The models have to provide explainable answers, with plausible and persuasive explanations of potential root causes, such that can be validated by technicians.
The work involved a large variety of activities and techniques, including but not limited to:
- Requirement elicitation and analysis
- Review of technical documentation
- Stakeholder analysis
- Knowledge capture interviews
- Literature review
- Data cleaning and exploratory analysis
- Process modelling and mining
- Associative rule mining
- Causal model development
- Proof of concept development
- Providing workflow technology training and support
- Documentation, reporting, workshops