PhD Data-Driven Predictive Maintenance of Complex Engineering Systems

Applicants should have completed (or be close to completion of) a Master degree in mathematics, operations management, operations research, econometrics, industrial engineering, or a closely related discipline, with a solid background in mathematical methods. Fluency in English is required.

The project

The high-tech industry is faced every day with the challenge of keeping their systems operational and maximize their availability whilst minimizing maintenance and operational costs. Predictive maintenance enables system downtime and costs to be minimized by acting before failures occur and grouping interventions to share set-up costs and possession time. By developing data-driven decision tools capable to extrapolate knowledge from different data sources and enable more reliable maintenance decisions based on data, with this project we aim at advancing knowledge in data-driven maintenance decision making.

In this project we aim to develop a smart maintenance decision framework for complex multi-component systems, specifically complex machines consisting of many heterogeneous maintainable units (components), which operate in an uncertain environment. Degradation, failure and repair are stochastic processes affected by uncertainty around operating conditions including environmental and usage factors. We envision the framework to be smart and deal with uncertainty by combining learning and updating methods with decision models based on robust optimization to support predictive maintenance planning driven by data from alarms, sensors and process logs. Such a decision framework shall enable robust maintenance policies to be developed so as to mitigate the effects of uncertainty which characterize real-life operation of high-tech systems.

We expect the Ph.D. student to:

• develop the project proposal based on the most up to date relevant academic literature;
• combine data-driven approaches with robust optimization into mathematical models to support maintenance decision making under uncertainty;
• design algorithms to solve the built models;
• present the findings at conferences and publish papers in internationally renowned journals;
• communicate the results at events of EAISI.