​Risk-based Inspection Intervals: A Practical Approach

The Challenge

In recent years Failure Assessment, Remaining Useful Life Predictions and Predictive Maintenance have been fruitful areas of research and development for asset intensive industries. Expanded capabilities of cloud computing and the Internet of Things technologies contribute to increase the amount of data available as well as supporting the introduction of Machine Learning and Deep Learning Techniques with proven improvements in prediction accuracy, failure diagnosis and pattern recognition.

Research has been carried in a broad range of applications and industries with different levels of results. Relevant studies have been conducted in two ways. The first one has focused on organising a complete and structured framework of work under the name of Prognostics Health Management The second approach explores more complex techniques of modelling and data analytics, with the purpose of incorporating new contextualising variables such as image/video recognition and state-of-the-art statistical techniques.

Notwithstanding the efforts being made, many complexities remain unsolved that are needed to achieve reliable and automated maintenance decision tools. In particular, aspects related to integration and developing new techniques and technologies that assist the field and managerial engineers to achieve their maintenance goals must be improved.

Proposed Solution

The aim of this research is to contribute to the development of advanced and improved tools for failure assessment (Remaining Useful Life and Failure Diagnosis) with a strong focus in increasing the adoption of these technologies in Industrial Environments.

Objectives

1. To study the impact of multimodal dataset (sensors, quality reports, maintenance records) on prognostics algorithms performance.
2. To assess the adoption of predictive algorithms and proposing remedial methods for improving real industrial applications.
3. To elaborate suitable mechanism for outcomes deployment in constantly-changing environments.

Proposed Benefits to WA

Gabriel’s research will support leaner operating strategies that could reduce maintenance costs in Western Australian mineral processing facilities by up to 30%. By improving the ability of engineers to assess the health of critical engineering infrastructure, his work will reduce unnecessary inspections and plant downtime while maintaining safety and performance at the high levels required in the mining industry.

Project Duration

3.5 years, commenced December 2019.