Development of high-dimensional robust split regression models to measure data quality for improved use of sensors and diagnostics in a heavy industrial setting

We propose the development of novel statistical methods to improve upon detection of normal and abnormal behaviours in industrial sensors. The ability to detect subtle changes in sensor performance in a wide range of operational conditions has significant potential for MineSense Technologies Ltd., which operates sensors in challenging conditions. It can be very difficult to perform reactive sensor system maintenance in an active mine, as it may require shutdown or suspension of critical operations. Therefore, the ability to detect abnormal behaviours and thus perform preventative maintenance is valuable. The main objectives of this project are to (1) characterize and monitor system data quality, (2) develop high-dimensional, robust split regression models for sensing system telemetry data, and (3) develop approaches to integrate these models into broader system health monitoring. TO BE CONT’D

Faculty Supervisor:

Ruben Zamar

Student:

Partner:

MineSense Technologies Inc.

Discipline:

Engineering

Sector:

Mining; Professional, scientific and technical services

University:

The University of British Columbia

Program: