Predictive maintenance expert

The idea behind the role is that failures announce themselves. A bearing, a motor or a gearbox changes how it vibrates, how warm it runs and how much current it draws before it breaks, and those changes are measurable long before anything is audible. The expert works from predictive maintenance as a discipline, which is the body of knowledge about which signals precede which failures.

Finding the signal is a data problem. The expert performs data analysis, analyses large datasets, applies statistical analysis techniques, and works from statistics, mathematics and computer programming. Sensor data arrives continuously, in quantity, and mostly describes equipment that is perfectly fine, so the task is detecting a small change against a large and noisy background.

Understanding the machine is what keeps the analysis honest. The expert works from electrical engineering. Reporting the need for maintenance is the output, and a false alarm costs a shutdown that was not necessary.