Physics-based techniques assume that you have a pretty detailed knowledge of the behavior of the equipment. Physics-based models join this knowledge about the equipment with the measured data to measure the performance of a piece of equipment and/or to predict its future behavior.

Compared to data-driven methods, physics-based analytics have several advantages.

• First, physics-based methods require a relatively small dataset
• Second, physics-based methods can make a long-term prediction
• Finally, physics-based models are normally more reliable

However, a disadvantage of this approach is that it requires in-depth knowledge of the problem in order to catch a physical failure.