(164c) Predictive Maintenance with Machine Learning

In the process industries, many companies are trying to reduce unexpected failures in their equipment by developing predictive maintenance algorithms. Some of the difficulties associated with developing these algorithms include the lack of failure data from the machine, as well as being unable to incorporate their engineering and modeling expertise into the final solution.

In this session, MathWorks will show you how to use physical models of your system to generate failure data by simulating fault conditions, and then using this data to train a predictive model in MATLAB. We also show you how to integrate these algorithms with your embedded devices so that your algorithms operate directly on real data coming from your machine. Specifically, we will concentrate on a triplex reciprocating pump – an example that was requested by several industry practitioners.