In engineering there are two main areas to use machine learning currently:
Image recognition and predictive maintenance. For each area one example application is described below.
1) Image recognition for sorting machines
A sorting machine must identify different objects on an assembly line and direct or put objects of different classes in different areas. Cameras can film the approaching objects continuously and machine learning algorithms can analyze these pictures and classify different object classes. Such algorithms, which recognize different object classes on pictures, are based on deep neural networks and can be programmed today with relatively little work.
The performance of state of the art image recognition algorithms is as good as the performance of humans,
often it is even better.
There are countless possible applications of image recognition in engineering. Essentially, image recognition algorithms can replace the human eye.
2) Predictive maintenance
Predictive maintenance is the automatized continuous surveillance of machines in order to predict when they are about to fail and stop working.
Sensors continuously produce data about the state of a machine like temperature, chemical composition of
air or fumes, electrical data, ...
Using such data of the past together with the information when machines failed in the past a predictive model can be constructed to predict the probability of failure in the near future of a machine on the basis of the sensor data. Having implemented such a predictive maintenance mechanism the machine maintenance operations can be optimized.
With ever more machines being connected via the internet predictive maintenance will become more and more widespread.