Can someone please help me out and tell me whether or not it is possible to implement an Artificial Neural Network on a S7-1200 PLC.

At the moment the PLC controls the heating cycles of a heat-pump that provides hot water for a household. My objective to complete is to recognize hot water usage patterns by the household and to use those patterns in order to switch on the heat-pump at optimal times.

Other suggestions in order to achieve my goal will also be greatly appreciated. 

Thanks

You do not need a "Artificial Neural Network" if you are not a borg.

What you may do:

An Archive for any day (MO...SU) that store the approx. value of water needed.

With resolution on hour - base you can do an update every day ( Dark&Dirty: Value := Value + newValue / n )

After a while, you have a value for approximate need of hot water.

Hello Brendan.

The PLC is an empty canvas. You can do what you want within the limitation of the resources and your own knowledge.

Yes it is possible to build a quasi neural network. Not very easy - but it is possible.

You will have to dig deep to what the definition is. Mr. Kirk Zurell made some usable commentary here:
https://www.quora.com/What-are-the-differences-between-fuzzy-logic-and-artificial-intelligence

Focus on:

Kirk Zurell

Contrast this with, say, a neural net that “learns” by adjusting its own weight values.

 In your case you have a fuzzy logic or rule based control system as basis. IBN-Service has already spoken about such idea. The variables that control the system (example: when to start heating) is "adjustable".

As designer you choose some start values for those control variables. For example - keep the water warm and cycling all the time (except dead night hours). Or you have the customer tell you what the starting point should be.

The system then builds up a data base of actual usage by the customer. Weekdays and weekends may differ, but high possibility is that you will be able to see a trend. This trend is what you use to make adjustment to the control variables.

The system is then self adaptive to the average usage of the customer. The system ensures that the water is kept warm at the times of requirement. The water will not be hot at all times of the day.

How you build the database and convert the data to adjust the control variables, is up to you. It is possible.

Regards.
- Dan Padric