Virtual Sensors - Time Series Prediction - Time to failure

[skef]

Hello,

is there any tutorial available with an example dataset, which shows how to implement an TTF by using the concept of virtual sensors? I tried to use TWX Analytics with this data set (https://c3.nasa.gov/dashlink/resources/139/) in order to predict the TTF/RUL of the engines, but I never could reach the correct RUL which are also provided in this dataset. The scored results of TWX Analytics where always wrong (with and without the virtual sensor concept).

I am also aware of this document (https://www.ptc.com/es/support/article?n=CS271176) and I used the described approach there to predict the RUL. 

Any help is appreciated.

[nsampat]

Hello,

Thank you for posting your question.

At this time we do not have a TTF sample dataset, and we refer users to that article you have linked as a starting point.

Please let me know if you have any additional questions.

Regards,

Neel

[skef]

Thank you for your reply.

Is the approach described in the linked article the right one, to fit with the needs in a real life scenario?

Why could TWX Analytics not predict the TTF/RUL of the simulate data of the training dataset? Which parameter can I change in order to improve the results for this training set?

I modified the training data this way:

I added a Boolean flag which indicates an error (1 or true for every last row of an engine id) and added the TTF according to the document above.