Building a time series predictive model using machine learning or deep learning for, intermittently sampled, vehicle diagnostic data

I’m relatively new to machine learning and deep learning subjects but have experience in matlab programming. I’m looking for a fundamental method of predicting the occurance of diagnostic data from a complex vehicle system such as a plane or train. Diagnostic data is recorded when an event with a particular code is triggered; that code then retrieves a set of environment variables such as speed, pressure, temperature etc. The data is sampled only when an event it triggered so the sampling, more often then not, does not have a constant frequency. I have data that tells me that the accumulation of an event and its environment variable leads to a maintenance action. Any pointers towards building a predictive model would be nice. Thanks Faz.

Answer 1

[Dheeraj Singh]

You can use the Diagnostic Feature Designer for extracting features of your data .

Refer to the following video for getting started.:

https://in.mathworks.com/videos/predictive-maintenance-part-4-how-to-use-diagnostic-feature-designer-for-feature-extraction-1554458327719.html

It is part of the Predictive maintenance toolbox . Refer to the following videos for getting started with the toolbox:

https://in.mathworks.com/videos/series/predictive-maintenance-tech-talk-series.html