Classifying a set of signals with Deep Learning

I have several measurements (1D) taken during the functioning of an instrument, such as temperature, pressure etc. They are not all equal in length. I would like to use these measurements to classify the "run" as good, iffy, ugly. Can someone advice on what type of deep learning architecture I should use? Will matlab's functions support this easily? I've been toying with ideas like convert to spectrogram and then tile the different signals and use that for input. Is there a way to directly use the 1D signals? Can anyone point me related blogs? Thanks a ton.