Segmentation of CMAPSS trajectories into states

For time-series classification, clustering and sequence prediction

I have been using CMAPSS datasets (initially developed for prognostics, forecasting and prediction, available here: http://ti.arc.nasa.gov/tech/dash/pcoe/prognostic-data-repository/) for a while and remarked that in some papers, it is used for classification or clustering. I put here a simple code to generate "states/clusters/classes" which can be used by other researchers for comparison purposes (if no parameters are changed in the file). It could work for other datasets even though i did try it a lot. The condition is that the signal should be "rather" monotonic (there is filtering in the code) and in [0,1].