I am trying to extract features from 3 time-dependent principal components signals derived from 15 pressure and temperature sensors of a filling plant. The plant is working batchwise and following a recipe, steps of which depending on the process part have either duration or pressure limits .
I already tried frequency analysis, although it shown some results I am still not able to classify the frequency in classes (frequencies of a batch which is about to fail and good frequencies).
Is there somebody out there, who can give me ideas how the define features of a time-dependent signal without frequency analysis?
Hello Giole,
Are you able to provide any further information on some properties of what you refer to as the 'features'? With more details I am sure it will be easier to provide a more appropriate answer.
A colleauge of mine developed a neat centroid-based filter for determining spikes in the time domain in the presence of noise. I will attach the article.
- Badges
- Science topic
- Similar topics
- Methodology
- Sensors