I have a time series welding data I wanted to create a model which can predict some weld parameters but extracting those parameters from time series data is being so difficult. Currently I tried taking two pointers and sliding window on the time series data and calculating slope and detecting the state and extracting those features here but that doesn't working properly it relays on slope and it's too complex and it's too difficult to hard code all condition for different wirefed speeds. What is the best and accurate way to achieve this (I'm relatively new to ml and stuff).

Features I'm trying to extract:

  • background time
  • background current
  • slope up time
  • peak current
  • peak time
  • slope down time
  • detachment time

I wanted to calculate these parameters for each of the wave and take the average value and build a model which takes these as features.

Weld Params



You must log in to answer this question.

Browse other questions tagged .