Modeling of deep-learning applications for chatter detection in the milling process

Sorry you have no rights to view this entry!

5 Comments

  1. Where is the borderline between medium and high chatter conditions? If chatter occur then cutting process is not stable. What is the reason to introduce one more level of chatter?

    • khairul.jauhari

      Thanks for the response Mike Kushnir, the boundary between medium to high we consider the magnitude of the cutting force of about 700 N. and after this there is a significant increase in the cutting force of about 1000 N, so this is the reason why it is necessary to observe the transition conditions.

  2. khairul.jauhari

    Thanks for the response Mike Kushnir, the boundary between medium to high we consider the magnitude of the cutting force of about 800 N. and after this there is a significant increase in the cutting force of about 1000 N, so this is the reason why it is necessary to observe the transition conditions.

  3. Thank you for the presentation Mr. Jauhari! It is a challenging task to do convergence of CNNs. What is training dataset size to train your model?

    • khairul.jauhari

      Miss Gamze Kecibas, Thank you for your responses and questions. That’s right what you’re saying. Too little data is less than optimal accuracy and too much data will take a lot of time towards converging. The dataset is created as a structure array with two fields: dynamic and static data, and labels. The Data field is a 98-by-13040 matrix where each row is a slot milling cutting process sampled at 10 kHz. Labels is a 98-by-1 cell array of condition status labels, one for each row of Data. The three condition status categories are: Stable Condition, Medium Chatter, and High Chatter. In this case, we pre-processed each signal with a sampling length of 13040 in order to increase the size of dataset and to make it suitable for training CNNs. Therefore, each category will provide 340, 300, 340 signals and total will be 980 signals.

Leave a Reply

Your email address will not be published. Required fields are marked *