/RMG_NDE_Disssertation

Code to support non-destructive evaluation of RMG cranes PhD dissertation

Primary LanguageJupyter Notebook

RMG_NDE_Disssertation

Code to support non-destructive evaluation of RMG cranes PhD dissertation

Overal Goals for Crane Data:

  1. Show comparison of Beta Wavelet with other standard for catagorizing fualts
  2. Compare standard Wavelet, with Fingerprint, with video, with image, with expert system, with straight signal

To do list for Beta Wavelet:

  1. make it work
  2. make real time analysis
  3. compare against all options with real time analysis

To Do list for Crane Data Analysis:

  1. Choose Random set from Repository
  2. Parse Name to get accepted output vector
  3. Filter out non-move data set
  4. Train Models on data
    • Data Prep:
      • Try un smoothed
      • Try smoothed with rolling average
      • Make fingerprints by dimmension
      • Make fingerprints by r
      • try smooth with kalman
      • set up wavelet low-pass filter
    • Model TypesVideo and Image
      • Video:
        • Rolling, 3x100, 3x changing frame sizes
        • overlapping segments, 3x?? hoping down line
        • rolling fingerprints
          • Try many fingerprints
          • look at Beta, and classification as it is going
          • Features from live fingerprints
      • Image:
        • Fingerprint
        • 3 x 60k image
  5. Examine differences of results

Rail LASER

  1. Start data
  2. Figure frequency needed
  3. Specific test cases
    • 'Singing track'
    • Over center anchor rise
    • Old, new, ground, rough, tamped, shakey

EFIT and related

  1. Write code
  2. Get code working
  3. GPU parrallelization
  4. Inner simulation space bundary conditions
  5. Compare LASER measurement to EFIT