Presentation of our Hackathon case decision „Windfarm
Installation case“

  • Crane
  • Navigational equitpmen
  • => Jacking System

Excuse and excerpt out of our CRISP-DM “Cross-Industry
Standard Process for Data Mining” Methodology

Continuous Variables to describe the currents systems state

  • Jacking Up
    • Leg1 Load
    • Bottom Contact
    • Leg 1 Extension
    • Leg 2 Load
  • Jacking Down

Discretize the Problem to get a “Markov Chain”

Keywords are “Vector Space”, Matrices, Multivariant
Distributions

Can’t observe the states directly => Hidden Markov Model
and Observables are the variables

LEARNING: Maximum Likelihood + Baum-Welch Algorithm

PREDICTION: Viterbi Algorithm

Results: Discretion results as Event Frames (4 States)

Phases: Leg loading and Leg Lifting as an Excel Data export

Using SEEQ as presentation layer for our Hidden Markov Model



osihack2018.jpg

(http://www.seeq.com/)