AUTOMATIC ELIMINATION OF SHIP DESIGN PARAMETERS BASED ON DATA ANALYSIS FOR SEAKEEPING PERFORMANCE


ŞAYLI A. , ALKAN A. D. , UYSAL A. O.

BRODOGRADNJA, cilt.65, ss.15-33, 2014 (SCI İndekslerine Giren Dergi) identifier identifier

  • Cilt numarası: 65 Konu: 4
  • Basım Tarihi: 2014
  • Dergi Adı: BRODOGRADNJA
  • Sayfa Sayıları: ss.15-33

Özet

In this paper, we are proposing a computer-based system which makes the automatic elimination of ship design parameters based on data analysis for seakeeping performance. Usually engineers do not have enough time to analyse the data. In this case it can be better to use less parameters in the data analysis. But if the investment has the high commercial worth then the engineers must consider and analyse all variables and their effects in the concept design to minimise the risks of further stages of the design. We are mainly focused on ship motions to identify their most influential parameters. By the use of statistics, the backward elimination method is constructed in a software based on the SQL Server Database. The system contains two modules named as "Identification" and "Elimination". Identification module is used to find out the weakest parameters by the method and then the elimination module avoids these parameters from the final model. In fact the most engineering areas concern with the problem of different parameters and physical issues to construct meta-models to calculate the closest prediction to real values.