Elucidating families of ship designs using clustering algorithms

Abstract

This paper proposes a method to elucidate families of ship designs generated by the TU Delft packing approach using data clustering algorithms. The authors explore whether commonly used data science techniques can extract new information from the existing data. To test this hypothesis this paper applies data clustering algorithms to a test case of layouts of a Mine Counter Measures Vessel (MCMV) generated by the packing approach. Results look to improve the understanding of the multidimensional structure of the data, as well as to improve the comprehension and visualization of the complex interactions between the design and performance space.