Research on knowledge-extraction technology in optimisation of ship- resistance performance

摘要

At present, most hull-form optimisation methods adopt some optimisation algorithms to search the design space to obtain an optimal solution. These optimisation methods ignore the implicit relationship between optimisation variables and objective functions, which often leads to low efficiency and difficulty in obtaining a global optimal solution. If we can learn in advance the relevant design knowledge of a design space, we can effectively determine the ship-form optimisation and save time. This paper proposes a knowledge-extraction method based on two data-mining techniques-partial correlation analysis and self-organising map (SOM). By considering numerical functions and DTMB5415 as the research objectives, knowledge extraction is performed by combining the partial correlation analysis and SOM based on optimisation data. Through a comparative analysis of results, the effectiveness of the data-mining technology is verified, and the obtained knowledge plays an important role in directing the ship-form optimisation.