Studies Of Data Mining Based On Ocean Circumstance And Ship Performance

With the proposals of the National's 11th five-year plan, there is an extensive concern about ocean and shipbuilding industry. How to utilize information technology in order to make a rapid progress in marine industry has attracted attention to research for both ship engineers and computer scientists.Nowadays, the most common ways of ship performance test and verification is based on the physical method. However, this physical method is incompetent for the whole ship, together with problems such as human errors, inefficiencies and costliness.The rapid development of computer science and technology brings a great opportunity for ship performance test and verification. On one hand, the growth of data warehouse and data mining technology provided abundant theories and algorithms for knowledge discovery in large historical data sets;on the other hand, the ship industry has already accumulated plenty of ship-experiment and trial-voyage data, which contributes the most valuable materials for data mining. The aim of this paper is to build a common marine (ship-experiment and trial-voyage) mining model based on a comprehensive study on the marine data, which remarkably reduce the cost, shorten the period of shipbuilding and improve the quality of ship design.The novelty of this paper firstly lies on the combination of marine historical data and data warehouse technology, which convenient for both high-level analysis and decision making. Second, we present a fast ship comparison algorithm based on data mining. Third, we proposed a heuristic strategy to build a performance estimate model by using expert knowledge. An inspiring experimental result demonstrates that our approach is competitive in both accuracy and time complexity.