Rough Set And Neural Networks-based Study Of The Competency For Seafarers

Aiming at the weakness of current study of the competency for seafarers, the statistic analysis and rough set, as well as Neural Network are introduced to study the competency for seafarers. Seafarer' information system based study shows that the current number of certificate holders is 186795, and the numbers of J, Y, B, Dcertificate holders are 95487, 5340,44862 and 41070 separately. However it shares about 4% of international labour market. The seafarers'age distribution and education degree have been improved greatly. The results of rough set based data mine on 50 well-selected maritime investigation reports in Liaoning Maritime Safety Administration indicate that the maritime education extent is the most important to seafarer'competency, whereas the type of certificate has least importance. The confidential factors of 8 assessment indexes to general and lower marine accidents are between 0.64 and 0.66. The general rough membership degree of maritime accidents is an inverse ratio to seafarer's age, and the membership degree of maritime accidents is also an inverse ratio to seafarer's sea experience. Highly educated seafarers have lower membership degrees, however ill-educated seafarers have higher ones. Type J certificate holders have higher membership degrees and type Y certificate holders have lower ones and the master has highest degree. The watch-keeping time periods in 0000～1200 have highest degrees, whereas in 1200～2400, the membership degrees are relatively low. In fatigue aspect, the membership degree of those seafarers with watch keeping duration less than 2 hours is a bit lower than those with more than 2 hours. The questionnaires from 158 management level officers confirm that the research results are reliable. Two reduction sets are introduced to train 2 BP Neural Networks separately. Based on those two well trained BP Neural Networks, the study on the relationships between 8 assessment indexes and the marine accidents is conducted. The results reveal that, basically, the trend of probability is consistent with the change of membership degree. The recommendations on how to establish the human elements related data collection mechanism are proposed:The awareness of seafarers'sustainable development should be improved, and the unified collection standards and procedures should be established. It is also important to train qualified officers to take the collection and analysis responsibilities.