Abstract: In order to construct the indicator system for inland unmanned ships navigation safety scientifically, the characteristic indicators reflecting the safety status of inland unmanned ships were collected according to the principle of selection of indicators. Then the entropy theory and the improved correlation coefficient method were used to screen indicators based on the importance and global correlation, and to eliminate indicators with less effective information and high information overlap. Finally, the concept of overall identification of the index system was proposed based on the entropy theory to verify the rationality of the established index system. The results show that the above research idea can comprehensively consider the various factors influencing inland unmanned ship navigation safety and improve efficiency of indicator screening, and that the value of indicator system overall identification is greater than 0.8, which indicates that the established indicator system is good.

Key words: inland unmanned ships, navigation safety status, overall identification, improved correlation coefficient method, entropy theory, indicator screening