In　order　to　realize　that　unmanned　cargo　ship　can　move　forward　according　to　the　optimal　route　and　reduce　the　risk　in　the　harsh　environment　of　wind　and　wave,　a　D∗　algorithm　for　dynamic　route　planning　based　on　A∗　static　route　planning　is　proposed.　Firstly,　spark　big　data　processing　platform　is　built.　The　big　data　platform　is　used　to　preprocess　data,　extract　effective　features　and　build　weather　forecast　models.　In　the　construction　of　weather　forecast　model,　the　commonly　used　forecasting　models　are　used　to　predict　and　optimize　the　optimal　model　parameters.　Select　the　F1　value　of　each　model　for　comparison,　and　then　select　the　model　with　the　highest　F1　value　in　each　period　as　the　prediction　model　of　the　period　according　to　the　characteristics　of　weather　data.　Finally,　the　results　of　multiple　models　are　summarized　to　form　the　final　weather　forecast　data　set.　Then,　according　to　the　starting　point　and　terminal　point,　combined　with　D∗　dynamic　programming　algorithm,　a　safe　and　shortest　route　is　planned　when　the　threat　changes　constantly.　©　2021　IEEE.