To solve the problem of dynamic path planning of ships in polar waters, an improved D^* Lite path planning algorithm was proposed combined with the Polar Operational Limit Assessment Risk Indexing System (POLARIS) to adapt to polar navigation. First, the POLARIS risk model was introduced,and by analyzing the grid-based sea ice concentration and sea ice thickness data, a polar navigation risk model was established combined with ship types. The grid environment map with navigation risk index was constructed, and the cost function of the D*Lite algorithm was reconstructed. Secondly, by extending the search neighborhood, the ship's turning Angle was not limited to integer multiples of π/2 or π/4,as a result  improving path smoothness and shortening  path length. Finally,  by introducing the navigable index of grids near non-navigable  areas due to safety factor expansion, the probability of pre-planned and re-planned routes moving away from immobilized obstacles was increased. Simulation experiments show that the improved algorithm of D^* Lite algorithm not only reduces sea ice risk by 15.49 % to 40.3 % compared to traditional algorithms , but also reduces path length by 2.99% to 5.32% and path inflection points by 45.95% to 60.98%.