Classification Prediction Study of Underwater Vehicle Fitness Zone Based on Random Forest Algorithm

In this paper, under the background of China's “Ocean Power” strategy, focusing on improving the navigation and positioning accuracy of underwater vehicles, a set of prediction models is established by processing gravity anomaly data, and the prediction technique of fitness zone classification is optimized. Firstly, the Kriging method is applied to improve the fineness of the gravity anomaly reference map, and the K-Means algorithm is used to realize the area division, and the polynomial regression model is used for the fitness calibration, which verifies its accuracy and reliability; secondly, the random forest algorithm is used to train the classification prediction model by combining the correlation analysis and feature importance assessment, which shows high accuracy and good generalization ability; finally, the model is validated. Finally, the applicability and prediction ability of the model on new datasets are verified by migration learning. Finally, the applicability and prediction ability of the model on new data sets are verified by migration learning. The research results in this paper are highly efficient and practical in the classification and prediction of marine gravity anomaly fitness zones, which is of great significance to the improvement of the underwater navigation accuracy and the marine environment management.