Data-driven prediction of underwater navigation fitness zone classification

• 1、Institute of Intelligent Engineering and Mathematics, Liaoning Technical University, Fuxin 123000, China

Abstract：How to accurately identify and classify adaptive zones for complex underwater geological features and underground gravity anomaly distribution in navigation systems has become an urgent problem to be solved. Traditional classification methods have limitations in dealing with the complexity of geological and gravity data, resulting in inaccurate and unreliable prediction results. Developing an effective regional adaptation zone classification prediction model to assist navigation systems in making decisions in different underwater areas has become crucial. Selecting the standard deviation of gravity field slope, gravity field roughness, and gravity anomaly difference entropy as initial feature attribute indicators, Pearson correlation coefficient analysis was conducted. It was found that there is a high correlation between gravity anomaly values, gravity field slope standard deviation, and gravity field roughness. Therefore, these two features were selected as model inputs. A golden sine search strategy, reverse learning, and Cauchy mutation perturbation optimized sparrow search algorithm were proposed, and combined with the XGBoost ensemble learner to construct an XGBoost adaptation area classification prediction model based on the optimized sparrow search algorithm. The optimization process enables the model to better adapt to complex underwater geological features and underground gravity anomaly distribution. The results indicate that the model can effectively classify different underwater areas as adaptive or non adaptive zones. The classification accuracy of the model has significantly improved in practical applications, providing more reliable decision support for navigation systems in complex environments.

Keywords： Pearson correlation coefficient Optimized sparrow search algorithm XGBoost Underwater navigation