Research on Underwater Target Attitude Estimation Model Based on Multilayer Neural Networks

In marine operations and equipment deployment, the attitude measurement of underwater targets is a critical factor in ensuring precision and stability. However, traditional methods are constrained by limitations in real-time performance and adaptability in complex ocean environments, failing to meet the demands for high accuracy. To address these challenges, this study proposes a deep learning model based on multilayer neural networks for classifying and predicting the angles of underwater target attitudes. MATLAB simulations were used to generate sonar echo data for rigid ellipsoid targets, extracting frequency response features under varying aspect ratios and attitude angles to construct a high-quality training dataset. The model architecture comprises an input layer, multiple hidden layers, and classification and regression output layers, enabling simultaneous classification of aspect ratios and prediction of attitude angles. During training, the model's performance was optimized by adjusting hyperparameters such as the number of layers, neurons per layer, activation functions, and learning rates, significantly enhancing prediction accuracy and efficiency. Experimental results demonstrate that the model achieves 100% accuracy in classification tasks and minimal error in regression tasks, reflecting excellent generalization and robustness. This research provides an efficient and reliable technical solution for underwater target monitoring and attitude estimation in complex marine environments, laying a solid foundation for further studies in related fields.