Classification Prediction of Underwater Navigation Fitness Zones Based on BP Neural Network Modeling

In order to realize high-precision underwater navigation, it is necessary to establish a classification prediction model of the fitness area to realize the fitness assessment of different sea areas. The data are finely gridded, and the variance of the gravity anomalies at the four vertices of the gridded area is calculated as the classification feature to measure the significant degree of change of the gravity anomalies in the area, and the C-mean cluster analysis is applied to cluster the classification feature values, and all the areas are calibrated into seven fitness classes. And the visualization was carried out. Applying the BP neural network model, the latitude, latitude and corresponding gravity anomalies of the data were used as inputs, and the calibrated grades were used as the results, and the final fitting effect of the trained model was basically in line with the expected values. And the model is applied to the prediction of fitness zone classification in other sea areas, comparing the predicted value with the real value, through the visualization analysis and the error analysis of the model, the predicted fitness calibration result has a smaller error with the actual value, and the fitness grade distribution map of the predicted value and the actual value has an obvious consistency, so that the established BP neural network model has applicability for the classification of fitness zones for underwater navigation in different regions.