Research On Underwater Source Ranging Algorithm Based On Domain Adaptation

Underwater acoustic source target localization is an important research content in the field of underwater acoustic.It is a technology to infer the distance,azimuth and position of unknown underwater acoustic source target with the help of underwater acoustic equipment according to the acoustic principle.Up to now,the main method of sound source localization is based on geometric theory.The underwater sound source ranging algorithm based on domain adaptive in this paper can effectively estimate the distance between the source and the target,and the distance estimate error is smaller than that of traditional matching field and traditional machine learning.Additionally,the distance estimation model is better.Underwater source passive ranging is based on the pressure radiated by the source in the received data.It is a parameter estimation problem to search for source position parameters in the airspace through the method.Parameter estimation problems are usually converted into classification problems by machine learning methods,which have more accurate estimation capabilities than traditional Matched Field Processing(MFP)and with needless prior sound field information.However,when the probability density of training data and test data follow different distributions or the training data is insufficient,the effect of the classifier under traditional machine learning methods is usually poor.Therefore,the underwater target source ranging algorithm based on Domain Adaptation(DA)is proposed to find an appropriate transformation matrix for data mapping,thereby reducing the distribution differences and realizing the migration between source and target.Or to address the issue of class imbalance,one possible solution is to focus on the confusion matrix to mitigate the impact of poor prediction caused by the limited number of samples in certain classes.For the same task of underwater source ranging,this paper proposes two underwater source ranging algorithms based on Joint Distribution Adaptation(JDA)and Minumum Class Confusion(MCC).The two algorithms can effectively analyze the newly acquired unknown data under the condition of known historical information.Also,they can effectively reduce the differences between the track data obtained in the underwater acoustic field at different times and orientations.The former approach facilitated a more consistent data distribution across different underwater acoustic environments,while the latter approach alleviated the adverse effects of imbalanced sample classes.In this study,we trained three machine learning models to classify and predict domain-adapted data.By comparing the prediction results of our proposed models with those of traditional machine learning methods,we found that our algorithms significantly reduced the prediction error of classifiers trained on the source domain by more than 30% when predicting on the target domain.Furthermore,our proposed algorithms addressed the issue of traditional machine learning methods failing to classify within a local distance interval,which led to more accurate distance estimation of sound sources in different underwater acoustic environments.Additionally,our algorithms improved the approximation of the distribution between different datasets and reduced the degree of dispersion among samples of the same class.This led to more accurate distance estimation of sound sources in different underwater acoustic environments.