*Article* **A Classifying-Inversion Method of Offshore Atmospheric Duct Parameters Using AIS Data Based on Artificial Intelligence**

**Jie Han 1,2, Jiaji Wu 1,\*, Lijun Zhang 2, Hongguang Wang 2, Qinglin Zhu 2, Chao Zhang 2, Hui Zhao 2 and Shoubao Zhang 2**


**Abstract:** Atmospheric duct parameters inversion is an important aspect of microwave-band radar and communication system performance evaluation. AIS (Automatic Identification System) is one of the signal sources used for atmospheric duct parameters inversion. Before the inversion of atmospheric duct parameters, determining the type of atmospheric duct plays an important role in the inversion results, but the current inversion methods ignore this point. We outlined a classifying-inversion method of atmospheric duct parameters using AIS signals combined with artificial intelligence. The method consists of an atmospheric duct classification model and a parameter inversion model. The classification model judges the type of atmospheric duct, and the inversion model inverts the atmospheric duct parameters according to the type of atmospheric duct. Our findings demonstrated that the accuracy of the atmospheric duct classification model based on deep neural network (DNN) even exceeds 97%, and the atmospheric duct parameters inversion model has better inversion accuracy than that of the traditional method, thereby illustrating the effectiveness and accuracy of this novel method.

**Keywords:** classifying-inversion method; AIS; atmospheric duct; artificial intelligence
