Optimizing Matching Area for Underwater Gravity-Aided Inertial Navigation Based on the Convolution Slop Parameter-Support Vector Machine Combined Method

This paper focuses on the selection of matching areas in gravity-aided inertial navigation system. Firstly, Sobel operator was used convolution gravity anomaly map to obtain feature map. The slope parameters were constructed by combining and characteristic parameters, such as difference between rows columns, variance map, pooling difference, range combined. Based support vector machine algorithm, parameter-support combined method is proposed. Second, we selected appropriate training sample set verify. results show that compared with pre-calibration results, classification accuracy test more than 92%, which proves can effectively distinguish suitable unsuitable area. Thirdly, applied this another region. experiment performed split-matching average positioning error better 100 m, correct rate 90%. sailing area accurately match trajectory reduce error.