DVL Model Prediction Based on Fuzzy Multi-Output Least Squares Support Vector Machine in SINS/DVL

For an underwater Strapdown Inertial Navigation System/Doppler velocity log (SINS/DVL) integrated navigation system, the short-term failure of DVL may lead to loss reliable external information from DVL, which will cause SINS errors accumulate. To circumvent this problem, paper proposes a predictor based on fuzzy multi-output least squares support vector machine (FMLS-SVM) predict measurements when malfunctions occur. Firstly, single-output (LS-SVM) model is extended LS-SVM (MLS-SVM), and self-adaptive membership introduced fuzzify input samples overcome over-fitting problem caused by excessive sensitivity outlier points. Secondly, function designed idea K nearest neighbor (KNN) algorithm. Finally, considering influence vehicle maneuver prediction dynamic attitude angles are improve adaptability under large conditions. The performance method verified lake experiments. comparison results show that FMLS-SVM can correctly provide estimated measurements, effectively prolong fault tolerance time faults, accuracy reliability SINS/DVL system.