FBLS-Based Fusion Method for Unmanned Surface Vessel Positioning Considering Denoising Algorithm

Although a USV navigation system is an important application of unmanned systems, combining Inertial Navigation System (INS) with Global Positioning (GPS) can provide reliable and continuous solutions positioning based on its several advantages; the random error characteristics INS instability derived from GPS signal blockage represent potential threat to INS/GPS integration USV. Under this background, composition framework nonlinear generalization capability support vector machines (SVM) multi-resolution ability wavelet transform used solve difficulty that suffers interference stochastic errors, dynamic information not influenced. An innovative fuzzy broad learning structure (BL) method utilized in USV, which are deemed as input Fuzzy Broad Learning (FBLS) train network, then trained network FBLS applied for estimating optimal solution during blockage. Based platform, sea trial was carried out confirm validity feasibility proposed by comparing existing algorithms integration. The experimental results show approach could achieve better denoising effect errors high-accuracy