基于熵权重的水下载体导航信息融合方法

针对水下载体动态导航定位中状态方程和观测方程噪声增加引起的卡尔曼滤波发散问题，提出了一种以高斯混合模型为框架，基于信息熵计算导航融合权重的新方法。首先给出了水下组合导航系统的整体结构和各子滤波器的状态方程以及观测方程；然后研究了各子滤波器输出信息熵值的计算方法，并且定义了熵积的概念用于计算高斯混合模型中各分量的权重；最后总结出了用于水下载体导航信息融合的熵权高斯混合模型滤波算法的计算流程。仿真实验表明，相比于传统的加权卡尔曼滤波算法，新方法的计算精度更高，对噪声引起滤波发散的抑制能力也更强。

A New Navigation Data Fusion Method Based on Entropy Coefficient Algorithm for Underwater Vehicles

This paper is focused on the problem that the increasing noise of state equations and observation equations lead to the divergence of Kalman filtering in the dynamic positioning navigation. A new method is put forward to calculate navigation fused weight. The new method is based on information entropy and adopts Gaussian mixture model as the framework. First, the main structure of underwater integrated navigation system and related state and observation equations of every sub-filters are provided. Then we studied the calculation method of the information entropy of each sub-filter, and the concept of entropy product is defined to calculate the weight of each component in the Gaussian mixture model. Finally, the computational process of the Gaussian mixture model filtering algorithm for entropy weighted underwater navigation information fusion is summarized. The simulation experiments show that the precision of the new method is much higher and the inhibiting ability against filtering divergence caused by noise is stronger than weighted Kalman filtering algorithm.