神经网络算法及其在GPS导航中的应用

GPS导航解算中常采用最小二乘算法。由于导航解算中存在观测异常（粗差），采用的模型精度很难满足用户需求，而高动态用户需求的精度却不断提高，为此文中提出一种利用神经网络的非线性逼近能力，在顾及观测信息权阵的条件下，利用Hopfieid神经网络进行GPS导航解算的新算法。该算法具有很好的抗差性和自适应性，能较好地抑制观测粗差对导航解的影响。计算结果证明了该算法的有效性。

Application of neural network algorithm in GPS navigation

LS algorithm is usually used in GPS dynamic data processing. With the improving of high precision demanding of kinematic users and the existing of measurement outliers in the measurements, the selecting of the appropriately functional models is usually difficult to meet the needs of the users. So a new robust Hopfield neural network, by using the neural network and considering the weigh matrix of observations, is presented in this paper, which, with a high degree of fault tolerance, uses the robust estimator to control the influences of the outliers of observational vectors, and can effectively improve the navigation reliability. It is shown, by calculations and analysis, that the Hopfield neural network algorithm based on robust estimation can control the influences of measurement outliers on navigation results.