基于神经网络技术的异类传感器信息融合方法

Methods of Information Fusion by Heterogenous Sensors Based on Neural Network Technical

将遗传—BP 算法应用于异类传感器信息融合系统中,利用异类传感器信息的互补性,得到比同类传感器信息融合更精确、更完全、更合理的估计和判断。并用神经网络滤波器代替传统的卡尔曼滤波器,对INS/GPS 组合导航系统进行了仿真研究,研究结果表明,基于神经网络滤波器的 INS/GPS 组合导航系统的精度和实时性都优于传统的卡尔曼滤波器。

The Genetie-BP algorithm is applied to Heterogenous Sensors information fusion system in this paper.Using the complementary property of Heterogenous Sensors information,we have obtained the more accurate,perfect and ra- tional estimation and predication than that of homogeneous sensors information fusion.We attempted to substitute Kal- man filter with neural network filter.The simulation results of INS/GPS integrated navigation system are analyzed. The results show that the accuracy and instantaneity of the neural network filter based on INS/GPS integrated naviga- tion is better than that of traditional Kalman filter.