神经网络在SINS/GPS组合定位中的应用

Application of neural network in SINS/GPS combined positioning

地籍测量中,单一系统无法满足定位要求,组合定位技术应运而生.其中,捷联惯性导航系统(SINS)和GPS组合定位应用最为广泛.在卫星信号受到干扰失效区域,系统进入纯SINS解算,定位误差会逐渐累积,无法满足定位精度要求.针对此问题,提出一种长短期记忆(LSTM)神经网络辅助的组合定位算法.根据LSTM神经网络能够有效运用于长距离时间序列的特性,在GPS有效区域,用卡尔曼滤波(KF)算法对SINS/GPS信号进行数据融合得到精确定位信息,同时利用惯性测量单元(IMU)、GPS和SINS输出信息对神经网络进行训练;在GPS失效区域,利用训练好的神经网络预测GPS位置信息,使得系统能继续用卡尔曼滤波器滤波.最后结合地籍测量特点,设计了仿真实验,证明了该算法在GPS信号失效时可以有效抑制系统误差发散、提高定位精度,在不同运动状态下依然可以满足定位精度要求、鲁棒性强.

In cadastral surveying,a single system cannot meet the positioning requirements,and combined positioning technology has emerged.Among them,the strapdown inertial positioning system(SINS)and the GPS combined positioning are most widely used.In areas where satellite signals are interfered and failed,the system enters the pure SINS solution,and the positioning error will gradually accumulate and cannot meet the positioning accuracy requirements.In response to this problem,this paper proposes a combined positioning algorithm assisted by long and short-term memory(LSTM)neural network.According to the characteristics of LSTM neural network that can be effectively applied to long-distance time series,in the GPS effective area,the Kalman filtering(KF)algorithm is used to compare SINS/GPS signal data fusion to obtain precise positioning information,while using inertial measurement unit(IMU),GPS and SINS output information is used to train the neural network;in the GPS failure area,the trained neural network to predict GPS location information is used,so that the system can continue filter with Kalman filter.Finally,combined with the characteristics of cadastral measurement,a simulation experiment was designed to prove that the algorithm can effectively suppress system error divergence and improve positioning accuracy when GPS signal fails,and it can still meet positioning accuracy requirements under different motion states with strong robustness.