【摘要】 随着智能手机的快速发展，使用智能手机进行室内定位成为近几年的研究热点之一。使用智能手机进行定位时，针对单纯MEMS惯导推算误差发散速度过快和累积误差会造成定位精度过低的问题，建立了行人运动曲率模型，提出了一种融和卡尔曼滤波和长短时记忆网络(LSTM)的是高精度定位算法。方法包含：使用手机内置传感器采集加速度传感器和陀螺仪数据，建立数据集。利用卡尔曼滤波剔除陀螺仪数据中高斯白噪声，建立LSTM深度神经网络模型预测陀螺仪数据，抑制数据中含有的常值漂移。通过实验验证表明：相比于直接使用MEMS传感器数据进行定位，基于LSTM的室内定位方法可以明显提高定位精度，平均误差在1.33M,满足了人们位置服务需求。更多还原

【Abstract】 With the rapid development of smart phones, indoor positioning using smart phones has become one of the research hotspots in recent years. When using a smart phone for positioning, in view of the problem of too fast divergence of the error divergence and the accumulated error of the simple MEMS inertial navigation, the positioning accuracy will be too low. This paper establishes a pedestrian motion curvature model. Based on this model, a fusion of Karl Mann filtering and long short-term memory network(LSTM) are high-precision positioning algorithms. The method includes: using the built-in sensor of the mobile phone to collect acceleration sensor and gyroscope data and establish a data set. The Kalman filter is used to eliminate the Gaussian white noise in the gyroscope data, and the LSTM deep neural network model is established to predict the gyroscope data and suppress the constant drift contained in the data. Experimental verification shows that compared to directly using MEMS sensor data for positioning, the indoor positioning method based on LSTM can significantly improve the positioning accuracy, with an average error of 1.33 M,which meets the needs of people’s location services.更多还原