基于随机森林权重补偿的无人机高精度定位算法

为了减小室外无人机(unmanned aerial vehicle，UAV)监测过程中的定位误差，对室外UAV进行实时定位，提出了一种基于随机森林的Chan-Taylor三维定位算法。通过K近邻对定位数据扩展后，根据Chan-Taylor算法将随机信号多径噪声转化为高斯分布，便于模型提取信号特征。使用交叉验证，实现随机森林特征参数与混淆矩阵阈值的自适应确定，并用该阈值衡量模型的一致性。利用分类结果更新UAV定位权值矩阵，有效地补偿目标高度数据。此外，使用标定UAV对设备误差进行估计，校正定位结果。理论分析与仿真结果表明，该算法能够有效地提高UAV定位精度，实现利用移动通信基站对UAV进行无源定位。

High-precision positioning algorithm for UAV based on random forest weight compensation

In order to reduce the positioning error in the process of outdoor unmanned aerial vehicle (UAV) monitoring and realize real-time positioning of UAV, a Chan-Taylor three-dimensional positioning algorithm based on random forest is proposed. After the positioning data is expanded by the K-nearest neighbor pair, the random signal multipath noise is converted into Gaussian distribution according to the Chan-Taylor algorithm, which is convenient for the model to extract signal features. Cross validation is used to realize the adaptive determination of random forest feature parameters and confusion matrix threshold, and this threshold is used to measure the consistency of the model. Using the classification results to update the UAV positioning weight matrix, and effectively compensate the target height data. In addition, the calibration UAV is used to estimate the device error and correct the positioning result. Theoretical analysis and simulation results show that the algorithm can effectively improve the UAV positioning accuracy, and realize the passive location of UAV using the mobile communication base stations.