基于深度学习的无人机飞行轨迹及地质勘测研究

Flight Trajectory of Unmanned Aerial Vehicle and Geological Survey Based on Deep Learning

为了对无人机的飞行轨迹及地质灾害勘测进行研究,首先基于深度学习、无人机等相关理论,对无人机飞行高度、图像分辨率以及地质勘测所消耗时间三者之间的关系进行分析;其次将卷积神经网络模型应用于地质勘测图像精度研究中;最后引入Sigmoid算法对无人机的飞行轨迹进行分析。结果表明:无人机的飞行高度与图像分辨率成反比、与地质勘测所用时间成正比。在同一高度、不同控制点的情况下,卷积神经网络模型能够缩小数据的点位误差以及高程误差,并且随着控制点的增多,精确度也会越来越高。将Sigmoid算法引入无人机的姿态控制以及速度控制中,能够将姿态控制的误差限制在-0.5~1之间,速度控制误差限制在-0.3~0.3之间,可见不管是卷积神经网络模型还是Sigmoid算法都对无人机的发展具有优化作用。因此,深度学习下飞行轨迹规划以及地质灾害勘测,对无人机的快速发展具有很大的参考意义。

To explore the flight trajectory of Unmanned Aerial Vehicle(UAV) and geological disaster survey,firstly,the work analyzes the relationship among UAV flight height,image resolution,and the time consumed in geological survey based on the relevant theories of deep learning and UAV.Secondly,the convolution neural network(CNN) model is applied to the study of geological survey image accuracy.Finally,Sigmoid algorithm is introduced to analyze the flight trajectory of UAV.The results show that the flight altitude of UAV is inversely proportional to the image resolution and directly proportional to the time spent in geological survey.In the case of the same height and different control points,the CNN model can reduce the point position error and elevation error of the data,and the accuracy will be higher with the increase of control points.Sigmoid algorithm is introduced into the attitude control and speed control of UAV,which can limit the attitude control error to-0.5~1 and the speed control error to-0.3~0.3.Both CNN model and Sigmoid algorithm can optimize the development of UAV.Therefore,studying flight trajectory planning and geological hazard survey under deep learning has great reference significance for the rapid development of UAV.