基于人工神经网络大气折光系数实时改正方法

Real-time Correction with Atmospheric Refraction Coefficient Based on Artificial Neural Network

根据大气折光的变化特点,导出了大气折光系数中误差的计算公式,分析了影响大气折光系数测定精度的主要因素。结合南里渡特大桥施工监测监控实例,以地形、时段、气温和气压为网络输入,大气折光系数为网络输出,建立了大气折光系数的人工神经网络模型,实现大气折光系数的实时改正。实施有效的大气折光系数实时改正。

According to the varying characteristics of atmospheric refraction, we deduced to the square-error's calculated formula and analyzed the main factors that influence the measuring accuracy of atmospheric refraction coefficient. Combining the constructional monitor and control in Nanlidu Bridge, the geography, time, temperature and atmospheric press were taken as network input, and the atmospheric refraction coefficient was taken as network output, then the neural network model of atmospheric refraction coefficient was established to realize its real-time correction. Implementing the valid real-time correction of atmospheric refraction coefficient has become the key to increase the precision of total-station instrument's triangle elevation.