基于反向传播神经网络估算大气光学湍流廓线

Estimation of Atmospheric Optical Turbulence Profile Based on Back Propagation Neural Network

大气光学湍流是与光电工程系统设计、应用密切相关的基本参数。通过仪器实地测量大气光学湍流廓线时会受到人力、物力、财力等多种条件的限制,因此依据常规气象参数估算大气光学湍流强度具有十分重要的意义。提出一种结合遗传算法的反向传播(GA-BP)神经网络。首先,基于Tatarski大气光学湍流参数化方案,利用HMNSP99外尺度模式估算了大气光学湍流廓线;其次,尝试基于实测数据训练BP神经网络,并结合遗传算法估算大气光学湍流廓线。将两种方法估算的大气光学湍流廓线与实测的廓线进行对比,结果表明:GA-BP神经网络模式估算值与实测值的均方根误差(RMSE)比HMNSP99模式的RMSE小,表明利用GA-BP人工神经网络模式估算大气光学湍流廓线是一种可行的方法。

Atmospheric optical turbulence is the fundamental parameter closely related to the design and application of optoelectronic systems.The field measurements of atmospheric optical turbulence profiles by instruments are limited by labor,materials,financial resources,and other conditions.Therefore,it is of great significance to estimate the intensity of atmospheric optical turbulence according to the conventional meteorological parameters.A back propagation combined with genetic algorithm(GA-BP)neural network is proposed.First,based on Tatarski atmospheric optical turbulence parameterization scheme,the HMNSP99 outer-scale model is used to estimate the optical turbulence profiles;second,attempting to construct BP artificial neural network combined with genetic algorithm,which are trained by measured data to predict atmospheric optical turbulence profiles.The atmospheric optical turbulence profiles estimated by the two methods are compared with the measured profiles.The results show that the root mean square error(RMSE)between the estimated values of GA-BP neural network and measured values is smaller than that of HMNSP99 model,which proves that it is a feasible method to use GA-BP artificial neural network model to estimate the optical turbulence profiles.