Application of the neural network approach for retrieving of gas concentration from CO_2 - laser data

摘要

In the report a method of atmospheric gases concentration retrieving from the CO2-laser gasanalyser data on the basis of neural networks (NN) is description. The method of neural networks is compared to the known method of the least squares most frequently meeting at processing of laser signals. One of the problems arising at processing of the lidar signals is stability of the solving (gas concentration) depending on random mistakes of measurement. A method of the neural network as have shown results of numerical modeling, it is possible to relate to a stable method of retrieving of gases concentration from CO2-laser data.