Research On Key Technologies Of Signal Processing For Atmospheric Remote Sensing Lidar

Due to the great advantage of continuous detection,measurement accuracy,temporal and spatial resolution,lidar has become a hot spot in the field of atmospheric remote sensing detection.The dynamic range and signal to noise ratio of the echo signal in the actual detection of the lidar atmospheric sensing limits the detection range and the signal quality of the lidar.In this paper,the lidar dual-channel data gluing technology and based on photon counting detection data in the noise reduction inversion to explore and improve the lidar detection range and signal quality.Since the laser beam emitted from the laser beam is inversely proportional to the square of the distance after the atmospheric response,the signal of the low-level boundary layer in the echo signal is very strong,and the higher troposphere and the stratosphere will gradually weaken the signal due to the increase of the detection distance and cannot detection at last.Lidar dual-channel data gluing technology can effectively solve the aforementioned problems.This paper presents a dual-channel data gluing algorithm for lidar based on fast non-dominated sorting algorithm(NSGA-?)and neighborhood rough set theory(NRS).The algorithm NRSWNSGA-? designs three functions to evaluate the fitting region,and obtains the weight coefficient through NRS training samples,and then uses NSGA-? to select the optimal fitting region linearly after the global random search.The technology can improve the detector detection range and dynamic range of limited problems,and to achieve the adaptive data stitching,access to a stable large dynamic range of the gluing signal.As the laser echo signal is weak,the laser radar system scattering and the circuit noise will further pollute the original signal.How to suppress the background noise and improve the signal-to-noise ratio is especially necessary for the 1064nm high-spectral-resolution lidar.In this paper,the noise reduction inversion technique is studied,and the noise reduction inversion based on photon counting detection data is realized by using the total variation(TV)penalized maximum likelihood estimator(PMLE).The algorithm TV-PMLE uses the features that noise in the photon counting data meets the characteristics of Poisson distribution,and the maximum likelihood probability is used to achieve the signal reduction.And the algorithm adds the full variation penalized mechanism to prevent the damage of the algorithm from the anomaly solution,and obtains the final noise reduction signal by using iterative and training to select the optimal penalty weight.Finally,the above two method are experimentally verified.The single-profile gluing experiment results show that the gluing effect is good and the error of the gain ratio of the repetition of the experiment is less than 0.04%.Compared with the existing algorithms,the gluing results are more stable in the three evaluation functions in all day gluing experiment.The noise reduction experiment is carried out by using pulse cumulative average,moving average,wavelet analysis,Savitzky-Golay filtering and empirical mode decomposition respectively.The results are compared with the total variation penalty likelihood estimation algorithm.The experiment results show that the algorithm has good de-noising effect and prevents the signal from being smoothed.