| Horizontal scanning lidar is an important technical means to carry out scientific and technological traceability of air pollution.In actual observation operations,due to the complexity and variability of the real atmospheric environment,the signal-to-noise ratio of the detection signal and the characteristics of the atmospheric structure will show great changes.In addition to the excellent hardware configuration of the laser radar itself,a set of radar detection signal analysis methods with strong applicability and continuous automatic analysis is the key to whether the laser radar can play an effective role in scientific and technological pollution traceability.In real application scenarios,the existing analytical methods of lidar air pollution detection signal are not suitable for different air quality,and lack of effective automatic hot spot extraction methods.Hot spot extraction depends on manual work.Under this background,the key links of the analysis and hot spot extraction of lidar air pollution detection signal are studied in this thesis.In the detection signal de-noising link,the optimal de-noising method and parameters are studied.Combined with the characteristics of lidar signal,the application effect of common denoising methods such as sliding filter and wavelet denoising is studied,and the evaluation method of denoising effect is established based on the business needs.From the comprehensive subjective and objective index evaluation,wavelet threshold denoising is most suitable for the lidar signal denoising studied in this thesis.In the process of inversion of atmospheric extinction coefficient by solving the lidar equation,the boundary value problem is found and improved to improve the applicability of the inversion method.In this thesis,the applicability of slope method,Klett method and Fernand method for continuous analysis is experimentally studied by using the measured lidar data.It is found that the inversion results of the Fernald method can give the aerosol extinction information close to the practical application needs.At the same time,the selection strategy of the boundary value of the Fernald method will greatly affect the stability of the continuous inversion results.There are some differences in the degree of influence under different air quality.Improper selection of the boundary value will also cause a large deviation in the inversion results of the angle of obstacles.On the basis of this research,this thesis proposes the far and near end calibration combination method by optimizing the boundary value selection strategy to solve the applicability problem of continuous inversion of mainstream methods,and fully proves the effectiveness of the optimization method through comparative verification.Finally,the research on hot spot information extraction and recognition methods based on the analysis results of scanning lidar is carried out in this thesis.Study the automatic generation method of hot spot distribution map and the calculation method of hot spot area,so that the scanned pseudo color map generated by computer analysis can directly show the contour of hot spot high value area distribution and the total hot spot area information without human adjustment.At the same time,the method of automatically obtaining the longitude and latitude of the peak hot spot in the whole layer scanning to locate the source of the hot spot is studied,and the feasibility of the method of extracting the radar detection hot spot information is fully verified by verifying the hot spot source information with manual troubleshooting in practical application.In this thesis,it is of great significance to improve the applicability of the analytical method of lidar atmospheric pollution detection signal in the real complex atmospheric environment by studying and eliminating the defects in the practical application of lidar atmospheric pollution detection signal.At the same time,the application of intelligent hot spot automatic extraction method will greatly improve the efficiency of lidar technology traceability. |
PDF Full Text Request