Signal Processing And Algorithm Research For Coherent Wind Lidar

Among many atmospheric wind field detection methods,coherent wind lidar based on fiber laser structure has shown the advantages of compact structure,stable performance,high accuracy of wind measurement,and fast time response,which has become significant detection device for atmospheric scientific research,weather forecast,wind farm management and atmospheric environment monitoring.The weak signal detection of coherent wind lidar requires strict estimation accuracy of the wind velocity estimation algorithms,and needs to quantitatively analyze the estimation error of those algorithms.The paper studies the performance evaluation of echo signal processing and wind velocity inversion algorithms of coherent wind lidar.It has important reference value for improving the accuracy of wind velocity inversion algorithms.The thesis firstly introduces the basic principle of coherent wind measurement lidar.By introducing the lidar equation,carrier-to-noise ratio,and antenna efficiency to evaluate the performance of coherent wind lidar system,which provides theoretical basis for the subsequent comparison of wind velocity estimation algorithms.Then based on the covariance matrix echo signal model and the hierarchical structure echo signal model with turbulence perturbation,employing maximum likelihood estimation(ML),maximum likelihood discrete spectrum peak estimation(ML DSP),and Gaussian fitting estimation(Gauss)processing different types of echo signals,which obtains the estimated standard deviation and detection probability that characterize the wind velocity estimation error.The results of simulation show that ML estimation has the best performance,and Gauss estimation is better than ML DSP estimation.Under the conditions of wider pulse width,the performance of the wind velocity estimations is better.And also,atmospheric turbulence seriously interferes with the accuracy of the wind speed estimation algorithm.Using the wind field data measured by a coherent wind lidar set up in our laboratory,the wind velocity estimation error by ML DSP and Gauss estimation was analyzed and compared.The results show that Gauss estimation performance is little better than that of corresponding ML DSP estimation in the low detection distance region.In the higher distance region,the wind velocity error difference between the two algorithms increases significantly with the distance,indicating that the accuracy of Gauss estimation is significantly better than ML DSP estimation.The statistical analysis of the two estimation algorithms using autocorrelation coefficients shows that the wind speed data obtained by Gauss estimation is more stable in the time domain.