LCVR phase retardation characteristic calibration method based on SVM regression

Abstract

To establish mathematical model between driving voltage and phase retardation of liquid crystal variable retarders (LCVR) attached with a compensator, A new setup based on Stokes parameters is built to measure the LCVR retardation. Under 25 °C, the 532nm wavelength vertical incident light, adjusting the driving voltage, collecting LCVR output power by polarization analyzing system. Using photoelectrical detection theory, the total of 121 experimental samples were calculated, the samples was divided equally, the one was training set and the other one is prediction set. The radial basis function is used as the kernel function in SVM(Support Vector Machines), the K-fold cross validation is used to optimize the penalty parameter c and the parameter g. Test results show, comparing with LS(least squares), For prediction model based on epsilon-SVM, the squared correlation coefficient reached 96%, the mean square error is increased. The model could rapidly and accurately obtained the phase retardation of all the wavelengths and all the voltage values of LCVR in the effective working range, prediction model based on £-SVM was an effective mean for LCVR phase calibration.