Prediction method of crystal resonator storage life based on LS-SVM

A method of data processing - least squares support vector machine (LS-SVM), has been introduced to predict the storage life of crystal resonator. Several environmental factors have been investigated for their effects on the degradation of the function of quartz crystal resonator, the main degradation mechanisms of crystal resonator are studied and analyzed to figure out the trend of the performance parameters in the long-term storage, the frequency deviation is determined as the sensitive degradation parameter of crystal resonator. When temperature is the only accelerated stress, the gradation model of frequency deviation has been combined with the Arrhenius model to obtain the regression function about frequency deviation, and by means of LS-SVM, the storage life prediction of crystal resonator can be modeled and computed. The accelerated storage life test of the crystal resonator JA8 is designed, the degradation data collected of frequency deviation are put into the LS-SVM model, and with the result that storage life of crystal resonator can be predicted successfully. This model takes advantage of LS-SVM to solve how to build small samples and nonlinear model and process test data in storage life prediction of crystal resonator, which improves the prediction accuracy and computing efficiency and will be very practical and promotional.