Hydrogen atomic clock difference prediction based on the LSSVM

Abstract

The atomic clock prediction is the key step for constructing atomic clock time scales and implementing atomic clocks steering, and its prediction level directly affects the accuracy and stability of atomic clock time scales. Aiming at the non-linear and non-stationary characteristics of the hydrogen atom clock, the least squares support vector machine (LSSVM) hydrogen clock prediction algorithm is proposed, which is verified by the actual hydrogen atomic clock data of the timekeeping laboratory in National Institute of Metrology, China. The results show that compared with the linear prediction algorithm and the support vector machine (SVM) prediction algorithm, the least squares SVM prediction algorithm improves the accuracy of the clock prediction, and its root mean square error is, respectively, reduced by 50% and 29%.

Keywords

• atomic clocks
• support vector machines
• least squares approximations
• support vector machine prediction algorithm
• linear prediction algorithm
• actual hydrogen atomic clock data
• vector machine hydrogen clock prediction algorithm
• hydrogen atom clock
• prediction level
• implementing atomic clocks steering
• atomic clock time scales
• atomic clock prediction
• lssvm
• hydrogen atomic clock difference prediction
• squares svm prediction algorithm