Adaptive Correction Algorithm For High Precision Time Maintenance Of Oven Controlled Crystal Oscillator

Whether it is the daily life of mankind or scientific research activities are inseparable from the clock system.With the development of science and technology,people put forward higher and higher requirements on the time unification and precision of the clock system.Oven controlled crystal oscillator is widely used for its advantages.However,the long-term stability of crystal poor so that it can not adapt to the application of high precision time to maintain for a long time.In order to make the crystal suitable for this situation,researchers have made unremitting efforts,let the crystal worked in the supervision of the reference clock source.In the case of the reference clock source,crystal worked in tame state;when the reference clock source is not available,it must be adopted to keep technology to maintain the frequency stability of crystal oscillator.How to maintain high precision stability of the crystal for a long time,when system entry holding stage?the main work of this paper is as follows:First,analyzes the structure of the crystal oscillator,factors affecting frequency and crystal clock error characteristics,reveals the inner reason of the poor long-term stability of crystal oscillator.Secondly,several adaptive correction algorithms for crystal oscillators are studied.In this paper,the construction principle and identification algorithm of the corresponding control model are derived.Finally,the performance of each control algorithm is analyzed by simulation experiment,including:in the adaptive control algorithm based on state space model,the interference caused by the parameter identification of the filter and the limitations of the model based on the adaptive control algorithm of the differential equation model etc..Finally,in order to solve the influence of noise from the crystal oscillator tame system to clock error,an improved model is presented.In this model,the system is ed as the noise source,and the noise from the noise source obeys the Gauss distribution.Modeling the noise to provide a more accurate characterization of the system.Using the extended least squares principle to identification model’s parameter.In this paper,we use the Simulink toolbox in Matlab to design two sets of experimental environment to test the performance difference between the improved model and ARX model,ARMAX model and state space model.The experimental results show that the improved model has stronger approximation ability,higher time accuracy and longer maintenance time ability.