Research On Output Signal Processing Algorithm Of Cesium Atomic Fountain Clock

The cesium atomic fountain clock is currently the most accurate atomic clock.But as a time reference,the cesium atomic fountain clock must also have a better frequency stability index.At present,most of the methods to improve the frequency stability index of cesium atomic fountain clock are to realize the conversion from frequency control to phase control by analyzing the relationship between frequency and phase in the cesium atomic fountain clock,and then remove the frequency modulation white noise,so as to realize the frequency stability curve from Change in accordance with the law of 1/(?)into change in accordance with the law of 1/τ.From the perspective of error processing,this article analyzes the error components contained in the frequency difference data output by the open loop of the cesium atomic fountain clock,and realizes the improvement of the frequency stability of the cesium atomic fountain clock while ensuring the accuracy of the cesium atomic fountain clock.At the same time,the power spectrum information of the frequency difference data of the cesium atomic fountain clock and its relationship with the frequency stability of the cesium atomic fountain clock are also analyzed.In order to remove the gross error in the frequency difference data,the traditional average method and median method are improved.First,the working principle of the cesium atom fountain clock and its frequency discrimination principle are analyzed,and two characteristics of it are obtained:one is that it has non-stationary characteristics,and the other is that the method of collecting data does not satisfy the Nyquist law.Then,according to these two characteristics,two methods are used to remove gross errors:one is based on the average value improvement method.The improvement method is to calculate the average value within the smoothing time by setting different smoothing times,and replace the gross error with the average value.;The other is based on the median improvement method,the improvement method is to use the median instead of the gross error found by the difference with the frequency difference.Both of these improved methods have achieved the effect of removing gross errors.In order to remove the non-stationary noise in the frequency difference data,a threshold method based on wavelet analysis is used to denoise the preprocessed frequency difference data.The frequency stability calculated by using the frequency difference data after denoising has improved the overall frequency stability index compared with the case without error processing.By comparing the frequency stability data,it can be seen that the improved average method is more effective than the improved median method,and the long-term stability of the frequency stability has been improved.However,their frequency stability curve still has a convex phenomenon,and there is always an abnormal value per second.Analyze the cause of frequency stability bump and second stability abnormality,and use data smoothing method to correct the abnormal value of frequency stability second stability to eliminate the phenomenon of convex frequency stability curve.According to the characteristics of the segmented calculation of frequency stability,first use the distribution histogram of the frequency difference to determine the range of correction required for the frequency difference,and then use the arithmetic average of the data and the sequence average to replace the data for data outside the range.Finally achieve data smoothing.The overall frequency stability index of the final corrected frequency difference data has been greatly improved,the ten-thousand-second stability has also entered 10-16,and the trend of frequency stability is also closer to the change trend of 1/τ.By analyzing the power spectrum of the open-loop frequency difference data of the cesium atom fountain clock,the relationship of the frequency stability of the cesium atom fountain clock is studied.Since the frequency difference data of the cesium atom fountain clock has two characteristics,the Burg method in modern spectrum analysis is used to estimate the power spectrum of the frequency difference data of the cesium atom fountain clock.The horizontal axis of the power spectrum can be interpreted as the power of the difference between the frequency difference data and the mean value of the frequency difference data,and the vertical axis is the power spectrum of the frequency difference value.The study found that the closer the total power of the frequency difference data is to 0,the closer the result of data smoothing is to the mean value of the frequency difference,and the closer the frequency stability of the cesium atomic fountain clock is to the change trend of 1/τ.