Cesium fountain clock and hydrogen clock frequency difference estimation method for optimizing wavelet neural network on basis of genetic algorithm

Abstract

The invention discloses a cesium fountain clock and hydrogen clock frequency difference estimation method for optimizing a wavelet neural network on the basis of a genetic algorithm, and belongs to the technical field of atomic time scales. Accordingly, cesium fountain clock and hydrogen clock frequency difference data is subjected to preprocessing which comprises outlier detection and missing data quasi complement. Input layers and hidden layers of the wavelet neural network, the number of the hidden layers and wavelet base selection are determined according to the frequency difference data, and a fountain clock data estimation model for optimizing the wavelet neural network on the basis of the genetic algorithm is built for further improving the fountain clock data estimation precision and estimation stability. In the fountain clock controlled hydrogen clock set estimation process, prediction is conducted through the genetic wavelet neural network for the first time, the prediction precision is greatly improved compared with that of an existing linear prediction, data is more stable, and therefore the control precision of a fountain clock controlled hydrogen clock set is improved, and a more accurate basis is provided for generating TA(NIM) higher in stability and accuracy.

Description

technical field [0001] The invention belongs to the technical field of time and frequency, and mainly uses the historical data of the frequency difference between a cesium atomic fountain clock and a hydrogen clock to obtain the data when the fountain clock is out of service through a genetic algorithm optimization wavelet neural network prediction method. Background technique [0002] Time frequency is an important branch of basic scientific research, and plays a pivotal role in scientific research, positioning systems, power systems, military and national security. For this reason, most countries in the world have built their own punctual laboratories to produce their own atomic time scales and participate in international comparisons. [0003] The Timekeeping Laboratory of the China Institute of Metrology is the first timekeeping laboratory in the world that uses a fountain clock to drive a hydrogen clock group to generate local atomic time. Since 2007, the local atomic ...

AI Technical Summary

Problems solved by technology

The idea of ​​the linear regression prediction algorithm is to regard the frequency difference data as a set of linear sequences, and use the least square method to obtain the unary linear coefficients to realize the prediction. The method is simple, but it can only predict the change trend of the frequency difference data; the real data is not ideal The straight line, resulting in insufficient data accuracy of the straight line fitting, and there is a large deviation from the real data

Images