Research On Improved Algorithm Of Time-Varying Correlation Estimation And Its Application In Hydrological Data Analysis

The Hurst parameter in the traditional sense is a fixed constant,and the self similarity described can only be limited to the whole data,but a large number of data show local self similarity,so the traditional Hurst parameter can not accurately describe the local mutation information of time series.For the actual time series data,local mutation may lead to different Hurst parameter estimation results,that is,different Hurst values may exist in different periods of a time series.Therefore,more and more attention has been paid to the local self similarity of random signals.The research and analysis of local self similarity of time series will provide important significance for the accurate establishment of data model and data prediction.In order to analyze the local self similarity of data and the stability of time-varying Hurst parameter estimation more accurately,this thesis introduces alpha stable distribution noise time series with obvious peaks to simulate the local mutation environment of data,and analyzes and evaluates the performance of five traditional Hurst algorithms.Aiming at the shortcomings of traditional periodogram algorithm,such as missing information and poor fitting effect,this thesis proposes an improved periodogram estimation algorithm based on ANN model,and analyzes the accuracy and stability of time-varying Hurst parameters in the presence of impulsive noise by combining with sliding window function.In order to further verify the effectiveness of the improved algorithm,it is applied to the analysis and prediction of daily precipitation in Shenyang and Zunyi in recent 60 years.By improving periodogram estimation algorithm,this paper analyzes its self similarity characteristics and predicts the future development trend according to FARIMA model,and obtains good experimental results.The improved algorithm proposed in this thesis effectively solves the defects of traditional Hurst parameter estimation algorithm,improves the accuracy of Hurst parameter estimation algorithm and expands its application scope,and also provides a reliable and effective method for future hydrological data analysis and prediction.