Analysis Of Chaotic Characteristics Of Hydrological Time Series And Traffic Flow Time Series Based On PE

Whether the time series has chaotic characteristics determines the exploration of prediction methods and the discussion of its related properties,which affects the accuracy of time series prediction.Chaotic identification provides an important theoretical premise for the prediction of time series.Therefore,chaotic identification play extremely important roles in time series analysis.In this paper,we use the ordered model,sorting entropy and other research methods and chaos theory to study the rainfall time series,runoff time series and traffic flow time series.Firstly,data preprocessing is performed on rainfall time series and runoff time series and traffic flow time se ries.Secondly,the three-state detection method based on ordered mode is introduced to study hydrological data and traffic flow data.Considering the actual situation,this paper mainly studies the hydrological time series with noise and the traffic flow time series based on the Bierley following model.The chaotic identification of the above time series is analyzed by three-state test and the influence of parameters on chaos identificati on is analyzed.The main conclusions are as follows:The three-state test can effectively identify the periodic,quasi-periodic and chaotic states.This paper uses the three-state test method to identify the rainfall and runoff and the simulated traffic flow time series,and further determine the rainfall and runoff and the traffic flow time series has long-term memory and chaotic characteristics,there is no quasi-periodic situation.Through time series analysis of hydrological time series and traffic flow time series with large length,when the number of groups Q is fixed,a larger step p length and a smaller initial value n₀ should be selected,that is,the shorter the initial subset in the grouping,the smaller the fluctuation o f the ??n? value and the gradual stabilization;When the step p length is fixed,a larger number of groupings Q should be selected,the shorter the length of the initial subset in the grouping,the smaller the fluctuation of the ??n?,most of the data points concentrated in a fix ed range,and the range will increase as the value increases.Considering the influence of the number of groups on the judgment Index,when the number of groupings Q is small,there is a certain error in the periodic index ??n?,which affects the accuracy of the results,and the more the grouping number Q,the more stable the periodic index ??n?.In practical application,in order to ensure the ac curacy of chaotic identification,a large number of groupings should be selected as far as possible.Thus,it is shown that the three-state test method is not affected by the length of the identified time series.By analyzing the hydrological time series of noise-containing,the three-state detection algorithm is more accurate than HUST algorithm time series chaotic identification.