Genetic Algorithm Based Change-point Analysis Method for Hydrological Time Series
JIN Ju-Liang~1,WEI Yi-Ming~2,DING Jing~3(1.School of Civil Engineering,Hefei University of Technology,Hefei,Anhui 230009;2.Institute of Policy & Management,Chinese Academy of Sciences,Beijing 100080;3.College of Hydraulic Engineering,Sichuan University,Chengdu,Sichuan 610065)
Analysis of stages and aberrance points of hydrological time series conduces to understand and manage the complex characteristics of hydrological system evolution process,which can be applied in many fields,such as hydrological frequency analysis,hydrological prediction,hydrological computation and so on.In order to overcome shortage of common change-point methods,such as computation complex and the difficulty of diagnosing all change points,a new method of change-point analysis based on accelerating genetic algorithm developed by the authors,named AGA-CPAM for short,is presented for hydrological time series.The modeling of AGA-CPAM is the key in this paper,which includes three steps as follows.Step 1 is to determine the search ranges of change point number and position of hydrological time series according to scatter point plan and dot value figure of hydrological time series.Step 2 is to optimize the parameters of change point positions and jumping values based on the criterion of least square of the subsection fitting errors with accelerating genetic algorithm.Step 3 is to analyse the stages and aberrance points of hydrological time series obtained from Step 2 based on cause of formation,which results can be used as scientific foundation of prediction,simulation,regulation and control of hydrological time series.The computation results of the case study can include two terms as follows:(1)With the accelerating genetic algorithm developed by the authors,both of change-point position values and jumping values can be optimized at the same time,and the difficulty problem of much computation of common change-point analysis methods is solved.(2)The example results show that AGA-CPAM is visual,simple,practical and efficient,and that AGA-CPAM can also be applied to cataclysm change analysis of different nonlinear time series.
【Fund】： 教育部优秀青年教师资助计划[教人司(2002)350];; 安徽省优秀青年科技基金;; 安徽省自然科学基金(01045102 01045409);; 国家自然科学基金(70425001 70471090)项目资助
【CateGory Index】： P333
【CateGory Index】： P333