摘要
针对目前月降雨序列混沌特性研究中存在的问题,以广东省西江流域高要站月降雨序列为例,运用功率谱方法、主成分分析法、饱和关联维数法、C-C方法进行了混沌特性的判定及特征参数的求取,同时分析了数据长度和噪声对混沌研究的影响。研究结果表明,利用功率谱方法进行混沌判定时,单纯的根据连续多峰的噪声背景作为判定混沌存在的依据并不可靠;饱和关联维数法仅从能量角度对混沌序列进行判定,此外,对混沌序列进行滤波会导致此法判定结果的稳健性降低,C-C方法证明了其计算结果的可靠性;为计算出相对稳定的饱和关联维D2,计算数据的长度至少应为450个点;递归图及相应的各种定量判定标准验证了改进的双小波空域降噪方法可有效去除混沌序列中噪声的影响。
A discussion is conducted due to several problems presented in chaotic research of monthly rainfall series.As an example,monthly precipitation data of Gaoyao Station in West River basin in Guangdong Province is utilized for chaotic identification and calculation of characteristic parameters via power spectrum method,PCA,G-P Algorithm and C-C method.Impacts of data length and noise in the dataset are also studied.The results show that it is not reliable to judge a chaotic feature reflected on the power spectrum as successive multi-peaks;although chaotic series and noise can be distinguished directly with PCA method,it can only be held in the way of energy;besides,filtering process on chaotic series will make identification results derived from PCA method unstable;embedded parameters calculated with C-C method prove the validity of results from G-P Algorithm;take dataset used in this paper for example,a relatively stable D2 can be calculated on the premise that at least 450 data points are considered;recurrence plot and corresponding quantitative analysis indices validate effectiveness of noise rejection on chaotic series with improved dual-wavelet spatial correlation method.
出处
《水电能源科学》
北大核心
2011年第3期9-12,193,共5页
Water Resources and Power
基金
国家自然科学重点基金资助项目(50839005)
2009年广东省水利创新研究基金资助项目(2009-41)
关键词
月降雨序列
混沌
功率谱
PCA
相空间
递归图
monthly rainfall series
chaos
power spectrum
PCA
phase space
recurrence plot
