论文部分内容阅读
为了克服递归趋势(recurrence trend,RT)指标对不同信号非平稳度估计存在误判的缺陷,分别采用互信息法和伪临近法确定了递归量化分析的最佳延迟时间和最小嵌入维数,然后在递归量化分析基础上,提出了归一化局部递归率标准差(standard deviation of normalized local recurrence rate,SDNLRR)作为信号非平稳度量化指标.利用该指标,通过递归量化分析方法分析了白噪声信号、正弦信号、调幅信号、线性调频信号4个基本信号和2个实测台风场脉动风速信号的非平稳特性,并与传统的递归趋势指标分析结果进行了对比.研究结果表明:利用SDNLRR指标对6个信号的非平稳度的量化比较准确率达100%,比RT指标的准确率提高了33.33%,消除了RT指标对正弦信号和平稳脉动风速信号的错误估计.
In order to overcome the shortcoming that the recurrence trend (RT) index misjudges the non-stationary signal estimation, the best delay time and the minimum embedding dimension of recursive quantitative analysis are respectively determined by mutual information method and pseudo-approach method. Then, Based on the recursive quantitative analysis, a standard deviation of normalized local recurrence rate (SDNLRR) is proposed as a measure of signal nonstationarity. By using this indicator, the white noise signal , Four basic signals of sinusoidal signal, amplitude modulation signal and chirp signal, and two non-stationary characteristics of fluctuating wind speed signal of two measured typhoon fields were compared with the traditional recursive trend index analysis results.The results show that the use of SDNLRR index 6 The accuracy of the non-stationary signal quantization is 100%, which is 33.33% higher than that of the RT, which eliminates the misjudgment of the error of the RT signal on the sinusoidal signal and the smoothed fluctuating wind speed signal.