时频域分形维数分析的光谱信号重叠峰解析算法

发布时间：2018-04-02 04:00

本文选题：分形　切入点：小波　出处：《光谱学与光谱分析》2017年12期

【摘要】：由于光谱谱线存在自然展宽、多普勒展宽、碰撞展宽等,使混合气体中多种成分的吸收光谱信号出现相邻谱峰重叠现象,给混合气体组成成分的定性或定量检测带来较大的困难。现有的方法在获取先验知识、处理精度、运算效率等方面存在不足。提出基于时频域分形维数分析的光谱信号重叠峰解析算法,结合小波的多尺度观测能力和分形的自相似度的度量能力,识别、定位和解析光谱信号中的重叠峰。首先利用小波对具有重叠谱峰的光谱信号进行光谱频率域和尺度域的分析,然后对该时频域的光谱信号在同一光谱频率下的多尺度数据进行自相似性度量和分形计算。逐频率计算后得到光谱信号在频率域的分形维数曲线。该曲线体现了光谱信号在不同尺度的自相似性,其极值位置与光谱信号的各独立峰的位置具有相关性。依据此特性,结合分形曲线的特征参数,最后利用神经网络解析出对应混合气体成分的混叠在一起的各个独立谱峰。该方法利用小波的多分辨率特性,对信号进行不同尺度的精细度量。分形模型则提高了系统解析复杂信号的能力,对重叠程度高的多谱峰重叠信号也有很强的处理能力。借助人工神经网络,实现了整个算法的自动测量。通过实验结果分析,验证了算法的有效性,并讨论影响算法效果的主要因素。
[Abstract]:Due to the natural broadening of spectral lines, Doppler broadening, collision broadening and so on, the absorption spectrum signals of various components in the mixed gases overlap with each other. It brings great difficulties to the qualitative or quantitative detection of the composition of mixed gases. The existing methods are used to obtain prior knowledge and deal with the accuracy. The algorithm based on fractal dimension analysis in time-frequency domain is proposed to analyze the overlapped peaks of spectral signals, which combines the multi-scale observation ability of wavelet and the measurement ability of fractal self-similarity. The overlapping peaks in the spectral signals are located and analyzed. Firstly, the spectral signals with overlapping spectral peaks are analyzed in the spectral frequency domain and the scale domain by wavelet transform. Then self-similarity measurement and fractal calculation of the multi-scale data of the spectral signal in the same spectral frequency are carried out. The fractal dimension curve of the spectral signal in the frequency domain is obtained by frequency calculation. The self-similarity of spectral signals at different scales, The position of the extreme value is correlated with the position of each independent peak of the spectral signal. According to this characteristic, the characteristic parameters of the fractal curve are combined. Finally, the neural network is used to analyze the independent spectral peaks corresponding to the mixed gas components. The method utilizes the multi-resolution characteristic of wavelet. The fractal model improves the system's ability to analyze complex signals, and it also has a strong ability to deal with multi-spectral peaks overlapping signals with high overlap degree. Through the analysis of experimental results, the validity of the algorithm is verified, and the main factors affecting the effect of the algorithm are discussed.
【作者单位】：武汉大学电子信息学院;电网环境保护国家重点实验室中国电力科学研究院;
【基金】：国家科技支撑计划课题(2011BAF02B02)资助
【分类号】：TN911.74

【相似文献】