摘要
光谱解析方法是一种常见的光谱分析方法,广泛用于各种化学计量学领域。现有的解析方法无法处理纯组分未知的应用体系,本文针对该情况,提出了基于遗传算法和最小二乘法的多元组分光谱解析定量分析方法。该方法首先通过遗传算法在混合光谱上寻找未知组分的最优峰位置和最优峰形,得到一组的最优纯组分光谱矩阵,再利用最小二乘拟合曲线,能够快速有效地解析混合光谱。在实验中,对纯组分光谱全未知、纯组分光谱部分未知及不同参数设置下算法的表现进行了讨论,分析其对算法收敛速度及计算结果精确性和稳定性的影响。利用该方法对流式细胞仪光谱数据进行处理,解析效果良好,谱线的契合程度高,验证其用于多组分流式细胞仪光谱数据分析的可行性、有效性和精确性。
Spectral unmixing method is a common method of spectral analysis and widely used in various fields of stoichiometry.The existing analytical methods cannot deal with the unknown application system of pure components,aiming at this situation,the proposed algorithm based on genetic algorithm and least square method can analyze the full spectrum data and separate the overlapping spectra automatically.Genetic algorithm(GA)is adopted to search the optimal positions and peak sharps of the basic spectra derived from the unknown components,and then the fluorescence intensity of each component can be estimated simultaneously by least squares method.In the experiment,the convergence rate,accuracy and stability of the proposed algorithm are evaluated under the conditions of total and part unknown components.The flow spectra data are processed,and the results demonstrate the feasibility and effectiveness of the proposed algorithm.
作者
范贤光
巫梅琴
陈宇欣
方晓玲
胡雪亮
王昕
FAN Xian-guang;WU Mei-qin;CHEN Yu-xin;FANG Xiao-ling;HU Xue-liang;WANG Xin(Department of Instrumental and Electrical Engineering,School of Aerospace Engineering,Xiamen University,Xiamen Fujian 361005;Fujian Key Laboratory of Universities and Colleges for Transducer Technology,Xiamen Fujian 361005;Xiamen Key Laboratory of Optoelectronic Transducer Technology,Xiamen Fujian 361005)
出处
《数字技术与应用》
2018年第1期134-137,139,共5页
Digital Technology & Application
基金
国家自然科学基金(NO.21503171)
国家重大科研仪器研制项目(NO.21627811)
关键词
光谱解析
遗传算法
最小二乘法
流式细胞仪
spectral unmixing
genetic algorithm
least squrares
flow cytometer
