基于自适应神经模糊推理系统的光谱预测

Adaptive Network-Based Fuzzy Inference System for the Forecast of Spectra

由于神经网络具有自适应以及自学习功能,模糊系统具有很好的非线性推理能力。将二者相结合的自适应神经模糊推理系统吸收了二者的优点,将其应用于对模型特征的分析和建模上,通过对数据的学习可以预测特性非常复杂的系统,如荧光光谱。荧光光谱分析法操作简单,精度高,分析速度快,是研究分子内部结构的重要手段。预测荧光光谱更具有重要意义。文章以N2分子的脉冲放电的发射谱为例进行了预测,预测结果显示,该方法可以预测光谱谱线的重要信息,误差均小于1.66%,达到了此次实验的精度要求,具有满意的效果。对光谱的预测是切实可行的。

Because of the function of independent adaptation and study of NN and the best nonlinear speculating ability of fuzzy system, the adaptive network-based fuzzy inference system （ANFIS） which is composed of them absorbs their virtue. When applying it to the data analysis and model construction, the authors got the good nonlinear forecast by learning data, such as fluo- rescence spectrum. Fluorescence spectrometry is an important means of researching inside structure of molecule, which works easy, and features rapidness and high precision. So its forecast is more important. In the present paper, the authors utilize the spectra of the N2 to calculate, and prove that the means can forecast important spectrum information, the error is very small, less than 1.66 percent, which is conformed to meet the demand of the experiment. In short, the approach is workable.