支持向量回归同时测定邻苯二酚、间苯二酚和对苯二酚稳健模型的研究

Study on simultaneous determination of pyrocatechol,resorcinol and hydroquinone by supporting vector regression robust modeling

在波长200～400nm范围内,测定邻苯二酚、间苯二酚和对苯二酚的吸收光谱。用连续小波变换(CWT)对光谱数据进行预处理,消除光谱信号中的背景和噪音后,再用偏最小二乘法(PLS)和支持向量回归方法(SVR)建立校正模型,并对水样中的邻苯二酚、间苯二酚和对苯二酚进行预测。结果表明,CWT-SVR方法优于PLS、SVR和CWT-PLS等方法,测定结果满意。

Within the wavelength range of 200 - 400 nm, the absorption spectra of pyrocatechol, resorcinol and hydroquinone were measured.The absorption spectra were firstly preprocessed using continuous wavelet transformation （CWT）for denoising and removing the spectral background, and then the partial least squares （PLS） and support vector regression （SVR） techniques were used for building the calibration model, which was applied in the determination of pyrecatechol, resorcinol and hydroquinone com- pounds in polluted water samples. The results had shown that CWT-SVR was better than those methods including PLS, SVR and CWT-PLS, when it was used to determine pyrocatechol, resorcinol and hydroquinone simultaneously.