【摘要】：為了快速、準(zhǔn)確地測定強化生物除磷(EBPR)過程中污泥胞內(nèi)糖原的含量,采用4種預(yù)處理方法分別對污泥近紅外光譜進(jìn)行預(yù)處理,并結(jié)合聯(lián)合區(qū)間偏最小二乘(si PLS)進(jìn)行變量優(yōu)選,建立光譜吸光度數(shù)據(jù)與糖原含量的定量分析模型。結(jié)果表明,將一階S-G(Savitzky-Golay)平滑處理后的光譜等分為20個子區(qū)間,聯(lián)合子區(qū)間[10 13 16 19]建立的si PLS模型預(yù)測效果最優(yōu),預(yù)測集的均方根誤差(RMSEP)和相關(guān)系數(shù)(rp)分別達(dá)到0.0048、0.9105,且該模型的交互驗證和外部驗證相對分析誤差(RPD)均大于3.0。一階平滑處理后的光譜si PLS模型預(yù)測精度高、建模變量少,可實現(xiàn)糖原含量的快速測定。
[Abstract]:In order to quickly and accurately determine the content of intracellular glycogen in sludge during enhanced biological phosphorus removal (EBPR), four pretreatment methods were used to pretreat sludge near infrared spectrum, and combined with combined interval partial least square (si PLS) for variable optimization, a quantitative analysis model of spectral absorbance data and glycogen content was established. The results show that the spectrum of the first order S 鈮，

本文編號：2522211