【摘要】 采用透射法采集了污水处理厂废水120个样品的近红外光谱数据,以采用标准稀释法测得的5 d废水生化需氧量为参考值建立废水BOD近红外预测模型.研究比较了以逐步多元线性回归(SMLR)、主成分回归(PCR)、偏最小二乘法(PLS)3种校正方法建立的废水生化需氧量(BOD)预测模型,发现PLS校正模型优于其余2种,其校正相关系数(rc)达0.763,校正标准差为27.7 mg.L-1,预测标准差为29.1 mg.L-1.结果表明,近红外光谱技术可成为废水BOD快速检测的新手段.更多还原

【Abstract】 The spectral data in the region of 800-2500 nm of 120 samples collected from Hangzhou wastewater treatment plant were obtained by means of transmission method using the nexus FT-IR thermo nicolet instruments.The models for the wastewater near-infrared forecast were set up based on 5 d biochemical oxygen demand in wastewater determined by the method of standard dilution.The forecast models of biochemical oxygen demand(BOD) in wastewater set by partial least squares(PLS),principal component regression and stepwise multiple linear regression were studied and compared.The results were as follows: PLS proofreaded model was the best of 3 models;the proofreaded correlation coefficient(rc) was 0.763;the standard error of prediction was 27.7 mg·L-1.It suggested that the technology of near-infrared spectrum could be a new mean for the rapid determination for BOD in wastewater.更多还原