摘要
为评估近红外光谱分析技术在噻嗪酮生产过程质量控制中应用的可行性,使用近红外光谱分析仪采集样品近红外光谱,采用偏最小二乘法建立噻嗪酮中间体生产过程中N-氯甲基-N-苯基氨基甲酰氯、N-甲基-N-苯基氨基甲酰氯和N-二氯甲基-N-苯基氨基甲酰氯的定量分析模型.最优模型的决定系数R2分别为 0.99、0.99和 0.93,交互验证标准偏差SECV分别为 0.40、0.38和 0.23,理论预测绝对误差范围分别为±0.80%、±0.76%和±0.46%.在实际使用过程中,3项指标的平均预测误差分别为 0.46%、0.38%和0.27%,均能满足客户要求.已在该产品的日常质控过程中使用近红外光谱分析技术替代大部分气相色谱的检测工作.新的分析手段降低了检测部门的工作量,提升了分析效率,提高了分析结果对生产的指导价值.此外,近红外技术在离线场景下的成功应用也为将来实施近红外在线分析奠定了基础.
Abstract
To evaluate the feasibility of applying near-infrared spectroscopy analysis technology in the quality control of the buprofezin production process,the samples spectra were collected using a near-infrared spectroscopy analyzer and the partial least squares method was utilized to establish quantitative analysis models for N-chloromethyl-N-phenyl carbamoyl chloride,N-methyl-N-phenyl carbamoyl chloride and N-dichloromethyl-N-phenyl carbamoyl chloride in the intermediate production process of buprofezin.The performance of optimal models are as follows:determination coefficients(R2)of 0.99,0.99,and 0.93,the cross-validated standard errors of calibration(SECV)of 0.40,0.38,and 0.23,and the theoretical prediction error ranges of±0.80%,±0.76%,and±0.46%,respectively.In practical use,the average prediction errors for the three models were 0.46%,0.38%,and 0.27%,which meeting customer requirements.The customer has used NIR spectroscopy analysis technology to replace most of the gas chromatography testing work in the daily quality control process of this product.The new analysis tool reduced the workload of the testing department,improved both the analysis efficiency and the guidance of the analysis results on production.Moreover,the successful application of near-infrared technology in the offline scenario also laid the foundation for the implementation of NIR online analysis in the future.
维普
万方数据