基于参数标准化的单样本定标方法在微波等离子体炬水泥快速分析中的应用

A Parameter Standardization-based Single Sample Calibration Method and Quantitative Analysis of Cement Based on Microwave Plasma Torch

流程工业颗粒物成分的在线检测具有低延时、安全可靠及低成本等要求,目前没有适合的技术手段.微波等离子炬(Microwave plasma torch,MPT)原子发射光谱法具有直接进样、成本低及装置安全可靠等优点,有潜力成为工业在线检测的重要手段.本文基于MPT装置对水泥样品进行直接进样分析,并针对工业应用场景中标准样品受限的问题,采用单样本定标技术(Single sample calibration,SSC)进行定量分析.SSC法中的线性假设容易受到等离子体参数波动的影响,导致精度下降.为此,提出了一种基于参数标准化的SSC算法PS-SSC(SSC based on parameter standardization),通过激发温度和电子数密度对谱线强度进行修正以提高SSC的定量精度.为了评估MPT和PS-SSC方法联用在水泥成分快速分析中的有效性,将GSB 08-2985-2013标准水泥粉末气溶胶直接引入MPT中进行定量分析,并将PS-SSC与现有定量方法进行比较.与传统SSC方法相比,PS-SSC方法的决定系数R2由‒0.81~0.81提高到0.39~0.88,平均相对误差由4.39%~10.33%提高到1.55%~5.83%,平均相对标准偏差由2.89%~9.40%提高到2.28%~6.50%,展现了该方法在工业在线成分检测中的应用潜力.

Owing to the stringent requirements for low latency,safety,reliability and cost-effectiveness,there is no suitable technical approach available for online detection of industrial components currently.However,microwave plasma torch(MPT)has the advantages of direct injection,low cost,safety and reliability,which qualify for becoming an essential technique for industrial product online detection.Consequently,we designed experiments on cement composition analysis based on MPT.Meanwhile,as to limited standard samples we use single sample calibration(SSC)for quantitative analysis.However,the simple linear assumption underlying SSC is susceptible to fluctuations in plasma parameters leading to reduced accuracy.To address this issue,we propose an improved SSC algorithm based on parameter standardization(PS-SSC),which enhances the accuracy of SSC by modifying spectral line intensity through excitation temperature and electron number density adjustments.To evaluate the effectiveness of MPT-based rapid analysis combined with PS-SSC for cement composition analysis,we directly introduced GSB 08-2985-2013 standard cement powder aerosol into MPT and contrasted the effects of different methods.Compared to traditional SSC methodology,the PS-SSC method exhibited significant improvements:the determination coefficient R^(2) of the PS-SSC method was increased from‒0.81—0.81 to 0.39—0.88,and the average relative error was increased from 4.39%—10.33% to 1.55%—5.83%.The average relative standard deviation increased from 2.89%—9.40% to 2.28%—6.50%,thus demonstrating its potential for online cement component detection.