CAJ | 학술논문

传统的石油产品水分检测存在试剂有毒、操作复杂、测量时间长、受环境影响大、满足不了在线快速检测等问题；在此利用近红外光谱快速测定透平油中微量水分，利用关联向量机对测定的光谱数据进行建模，实验结果表明：最大绝对误差和均方根误差分别为20．99×10^-6（V／V）、11．1l×10^-6（V／V），说明该模型具有较高的预测能力，对实现透平油微量水分的快速检测具有重要意义。
전통적석유산품수분검측존재시제유독、조작복잡、측량시간장、수배경영향대、만족불료재선쾌속검측등문제；재차이용근홍외광보쾌속측정투평유중미량수분，이용관련향량궤대측정적광보수거진행건모，실험결과표명：최대절대오차화균방근오차분별위20．99×10^-6（V／V）、11．1l×10^-6（V／V），설명해모형구유교고적예측능력，대실현투평유미량수분적쾌속검측구유중요의의。
There are evident limitations in traditional measurement of moisture content in oils, such as poisonous reagents, complicated operations, time-consuming, susceptible to environmental interferences and unable on-line quick measurement. A fast determination of the moisture content in turbine oil by near infrared spectroscopy has been introduced in this paper. The relevance vector machine was employed to model the spectral data. The Maximum Absolute Error and the Root Mean Square Error of the experimental results are respectively 20.99 × 10^-6 （V/V） and 11. 11 × 10^-6 （V/V）. The results indicate that the proposed model features good prediction performance and thus plays an important role in fast measurement of moisture content in turbine oil