摘要
将近红外光谱法与偏最小二乘法(PLS)相结合,快速检测了水体成分相对稳定的城市湖泊水体化学需氧量,通过主成分分析再结合欧氏空间距离聚类法筛选具有代表性的样本,并从系统独立变量数判断了主成分数,构建了数学校正模型,并对预测集样品含量进行预测。试验结果表明,该模型预测效果较好,为化学需氧量检测提供了一种软测量方法。
A method for rapid detection of chemical oxygen demand （COD） in city lake water of relatively stable com- ponents is investigated by combining the near infrared spectrum with partial least squares. Principal component analysis and Euclidean distance cluster are applied to select the representative samples. The number of system independent varia- bles is used to determine the number of principal components and the mathematic calibration model is built. At the same time the sample content of prediction set is predicted. The experimental result shows that the model has a better predic- tive effect, which provides a soft measurement method for COD detection.
出处
《水电能源科学》
北大核心
2011年第8期29-30,186,共3页
Water Resources and Power
基金
广西自然科学基金资助项目(桂科自0832064)
广西教育厅桂教科研基金资助项目(200708MS067)
关键词
化学需氧量
近红外光谱
偏最小二乘法
模型
主成分数
chemical oxygen demand
near infrared spectrum
partial least squares
model
number of principal components
