基于PLS的火电二氧化硫减排因素分析研究

Analysis of SO_2 from Thermal Power Plant based on PLS

以四川省为例,以火电二氧化硫排放量作为因变量,对不同时间、不同地区的样本采样,采用偏最小二乘回归分析方法,建立自变量的偏最小二乘回归模型。该模型综合了多元回归分析、主成分分析及典型相关分析,很好地解决了由变量多重相关性、样本点少于自变量个数等引起的模型预测精度不高的问题,同时也避免了使用普通最小二乘回归方法所引起的模型回归系数异常的问题。结果显示,所建模型解释能力强、预测精度高,能为相关部门制定二氧化硫减排措施提供科学的依据。

This paper takes Sichuan Province as an example. By taking thermal SO2 discharge as the dependent variable, the model of PLS of independent variables was establish through using samples at different time in different areas. The model synthesizes multiple regression analysis, principal component analysis and canonical correlation analysis, which well solves the problems of low accuracy of the model prediction that is caused by variable multiple correlations, and of the sample points less than the independent number and so on. Meanwhile, it avoids the problem of abnormality of the model coefficients which is caused by using the ordinary PLS methods. The results shows that the model established has strong explaining ability and high prediction accuracy, which provides scientific foundation for the relative departments to lay down SO2 discharge measures.