基于支持向量机的汽油辛烷值预测研究

Research on gasoline octane number prediction based on support vector machine

针对工业汽油成品辛烷值含量难以实时获取的问题,提出一种辛烷值预测模型,对成品汽油辛烷值含量进行精确预测。首先以某一大型石化企业真实采集数据为基础,提出一种基于孤立森林的数据清洗方式,对原始数据中异常值及缺失值进行预测及填充;然后通过主成分分析法选取与辛烷值含量相关系数较高的36个特征变量基于支持向量机训练辛烷值含量预测模型,在原始模型基础上采用高斯核函数并采用学习曲线对模型最优参数进行选择。结果表明,改进后的模型决定系数为84.36 %,平均误差为0.169,可实现对汽油成品辛烷值的有效预测。

Aiming at the problem that it is difficult to obtain the octane number content of industrial gasoline in real time,an octane number prediction model is proposed to accurately predict the octane number content of finished gasoline.Firstly,based on the real data collected by a large petrochemical enterprise,a data cleaning method based on isolated forest is proposed to predict and fill in the abnormal and missing values in the original data;Then,36 characteristic variables with high correlation coefficient with octane number content are selected by principal component analysis,and the octane number content prediction model is trained based on support vector machine.On the basis of the original model,Gaussian kernel function and learning curve are used to select the optimal parameters of the model.The results show that decisive coefficient of the improved model is 84.36 %and the average error is 0.169 which can effectively predict the octane number of gasoline products.