数据挖掘方法在汽油辛烷值损失计算中的应用

针对汽油清洁化中降低辛烷值的损失这一重点问题,提出了基于数据挖掘的辛烷值损失预测方法.首先,对影响辛烷值损失的各类因素进行了分析;然后以某石化企业为例,应用数据挖掘方法对其提供的数据进行有效的数据清洗;其次对多种复杂的影响因素进行合理的特征提取,成功提取出28个影响辛烷值损失特性的代表因素;接着利用如支持向量机回归、神...

Application of Data Mining Method in Calculating the Loss of Gasoline Octane Number

Aiming at the key problem of reducing the octane loss in gasoline cleaning, a data mining-based octane loss calculation method is proposed. Firstly, various factors affecting the octane loss were analyzed; then as an example a petrochemical company, the data provided by it was preprocessed reasonably and effectively through applying data mining methods. Secondly, feature extraction for multiple complex influencing factors was performed reasonably, and successfully extracted that 28 representative factors of affecting octane loss; Then mining modeling methods such as support vector machine regression, neural networks and random forests and cross-validation to train models were used to predict octane loss. Finally, through experiments and result analysis, it is shown that the random forest model based on data mining method can more accurately predict the octane loss, and has shown strong ability in feature extraction of influence factors and the prediction calculation on octane loss and can better serve the gasoline cleaning.