基于极限学习机的汽油辛烷值含量回归预测建模研究  

Regression Modeling Prediction of Octane Content in Gasoline Based on Extreme Learning Machine

作　　者：付学敏[1] 王辉[1] FU Xue-min;WANG Hui(Anhui Vocational College of Press and Publishing,Hefei 230601,Anhui Province,China)

机构地区：[1]安徽新闻出版职业技术学院,合肥230601

出　　处：《景德镇学院学报》2021年第3期73-76,共4页Journal of JingDeZhen University

基　　金：安徽省高校自然科学研究重点项目(KJ2018A0924)。

摘　　要：本文引入极限学习机(ELM),以解决汽油辛烷值含量建模计算问题。首先,对汽油辛烷值含量数据集进行预处理,随机选取出训练集与测试集;其次,引入ELM机器学习算法,建立汽油辛烷值含量的ELM回归计算模型,选取合适参数,并将训练集数据代入该模型进行训练,拟合出输入参数与输出参数之间的非线性关系;然后,使用测试集的真实值与ELM回归建模预测值进行对比,并选取确定系数(R^(2))作为精度校验指标,选取耗时作为计算效率校验指标,对ELM模型进行测试。最后与支持向量机BP神经网络等回归建模方法,进行对比实验,结果表明ELM方法在汽油率烷值含量回归建模预测方面具有较好效果。In this paper,Extreme Learning Machine(ELM)is introduced to solve the problem of calculating the octane content in gasoline by modeling.Firstly,the data set of the octane content in gasoline is pretreated and the training set and test set are selected randomly.Secondly,the ELM learning algorithm is introduced to establish the ELM regression calculation model of the octane content in gasoline,and appropriate parameters are selected.The training set data are substituted into the model for training,and the nonlinear relationship between the input parameters and output parameters are fitted.The real value of the test set is compared with the predicted value of ELM regression model in order to test the ELM model by selecting the determination coefficient(R^(2))as the index of checking accuracy,and consuming time as the index of checking calculation efficiency.Finally,comparing with the regression modeling methods such as support vector machine and BP neural network,we find that ELM is efficacious in regression modeling prediction of octane content in gasoline.

关 键 词：极限学习机 汽油辛烷值含量 回归建模 支持向量机 BP神经网络 

分 类 号：TE626.21[石油与天然气工程—油气加工工程]