Mingfei Chen , Jiaying He , Xuan Zhao , Runtian Yu , Kaixuan Yang , Dong Liu
{"title":"Combustion condition predictions for C2-C4 alkane and alkene fuels via machine learning methods","authors":"Mingfei Chen , Jiaying He , Xuan Zhao , Runtian Yu , Kaixuan Yang , Dong Liu","doi":"10.1016/j.fuel.2024.132375","DOIUrl":null,"url":null,"abstract":"<div><p>The accurate and rapid prediction of hydrocarbon type was a precondition for the utilization of fossil fuels with high efficiency and safety. In this study, machine learning based techniques were used to predict the type and equivalence ratio of flames of C<sub>2</sub>-C<sub>4</sub> alkane and alkene fuels based on the differences in flame morphology between various combustion conditions. The test results of different machine learning algorithms, including ANN, SVM, SVR, KNN, MLR, and RF were compared in detail using statistical methods. Results indicated that ANN, SVM, KNN, and RF all exhibited an outstanding performance in predicting the types of C<sub>2</sub>-C<sub>4</sub> alkane and alkene flames, achieving accuracies of 95.7 %, 96.3 %, 93.8 %, and 96.5 %, respectively. For the prediction of the equivalence ratio among these fuels, the mean absolute percentage errors of the ANN, SVR, MLR, and RF were only 5.6 %, 3.8 %, 8.2 %, and 3.8 %, respectively. The performance of SVM, SVR, and RF algorithms was significantly superior to that of ANN, MLR, and KNN algorithms for flame prediction. Moreover, the data of feature analysis revealed that the importance level of designed features exhibited a significant distinction between different prediction targets. For predicting the type of C<sub>2</sub>-C<sub>4</sub> alkane and alkene fuels, the features associated with blue region showed a stronger importance level. However, the yellow region related features played a more significant role for the prediction of the equivalence ratio.</p></div>","PeriodicalId":325,"journal":{"name":"Fuel","volume":null,"pages":null},"PeriodicalIF":6.7000,"publicationDate":"2024-07-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Fuel","FirstCategoryId":"5","ListUrlMain":"https://www.sciencedirect.com/science/article/pii/S0016236124015230","RegionNum":1,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q2","JCRName":"ENERGY & FUELS","Score":null,"Total":0}
引用次数: 0
Abstract
The accurate and rapid prediction of hydrocarbon type was a precondition for the utilization of fossil fuels with high efficiency and safety. In this study, machine learning based techniques were used to predict the type and equivalence ratio of flames of C2-C4 alkane and alkene fuels based on the differences in flame morphology between various combustion conditions. The test results of different machine learning algorithms, including ANN, SVM, SVR, KNN, MLR, and RF were compared in detail using statistical methods. Results indicated that ANN, SVM, KNN, and RF all exhibited an outstanding performance in predicting the types of C2-C4 alkane and alkene flames, achieving accuracies of 95.7 %, 96.3 %, 93.8 %, and 96.5 %, respectively. For the prediction of the equivalence ratio among these fuels, the mean absolute percentage errors of the ANN, SVR, MLR, and RF were only 5.6 %, 3.8 %, 8.2 %, and 3.8 %, respectively. The performance of SVM, SVR, and RF algorithms was significantly superior to that of ANN, MLR, and KNN algorithms for flame prediction. Moreover, the data of feature analysis revealed that the importance level of designed features exhibited a significant distinction between different prediction targets. For predicting the type of C2-C4 alkane and alkene fuels, the features associated with blue region showed a stronger importance level. However, the yellow region related features played a more significant role for the prediction of the equivalence ratio.
期刊介绍:
The exploration of energy sources remains a critical matter of study. For the past nine decades, fuel has consistently held the forefront in primary research efforts within the field of energy science. This area of investigation encompasses a wide range of subjects, with a particular emphasis on emerging concerns like environmental factors and pollution.