Hybrid Phenomenological and Mathematical-Based Modeling Approach for Diesel Emission Prediction

Rezaei, Reza; Hayduk, Christopher; Alkan, Emre; Kemski, Thomas; Delebinski, Thaddaeus; Bertram, Christoph

doi:10.4271/2020-01-0660

Cited by 8 publications

(13 citation statements)

References 15 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Results showed that the prediction error is generally larger for soot emissions in comparison with NOx emissions. The same trend observed in our previous works [16,26]. The gray-box emission modeling for a wide range of emissions was investigated in [16].…”

Section: Introductionsupporting

confidence: 67%

“…The same trend observed in our previous works [16,26]. The gray-box emission modeling for a wide range of emissions was investigated in [16]. A physical model was used, and different data-driven algorithms with fixed input feature sets for different emissions were used.…”

Section: Introductionmentioning

confidence: 80%

“…The complexity of soot formation and oxidation processes [2] makes it difficult for soot emissions formation and oxidation process to be adequately represented by 1D physical models [2]. Modelbased studies for soot emissions prediction show the same trend [16], and have motivated the data-driven methods of soot emissions prediction. According to Table 5, GPR and SVM are the most accurate methods for this data set.…”

Section: Resultsmentioning

confidence: 99%

“…In addition, soot emissions are more accurately modeled with SVM in comparison with ANN. In both of these studies [16,26], input feature sets have not been analysed and only physical knowledge about the emissions formation and oxidation process were used to choose the fixed input feature sets for emission modeling. Using physical knowledge to select the input feature set, some of the crucial parameters might be missed because our physical knowledge about soot emissions is not complete.…”

Section: Introductionmentioning

confidence: 99%

“…An alternative way for choosing the input feature set is using ML feature selection methods which was the main focus of our last study [23], where a new gray-box mechanism and black-box emission model for a different diesel engine was developed. Compared to the previous studies [16,26], a new platform is introduced in this work in terms of the number of applied ML methods (RT, ERT, SVM, ANN and BNN methods are tested), and a new feature selection process (LASSO). Additionally, more advanced algorithms (including Bayesian and grid search) are used to optimize the hyperparameters of the ML methods, which resulted in improved performance.…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

Hybrid Machine Learning Approaches and a Systematic Model Selection Process for Predicting Soot Emissions in Compression Ignition Engines

et al. 2021

Self Cite

View full text Add to dashboard Cite

The standards for emissions from diesel engines are becoming more stringent and accurate emission modeling is crucial in order to control the engine to meet these standards. Soot emissions are formed through a complex process and are challenging to model. A comprehensive analysis of diesel engine soot emissions modeling for control applications is presented in this paper. Physical, black-box, and gray-box models are developed for soot emissions prediction. Additionally, different feature sets based on the least absolute shrinkage and selection operator (LASSO) feature selection method and physical knowledge are examined to develop computationally efficient soot models with good precision. The physical model is a virtual engine modeled in GT-Power software that is parameterized using a portion of experimental data. Different machine learning methods, including Regression Tree (RT), Ensemble of Regression Trees (ERT), Support Vector Machines (SVM), Gaussian Process Regression (GPR), Artificial Neural Network (ANN), and Bayesian Neural Network (BNN) are used to develop the black-box models. The gray-box models include a combination of the physical and black-box models. A total of five feature sets and eight different machine learning methods are tested. An analysis of the accuracy, training time and test time of the models is performed using the K-means clustering algorithm. It provides a systematic way for categorizing the feature sets and methods based on their performance and selecting the best method for a specific application. According to the analysis, the black-box model consisting of GPR and feature selection by LASSO shows the best performance with test R2 of 0.96. The best gray-box model consists of SVM-based method with physical insight feature set along with LASSO for feature selection with test R2 of 0.97.

show abstract

Section: Introductionsupporting

confidence: 67%