Hybrid Physical and Machine Learning-Oriented Modeling Approach to Predict Emissions in a Diesel Compression Ignition Engine

Mohammad, Aran; Rezaei, Reza; Hayduk, Christopher; Delebinski, Thaddaeus; Shahpouri, Saeid; Shahbakhti, Mahdi

doi:10.4271/2021-01-0496

Cited by 10 publications

(17 citation statements)

References 22 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%

“…Therefore, 219 data points in Figure 2 covers most of the possible operating conditions. It is worth mentioning that for highway truck application, due to various driving cycles, 220 data points might not be sufficient as in the literature for such an application, more than 900 data points were used [26]. To analyze the main features of the diesel engine that play an important role both in soot emissions modeling, the histogram of them are plotted in Figure 3.…”

Section: Methodsmentioning

confidence: 99%

“…Although a more advanced ML method (ANN) was used in this study, there are still other ML methods that could be used for the data driven part. For gray-box and black-box emission modeling, ANN and SVM methods were trained with the selected features [26]. This study also showed that soot is a challenging emission to model.…”

Section: Introductionmentioning

confidence: 94%

“…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%

Section: Methodsmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 94%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 3 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

show abstract

Predictive modeling of energy‐related greenhouse gas emissions in Ghana towards a net‐zero future

Sokama‐Neuyam,

Amezah,

Adjei

et al. 2023

Greenhouse Gases

View full text Add to dashboard Cite

Ghana is determined to reduce greenhouse gas (GHG) emissions by at least 15% by 2030 and attain net‐zero emissions by 2070. However, like many developing countries, Ghana must utilize its limited resources effectively to actualize its climate goals. Currently, climate policies in the country are not driven by emission data, which has important implications on effective utilization of resources and selection of efficient mitigation techniques. We analyzed energy consumption and GHG emission data between 1990 and 2016 from Ghana's energy sector which is responsible for about 36% of the country's total emissions. Predictive models were then developed using machine learning to forecast energy related emissions up to 2030. Based on the analysis and projections, attainable data‐driven recommendations were proposed to direct climate policies in the country. We found that between 1990 and 2016, petroleum fuel consumption increased by about 64.5% and the corresponding GHG emissions increased by 303%. The projections suggests that by 2030, energy sector emissions could increase by 131% compared to 2016 levels. Transport sector emission is also projected to increase by a whopping 219% and fuel consumption could hit 6742 ktoe by 2030, which is about 106% increase from the 2016 benchmark. The findings from this work will direct policy for effective mitigation of GHG emissions in the country while ensuring effective utilization of climate resources to pursue its net‐zero targets. © 2023 Society of Chemical Industry and John Wiley & Sons, Ltd.

show abstract

Effect of hybrid nanoparticles MgO and Al₂O₃ added water‐in‐diesel emulsion fueled diesel engine using hybrid deep neural network‐based spotted hyena optimization

Patel¹,

Dhiman

2022

Heat Trans

View full text Add to dashboard Cite

The automotive sector is one of the top energy consumers globally compared with any other sector where oil plays the main role. According to http://statista.com (an online platform that displays the stats on oil consumption and reserves), in 2016, the demand for crude oil was about 85.3 million barrels per day worldwide. In 2019, the demand almost reached 100 barrels per day, and then it was reduced to 91 million per day during the 2020 global COVID‐19 crisis. However, it was predicted that the global crude oil requirements will exceed 100 million barrels per day by 2023 and will continue to raise. Also, geographically, not all countries worldwide have the oil reservoirs or the technology to extract oil from the reservoirs. Although the oil demand worldwide is increasing every year, researchers have estimated that there are only 47 years of oil left on the planet earth if the current oil consumption will raise this way. Researchers are looking for finding alternative fuels, such as electrochemical energy (electric vehicles), biodiesel, electricity, ethanol, hydrogen, natural gas, propane, emerging fuels, or the use of such fuels in the existing engines to minimize the use of nonrenewable energy resources. However, researchers are also looking more at fuels based on hybrid nanoparticles added to emulsified fuels. Due to the limitations of utilizing electrochemical energy or biofuels include high charging time, limited millage, harmful exhaust emissions, complicated production process, engine efficiency, the cost of fuel, and the advantages over hybrid nanoparticles added emulsified fuels than biofuels. Hence, after going through numerous researches, in this present experimentation, “hybrid nanoparticles (MgO and Al2O3) added water‐in‐diesel emulsion” are formulated to enhance the quality of emission and improve the performance of the compression ignition engine. The best combination of diesel, MgO, and Al2O3, a water‐in‐diesel (W/D) emulsion blend, is proposed. The results were further validated using deep neural network‐based spotted hyena optimization (DNN‐SHO) prediction and compared with traditional machine learning approaches artificial neural network (ANN), convolution neural network (CNN), regression‐based network (RBN), recurrent neural network (RNN), and DNN. As a result, the best proportions of the proposed nanoparticles added into the W/D bend are identified to be 10% W/D, 50 ppm of MgO, and 50 ppm of Al2O in terms of engine performance and emission characteristics brake thermal efficiency 30.7%, brake‐specific fuel consumption 0.29 kg/kW‐h, CO 0.027 vol%, NOx 855 ppm vol, and HC 12 ppm vol. Besides this, the DNN‐SHO‐based validated outcomes are in good agreement with the experimental values and outperformed other traditional approaches ANN, CNN, RBN, RNN, and DNN used in this study.

show abstract

Hybrid Physical and Machine Learning-Oriented Modeling Approach to Predict Emissions in a Diesel Compression Ignition Engine

Cited by 10 publications

References 22 publications

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

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

Predictive modeling of energy‐related greenhouse gas emissions in Ghana towards a net‐zero future

Effect of hybrid nanoparticles MgO and Al₂O₃ added water‐in‐diesel emulsion fueled diesel engine using hybrid deep neural network‐based spotted hyena optimization

Contact Info

Product

Resources

About

Hybrid Physical and Machine Learning-Oriented Modeling Approach to Predict Emissions in a Diesel Compression Ignition Engine

Cited by 10 publications

References 22 publications

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

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

Predictive modeling of energy‐related greenhouse gas emissions in Ghana towards a net‐zero future

Effect of hybrid nanoparticles MgO and Al2O3 added water‐in‐diesel emulsion fueled diesel engine using hybrid deep neural network‐based spotted hyena optimization

Contact Info

Product

Resources

About

Effect of hybrid nanoparticles MgO and Al₂O₃ added water‐in‐diesel emulsion fueled diesel engine using hybrid deep neural network‐based spotted hyena optimization