Exploring the potential of machine learning in reducing the computational time/expense and improving the reliability of engine optimization studies

2018 ◽  
Vol 21 (7) ◽  
pp. 1251-1270 ◽  
Author(s):  
Chaitanya Kavuri ◽  
Sage L Kokjohn
Keyword(s):  
Machine Learning ◽  
Fluid Dynamics ◽  
Genetic Algorithm ◽  
Computational Fluid Dynamics ◽  
Regression Model ◽  
Gaussian Process ◽  
Gaussian Process Regression ◽  
Computational Time ◽  
Dynamics Modeling ◽  
Initial Dataset

Past research has shown that multidimensional computational fluid dynamics modeling in combination with a genetic algorithm method is an effective approach for optimizing internal combustion engine design. However, optimization studies performed with a detailed computational fluid dynamics model are time intensive, which limits the practical application of this approach. This study addresses this issue by using a machine learning approach called Gaussian process regression in combination with computational fluid dynamics modeling to reduce the computational optimization time. An approach was proposed where the Gaussian process regression model could be used instead of the computational fluid dynamics model to predict the outputs of the genetic algorithm optimization. In this approach, for every nth generation of the genetic algorithm, the data from the previous n − 1 generations was used to train the Gaussian process regression model. The approach was tested on an engine optimization study with five input parameters. When the genetic algorithm was run solely with computational fluid dynamics, the optimization took 50 days to complete. In comparison with the computational fluid dynamics and Gaussian process regression approach, the computational time was reduced by 62%, and the optimization was completed in 19 days using the same amount of computational resources. Additional parametric studies were performed to investigate the impact of genetic algorithm + Gaussian process regression parameters. Results showed that either reducing the initial dataset size or relaxing the error criterion resulted in increased Gaussian process regression evaluations within the genetic algorithm. However, relaxing the error criterion was found to impact the model predictions negatively. The initial dataset size was found to have a negligible impact on the final optimum design. Finally, the potential of machine learning in further improving the optimization process was explored by using the Gaussian process regression model to check for the robustness of the designs to operating parameter variations during the optimization. The genetic algorithm was repeated with the modified procedure and it was shown that adding the stability check resulted in a different, more reliable and stable optimum solution.

scholarly journals Computational analysis of airflow dynamics for predicting collapsible sites in the upper airways: machine learning approach

2019 ◽  
Vol 127 (4) ◽  
pp. 959-973 ◽  
Author(s):  
Seung Ho Yeom ◽  
Ji Sung Na ◽  
Hwi-Dong Jung ◽  
Hyung-Ju Cho ◽  
Yoon Jeong Choi ◽  
...  
Keyword(s):  
Machine Learning ◽  
Fluid Dynamics ◽  
Computational Fluid Dynamics ◽  
Obstructive Sleep Apnea ◽  
Sleep Apnea ◽  
Gaussian Process ◽  
Gaussian Process Regression ◽  
Support Vector ◽  
Data Set ◽  
Obstructive Sleep

Obstructive sleep apnea (OSA) is a common sleep breathing disorder. With the use of computational fluid dynamics (CFD), this study provides a quantitative standard for accurate diagnosis and effective surgery based on the investigation of the relationship between airway geometry and aerodynamic characteristics. Based on computed tomography data from patients having normal geometry, 4 major geometric parameters were selected and a total of 160 idealized cases were modeled and simulated. We created a predictive model using Gaussian process regression (GPR) through a data set obtained through numerical method. The results demonstrated that the mean accuracy of the overall GPR model was ~72% with respect to the CFD results for the realistic upper airway model. A support vector machine model was also used to identify the degree of OSA symptoms in patients as normal-mild and moderate and severe. We achieved an accuracy of 82.5% with the training data set and an accuracy of 80% with the test data set. NEW & NOTEWORTHY There have been many studies on the analysis of obstructive sleep apnea (OSA) through computational fluid dynamics and finite element analysis. However, these methods are not useful for practical medical applications because they have limited information for OSA symptom. This study employs the machine learning algorithm to predict flow characteristics quickly and to determine the symptoms of the patient's OSA. The overall Gaussian process regression model's mean accuracy was ~72%, and the accuracy for the classification of OSA was >80%.

Computational Fluid Dynamics Modeling of a Self-Recuperative Burner and Development of a Simplified Equivalent Radiative Model

2012 ◽  
Vol 134 (12) ◽  
Author(s):  
Haytham Sayah ◽  
Maroun Nemer ◽  
Wassim Nehmé ◽  
Denis Clodic
Keyword(s):  
Heat Transfer ◽  
Fluid Dynamics ◽  
Computational Fluid Dynamics ◽  
Computational Time ◽  
Dynamics Modeling ◽  
Gas Combustion ◽  
Experimental Database ◽  
Transfer Mode ◽  
High Temperature Processes ◽  
Burner Model

The solution for dynamic modeling of reheating furnaces requires a burner model, which is simultaneously accurate and fast. Based on the fact that radiative heat transfer is the most dominant heat transfer mode in high-temperature processes, the present study develops a simplified flame representation model that can be used for dynamic simulation of heat transfer in reheating furnaces. The first part of the paper investigates, experimentally and computationally, gas combustion in an industrial burner. Experiments aim at establishing an experimental database of the burner characteristics. This database is compared with numerical simulations in order to establish a numerical model for the burner. The numerical burner model was solved using a commercial computational fluid dynamics (CFD) software (FLUENT 6.3.26). A selection of results is presented, highlighting the usefulness of CFD as a modeling tool for industrial scale burners. In the second part of the paper, a new approach called the “emissive volume approach” is established. This approach consists of replacing the burner flame by a number of emissive volumes that replicates the radiative effect of the flame. Comparisons with CFD results show a difference smaller than 1% is achieved with the emissive volume approach, while computational time is divided by 40.

scholarly journals Application of an automated machine learning-genetic algorithm (AutoML-GA) coupled with computational fluid dynamics simulations for rapid engine design optimization

2021 ◽  
pp. 146808742110234
Author(s):  
Opeoluwa Owoyele ◽  
Pinaki Pal ◽  
Alvaro Vidal Torreira ◽  
Daniel Probst ◽  
Matthew Shaxted ◽  
...  
Keyword(s):  
Machine Learning ◽  
Fluid Dynamics ◽  
Genetic Algorithm ◽  
Computational Fluid Dynamics ◽  
Design Optimization ◽  
Internal Combustion Engines ◽  
Bayesian Optimization ◽  
Cfd Simulations ◽  
Engine Design ◽  
Design Optimum

In recent years, the use of machine learning-based surrogate models for computational fluid dynamics (CFD) simulations has emerged as a promising technique for reducing the computational cost associated with engine design optimization. However, such methods still suffer from drawbacks. One main disadvantage is that the default machine learning (ML) hyperparameters are often severely suboptimal for a given problem. This has often been addressed by manually trying out different hyperparameter settings, but this solution is ineffective in case of a high-dimensional hyperparameter space. Besides this problem, the amount of data needed for training is also not known a priori. In response to these issues that need to be addressed, the present work describes and validates an automated active learning approach, AutoML-GA, for surrogate-based optimization of internal combustion engines. In this approach, a Bayesian optimization technique is used to find the best machine learning hyperparameters based on an initial dataset obtained from a small number of CFD simulations. Subsequently, a genetic algorithm is employed to locate the design optimum on the ML surrogate surface. In the vicinity of the design optimum, the solution is refined by repeatedly running CFD simulations at the projected optima and adding the newly obtained data to the training dataset. It is demonstrated that AutoML-GA leads to a better optimum with a lower number of CFD simulations, compared to the use of default hyperparameters. The proposed framework offers the advantage of being a more hands-off approach that can be readily utilized by researchers and engineers in industry who do not have extensive machine learning expertise.

Exchange Spin Coupling from Gaussian Process Regression

2020 ◽  
Author(s):  
Marc Philipp Bahlke ◽  
Natnael Mogos ◽  
Jonny Proppe ◽  
Carmen Herrmann
Keyword(s):  
Machine Learning ◽  
Gaussian Process ◽  
Gaussian Process Regression ◽  
Molecular Magnets ◽  
Molecular Structures ◽  
Spin Coupling ◽  
Structure Property ◽  
Data Set ◽  
Uncertainty Estimates

Heisenberg exchange spin coupling between metal centers is essential for describing and understanding the electronic structure of many molecular catalysts, metalloenzymes, and molecular magnets for potential application in information technology. We explore the machine-learnability of exchange spin coupling, which has not been studied yet. We employ Gaussian process regression since it can potentially deal with small training sets (as likely associated with the rather complex molecular structures required for exploring spin coupling) and since it provides uncertainty estimates (“error bars”) along with predicted values. We compare a range of descriptors and kernels for 257 small dicopper complexes and find that a simple descriptor based on chemical intuition, consisting only of copper-bridge angles and copper-copper distances, clearly outperforms several more sophisticated descriptors when it comes to extrapolating towards larger experimentally relevant complexes. Exchange spin coupling is similarly easy to learn as the polarizability, while learning dipole moments is much harder. The strength of the sophisticated descriptors lies in their ability to linearize structure-property relationships, to the point that a simple linear ridge regression performs just as well as the kernel-based machine-learning model for our small dicopper data set. The superior extrapolation performance of the simple descriptor is unique to exchange spin coupling, reinforcing the crucial role of choosing a suitable descriptor, and highlighting the interesting question of the role of chemical intuition vs. systematic or automated selection of features for machine learning in chemistry and material science.

scholarly journals Computational Fluid Dynamics Modeling of Rotating Annular VUV/UV Photoreactor for Water Treatment

Processes ◽  
2020 ◽  
Vol 9 (1) ◽  
pp. 79
Author(s):  
Minghan Luo ◽  
Wenjie Xu ◽  
Xiaorong Kang ◽  
Keqiang Ding ◽  
Taeseop Jeong
Keyword(s):  
Fluid Dynamics ◽  
Computational Fluid Dynamics ◽  
Water Treatment ◽  
Cfd Modeling ◽  
Flow Mode ◽  
Oh Radicals ◽  
Dynamics Modeling ◽  
Rotational Movement ◽  
Contaminant Degradation ◽  
Wide Range

The ultraviolet photochemical degradation process is widely recognized as a low-cost, environmentally friendly, and sustainable technology for water treatment. This study integrated computational fluid dynamics (CFD) and a photoreactive kinetic model to investigate the effects of flow characteristics on the contaminant degradation performance of a rotating annular photoreactor with a vacuum-UV (VUV)/UV process performed in continuous flow mode. The results demonstrated that the introduced fluid remained in intensive rotational movement inside the reactor for a wide range of inflow rates, and the rotational movement was enhanced with increasing influent speed within the studied velocity range. The CFD modeling results were consistent with the experimental abatement of methylene blue (MB), although the model slightly overestimated MB degradation because it did not fully account for the consumption of OH radicals from byproducts generated in the MB decomposition processes. The OH radical generation and contaminant degradation efficiency of the VUV/UV process showed strong correlation with the mixing level in a photoreactor, which confirmed the promising potential of the developed rotating annular VUV reactor in water treatment.

scholarly journals Thermal prediction of turbulent forced convection of nanofluid using computational fluid dynamics coupled genetic algorithm with fuzzy interface system

2021 ◽  
Vol 11 (1) ◽  
Author(s):  
Meisam Babanezhad ◽  
Iman Behroyan ◽  
Ali Taghvaie Nakhjiri ◽  
Mashallah Rezakazemi ◽  
Azam Marjani ◽  
...  
Keyword(s):  
Fluid Dynamics ◽  
Genetic Algorithm ◽  
Computational Fluid Dynamics ◽  
Kinetic Energy ◽  
Population Size ◽  
Turbulent Kinetic Energy ◽  
Forced Convection ◽  
Inlet Temperature ◽  
Fluid Temperature ◽  
Interface System

AbstractComputational fluid dynamics (CFD) simulating is a useful methodology for reduction of experiments and their associated costs. Although the CFD could predict all hydro-thermal parameters of fluid flows, the connections between such parameters with each other are impossible using this approach. Machine learning by the artificial intelligence (AI) algorithm has already shown the ability to intelligently record engineering data. However, there are no studies available to deeply investigate the implicit connections between the variables resulted from the CFD. The present investigation tries to conduct cooperation between the mechanistic CFD and the artificial algorithm. The genetic algorithm is combined with the fuzzy interface system (GAFIS). Turbulent forced convection of Al2O3/water nanofluid in a heated tube is simulated for inlet temperatures (i.e., 305, 310, 315, and 320 K). GAFIS learns nodes coordinates of the fluid, the inlet temperatures, and turbulent kinetic energy (TKE) as inputs. The fluid temperature is learned as output. The number of inputs, population size, and the component are checked for the best intelligence. Finally, at the best intelligence, a formula is developed to make a relationship between the output (i.e. nanofluid temperatures) and inputs (the coordinates of the nodes of the nanofluid, inlet temperature, and TKE). The results revealed that the GAFIS intelligence reaches the highest level when the input number, the population size, and the exponent are 5, 30, and 3, respectively. Adding the turbulent kinetic energy as the fifth input, the regression value increases from 0.95 to 0.98. This means that by considering the turbulent kinetic energy the GAFIS reaches a higher level of intelligence by distinguishing the more difference between the learned data. The CFD and GAFIS predicted the same values of the nanofluid temperature.

Artificial Intelligence, Machine Learning and Calculation of Intraocular Lens Power

2020 ◽  
Vol 237 (12) ◽  
pp. 1430-1437
Author(s):  
Achim Langenbucher ◽  
Nóra Szentmáry ◽  
Jascha Wendelstein ◽  
Peter Hoffmann
Keyword(s):  
Artificial Intelligence ◽  
Machine Learning ◽  
Regression Model ◽  
Prediction Error ◽  
Learning Algorithms ◽  
Anterior Chamber Depth ◽  
Gaussian Process Regression ◽  
Machine Learning Algorithms ◽  
Effect Sizes ◽  
Axial Position

Abstract Background and Purpose In the last decade, artificial intelligence and machine learning algorithms have been more and more established for the screening and detection of diseases and pathologies, as well as for describing interactions between measures where classical methods are too complex or fail. The purpose of this paper is to model the measured postoperative position of an intraocular lens implant after cataract surgery, based on preoperatively assessed biometric effect sizes using techniques of machine learning. Patients and Methods In this study, we enrolled 249 eyes of patients who underwent elective cataract surgery at Augenklinik Castrop-Rauxel. Eyes were measured preoperatively with the IOLMaster 700 (Carl Zeiss Meditec), as well as preoperatively and postoperatively with the Casia 2 OCT (Tomey). Based on preoperative effect sizes axial length, corneal thickness, internal anterior chamber depth, thickness of the crystalline lens, mean corneal radius and corneal diameter a selection of 17 machine learning algorithms were tested for prediction performance for calculation of internal anterior chamber depth (AQD_post) and axial position of equatorial plane of the lens in the pseudophakic eye (LEQ_post). Results The 17 machine learning algorithms (out of 4 families) varied in root mean squared/mean absolute prediction error between 0.187/0.139 mm and 0.255/0.204 mm (AQD_post) and 0.183/0.135 mm and 0.253/0.206 mm (LEQ_post), using 5-fold cross validation techniques. The Gaussian Process Regression Model using an exponential kernel showed the best performance in terms of root mean squared error for prediction of AQDpost and LEQpost. If the entire dataset is used (without splitting for training and validation data), comparison of a simple multivariate linear regression model vs. the algorithm with the best performance showed a root mean squared prediction error for AQD_post/LEQ_post with 0.188/0.187 mm vs. the best performance Gaussian Process Regression Model with 0.166/0.159 mm. Conclusion In this paper we wanted to show the principles of supervised machine learning applied to prediction of the measured physical postoperative axial position of the intraocular lenses. Based on our limited data pool and the algorithms used in our setting, the benefit of machine learning algorithms seems to be limited compared to a standard multivariate regression model.

Sign in / Sign up

Export Citation Format

Share Document