Download or read book A Machine Learning Methodology for Developing Microscopic Vehicular Fuel Consumption and Emission Models for Local Conditions Using Real world Measures written by Ehsan Moradi and published by . This book was released on 2021 with total page pages. Available in PDF, EPUB and Kindle. Book excerpt: "Road transport is a major contributor to world energy consumption and emissions. The validity of models developed for environmental assessment of transport projects when used out of their origins is questionable as they are only validated for the prevailing conditions at their origin. This study starts by the validation of one of the most popular transportation environmental assessment models, MOVES, for use in non-U.S. regions such as Canada through performing on-road measurements. Distinct differences between the ground-truth and MOVES predictions are revealed. MOVES underestimates fuel and CO2 rates by 17% and 35%, respectively. Nitrogen Oxides (NOx) and Particulate Matters (PM) predictions set overestimation records of up to +420%. Furthermore, MOVES output is biased for vehicle groups with specific attributes. The results of MOVES validation emphasized the need for using alternative local fuel and emission models. However, many of the existing vehicular fuel and emission modeling methodologies are criticized in aspects such as ignoring real-world training data, low diversity of test fleet, impracticality in real-world applications (such as instrument-independent eco-driving or use alongside with traffic microsimulation), and low prediction power in the non-linear multi-dimensional space of fuel consumption and emission generation. Hence, a machine learning modeling methodology relying on on-road data from a fleet of 35 vehicles is proposed. The accuracy of the proposed instrument-independent models is tried to be improved by introducing estimates of influential engine variables to the feature set through a cascaded modeling procedure. As a result, the R-squared metric reached 83%, while score improvements as high as 37% are achieved depending on the vehicle class and the machine learning technique used.Despite the considerable scores achieved by utilizing fully-connected neural networks architectures, use of techniques compatible with the serially-correlated nature of vehicular operation seems more promising in achieving higher accuracy and robustness. Moreover, generalizing the models developed for particular vehicles to more aggregate levels is a need for diversifying models’ use cases. To this end, a two-stage ensemble learning methodology based on vehicle-specific Recurrent Neural Network (RNN) models is proposed.Long Short-Term Memory (LSTM) cell architecture resulted in the best lag-specific modeling scores (compared to the other RNN cell types). Vehicle-specific ensemble models developed by combining predictions from lag-specific RNN models showed score improvement records of up to 28% compared to the best component model (4% on average). In addition, the category-specific ensembles developed on top of metamodels achieved score improvements of up to 32% compared to the best component metamodel (6% on average). Linear regression dominantly resulted in the best score improvements for NOx and PM rates at both forecast combination stages, while random forests and gradient boosting methods dominantly worked the best for fuel and CO2 rates"--

Download or read book An Efficient Soft Computing Based Method for Calibration of Vehicular Microscopic Simulation Models written by Hamed Shahrokhi Shahraki and published by . This book was released on 2014 with total page 87 pages. Available in PDF, EPUB and Kindle. Book excerpt: In recent years, due to the advances in computation technology, microscopic vehicular traffic simulation has become one of the main tools used by transportation professionals to solve various design and analysis problems (e.g. safety performance evaluation of highways, impact of different design scenarios in units of safety and efficiency, etc.). The effective use of any of the existing simulation models is limited by the calibration of specific parameters that are based on observed real-life conditions. However, because the calibration of the simulation models is a time consuming and resource intensive process, one might resort to using the default parameter values. In this study, a soft computing-based methodology which synergistically combines Artificial Neural Networks and Genetic Algorithm (GA) applications, is proposed as an alternative for calibration methodology that considerably reduces the computation time in comparison to other commonly used methods. First, a Latin Hypercube Sampling method is used to select representative sets of values for VISSIM’s main calibration parameters. Second, the effect of each set of parameter values on the simulated traffic stream speed is recorded. Third, a neural-network is trained to determine the relationship between the input parameter values and the output vehicular speed. Finally, a genetic-algorithm uses the trained neural-network in its fitness function to determine the appropriate set of values for the calibration parameters. The proposed methodology allows for the calibration of microscopic traffic models with fewer computational resources than is commonly used. The feasibility of the method and its applicability to real-world traffic conditions is proved by employing the model using a real-world High Occupancy Vehicle (HOV) lane along a freeway segment. The results of proposed calibration method are compared with those from GA-only based calibration method.. It is concluded that the proposed method performs faster than the GA based calibration method while maintinaing a certain level of accuracy. To highlight the potential benefits of the proposed calibration method, a before-and-after calibration conflict analysis is presented. It is recommended to apply the proposed method to urban environments and to consider other performance measures (travel time, queue length, etc.) to investigate proposed method’s generality.

Download or read book Emission estimation based on traffic models and measurements written by Nikolaos Tsanakas and published by Linköping University Electronic Press. This book was released on 2019-04-24 with total page 131 pages. Available in PDF, EPUB and Kindle. Book excerpt: Traffic congestion increases travel times, but also results in higher energy usage and vehicular emissions. To evaluate the impact of traffic emissions on environment and human health, the accurate estimation of their rates and location is required. Traffic emission models can be used for estimating emissions, providing emission factors in grams per vehicle and kilometre. Emission factors are defined for specific traffic situations, and traffic data is necessary in order to determine these traffic situations along a traffic network. The required traffic data, which consists of average speed and flow, can be obtained either from traffic models or sensor measurements. In large urban areas, the collection of cross-sectional data from stationary sensors is a costefficient method of deriving traffic data for emission modelling. However, the traditional approaches of extrapolating this data in time and space may not accurately capture the variations of the traffic variables when congestion is high, affecting the emission estimation. Static transportation planning models, commonly used for the evaluation of infrastructure investments and policy changes, constitute an alternative efficient method of estimating the traffic data. Nevertheless, their static nature may result in an inaccurate estimation of dynamic traffic variables, such as the location of congestion, having a direct impact on emission estimation. Congestion is strongly correlated with increased emission rates, and since emissions have location specific effects, the location of congestion becomes a crucial aspect. Therefore, the derivation of traffic data for emission modelling usually relies on the simplified, traditional approaches. The aim of this thesis is to identify, quantify and finally reduce the potential errors that these traditional approaches introduce in an emission estimation analysis. According to our main findings, traditional approaches may be sufficient for analysing pollutants with global effects such as CO2, or for large-scale emission modelling applications such as emission inventories. However, for more temporally and spatially sensitive applications, such as dispersion and exposure modelling, a more detailed approach is needed. In case of cross-sectional measurements, we suggest and evaluate the use of a more detailed, but computationally more expensive, data extrapolation approach. Additionally, considering the inabilities of static models, we propose and evaluate the post-processing of their results, by applying quasi-dynamic network loading.

Download or read book Bulletin of the Atomic Scientists written by and published by . This book was released on 1961-05 with total page 88 pages. Available in PDF, EPUB and Kindle. Book excerpt: The Bulletin of the Atomic Scientists is the premier public resource on scientific and technological developments that impact global security. Founded by Manhattan Project Scientists, the Bulletin's iconic "Doomsday Clock" stimulates solutions for a safer world.

Download or read book Vehicle Emission Prediction Using Remote Sensing Data and Machine Learning Techniques written by Jiazhen Chen and published by . This book was released on 2018 with total page 80 pages. Available in PDF, EPUB and Kindle. Book excerpt: More researchers are using remote sensing technology to measure real-world, on-road automobile emissions of nitric oxide (NO), one of the most important and frequently studied pollutants. Partnered with the National Institute of Water and Atmospheric Research (NIWA) in New Zealand, we aim to establish a robust NO emission factor prediction model using remote sensing data to forecast future emissions. We have conducted this research using real-world data that were collected over a 11-year span between 2005 and 2015. The experimental results have shown that the vehicle emission patterns are continuously changing and the relevance of remote sensing data for future predictions decays as they get older. We propose a three-step machine learning approach to establish this model. We use quantile regression forest (QRF) as the base algorithm and use random forests variable importance measure to validate and interpret the features. We have found empirically, the model is more accurate than models that are based on three other algorithms: linear regression, linear model based recursive partitioning, random forest. Lastly, we have extracted human-interpretable prediction rules from our quantile regression forest based model, using the decision based rule extraction algorithm. The rules are useful to generalise prediction logic from a black-box model such as our QRF based model.

Download or read book Application of Machine Learning Models in Agricultural and Meteorological Sciences written by Mohammad Ehteram and published by Springer Nature. This book was released on 2023-03-21 with total page 201 pages. Available in PDF, EPUB and Kindle. Book excerpt: This book is a comprehensive guide for agricultural and meteorological predictions. It presents advanced models for predicting target variables. The different details and conceptions in the modelling process are explained in this book. The models of the current book help better agriculture and irrigation management. The models of the current book are valuable for meteorological organizations. Meteorological and agricultural variables can be accurately estimated with this book's advanced models. Modelers, researchers, farmers, students, and scholars can use the new optimization algorithms and evolutionary machine learning to better plan and manage agriculture fields. Water companies and universities can use this book to develop agricultural and meteorological sciences. The details of the modeling process are explained in this book for modelers. Also this book introduces new and advanced models for predicting hydrological variables. Predicting hydrological variables help water resource planning and management. These models can monitor droughts to avoid water shortage. And this contents can be related to SDG6, clean water and sanitation. The book explains how modelers use evolutionary algorithms to develop machine learning models. The book presents the uncertainty concept in the modeling process. New methods are presented for comparing machine learning models in this book. Models presented in this book can be applied in different fields. Effective strategies are presented for agricultural and water management. The models presented in the book can be applied worldwide and used in any region of the world. The models of the current books are new and advanced. Also, the new optimization algorithms of the current book can be used for solving different and complex problems. This book can be used as a comprehensive handbook in the agricultural and meteorological sciences. This book explains the different levels of the modeling process for scholars.

Download or read book Large Scale Machine Learning in the Earth Sciences written by Ashok N. Srivastava and published by CRC Press. This book was released on 2017-08-01 with total page 314 pages. Available in PDF, EPUB and Kindle. Book excerpt: From the Foreword: "While large-scale machine learning and data mining have greatly impacted a range of commercial applications, their use in the field of Earth sciences is still in the early stages. This book, edited by Ashok Srivastava, Ramakrishna Nemani, and Karsten Steinhaeuser, serves as an outstanding resource for anyone interested in the opportunities and challenges for the machine learning community in analyzing these data sets to answer questions of urgent societal interest...I hope that this book will inspire more computer scientists to focus on environmental applications, and Earth scientists to seek collaborations with researchers in machine learning and data mining to advance the frontiers in Earth sciences." --Vipin Kumar, University of Minnesota Large-Scale Machine Learning in the Earth Sciences provides researchers and practitioners with a broad overview of some of the key challenges in the intersection of Earth science, computer science, statistics, and related fields. It explores a wide range of topics and provides a compilation of recent research in the application of machine learning in the field of Earth Science. Making predictions based on observational data is a theme of the book, and the book includes chapters on the use of network science to understand and discover teleconnections in extreme climate and weather events, as well as using structured estimation in high dimensions. The use of ensemble machine learning models to combine predictions of global climate models using information from spatial and temporal patterns is also explored. The second part of the book features a discussion on statistical downscaling in climate with state-of-the-art scalable machine learning, as well as an overview of methods to understand and predict the proliferation of biological species due to changes in environmental conditions. The problem of using large-scale machine learning to study the formation of tornadoes is also explored in depth. The last part of the book covers the use of deep learning algorithms to classify images that have very high resolution, as well as the unmixing of spectral signals in remote sensing images of land cover. The authors also apply long-tail distributions to geoscience resources, in the final chapter of the book.

Download or read book Real time Vehicle Emission Estimation Using Traffic Data written by Anjie Liu and published by . This book was released on 2019 with total page 142 pages. Available in PDF, EPUB and Kindle. Book excerpt: The current state of climate change should be addressed by all sectors that contribute to it. One of the major contributors is the transportation sector, which generates a quarter of greenhouse gas emissions in North America. Most of these transportation related emissions are from road vehicles; as result, how to manage and control traffic or vehicular emissions is therefore becoming a major concern for the governments, the public and the transportation authorities. One of the key requirements to emission management and control is the ability to quantify the magnitude of emissions by traffic of an existing or future network under specific road plans, designs and traffic management schemes. Unfortunately, vehicular traffic emissions are difficult to quantify or predict, which has led a significant number of efforts over the past decades to address this challenge. Three general methods have been proposed in literature. The first method is for determining the traffic emissions of an existing road network with the idea of measuring the tail-pipe emissions of individual vehicles directly. This approach, while most accurate, is costly and difficult to scale as it would require all vehicles being equipped with tail-pipe emission sensors. The second approach is applying ambient pollutant sensors to measure the emissions generated by the traffic near the sensors. This method is only approximate as the vehicle-generated emissions can easily be confounded by other nearby emitters and weather and environmental conditions. Note that both of these methods are measurement-based and can only be used to evaluate the existing conditions (e.g., after a traffic project is implemented), which means that it cannot be used for evaluating alternative transportation projects at the planning stage. The last method is model-based with the idea of developing models that can be used to estimate traffic emissions. The emission models in this method link the amount of emissions being generated by a group of vehicles to their operations details as well as other influencing factors such as weather, fuel and road geometry. This last method is the most scalable, both spatially and temporally, and also most flexible as it can meet the needs of both monitoring (using field data) and prediction. Typically, traffic emissions are modelled on a macroscopic scale based on the distance travelled by vehicles and their average speeds. However, for traffic management applications, a model of higher granularity would be preferred so that impacts of different traffic control schemes can be captured. Furthermore, recent advances in vehicle detection technology has significantly increased the spatiotemporal resolutions of traffic data. For example, video-based vehicle detection can provide more details about vehicle movements and vehicle types than previous methods like inductive loop detection. Using such detection data, the vehicle movements, referred to as trajectories, can be determined on a second-by-second basis. These vehicle trajectories can then be used to estimate the emissions produced by the vehicles. In this research, we have proposed a new approach that can be used to estimate traffic generated emissions in real time using high resolution traffic data. The essential component of the proposed emission estimation method is the process to reconstruct vehicle trajectories based on available data and some assumptions on the expected vehicle motions including cruising, acceleration and deceleration, and car-following. The reconstructed trajectories containing instantaneous speed and acceleration data are then used to estimate emissions using the MOVES emission simulator. Furthermore, a simplified rate-based module was developed to replace the MOVES software for direct emission calculation, leading to significant improvement in the computational efficiency of the proposed method. The proposed method was tested in a simulated environment using the well-known traffic simulator - Vissim. In the Vissim model, the traffic activities, signal timing, and vehicle detection were simulated and both the original vehicle trajectories and detection data recorded. To evaluate the proposed method, two sets of emission estimates are compared: the "ground truth" set of estimates comes from the originally simulated vehicle trajectories, and the set from trajectories reconstructed using the detection data. Results show that the performance of the proposed method depends on many factors, such as traffic volumes, the placement of detectors, and which greenhouse gas is being estimated. Sensitivity analyses were performed to see whether the proposed method is sufficiently sensitive to the impacts of traffic control schemes. The results from the sensitivity analyses indicate that the proposed method can capture impacts of signal timing changes and signal coordination but is insufficiently sensitive to speed limit changes. Further research is recommended to validate the proposed method using field studies. Another recommendation, which falls outside of this area of research, would be to investigate the feasibility of equipping vehicles with devices that can record their instantaneous fuel consumption and location data. With this information, traffic controllers would be better informed for emission estimation than they would be with only detection data.

Download or read book Automotive Model Predictive Control written by Luigi Del Re and published by Springer. This book was released on 2010-03-11 with total page 291 pages. Available in PDF, EPUB and Kindle. Book excerpt: Automotive control has developed over the decades from an auxiliary te- nology to a key element without which the actual performances, emission, safety and consumption targets could not be met. Accordingly, automotive control has been increasing its authority and responsibility – at the price of complexity and di?cult tuning. The progressive evolution has been mainly ledby speci?capplicationsandshorttermtargets,withthe consequencethat automotive control is to a very large extent more heuristic than systematic. Product requirements are still increasing and new challenges are coming from potentially huge markets like India and China, and against this ba- ground there is wide consensus both in the industry and academia that the current state is not satisfactory. Model-based control could be an approach to improve performance while reducing development and tuning times and possibly costs. Model predictive control is a kind of model-based control design approach which has experienced a growing success since the middle of the 1980s for “slow” complex plants, in particular of the chemical and process industry. In the last decades, severaldevelopments haveallowedusing these methods also for “fast”systemsandthis hassupporteda growinginterestinitsusealsofor automotive applications, with several promising results reported. Still there is no consensus on whether model predictive control with its high requi- ments on model quality and on computational power is a sensible choice for automotive control.

Download or read book Machine Learning and Deep Learning for Modeling and Control of Internal Combustion Engines written by Armin Norouzi Yengeje and published by . This book was released on 2022 with total page 0 pages. Available in PDF, EPUB and Kindle. Book excerpt: Internal Combustion Engines (ICEs) are ubiquitous; they power a wide range of systems. The broad use of ICEs globally causes more than 20% of the total greenhouse gas emissions. In many countries, emission legislation is transitioning from certification using only traditional chassis dynomometer testing to now requiring the inclusion of Real Driving Emissions (RDE). Complying with this legislation has led to increased challenges to meet emissions levels under on-road use of the engine. The stringent legislation governing emissions and fuel economy, in combination with the complexity of the combustion process, have led to requirements for significantly more advanced engine controllers than are currently used. Reducing the emissions of diesel engines while simultaneously increasing their thermal efficiency through online control optimization and Machine Learning (ML) are the main objectives of this thesis. ML techniques offer powerful solutions that help to address the existing challenges in ICE modeling, control, and optimization. ML can also help to reduce the time, cost, and effort required for ICE calibration for both vehicular and stationary applications. In this thesis, a four-cylinder medium-duty Cummins diesel engine and emission measurement system including an electrochemical fast Nitrogen Oxides (NOx) sensor, Pegasor Particle Sensor (PPS-M), and MKS Fourier-Transform Infrared Spectroscopy (FTIR) are used for experimental implementation. A dSPACE MicroAutoBox II, which is a rapid prototyping system, is used for control implementation. In order to compare the proposed control method with the existing Cummins calibrated engine control unit (ECU), all the production calibration tables are imported to the MicroAutoBox. The simulation results presented in this thesis are developed using a detailed physics-based model using the GT-power\(^{\copyright}\) software. A co-simulation of GT-power\(^{\copyright}\)/Matlab\(^{\copyright}\)/Simulink is used as an Engine Simulation Model (ESM). The application of ML in engine control can be divided into three main categories: i) ML in emission prediction, ii) Integration of ML and Model Predictive Control (MPC), and iii) ML in the learning-based controller. In the first category, a correlation-based order reduction algorithm is developed to model \nox, resulting in a simple and accurate model. This algorithm utilizes Support Vector Machine (SVM) techniques to predict \nox~emission with high accuracy. In addition, a comprehensive study involving eight ML methods and five feature sets is done for Particulate Matter (PM) modeling using gray-box techniques. Then using the K-means clustering algorithm, a systematic way to select the best method for a specific application is proposed. In the second category, two methods of combining ML and MPC were used: ML-based modeling and ML imitation control. First, ML is used to identify a model for implementation in MPC optimization problems. Additionally, ML can be used to replace MPC, where the ML controller learns the optimal control action by imitating the behavior of the MPC. Using the ESM to provide simulation data, SVM and deep recurrent neural networks, including long-short-term memory (LSTM) layers, are used to develop engine performance and emission models. Then based on these models, MPC is designed and compared to both a linear controller and the Cummins' calibrated ECU model in ESM. Then, a deep learning scheme is deployed to imitate the behavior of the developed controllers. These imitative controllers behave similarly to the online optimization of MPC but require significantly lower computational time. The LSTM-based MPC is then implemented on the real-time system using open-source software. Compared to the stock Cummins ECU, this controller has significant emission reduction, fuel economy improvement, and thermal efficiency. Reinforcement Learning (RL) and Iterative Learning controller (ILC) are developed to investigate learning-based controllers. Using the ESM, a model-free off-policy algorithm, Deep Deterministic Policy Gradient (DDPG), is developed. A safety filter is added to the deep RL to avoid damage to the engine. This filter guarantees output and input constraints for both RL and ILC. The developed safe RL is then compared with ILC and LSTM-NMPC.

Download or read book Prediction of Mobile Source Emissions and Fuel Consumption Using the TEXAS Model written by Tsai-Yun Liao and published by . This book was released on 1995 with total page 39 pages. Available in PDF, EPUB and Kindle. Book excerpt:

Download or read book Microscopic Vehicle Emission Modelling written by Hajar Hajmohammadi Hosseinabadi and published by . This book was released on 2019 with total page 188 pages. Available in PDF, EPUB and Kindle. Book excerpt: Vehicle emission models are widely used to estimate air pollution from road transport. This estimation can then be considered for transport management and traffic control policies, to quantify their impacts on urban air quality. The focus of this study is to investigate the relationship between vehicle dynamics and tailpipe emission by statistical methods. These methods are: log- polynomial and classified log-polynomial model based on acceleration and deceleration, lagged regression and transfer function model based on time series analysis, gear-based emission model based on estimated transmission gear components, and the general additive model for location, scale and shape (GAMLSS) based on spline functions. The dataset for this study is second-by-second emission laboratory measurements of four different vehicle types while following a driving cycle recorded in urban, suburban and motorway areas of London. The vehicles can be categorized by size (compact and saloon), fuel type (petrol and diesel) and transmission type (manual and automatic). For each vehicle type, CO2, CO and NOx emissions are estimated in each second of driving by the speed profile as the main explanatory variable. The six emission models developed in this study are: Log-polynomial (LP), classified log-polynomial (CLP), lagged regression (LR), transfer function (TF), gear-based and GAMLSS. These are evaluated using the BIC, total emission recovery and statistical time series analysis of the residuals. The GAMLSS model consistently has the best BIC values for all vehicle and emission types, while the recovery ratio of this model is within 1% for all vehicle types. In addition, statistical analysis of the ACF/PACF time series plots shows that the GAMLSS emission model is clearer from the significant lags compared to the parametric models (LP, TF, Gear-based, gear-based and CLP). Among the parametric models, the classified models represent the emission relationship better than others. The best BIC values (after GAMLSS) were achieved by the gear- based and the CLP emission models. These results indicate that the GAMLSS approach which uses spline functions and flexible error structure performs better than the other models investigated here. This model is validated by 10- fold cross-validation approach which shows that the prediction power of the GAMLSS emission model exceeds that of the parametric models. The models are evaluated by the BIC values, total emission recovery and analysis of the residuals. Based on these criteria, the GAMLSS emission model is the most effective, especially for CO and NOx emission modelling. This model is then validated by the K-fold cross-validation process. The suggestion for future research is to evaluate the performance of the developed models with track and real driving emission (RDE) tests. The calibrated model then will be implemented to a traffic microsimulation, where different transportation management and traffic policies can be simulated and evaluated by their impacts on air quality.

Download or read book Scientific Machine Learning for Dynamical Systems written by Abhinav (Mechanical engineering and computation expert) Gupta and published by . This book was released on 2022 with total page 0 pages. Available in PDF, EPUB and Kindle. Book excerpt: Complex dynamical models are used for prediction in many domains, and are useful to mitigate many of the grand challenges being faced by humanity, such as climate change, food security, and sustainability. However, because of computational costs, complexity of real-world phenomena, and limited understanding of the underlying processes involved, models are invariably approximate. The missing dynamics can manifest in the form of unresolved scales, inexact processes, or omitted variables; as the neglected and unresolved terms become important, the utility of model predictions diminishes. To address these challenges, we develop and apply novel scientific machine learning methods to learn unknown and discover missing dynamics in models of dynamical systems. In our Bayesian approach, we develop an innovative stochastic partial differential equation (PDE) - based model learning theory and framework for high-dimensional coupled biogeochemical-physical models. The framework only uses sparse observations to learn rigorously within and outside of the model space as well as in that of the states and parameters. It employs Dynamically Orthogonal (DO) differential equations for adaptive reduced-order stochastic evolution, and the Gaussian Mixture Model-DO (GMM-DO) filter for simultaneous nonlinear inference in the augmented space of state variables, parameters, and model equations. A first novelty is the Bayesian learning among compatible and embedded candidate models enabled by parameter estimation with special stochastic parameters. A second is the principled Bayesian discovery of new model functions empowered by stochastic piecewise polynomial approximation theory. Our new methodology not only seamlessly and rigorously discriminates between existing models, but also extrapolates out of the space of models to discover newer ones. In all cases, the results are generalizable and interpretable, and associated with probability distributions for all learned quantities. To showcase and quantify the learning performance, we complete both identical-twin and real-world data experiments in a multidisciplinary setting, for both filtering forward and smoothing backward in time. Motivated by active coastal ecosystems and fisheries, our identical-twin experiments consist of lower-trophic-level marine ecosystem and fish models in a two-dimensional idealized domain with flow past a seamount representing upwelling due to a sill or strait. Experiments have varying levels of complexities due to different learning objectives and flow and ecosystem dynamics. We find that even when the advection is chaotic or stochastic from uncertain nonhydrostatic variable-density Boussinesq flows, our framework successfully discriminates among existing ecosystem candidate models and discovers new ones in the absence of prior knowledge, along with simultaneous state and parameter estimation. Our framework demonstrates interdisciplinary learning and crucially provides probability distributions for each learned quantity including the learned model functions. In the real-world data experiments, we configure a one-dimensional coupled physical-biological-carbonate model to simulate the state conditions encountered by a research cruise in the Gulf of Maine region in August, 2012. Using the observed ocean acidification data, we learn and discover a salinity based forcing term for the total alkalinity (TA) equation to account for changes in TA due to advection of water masses of different salinity caused by precipitation, riverine input, and other oceanographic processes. Simultaneously, we also estimate the multidisciplinary states and an un- certain parameter. Additionally, we develop new theory and techniques to improve uncertainty quantification using the DO methodology in multidisciplinary settings, so as to accurately handle stochastic boundary conditions, complex geometries, and the advection terms, and to augment the DO subspace as and when needed to capture the effects of the truncated modes accurately. Further, we discuss mutual-information-based observation planning to determine what, when, and where to measure to best achieve our learning objectives in resource-constrained environments. Next, motivated by the presence of inherent delays in real-world systems and the Mori-Zwanzig formulation, we develop a novel delay-differential-equations-based deep learning framework to learn time-delayed closure parameterizations for missing dynamics. We find that our neural closure models increase the long-term predictive capabilities of existing models, and require smaller networks when using non-Markovian over Markovian closures. They efficiently represent truncated modes in reduced-order-models, capture effects of subgrid-scale processes, and augment the simplification of complex physical-biogeochemical models. To empower our neural closure models framework with generalizability and interpretability, we further develop neural partial delay differential equations theory that augments low-fidelity models in their original PDE forms with both Markovian and non-Markovian closure terms parameterized with neural networks (NNs). For the first time, the melding of low-fidelity model and NNs with time-delays in the continuous spatiotemporal space followed by numerical discretization automatically provides interpretability and allows for generalizability to computational grid resolution, boundary conditions, initial conditions, and problem specific parameters. We derive the adjoint equations in the continuous form, thus, allowing implementation of our new methods across differentiable and non-differentiable computational physics codes, different machine learning frame- works, and also non-uniformly-spaced spatiotemporal training data. We also show that there exists an optimal amount of past information to incorporate, and provide methodology to learn it from data during the training process. Computational advantages associated with our frameworks are analyzed and discussed. Applications of our new Bayesian learning and neural closure modeling are not limited to the shown fluid and ocean experiments, but can be extended to other fields such as control theory, robotics, pharmacokinetic-pharmacodynamics, chemistry, economics, and biological regulatory systems.

Download or read book Micro Scale On Road Vehicle Specific Emissions Measurement and Modeling written by and published by . This book was released on 2004 with total page pages. Available in PDF, EPUB and Kindle. Book excerpt: The main objectives of this work are to quantify and compare intra- and inter-vehicle variability in fuel use and emissions and to develop capabilities of measuring and estimating fuel use and emissions at the micro-scale. This dissertation developed methodology to achieve the objectives, including experimental design for on-road data collection using a portable emission measurement system (PEMS), road grade estimation, evaluation of measurement accuracy, quantification of intra- and inter-vehicle variability in emissions, and micro-scale emissions modeling. A Light Detection and Ranging (LIDAR)-based method for road grade estimation was shown to be accurate and reliable. Measurement accuracy on a trip or mode basis was shown to be adequate. Routes, drivers, road grade, and time of day are significant sources of intra-vehicle variability. Significant inter-vehicle variability in emissions was observed, although only a small number of vehicles were tested and all belong to the same vehicle class. Thus, for accurate emission inventory development, both intra- and inter-vehicle variability should be taken into account. Consecutive averages were used for micro-scale emissions modeling to account for the response time of the PEMS. Choice of averaging time determines the model spatial and temporal resolution of prediction. Models for all pollutants are generally accurate, and precise in fuel use and CO2 emission estimation and moderately precise for other pollutants for various averaging times. Furthermore, models are capable of capturing the micro-scale events in emissions. Thus, the modeling schemes developed here can be used for a variety of applications including identification of the hotspots in emissions, transportation improvement programs on a corridor or intersection level, and more representative and accurate regional emission inventories development.

Download or read book RouteE written by Jacob Holden and published by . This book was released on 2020 with total page 7 pages. Available in PDF, EPUB and Kindle. Book excerpt:

Download or read book An Empirical Comparison of Emissions Based on Vehicular Trajectories from Microscopic and Mesoscopic Simulation Models written by Jingjing Zang and published by . This book was released on 2013 with total page 69 pages. Available in PDF, EPUB and Kindle. Book excerpt: Following increasing public concerns, various policies have been implemented to reduce air pollution from the operation of motor vehicles. Unfortunately, estimating the effectiveness of these policies requires analyses that are often costly and time-consuming. The state-of-the-art approach for estimating vehicular emissions on a road network is to combine a vehicular microsimulation model, such as Paramics or TransModeler, with a microscopic emissions model such as EPA's MOVES. However, this approach has not yet been widely adopted because creating and calibrating microsimulation models of large networks is very time-consuming. A potential alternative to vehicular microsimulation is to rely on a mesoscopic traffic simulation model, but differences in air pollutant emissions between these two approaches are not yet well understood. In this context, this thesis contrasts vehicular emissions of mainstream air pollutants obtained by applying MOVES to results of both microscopic and mesoscopic traffic simulations in TransModeler. Traffic was simulated for 24 hours on a large network that extends between the San Pedro Bay Ports (Los Angeles and Long Beach) and downtown Los Angeles. Results show that for freeways near free-flow conditions, the difference in emission results obtained by combining microscopic and mesoscopic traffic simulation models with MOVES is under 10 percent, so mesoscopic simulation could replace microscopic simulation under these conditions. However, the difference between those two approaches can exceed 55% for congested arterial roads; in that case, microscopic traffic simulation should be clearly preferred for the evaluation of vehicular emissions.

Download or read book Modeling Size resolved Particle Number Emissions from Advanced Technology and Alternative Fueled Vehicles in Real operating Conditions written by Darrell Bruce Sonntag and published by . This book was released on 2010 with total page 0 pages. Available in PDF, EPUB and Kindle. Book excerpt: Two particle number emission datasets were analyzed in detail. The first data set contained particle number emissions from four transit buses, including two hybrid diesel-electric buses, under a variety of driving conditions and technological/fuel treatments including: diesel oxidation catalysts, diesel particle filters and ultra-low sulfur diesel fuel. A linear mixed model was used to control for multiple sources of variability in real-world particle measurements, and identified significant factors influencing particle number emissions. Subsequently, link-level particle number emission models were developed for the DOC-equipped conventional buses, using different sets of available predictive data. Principle component analysis was used to reduce the variability of engine parameters to three interpretable parameters: percent engine load, engine speed and exhaust temperature. Time-resolved particle emissions from the diesel transit buses were evaluated in detail to understand the relationship of particle emissions, operating modes, and the relationship among multiple pollutants. Particle number and mass emissions are generally well-correlated during real-world behavior, however number are emissions are more influenced by the storage and subsequent release of particles evident during high engine speeds, while particle mass emission are more consistent with fuel events. Acceleration events on a stop-and-go urban route caused the maximum particle emission rates at resolved spatial scales, while over large spatial scales the highest emission rates occurred on the freeway. The concept of emission modes was introduced to understand the variability of gaseous and particle pollution during transient operation of the transit bus. Six repeatable emission modes were identified as being capable of explaining more than 75% of the total variability in emissions. Functional data analysis was introduced to analyze particle size distributions collected on a flex-fuel vehicle. A non-parametric smoothing technique can optimally smooth particle size distribution data without imposing prior distributional assumptions. The relationship among particle concentrations, operation conditions, and fuel type was estimated as a function of particle size using a functional linear model. Future paths of research are identified which take into account the smoothness of particle-size distributions. In summary, this dissertation contributes data, understanding, and quantitative concepts and methods to advance both research and practice-oriented particle emission models.