SEARCH · Engineering Papers

Results for “efficiency improvement”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

SourceResearch connectionFrom yearThrough year

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 1,207 records · Page 67

All sources NASA NTRS DOE OSTI

DOE OSTI2024

A Step toward Process Electrification: Microwave-Assisted Co-Gasification of Waste Plastics and Biomass to Produce Clean Energy

Barely 9% of the 353 million metric tons of plastic waste produced worldwide over the last two decades has been recycled, with the majority of it being either landfilled or incinerated. Recycling and upcycling plastic waste is thus essential in mitigating environmental pollution. Gasification is the thermochemical conversion of solid carbon feedstock at high temperatures using a gasifying agent to produce hydrogen-rich syngas. Existing gasification technologies are energy-intensive and require high operating temperatures (>1000 °C), leaving them uncommercial. Microwave heating has sparked attention in a variety of chemical processes due to its specific advantages over traditional heating technologies, such as rapid and selective heating to improve process efficiency. NETL’s research team recently investigated the synergistic advantages of corn stover and mixed plastics in producing hydrogen-rich syngas via microwave heating. As a continuation of the research, this study provides an outlook over the non-catalytic versus catalytic effects of plastic-corn stover gasification via process electrification to enhance H2 production, minimize undesired tar, and improve process efficiency. Our aim is to provide complete electrification of the existing gasification process by using waste plastics as a starting block to produce valuable fuel and support the nation’s net zero goals.

Abedin, Ashraf↗

DOE OSTI2024

Transitioning from Simulation to Reality: Applying Chatter Detection Models to Real-World Machining Data

Chatter, a self-excited vibration phenomenon, is a critical challenge in high-speed machining operations, affecting tool life, product surface quality, and overall process efficiency. While machine learning models trained on simulated data have shown promise in detecting chatter, their real-world applicability remains uncertain due to discrepancies between simulated and actual machining environments. The primary goal of this study is to bridge the gap between simulation-based machine learning models and real-world applications by developing and validating a Random Forest-based chatter detection system. This research focuses on improving manufacturing efficiency through reliable chatter detection by integrating Operational Modal Analysis (OMA), Receptance Coupling Substructure Analysis (RCSA), and Transfer Learning (TL). The study applies a Random Forest classification model trained on over 140,000 simulated machining datasets, incorporating techniques like Operational Modal Analysis (OMA), Receptance Coupling Substructure Analysis (RCSA), and Transfer Learning (TL) to adapt the model for real-world operational data. The model is validated against 1600 real-world machining datasets, achieving an accuracy of 86.1%, with strong precision and recall scores. The results demonstrate the model’s robustness and potential for practical implementation in industrial settings, highlighting challenges such as sensor noise and variability in machining conditions. This work advances the use of predictive analytics in machining processes, offering a data-driven solution to improve manufacturing efficiency through more reliable chatter detection.

42 ENGINEERING↗

NASA NTRSDate not supplied

A New BRDF Hotspot Model and its Use in Improving Radiative Transfer Model Efficiency and Application to VLIDORT-based PCRTM Model

The term “hotspot” refers to the sharp increase of reflectance occurring when incident (solar) and reflected (viewing) directions almost coincide in the backscatter direction. The accurate simulation of hotspot directional signatures is important for many remote sensing applications. The RossThick-LiSparse-Reciprocal (RTLSR) Bidirectional Reflectance Distribution Function (BRDF) model is widely used in radiative transfer simulations, and the hotspot model mostly used is from Maignan- Bréon but it typically requires large values of numerical quadrature and Fourier expansion terms in order to represent the hotspot accurately. To improve its use in atmospheric radiative transfer (RT) model simulations, we have developed a modified version based on the Maignan-Bréon’s hotspot BRDF model that converge much faster numerically, making it more practical for use in RT models that require Fourier expansion of BRDF to simulate the top-of-atmosphere (TOA) hotspot signatures. On basis of the vector linearized discrete ordinate radiative transfer model (VLIDORT), we have built a line-by-line based simulation system that can simulate the TOA radiance for atmospheres with 26 gases, different aerosols, dust and clouds, and surface with different BRDF. Using simulated results under diverse atmospheric conditions, we have developed VLIDORT-PCRTM model.

Xiaozhen Xiong↗

DOE OSTI2021

Optical Characterization of Defects in High-Efficiency (Ag,Cu)(In,Ga)Se2

We applied time-resolved photoluminescence (TRPL) spectroscopy to study optimized chalcopyrite (Ag,Cu)(In,Ga)Se2 thin films. The device shows power conversion efficiency of 18.7%. The metastable defect VSe-VCu within ACIGS at EV+0.98 eV is detected in sub-bandgap TRPL excitation spectra. TRPL lifetime of 50 ns is limited by the density of mid-gap defects such as CuGa or CuIn. The similarity of TRPL dynamics before and after light soaking indicates the optimized ACIGS thin film is less metastable because the density of VCu-VSe defect is reduced to below 10^15 cm^-3. This study indicates that ACIGS has improved cell efficiency and reliability characteristics.

41 EE - Solar Energy Technologies Office (EE-4S)↗

DOE OSTI2025

Machine Learning for Anomaly Detection in Neural Network Security and SRF Cavities

This dissertation explores the development and deployment of machine learning approaches to address critical challenges in anomaly detection across two distinct domains: neural network security in federated learning settings and cavity behavior analysis in particle accelerator operations at Jefferson Lab in Newport News, Virginia. Anomaly detection identifies deviations from expected patterns, safeguarding systems in cybersecurity, industry, and research against malicious activities and failures. This dissertation demonstrates how our machine learning approaches enhance detection accuracy and efficiency in both neural network security and industrial applications. First, we investigate vulnerabilities in deep neural networks deployed in federated learning. Although federated learning preserves user privacy by training models locally, it remains vulnerable to backdoor attacks, in which malicious participants embed hidden triggers that induce targeted misbehavior. We propose a self-supervised contrastive learning framework to detect and mitigate such backdoor attacks. In our experiments, this method achieves higher detection accuracy and lower false positive rates than existing defenses, while operating without access to local model updates or original training data and thus preserving the privacy guarantees of the federated setting. Second, we address the operational reliability of superconducting radio-frequency (SRF) cavities at the Continuous Electron Beam Accelerator Facility (CEBAF). Our research leverages an unsupervised learning approach, combined with Principal Component Analysis (PCA) and k-means clustering, to identify anomalous behaviors in SRF cavities. Our method detects subtle anomalous behavior by analyzing SRF signal data. This knowledge allows for the early detection and resolution of potential faults, significantly improving the efficiency and reliability of operations. Third, we extend these insights to time-series anomaly detection more broadly. We design a contrastive-learning based model tailored to increasingly dynamic environments and academic research. This model improves detection accuracy in settings that require real-time monitoring and predictive maintenance. Our research underscores the broader applicability and impact of advanced machine learning techniques in anomaly detection. By extracting meaningful patterns from complex data, machine learning can significantly enhance security in distributed neural networks and improve the efficiency of particle accelerator operations. This dissertation serves as a stepping stone for future investigations into the vast possibilities of anomaly detection, inspiring further exploration and development of machine learning techniques in this field.

Ferguson, Hal [Old Dominion University]↗

DOE OSTI2022

Porous condensers can double the efficiency of membrane distillation

Among membrane distillation (MD) configurations, air gap MD (AGMD) has superior energy efficiency at high salinity. However, flooding of the gap prevents optimal thin gap depths as water removal is a challenge. To address these challenges, this work provides the first analysis of efficiency enhancement via porous wicking conductive condensers in MD. To do so, we examined performance with experiments and comprehensive thermo-fluid numerical models; we find that the porous wick condenser significantly improves thermal efficiency, gain output ratio (GOR), and flux. MD experiments show that the porous copper foam improves the thermal efficiency of air gap membrane distillation up to 40 % and water flux up to 144 %. Numerical modeling shows the porous copper foam improves the GOR of membrane distillation in high salinity by 98.7 %, outperforming air gap MD, permeate gap MD, and conductive gap MD. Furthermore, the optimized thickness and permeability of porous condensers significantly help eliminate the risk of flooding in AGMD.

42 ENGINEERING↗

DOE OSTI2025

National User Resource for Biological Accelerator Mass Spectrometry (Final Report)

The National User Resource for Biological Accelerator Mass Spectrometry (User Resource) will provide isotopic analysis (primarily radiocarbon or 14C) by accelerator mass spectrometry (AMS) for NIH- funded researchers across the United States and will be the only User Resource of its type in the United States. The User Resource will provide measurement capability and expertise to a research community that requires highly sensitive, quantitative isotope analyses. Since commissioning a new accelerator mass spectrometer in June 2014, we have measured over 4000 samples a year for collaborators and service users. The User Resource will enable us to continue to meet these research needs, as well as provide for new users whose research programs would benefit from AMS as a measurement tool. The User Resource’s forte will be ultra-high sensitivity quantitation of radiocarbon and selected other radioisotopes for research studies where isotopes are required. Radioisotope labeling studies have been and will continue to be an important tool for addressing many complex biomedical science problems. AMS is a specialized and unique type of mass spectrometry that provides absolute quantitation of radiocarbon and other relevant radioisotopes with extreme sensitivity, having limits of detection in real samples on the order of a few attomol/mg of sample at measurement precisions of ~3%. It is the only instrumental method capable of quantifying radioisotope-labeled agents routinely in real-world samples with such precision and sensitivity. The sensitivity of AMS allows for the quantification of radiolabeled metabolites in extremely complex matrices of cells and organisms at very low concentrations and in small samples. AMS allows studies to be conducted without perturbing metabolism leading to more relevant quantification of metabolic rates and pathways. In addition, it enables quantification of pharmacokinetic and metabolic properties of toxicants at environmentally relevant concentrations in model systems as well as the ability to quantify pharmacokinetics and other molecular endpoints directly in humans. Such quantitative assessments can 1) improve risk assessment for toxicants, 2) address safety and efficacy considerations for therapeutic entities, 3) deepen understanding of xenobiotic and intermediary metabolism, 4) help understand the interactions between critical molecular pathways, and 5) improve efforts to model and predict various metabolic and biological states. These capabilities have been applied in a number of areas including research in carcinogenesis, toxicology, nutrition, pharmacology/drug development and basic biological science. As a NIGMS National Resource the National User Resource for Biological Accelerator Mass Spectrometry will help NIH funded scientists achieve a deeper understanding of the etiology of human health concerns by (1) enabling the quantification of pharmacokinetics and other molecular endpoints directly in humans; (2) offering the ability to conduct quantitative studies using biologics such as proteins or lipids, and thereby reducing the amount of radioisotope usage in biomedical labs; and (3) enabling more relevant studies of metabolic pathways in health and disease through the use of much lower, more biologically-relevant, concentrations of metabolic substrates in cells and intact organisms. Such studies support NIGMS’s basic biomedical research areas that contribute to the understanding of fundamental cellular and physiological principles and enable research supported by the Biophysics, Biomedical Technology, and Computational Biosciences (BBCB); Genetics and Molecular, Cellular, and Developmental Biology (GMCDB); Pharmacology, Physiology, Biological Chemistry (PPBC) and Training, Workforce Development, and Diversity (TWD) Divisions. Over the next five years, our goals are to: 1. Improve the efficiency of operation for AMS measurements through installation of new interfaces to our AMS systems, technical modifications to improve gas accepting ion source efficiency and upgrading our data analysis software for improved ease of use and data reporting. 2. Increase the accessibility and visibility of ultra-sensitive 14C measurements for the biomedical research community by training of new investigators and expanding our national user base. 3. Provide high throughput, ultra-sensitive 14C analysis for the NIGMS and NIH user community.

47 OTHER INSTRUMENTATION↗

DOE OSTI2021

Modeling Spectrally-Selective Reflection for Thermal Management in Monofacial and Bifacial Modules

Parasitic absorption in photovoltaic modules is a major source of waste heat, which drives operating temperatures 20-30K above ambient. Spectrally-selective sub-bandgap reflection can reduce parasitic absorption, thereby improving module efficiency and power output. Here, we investigate the performance of 1-D spectrally-selective mirrors in monofacial Al BSF and PERC modules, and bifacial PERC modules. In monofacial modules, these mirrors offer >1.2% increase in energy yield compared to single-layer anti-reflection coatings, while cooling by over 1K on average. Mirrors reduced bifacial module parasitic absorption by up to 34 W/m2 out of 1240 W/m2 incident.

41 EE - Solar Energy Technologies Office (EE-4S)↗

NASA NTRS2004

Numerical Modeling and Testing of an Inductively-Driven and High-Energy Pulsed Plasma Thrusters

Pulsed Plasma Thrusters (PPTs) are advanced electric space propulsion devices that are characterized by simplicity and robustness. They suffer, however, from low thrust efficiencies. This summer, two approaches to improve the thrust efficiency of PPTs will be investigated through both numerical modeling and experimental testing. The first approach, an inductively-driven PPT, uses a double-ignition circuit to fire two PPTs in succession. This effectively changes the PPTs configuration from an LRC circuit to an LR circuit. The LR circuit is expected to provide better impedance matching and improving the efficiency of the energy transfer to the plasma. An added benefit of the LR circuit is an exponential decay of the current, whereas a traditional PPT s under damped LRC circuit experiences the characteristic "ringing" of its current. The exponential decay may provide improved lifetime and sustained electromagnetic acceleration. The second approach, a high-energy PPT, is a traditional PPT with a variable size capacitor bank. This PPT will be simulated and tested at energy levels between 100 and 450 joules in order to investigate the relationship between efficiency and energy level. Arbitrary Coordinate Hydromagnetic (MACH2) code is used. The MACH2 code, designed by the Center for Plasma Theory and Computation at the Air Force Research Laboratory, has been used to gain insight into a variety of plasma problems, including electric plasma thrusters. The goals for this summer include numerical predictions of performance for both the inductively-driven PPT and high-energy PFT, experimental validation of the numerical models, and numerical optimization of the designs. These goals will be met through numerical and experimental investigation of the PPTs current waveforms, mass loss (or ablation), and impulse bit characteristics.

Parma, Brian↗

DOE OSTI2025

ExtremeMETA: High-speed Lightweight Image Segmentation Model by Remodeling Multi-channel Metamaterial Imagers

Deep neural networks (DNNs) have heavily relied on traditional computational units, such as CPUs and GPUs. However, this conventional approach brings significant computational burden, latency issues, and high power consumption, limiting their effectiveness. This has sparked the need for lightweight networks such as ExtremeC3Net. Meanwhile, there have been notable advancements in optical computational units, particularly with metamaterials, offering the exciting prospect of energy-efficient neural networks operating at the speed of light. Yet, the digital design of metamaterial neural networks (MNNs) faces precision, noise, and bandwidth challenges, limiting their application to intuitive tasks and low-resolution images. In this study, we proposed a large kernel lightweight segmentation model, ExtremeMETA. Based on ExtremeC3Net, our proposed model, ExtremeMETA maximized the ability of the first convolution layer by exploring a larger convolution kernel and multiple processing paths. With the large kernel convolution model, we extended the optic neural network application boundary to the segmentation task. To further lighten the computation burden of the digital processing part, a set of model compression methods was applied to improve model efficiency in the inference stage. The experimental results on three publicly available datasets demonstrated that the optimized efficient design improved segmentation performance from 92.45 to 95.97 on mIoU while reducing computational FLOPs from 461.07 MMacs to 166.03 MMacs. The large kernel lightweight model ExtremeMETA showcased the hybrid design’s ability on complex tasks.

large convolution kernel↗

DOE OSTI2022

Efficient high-fidelity TRISO statistical failure analysis using Bison: Applications to AGR-2 irradiation testing

The ability of tri-structural isotropic (TRISO) fuel to contain fission products is largely dictated by the quality of the manufacturing process, since most of the fission product release is expected to occur due to coating layer failure in a small number of particles containing defects. The Bison fuel performance code has capabilities to predict failure in individual particles, accounting for the presence of defects, and to apply statistical analysis methods to compute the probability of failure in a set of fuel particles. Bison has recently undergone significant development both to improve its physical representations of fuel particle behavior and to improve the efficiency of its statistical failure calculations. Physical model improvements include new capabilities to account for the pressure generated by fission gases on inner pyrolytic carbon (IPyC) crack surfaces and to use local material coordinate orientation to accurately incorporate the anisotropy in the material properties in aspherical particles. To improve statistical modeling efficiency, a direct integration approach which involves directly integrating the failure probability function associated with statistically varying parameters has been developed. The direct integration approach is much more efficient than the Monte Carlo (MC) schemes commonly employed, and allows Bison to directly run high-dimensional fuel performance models, which improves the accuracy of failure probability calculations. Finally, a set of benchmark problems is considered here to compare the MC and direct integration approaches, and a statistical failure analysis of compacts in the Advanced Gas Reactor (AGR)-2 experiments is performed using the direct integration approach.

36 MATERIALS SCIENCE↗

DOE OSTI2021

Improving the accuracy and efficiency of quantum connected moments expansions

The still-maturing noisy intermediate-scale quantum technology faces strict limitations on the algorithms that can be implemented efficiently. In the realm of quantum chemistry, the variational quantum eigensolver (VQE) algorithm has become ubiquitous, with many variations. Alternatively, a promising new avenue has been unraveled by the quantum variants of techniques grounded on expansions of the moments of the Hamiltonian, notably the connected moments expansion (CMX) and the Peeters–Devreese–Soldatov (PDS) energy functional. Common to those approaches is that, upon preparing an approximate ground state used to compute the necessary moments, the accuracy of the estimated ground state energy depends on the degree of overlap between the prepared state and the true ground state. Thus, we use the ADAPT-VQE algorithm to test shallow circuit construction strategies for the purpose of increasing the overlap with the exact ground state, validated by the sizable accuracy improvement herein reported in the PDS and CMX ground state energies. We also show that we can take advantage of the fact that the terms to be measured are highly recurring in different moments, incurring a substantial reduction in the number of necessary measurements. Finally, by coupling this measurement caching with a threshold that determines whether a given term is to be measured based on its associated scalar coefficient, we observe a further reduction in the number of circuit implementations while allowing for tunable accuracy.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

DOE OSTI2020

Rational design, synthesis, and evaluation of uncharged, “smart” bis-oxime antidotes of organophosphate-inhibited human acetylcholinesterase

Organophosphate (OP) intoxications from nerve agent and OP pesticide exposures are managed with pyridinium aldoxime–based therapies whose success rates are currently limited. The pyridinium cation hampers uptake of OPs into the central nervous system (CNS). Furthermore, it frequently binds to aromatic residues of OP-inhibited acetylcholinesterase (AChE) in orientations that are nonproductive for AChE reactivation, and the structural diversity of OPs impedes efficient reactivation. Improvements of OP antidotes need to include much better access of AChE reactivators to the CNS and optimized orientation of the antidotes' nucleophile within the AChE active-center gorge. On the basis of X-ray structures of a CNS-penetrating reactivator, monoxime RS194B, reversibly bound to native and venomous agent X (VX)–inhibited human AChE, here we created seven uncharged acetamido bis-oximes as candidate antidotes. Both oxime groups in these bis-oximes were attached to the same central, saturated heterocyclic core. Diverse protonation of the heterocyclic amines and oxime groups of the bis-oximes resulted in equilibration among up to 16 distinct ionization forms, including uncharged forms capable of diffusing into the CNS and multiple zwitterionic forms optimal for reactivation reactions. Conformationally diverse zwitterions that could act as structural antidote variants significantly improved in vitro reactivation of diverse OP-human AChE conjugates. Oxime group reorientation of one of the bis-oximes, forcing it to point into the active center for reactivation, was confirmed by X-ray structural analysis. Finally, our findings provide detailed structure-activity properties of several CNS-directed, uncharged aliphatic bis-oximes holding promise for use as protonation-dependent, conformationally adaptive, “smart” accelerated antidotes against OP toxicity.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

DOE OSTI2020

A 150-kW 99% Efficient All-Silicon-Carbide Triple-Active-Bridge Converter for Solar-Plus-Storage Systems

Solar-plus-storage systems could effectively mitigate the uncertainties of the photovoltaic (PV) generation and improve system reliability by adding an integrated battery energy storage system. As a three-port bidirectional isolated dc–dc converter with soft-switching capability, the triple-active-bridge (TAB) converter inherently matches the requirements of the solar-plus-storage system. However, challenges still remain in the TAB converter design to further improve system efficiency. In this article, the detailed design, implementation, and demonstration for a silicon carbide (SiC) 150-kW TAB converter are presented. Starting from a brief review of the TAB converter, the modulation scheme, power characteristics, and soft-switching region are analyzed. Then, the detailed design of the H-bridge converter building block is given. Here, to improve the system efficiency, a comprehensive characterization of the SiC gate driver with various external gate resistances is performed to address tradeoffs between switching loss and voltage overshoot during transients, as well as the thermal performance of the H-bridge building block. In addition, the design and characterization of the 20-kHz three-port transformer are also given. Comprehensive experimental studies are conducted on a full-power prototype to verify the proposed design. With a measured 99.1% peak efficiency, the proposed TAB converter can fulfill the requirements for solar-plus-storage applications.

14 SOLAR ENERGY↗

DOE OSTI2025

Macro-micro multiscale modeling to assist the design of HPDC Al castings microstructure and alloys for EV super-large body structures (Phase 1)

99 GENERAL AND MISCELLANEOUS↗

DOE OSTI2025

Macro-micro multiscale modeling to assist the design of HPDC Al castings microstructure and alloys for EV super-large body structures (Phase 1)

Implementation of High Pressure Die Casting (HPDC) Aluminum (Al) body structures for high volume electrified vehicles (EV) to improve electric efficiency remains a key strategy within many original equipment manufacturer (OEM)s. In addition to high strength for safety requirements, superior Self-Piercing Riveting (SPR) performance is demanded for HPDC Al alloys to be compatible with high volume SPR joining. In this work, it is proposed to extend and validate an existing Contractor finite element multiscale macro-micro modeling approach to quantify the influence of the microstructure of HPDC alloys on the fracture strain/displacement under 3-point bend and clinch testing. The success of this work will allow to replace solution treatment stage with low energy consumption heat treatment (HT) processes, or to design new non heat treatable (NHT) HPDC Al alloys to eliminate HT requirements. Ultimately, this project will facilitate the application of HPDC Al alloys for super-large vehicle structures to significantly reduce vehicle weight, and thus improving energy efficiency. The purpose of this project is to extend and validate an existing finite element code, which is based on the Contractor developed macro-micro multi-scale modeling approach, to numerically simulate the three-point bending and clinch test and study the influences of material microstructural characteristics and phase properties on the rivetability. The macro-micro modeling approach begins with a sample scale model and identify the location, which is mostly prone to failure, the deformation history of the boundaries of that location calculated will be used to drive a microstructure-based sub-models where the material microstructure and microscale properties are considered. Using this approach, the wrap-bending failure for two Al alloys are correctly predicted for the first time. This will start with phase I effort of building a framework of macro-micro three point bending test and clinch test of Al10SiMgMn HPDC alloy in the as-cast and T7 heat treated conditions. Those results will then be validated with experimental test results. The phase II effort will involve the utilization of the knowledge learned in phase I to establish the quantitative correlation between the microstructure characteristics and the riveting performance, which will be further used to guide the optimization of HPDC Al alloy microstructure using heat treatment process to achieve sufficient rivetability to join large thin-wall HPDC alloys.

36 MATERIALS SCIENCE↗

DOE OSTI2023

Air quality and comfort constrained energy efficient operation of multi-zone buildings

Maintaining indoor air quality (IAQ) through effective ventilation is essential for the well-being and productivity of building occupants. Control strategies aimed at improving the efficiency of heating, ventilation and air conditioning (HVAC) systems must jointly determine ventilation and heating and cooling processes. Here, in this paper, we study the problem of minimizing the energy consumption of the HVAC system in a multi-zone building, while meeting thermal comfort and IAQ requirements. We first perform a steady state analysis of the zonal carbon dioxide (CO 2 ) concentration and the temperature dynamics. The resulting expressions are convex in the zonal mass flow rates and zonal temperatures. Guided by the steady state solutions for meeting the thermal comfort constraints, we develop two control policies for improving the energy efficiency of building HVAC systems while jointly satisfying indoor temperature and IAQ constraints. We compare the performance of our proposed approaches with those of multiple baseline approaches which implement separate regimes for controlling zonal temperature and IAQ for a typical work-day in a multi-zone campus building. We have evaluated the performance of our proposed approaches under varying levels of flexibility in zonal temperatures. We have shown that zonal temperature flexibility can result in energy savings up to 32% (for the same control strategies) as compared to the case where no such flexibility is permitted. Our proposed approaches were seen to offer potential savings of nearly 29% compared to the baseline.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗

DOE OSTI2024

Implications of stop-and-go traffic on training learning-based car-following control

Learning-based car-following control (LCC) of connected and autonomous vehicles (CAVs) is gaining significant attention with the advancement of computing power and data accessibility. While the flexibility and large model capacity of model-free architecture enable LCC to potentially outperform the model-based car-following (CF) model in improving traffic efficiency and mitigating congestion, the generalizability of LCC for traffic conditions different from the training environment/dataset is not well-understood. Herein, this study seeks to explore the impact of stop-and-go traffic in the training dataset on the generalizability of LCC. It uses the characteristics of lead vehicle trajectories to describe stop-and-go traffic, and links the theory of identifiability (i.e., obtaining a unique parameter estimation result using sensor measurements) to the generalizability of behavior cloning (BC) and policy-based deep reinforcement learning (DRL). Correspondingly, the study shows theoretically that: (i) stop-and-go traffic can enable the property of identifiability and enhance the control performance of BC-based LCC in different traffic conditions; (ii) stop-and-go traffic is not necessary for DRL-based LCC to generalize to different traffic conditions; (iii) DRL-based LCC trained with only constant-speed lead vehicle trajectories (not sufficient to ensure identifiability) can be generalized to different traffic conditions; and (iv) stop-and-go traffic increases variance in the training dataset, which improves the convergence of parameter estimation while negatively impacting the convergence of DRL to the optimal control policy. Numerical experiments validate the above findings, illustrating that BC-based LCC entails comprehensive training datasets for generalizing to different traffic conditions, while DRL-based LCC can achieve generalization with simple free-flow traffic training environments. This further suggests DRL as a more promising and cost-effective LCC approach to reduce operational costs, mitigate traffic congestion, and enhance safety and mobility, which can accelerate the deployment and acceptance of CAVs.

33 ADVANCED PROPULSION SYSTEMS↗