Search

The application of cross-dataset training in object detection tasks is complicated because the inconsistency in the category range across datasets transforms fully supervised learning into semi-supervised learning. To address this problem, recent studies focus on the generation of high-quality missing annotations. In this study, we first specify that it is not enough to generate high-quality annotations using a single model, which looks only once for annotations. Through detailed experimental analyses, we further conclude that hard-label training is conducive for generating high-recall annotations, whereas soft-label training tends to obtain high-precision annotations. Inspired by the aspects mentioned above, we propose a dynamic supervisor framework that updates the annotations multiple times through multiple-updated submodels trained using hard and soft labels. In the final generated annotations, recall and precision improve significantly through the integration of hard-label training with soft-label training. Extensive experiments conducted on various dataset combination settings support our analyses and demonstrate the superior performance of the proposed dynamic supervisor.

Boundary layers are omnipresent in fundamental kinetic experimental facilities and practical combustion engines, which can cause ambiguity and misleading results in kinetic target acquisition and even abnormal engine combustion. In this paper, using n-heptane as a representative large hydrocarbon fuel exhibiting pronounced low-temperature chemistry (LTC), two-dimensional numerical simulation is conducted to resolve the transient autoignition phenomena affected by a boundary layer. We focus on the ignition characteristics and the subsequent combustion mode evolution of a hot combustible mixture flowing over a colder flat plate in an isobaric environment. For cases with autoignition occurring within the boundary layer, similarity is observed in the first-stage ignition as manifested by a constant temperature at all locations. The first-stage ignition is found to be rarely affected by heat and radical loss within the boundary layer. While for the main ignition event, an obvious dependence of ignition process on boundary layer thickness is identified, where the thermal-chemical process exhibits similarity at locations with similar boundary layer thickness, and the main ignition tends to first occur within the boundary layer at the domain end and generates a C-shape reaction front. It is found that sequential spontaneous autoignition is the dominant subsequent combustion mode at high-pressure conditions. At low to intermediate pressures, auto-ignition assisted flame propagation is nevertheless the dominant mode for combustion evolution. This research identifies novel features of autoignition and the subsequent combustion mode evolution affected by a cold, fully developed boundary layer, and provides useful guidance to the interpretation of abnormal combustion and combustion mode evolution in boundary layer flows.

In direct-injection engines, the formation of fuel wall film on piston surface and liner wall is a primary cause for emissions of unburnt hydrocarbons and particulate matter. It is therefore important to investigate the behavior and effect of a wall fuel film under typical engine conditions. Following our previous study of wall film effects on flame propagation and quenching under constant thermodynamic conditions (Tao et al. IJER, 2018), more complexities rooted in real engine conditions have been considered in the current work, including two real engine in-cylinder thermodynamic trajectories occurring at catalyst warming (CW) and low-speed high-load (LSHL) conditions, varying fuel wall film thickness accounting for both vaporization and condensation, and the stagnation boundary layer flow over the wall film. To shed light on the role of wall heat transfer, parameter sweeping of wall temperature is conducted from 303 K to 363 K. Two representative wall film models using empirical vaporization rate and gas-liquid interface heat flux are compared with our numerical simulation. A good correlation between vapor boundary thickness and quenching distance has been found. Competition between enhanced vaporization due to tangential convection and aggravate condensation from elevated pressure is also demonstrated. The results lead to useful insights into the behavior of a wall film in real engine in-cylinder thermodynamic conditions and could be used to construct low dimensional empirical wall film models in three-dimensional engine combustion modeling.

The ignition delay times of toluene, n-heptane/iso-octane primary reference fuel blends, and toluene/n-heptane/iso-octane toluene primary reference fuel blends were studied in the miniature high repetition rate shock tube (HRRST). In the HRRST, the temperature and pressure change during the test time, complicating interpretation of ignition delay times. Two methods of chemical kinetic modeling were employed in order to account for the changing thermodynamic states. One was based on direct simulation using the pressure profile and the kinetic mechanism, and the other was based on a Livengood-Wu type approach using modeled constant state ignition delay times. Two different kinetic mechanisms were used in simulating the data. Both methods showed agreement with experimental data over a range of temperatures, pressures, and blends. While overall agreement was fair, agreement was poorer for one of the mechanisms for the primary reference fuel blends. An evaluation of the ignition delay iso-contours from these mechanisms for representative HRRST experiment thermodynamic trajectories revealed that the HRRST measured IDTs are controlled by the high temperature chemistry and are more sensitive to temperature. The different performance of mechanisms is therefore attributed to their different high temperature reactivity.

Substantial research effort has been spent on the development of advanced compression ignition (ACI) engine combustion strategies over the past decades, including homogeneous charge compression ignition (HCCI), reactivity controlled compression ignition (RCCI), and gasoline direct-injection compression-ignition (GDCI), etc. The behavior of gasoline-type fuels under compression ignition conditions has subsequently attracted extensive experimental and kinetic modeling interest. On the other hand, towards the development of future transportation fuels and engines, the evaluation of general fuel properties, instead of endless testing of specific fuels, should be the future research focus. In this study, the individual roles of the research and motor octane numbers (i.e., RON and MON) and fuel sensitivity (S) in characterizing the ignition performance of gasoline surrogates have been systematically investigated under a typical ACI engine condition using well-validated surrogate and kinetic models. The crank angle corresponding to 50% total heat release (CA50) was utilized as an indicator of the overall fuel reactivity, and iso-contours of CA50 were mapped out in the full engine operating domain characterized by the temperature and pressure at intake valve closing (IVC). By comparing the ignition performance of toluene primary reference fuel blends (TPRF) with the same RON, MON or S values, the distinctive effects and dominant ACI operating conditions of these fuel properties are clearly demonstrated. The role of equivalence ratio is also discussed by comparing the intrinsic stoichiometric condition of the standard octane rating tests and the lean mixture conditions required by ACI operation. The results show that octane sensitivity manifests itself in the high pressure low temperature operating regime for fuels with identical RON or MON, through different low temperature reactivities and heat release rates, while in low pressure high temperature conditions, combustion phasing is less sensitive to all fuel properties, including RON, MON and S. TPRFs with the same sensitivity show the largest variation of CA50 in the intermediate operating conditions, where the thermodynamic traces pass primarily through the ignition delay regime characterized by negative-temperature coefficient (NTC) behavior. Finally, by comparing the combustion phasing of five different gasoline surrogates with nearly identical RON and MON, potential compositional effects for gasoline fuels under ACI condition are further discussed via kinetic modeling. This study also demonstrates that the controlling fuel properties of gasoline-like fuels depend on ACI operating conditions (pressure and temperature trajectory), which should be carefully considered when constructing fuel metrics and comparing experimental results of ACI engines.

In the current work, we utilize a recently developed staged Livengood–Wu (L–W) correlation to evaluate and predict the two-stage ignition behaviors of different primary reference fuels (PRFs) under homogeneous reciprocating engine conditions, to demonstrate its capability in capturing detailed ignition kinetics in combustion simulation. Based on the understandings of the global and detailed kinetics of low-temperature chemistry, simplified Arrhenius-based global expressions were developed for both ignition stages, and a linear correlation was applied for the cool flame temperature rise at the end of the first stage, for PRFs with various octane number (ON) 0, 20, 60 and 87 under constant volume conditions. For their performance under homogeneous charge compression ignition (HCCI) conditions, it is shown that for the same engine speed, the inlet temperature range allowing for two-stage ignition shrinks with increased ON numbers, and eventually two-stage ignition turns into single-stage ignition with sufficiently high ON numbers. Based on the ignition database, the combustion phasing for all PRFs under a wide HCCI operation range are predicted with the staged L–W integral method with satisfactory accuracy, including the first- and second-stage ignition timing as well as the cool flame temperature rise. The sensitivities of each of the engine operation parameters are successfully captured, further demonstrating the applicability of the staged L–W method for realistic fuels with low temperature chemistry and its potential applications for engine combustion control. This study also provides an alternative approach of accommodating low temperature ignition kinetics in combustion simulation, especially useful for those under variable thermodynamic conditions.

hTid-1, a human DnaJ protein, is a novel cellular target for HTLV-1 Tax. Here, we show that hTid-1 represses NF-kappaB activity induced by Tax as well as other activators such as tumor necrosis factor alpha (TNFalpha) and Bcl10. hTid-1 specifically suppresses serine phosphorylation of IkappaBalpha by activated IkappaB kinase beta (IKKbeta), but the activities of other serine kinases including p38, ERK2, and JNK1 are not affected. The suppressive activity of hTid-1 on IKKbeta requires a functional J domain that mediates association with heat shock proteins and results in prolonging the half-life of the NF-kappaB inhibitors IkappaBalpha and IkappaBbeta. Collectively, our data suggest that hTid-1, in association with heat shock proteins, exerts a negative regulatory effect on the NF-kappaB activity induced by various extracellular and intracellular activators including HTLV-1 Tax.