Search Results (92)

In the recommendation system (RS) literature, a distinction exists between studies dedicated to fully operational (known users/items) and cold-start (new users/items) RSs. The warm-up phase—the transition between the two—is not widely researched, despite evidence that attrition rates are the highest for users and content providers during such periods. RS formulations, particularly deep learning models, do not easily allow for a warm-up phase. Herein, we propose two independent and complementary models to increase RS performance during the warm-up phase. The models apply to any cold-start RS expressible as a function of all user features, item features, and existing users’ preferences for existing items. We demonstrate substantial improvements: Accuracy-oriented metrics improved by up to 14% compared with not handling warm-up explicitly. Non-accuracy-oriented metrics, including serendipity and fairness, improved by up to 12% compared with not handling warm-up explicitly. The improvements were independent of the cold-start RS algorithm. Additionally, this paper introduces a method of examining the performance metrics of an RS during the warm-up phase as a function of the number of user–item interactions. We discuss problems such as data leakage and temporal consistencies of training/testing—often neglected during the offline evaluation of RSs. Full article

Hyperparameter Optimization (HPO) is an important and challenging problem in machine learning. Traditional HPO methods require substantial evaluations to search for superior configurations. Recent Large Language Model (LLM)-based approaches leverage domain knowledge and few-shot learning proficiency to discover promising configurations with minimal human effort. However, the repetition issues causes LLMs to generate configurations similar to context examples, which may confine the optimization process to local regions. Moreover, since LLMs rely on the examples they generate for a few-shot learning, a self-reinforcing loop is formed, hindering LLMs from escaping local optima. In this work, we propose EvoContext, which aims to intentionally generate configurations that differ significantly from examples via external interventions and actively breaks the self-reinforcing effect for a more efficient approximation of the global optimum. Our EvoContext method involves two phases: (i) initial example generation through cold or warm starting and (ii) iterative optimization that integrates genetic operations for updating examples to enhance global exploration capabilities. Additionally, it employs LLMs in-context learning to generate configurations based on competitive examples for local refinement. Experiments on several real-world datasets show that EvoContext outperforms traditional and other LLM-driven approaches on HPO. Full article

Methanol, a renewable and sustainable fuel, provides an effective strategy for reducing greenhouse gas emissions when synthesized through carbon dioxide hydrogenation integrated with carbon capture technology. The incorporation of hydrogen into methanol-fueled engines enhances combustion efficiency, mitigating challenges such as pronounced cycle-to-cycle variations and cold-start difficulties. A simulation framework was developed using Python 3.13 and the Cantera 3.1.0 library to model the combustion system of a four-stroke spark-ignited (SI) methanol–hydrogen engine. This framework integrates a fractal turbulent combustion model with chemical reaction kinetics, complemented by early flame development and near-wall combustion models to address limitations during the initial and terminal combustion phases. The model was validated by using experimental data measured from a spark-ignited methanol engine. The effects of varying Hydrogen Energy Rates (HER) on engine power performance, combustion characteristics, and emissions (like formaldehyde and carbon monoxide) were subsequently analyzed under different operating loads, whilst the knock limit boundaries were established for different operational conditions. Findings demonstrate that increasing HER improves the engine power output and thermal efficiency, shortens the combustion duration, and reduces the formaldehyde and carbon monoxide emissions. Nevertheless, under high-load conditions, higher HER increases the knocking tendency, which constrains the maximum permissible HER decreasing from approximately 40% at 15% load to 20% at 100% load. The model has been developed into a Python library and will be open-sourced on Github. Full article

Forecasting demand for newly introduced products presents substantial challenges within high-mix, low-volume manufacturing contexts, primarily due to cold-start conditions and unpredictable order behavior. This research proposes the Dynamic Dual-Phase Forecasting Framework (DDPFF) that amalgamates machine learning-based classification, similarity-driven analogous forecasting, ARMA-based residual compensation, and statistical process control for adaptive model refinement. The framework underwent evaluation through five real-world case studies conducted by a Taiwanese semiconductor tray manufacturer, encompassing a variety of scenarios characterized by high volatility, seasonality, and structural drift. The results indicate that DDPFF consistently outperformed conventional ARIMA and analogous forecasting methodologies, yielding an average reduction of 35.7% in mean absolute error and a 41.8% enhancement in residual stability across all examined cases. In one representative instance, the forecast error decreased by 44.9% compared to established benchmarks. These findings underscore the framework’s resilience in cold-start situations and its capacity to adapt to evolving demand patterns, providing a viable solution for data-scarce and dynamic manufacturing environments. Full article

The real driving emissions of gasoline and diesel vehicles are significantly influenced by altitude, temperature, and starting conditions. In this study, the real driving emissions (RDEs) of gasoline and diesel vehicles compliant with China V standards were investigated under various conditions. The adaptability of RDE testing in China was evaluated by analyzing vehicle emissions at different altitudes, ambient temperatures, and starting conditions. The results show that, with increasing altitude, CO, NOx, and PN emissions generally exhibit a downward trend, particularly for gasoline vehicles, whose conformity factors remain well below the China VI limit. However, for China V diesel vehicles relying solely on EGR technology, NOx emissions significantly exceed China VI standards, indicating that EGR alone is insufficient to meet regulatory requirements. Temperature variations have little effect on the emissions of China V PFI gasoline vehicles, while diesel vehicles continue to exhibit excessive NOx emissions under varying temperatures. Although the cold-start phase generates substantial pollutant emissions, the EMROAD evaluation method excludes this phase, resulting in limited differences between cold- and hot-start emission results. Nevertheless, the inclusion of cold-start emissions should be considered in future RDE assessments. Full article

Regarding the issue of high dependency on odometry in the adaptive Monte Carlo localization (AMCL) algorithm, an improved AMCL algorithm based on the normal distributions transform (NDT) and extended Kalman filter (EKF) is proposed. A virtual motion model is introduced into the AMCL framework to enable pose updates even when the robot has not moved. NDT is used for point cloud matching to estimate virtual displacement and calculate virtual control quantities, which are then fed into the motion model to predict and update particle states when the robot has not moved. Additionally, to avoid the negative impacts of encoder errors and wheel slippage on motion state estimation, the EKF algorithm integrates information from the wheel odometer and inertial measurement unit to estimate the robot’s displacement, thereby improving localization accuracy and stability. The performance of the proposed algorithm was experimentally validated in both simulated and real environments and compared with other localization algorithms. Experimental results show that the proposed algorithm can effectively improve localization speed during the cold start phase and enhances localization accuracy and stability throughout the localization process. The proposed method is a potential method for improving the performance of mobile robot localization. Full article

In this work, an effective route for achieving high-performance all-polymer materials through the proper manipulation of the material microstructure and starting from a waste material is proposed. In particular, recycled polyethylene terephthalate (rPET) fibers from discarded safety belts were used as reinforcing phase in melt-compounded high-density polyethylene (HDPE)-based systems. The formulated composites were subjected to hot- and cold-stretching for obtaining filaments at different draw ratios. The performed characterizations pointed out that the material morphology can be profitably modified through the application of the elongational flow, which was proven able to promote significant microstructural evolutions of the rPET dispersed domains, eventually leading to the obtainment of micro-fibrillated all-polymer composites. Furthermore, tensile tests demonstrated that hot-stretched and, especially, cold-stretched materials show significantly enhanced tensile modulus and strength as compared to the unfilled HDPE filaments, likely due to the formation of a highly oriented and anisotropic microstructure. Full article

Over 50% of an engine’s energy dissipates via the exhaust and cooling systems, leading to considerable energy loss. Effectively harnessing the waste heat generated by the engine is a critical avenue for enhancing energy efficiency. Traditional exhaust heat recovery systems are limited to real-time recovery of exhaust heat primarily for engine warm-up and fail to fully optimize exhaust heat utilization. This paper introduces a novel exhaust heat recovery system leveraging thermal battery technology, which utilizes phase change materials for both heat storage and reutilization. This innovation significantly minimizes the engine’s cold start duration and provides necessary heating for the cabin during start-up. Dynamic models and thermal management system models were constructed. Parameter optimization and calculations for essential components were conducted, and the fidelity of the simulation model was confirmed through experiments conducted under idle warm-up conditions. Four distinct operational modes for engine warm-up are proposed, and strategies for transitioning between these heating modes are established. A simulation analysis was performed across four varying operational scenarios: WLTC, NEDC, 40 km/h, and 80 km/h. The results indicated that the thermal battery-based exhaust heat recovery system notably reduces warm-up time and fuel consumption. In comparison to the cold start mode, the constant speed condition at 40 km/h showcased the most significant reduction in warm-up time, achieving an impressive 22.52% saving; the highest cumulative fuel consumption reduction was observed at a constant speed of 80 km/h, totaling 24.7%. This study offers theoretical foundations for further exploration of thermal management systems in new energy vehicles that incorporate heat storage and reutilization strategies utilizing thermal batteries. Full article

This study investigates the enhancement of vehicle warm-up performance using phase-change materials (PCMs) and various thermal storage methods. The primary objective is to utilize the thermal energy lost during engine cooling to improve the cold-start performance, thereby reducing fuel consumption and emissions. Thermal storage devices incorporating PCMs were developed and tested by measuring temperature changes and energy transfer over soaking periods of 4, 8, 16, and 24 h. The results show energy transfers of 591, 489, 446, and 315 kJ at 4, 8, 16, and 24 h, respectively. In terms of the warm-up time, the use of thermal storage devices reduced the time required to reach 70 °C by up to 24.45%, with significant reductions observed across all soaking periods. This reduction in the warm-up time directly contributes to faster engine stabilization, leading to proportional improvements in fuel efficiency and a corresponding decrease in exhaust emissions, including CO₂. The findings highlight the effectiveness of PCMs in improving the engine warm-up performance and emphasize the importance of optimizing thermal storage systems to balance energy efficiency and practical application considerations. Additionally, the experimental data provide useful benchmark information for computational simulation validation, enabling the further optimization of automotive thermal management systems. Integrating a PCM-based thermal storage device can significantly enhance a vehicle’s warm-up performance, leading to reduced fuel consumption and lower emissions. Full article

Traditional recommendation models grapple with challenges such as the scarcity of similar user or item references and data sparsity, rendering the cold-start problem particularly formidable. Meta-learning has emerged as a promising avenue to address these issues, particularly in solving the item cold-start problem by generating meta-embeddings for new items as their initial ID embeddings. This approach has shown notable success in enhancing the accuracy of click-through rate predictions. However, prevalent meta-embedding models often focus solely on the attribute features of the item, neglecting crucial user information associated with it during the generation of initial ID embeddings for new items. This oversight hinders the exploitation of valuable user-related information to enhance the quality and accuracy of the initial ID embedding. To tackle this limitation, we introduce the residual graph meta-embedding model (RGMeta). RGMeta adopts a comprehensive approach by considering both the attribute features and target users of both old and new items. Through the integration of residual connections, the model effectively combines the representation information of the old neighbor items with the intrinsic features of the new item, resulting in an improved initial ID embedding generation. Experimental results demonstrate that RGMeta significantly enhances the performance of the cold-start phase, showcasing its effectiveness in overcoming challenges associated with sparse data and limited reference points. Our proposed model presents a promising step forward in leveraging both item attributes and user-related information to address cold-start problems in recommendation systems. Full article

The cold-start problem in sequence recommendations presents a critical and challenging issue for portable sensing devices. Existing content-aware approaches often struggle to effectively distinguish the relative importance of content features and typically lack generalizability when processing new data. To address these limitations, we propose a content-aware few-shot meta-learning (CFSM) model to enhance the accuracy of cold-start sequence recommendations. Our model incorporates a double-tower network (DT-Net) that learns user and item representations through a meta-encoder and a mutual attention encoder, effectively mitigating the impact of noisy data on auxiliary information. By framing the cold-start problem as few-shot meta-learning, we employ a model-agnostic meta-optimization strategy to train the model across a variety of tasks during the meta-learning phase. Extensive experiments conducted on three real-world datasets—ShortVideos, MovieLens, and Book-Crossing—demonstrate the superiority of our model in cold-start recommendation scenarios. Compared to MetaCs-DNN, the second-best approach, CFSM, achieves improvements of 1.55%, 1.34%, and 2.42% under the AUC metric on the three datasets, respectively. Full article

This study investigates corner solidification in a closed cavity in which the left and bottom walls are kept at a temperature lower than its initial temperature. The liquid material in the cavity initially lies at its phase transition temperature and, due to cold boundary conditions at the left–bottom walls, solidification starts. The simulation of corner solidification was performed using a kinetic-based lattice Boltzmann method (LBM), and the tracking of the solid–liquid interface was captured through the evaluation of time. The present investigation addresses the effect of natural convection over conduction across a wide range of higher Rayleigh numbers, from 10⁶ to 10⁸. The total-enthalpy-based lattice Boltzmann method (ELBM) was used to observe the thermal profiles in the entire cavity with a two-phase interface. The isotherms reveal the relative dominance of natural convection over conduction, and the pattern of interface reveals the effective growth of the solidified layer in the cavity. To quantify the uniformity of cooling, a coefficient of variation (COV) for the thermal field was calculated in the effective solidified zone at a wide range of Ra. The results show that the value of COV increases with Ra and reduces with time. The thermal instability in the flow field is also quantified through FFT analyses. Full article

Vanadis 4E (V4E) is a powder metallurgical cold work tool steel predominantly used in application with demand for wear resistance, high hardness, and toughness. It is of interest to have a processing route that enables full density starting from clean gas-atomized powder allowing component shaping capabilities. This study presents a process involving freeze granulation of powder to facilitate compaction by means of cold isostatic pressing, followed by sintering to allow for capsule-free hot isostatic pressing (HIP) and subsequent heat treatments of fully densified specimens. The sintering stage has been studied in particular, and it is shown how sintering in pure nitrogen at 1150 °C results in predominantly closed porosity, while sintering at 1200 °C gives near full density. Microstructural investigation shows that vanadium-rich carbonitride (MX) is formed as a result of the nitrogen uptake during sintering, with coarser appearance for the higher temperature. Nearly complete densification, approximately 7.80 ± 0.01 g/cm³, was achieved after sintering at 1200 °C, and after sintering at 1150 °C, followed by capsule-free HIP, hardening, and tempering. Irrespective of processing once the MX is formed, the nitrogen is locked into this phase and the austenite is stabilised, which means any tempering tends to result in a mixture of austenite and tempered martensite, the former being predominate during the sequential tempering, whereas martensite formation during cooling from austenitization temperatures becomes limited. Full article

Pulsating heat pipes (PHPs) are widely used in the heat dissipation of electronic components, waste heat recovery, solar energy utilization, etc., relying on the pulsating flow of the work material in the pipe and the heat transfer by phase change, and they have the advantages of high heat-transfer efficiency, simple structure, and low cost. In this paper, an experimental method is used to adjust the length of local pipes in the PHP structure, so that the PHP forms a high- and low-staggered asymmetric structure, and to study the effects of different liquid charging rates and heat-source temperatures on the vibration, startup, and operation of the PHP in the asymmetric structure. We found the following: it is difficult to start up and operate the workpiece at 10%, 68%, and 80% liquid charging rates; the effect of the oscillating impact is worse; the temperature difference between the evaporation section of the pulsating heat pipe and condensation section is larger; and the temperature difference between the evaporation section and condensation section is larger. The temperature difference between the evaporation section and condensation section of the pulsating heat pipe is large, the temperature difference is between 10~25 °C, and it is difficult to achieve a small temperature difference in heat transfer. When the liquid charging rate is 30% and 50%, the pulsating heat pipe oscillates better; the pulsation frequency is relatively high; and the temperature difference between the end of the cold and hot sections is small, the temperature difference is between 3 and 7 °C, and the performance of heat transfer is better. However, when the liquid charging rate is 30% and the heat source is 70 °C, the thermal resistance is increased to 0.016 K/W, and the equivalent thermal conductivity is reduced. When the performance of heat transfer is changed to 0.016 K/W and the equivalent thermal conductivity is reduced, the coefficient decreases, and the heat-transfer performance becomes weaker. Full article

In order to solve the cold-starting problem and improve the intake and exhaust pipe temperatures of diesel engines under cold-starting and low- and medium-speed conditions, this paper proposes a twice-open control method for a hydraulic variable valve system. First, a hydraulic variable valve system that can realize a fully variable valve lift and phase angle is applied to replace the original intake system in order to meet the air intake requirements of different conditions. Then, a twice-open control method in which the intake valve opens two times at the exhaust stroke and intake stroke is proposed to improve the intake pipe temperature and solve the cold-starting problem. This paper contains a numerical work analysis. A GT-POWER model is constructed to validate the intake valve twice-open control method. The cylinder pressure, cylinder temperature, intake pipe pressure, and intake pipe temperature are obtained and compared between the original intake valve system and the hydraulic variable valve system with the proposed intake valve twice-open control method. The results show that the twice-open control method can increase the intake pipe temperature to 260 K or even higher, which can improve the cold-starting performance and the exhaust temperature at low and medium speeds. At the same time, the performance under low- and medium-speed conditions is improved. Full article