In recent years， despite significant advancements in drilling and completion technologies for deep and ultra-deep wells， lost circulation remains one of the most prevalent downhole complexities encountered during the development of deep and ultra-deep oil and gas reservoirs. In drilling operations， economic losses attributable to lost circulation account for approximately 25% to 40% of the total costs. This paper proposes a continuous plugging and drill-out operation method for cement slurry applications. This method can reduce the number of round trips required for cement slurry plugging and subsequent drill-out of cement plugs， thereby significantly shortening the operational cycle of cement slurry plugging operations. On this basis， a comparative analysis was conducted on the advantages and disadvantages of four technical principles， i.e. hydraulic drive， mechanical drive， mechanically controllable throttling + hydraulic drive， and hydraulically controllable throttling + hydraulic drive. The mechanically controllable throttling + hydraulic drive mode was selected as the optimal target in this study. According to the <italic>Hydraulic Calculation Manual</italic> and the <italic>Drilling Engineering Technical Manual</italic>， and through fluid simulation analysis， a theoretical formula for hydraulic thrust was established. Then， a multifunctional tool for continuous plugging and drill-out was developed by adopting the design of cutter block mis-expansion structure， 45° expansion angle cutter block， cement slurry accumulation and solidification control structure， three long centrally-connected blades， integrated bit structure， and large-diameter water hole. The results of laboratory experiments and field tests demonstrate that the theoretical formula for hydraulic thrust yields relatively accurate calculation， providing a theoretical support for the serialization of the multifunctional tool for continuous plugging and drill-out. Moreover， the multifunctional tool facilitated a successful job of cement slurry pumping and cement plug drill-out in a single trip at Well ZS-X， with the wellbore diameter achieved after drilling out the cement plug meeting the design specification. Thus， it enables a continuous plugging and drill-out process without tripping out， which features one trip less than conventional jobs. Its application is expected to significantly shorten the operational cycle of cement slurry plugging.
Chemical engineering, Petroleum refining. Petroleum products
黎明珠1,2，宋菲2,3，唐敏敏2,3，王挥2,3，阚金涛2,3，张玉锋2,3 LI Mingzhu1,2，SONG Fei2,3，TANG Minmin2,3，WANG Hui2,3， KAN Jintao2,3，ZHANG Yufeng2,3
旨在解决植脂搅打稀奶油反式脂肪酸含量较高的问题，以初榨椰子油为油相制备零反式脂肪酸植脂搅打稀奶油，研究其添加量（28%~36%）对稀奶油搅打前后品质的影响，并与市售植脂稀奶油打发后的样品进行质构特性、裱花形态及感官品质方面的对比。结果表明，随着初榨椰子油添加量的增加，稀奶油搅打时间逐渐缩短，搅打起泡率和泡沫稳定性呈现先增大后减小的趋势，表观黏度增大，搅打稀奶油脂肪球体积平均粒径、脂肪部分聚结率增大。最适初榨椰子油添加量为32%，此时稀奶油搅打时间为122 s，搅打起泡率最高（95.86%），泡沫稳定性最佳（94.58%），搅打稀奶油质构特性（硬度、内聚性、弹性、胶黏性、咀嚼性）和感官得分与市售植脂稀奶油打发后的样品均无显著差异，搅打稀奶油裱花后具有锋利的边缘和坚挺的峰，25 ℃放置1 h后无乳清析出现象，且未检出反式脂肪酸。综上，以初榨椰子油为油相制备植脂搅打稀奶油，克服了传统植脂搅打稀奶油反式脂肪酸含量较高的问题，可满足消费者对营养和健康的追求。In order to solve the problem of high trans fatty acid content in vegetable-fat whipped cream, the whipped cream with zero trans fatty acid was prepared with virgin coconut oil as the oil phase. The effects of virgin coconut oil dosage（28%-36%）on the quality of the cream before and after whipping were studied, and the texture characteristics, decoration performance and sensory quality of the whipped cream were compared with those of the commercial vegetable-fat whipped cream. The results indicated that with the increase of virgin coconut oil dosage, the whipping time of the cream gradually shortened, the foaming rate and foam stability of the cream showed a trend of first increasing and then decreasing, and the apparent viscosity of the cream increased. The average particle size of the fat globules, and the fat partial coaslescence rate of the whipped cream both increased. The optimum virgin coconut oil dosage was 32%. At this time, the whipping time of the cream was 122 s, the whipping foaming rate of the cream was the highest (95.86%), the stability of foam was the best (94.58%), and there were no significant differences in textural properties (hardness, cohesiveness, elasticity, gumminess, chewiness) and sensory scores between the whipped cream and the commercial vegetable-fat whipped cream. The whipped cream had sharp edges and firm peaks, and there was no whey separation after being placed at 25 ℃ for 1 h. And no trans fatty acids were detected in the cream. In conculsion, preparing vegetable-fat whipped cream with virgin coconut oil as the oil phase overcomes the problem of high trans fatty acids content in traditional vegetable-fat whipped cream, which can satisfy the consumer′s pursuit of nutrition and health.
Andre Juarez, Luis Salsavilca, Frida Coaquira
et al.
Deep learning models for SAR oil spill segmentation often fail to generalize across regions due to differences in sea-state, backscatter statistics, and slick morphology, a limitation that is particularly severe along the Peruvian coast where labeled Sentinel-1 data remain scarce. To address this problem, we propose \textbf{MORP--Synth}, a two-stage synthetic augmentation framework designed to improve transfer from Mediterranean to Peruvian conditions. Stage~A applies Morphological Region Perturbation, a curvature guided label space method that generates realistic geometric variations of oil and look-alike regions. Stage~B renders SAR-like textures from the edited masks using a conditional generative INADE model. We compile a Peruvian dataset of 2112 labeled 512$\times$512 patches from 40 Sentinel-1 scenes (2014--2024), harmonized with the Mediterranean CleanSeaNet benchmark, and evaluate seven segmentation architectures. Models pretrained on Mediterranean data degrade from 67.8\% to 51.8\% mIoU on the Peruvian domain; MORP--Synth improves performance up to +6 mIoU and boosts minority-class IoU (+10.8 oil, +14.6 look-alike).
Estelle Lu, Williams Flores Cisternas, Héloïse Uhl
et al.
Blood-mimicking fluids (BMFs) play a critical role in ultrasonic imaging and Doppler flow studies by replicating the physical and acoustic properties of blood. This study introduces a novel soybean oil-in-water emulsion as a BMF with particle size and deformability akin to red blood cells. Using a millifluidic device, we cross-validated flow profiles through both Doppler velocimetry and optical particle tracking, demonstrating compatibility with theoretical Poiseuille flow models. The millifluidic chip, fabricated via stereolithography, provided an optimized platform for dual optical and ultrasonic assessments. Results showed strong agreement between the two methods across a range of flow rates, affirming the suitability of the emulsion for velocimetry applications. Furthermore, the acoustic properties of soybean oil droplets support their potential as an echogenic and stable alternative to conventional BMFs.
Martín A. Díaz-Viera, Arturo Ortiz-Tapia, Joaquín R. Hernández-Pérez
et al.
A general 3D flow-and-transport model in porous media is derived using an axiomatic continuum-mechanics approach and implemented with the finite element method to simulate microbial enhanced oil recovery (MEOR) at core scale under laboratory conditions. The development pipeline (conceptual -> mathematical -> numerical -> computational) is detailed. The model captures clogging/declogging from biomass, changes in interfacial tension due to biosurfactant, and the resulting impact on relative permeability, capillary pressure, and residual oil saturation via a trapping-number framework. The flow model is validated (Buckley-Leverett and coreflood benchmarks); transport (microbes/nutrients/surfactant) is validated against Hendry et al. 1997 breakthrough data. Finally, the model accurately predicts a Berea-core MEOR case study using field microbes and brine, matching recovery histories with small RMS error. Owing to its generality, the framework can be extended to other EOR scenarios and constitutive laws.
尤悦1，王俏君2，秦雅丽1，杨礼学2，张逸人1，张晖1，吴港城1，王兴国1YOU Yue1, WANG Qiaojun2, QIN Yali1, YANG Lixue2, ZHANG Yiren1, ZHANG Hui1, WU Gangcheng1, WANG Xingguo1
旨在为不同风味需求火锅底料的研究提供参考依据，通过消费者偏好分析和挥发性风味成分的含量与种类从6种牛油（3种市售牛油、未精炼牛油、脱胶牛油、精炼牛油）中筛选出最受消费者喜爱的牛油，并采用顶空固相微萃取结合气相色谱-质谱联用技术、香味提取物稀释分析法、气味活度值法、风味重组和风味缺失实验对其风味物质进行鉴定。结果表明：精炼牛油的消费者偏好平均得分最高且挥发性物质组成最丰富，其含有23种风味物质，包括醛类10种、酸类5种、杂环类4种、醇类3种和酮类1种；根据牛油风味物质的气味活度值，确定牛油的7种关键风味物质为（E,E）-2,4-壬二烯醛、（E）-2-辛烯醛、乙酸、2-戊基吡啶、辛醛、1-辛烯-3-醇和2-戊基呋喃，通过风味重组和风味缺失实验得出（E,E）-2,4-壬二烯醛、（E）-2-辛烯醛、乙酸和辛醛对牛油风味的贡献最大，2-戊基吡啶、1-辛烯-3-醇和2-戊基呋喃对牛油风味的贡献次之。综上，确认了牛油中7种关键风味物质，可为牛油产品风味的合理开发提供参考。In order to establish a foundation for researching the diverse flavour requirements of substances in hotpot bases, six types of beef tallow(three types of commercially available beef tallow, unrefined beef tallow, degummed beef tallow, and refined beef tallow) were used as materials, the most popular beef tallow was selected based on customer preference analysis and the content and type of volatile flavour components, and the flavour substances of the most popular beef tallow were identified using headspace solid-phase microextraction-gas chromatography - mass spectrometry, dilution analysis of flavour extracts, odour activity values, flavour reconstitution and flavour deletion experiments. The results indicated that refined beef tallow received the highest average score in the consumer preference analysis and had the most substantial chemical composition with 23 volatile substances including 10 kinds aldehydes, 5 kinds acids, 4 kinds heterocyclic, 3 kinds alcohols and 1 kind ketone compounds. Based on the odour activity values of beef tallow flavour substances, 7 kinds key flavour substances of beef tallow were identified as (E,E)-2,4-nonadienal, (E)-2-octenal, acetic acid, 2-pentylpyridine, octanal, 1-octen-3-ol and 2-pentylfuran. Flavour reconstitution and deletion experiments showed that (E,E)-2,4-nonadienal, (E)-2-octenal, acetic acid and octanal contributed the most to beef tallow flavour, followed by 2-pentylpyridine, 1-octen-3-ol and 2-pentylfuran. In conclusion, the 7 kinds key flavour substances are determined in the beef tallow, which can provide a reference for the rational development of beef tallow product flavours.
Istoni da Luz Sant'Ana, Patricia Román-Román, Francisco Torres-Ruiz
Accurately charting the progress of oil production is a problem of great current interest. Oil production is widely known to be cyclical: in any given system, after it reaches its peak, a decline will begin. With this in mind, Marion King Hubbert developed his peak theory in 1956 based on the bell-shaped curve that bears his name. In the present work, we consider a stochasticmodel based on the theory of diffusion processes and associated with the Hubbert curve. The problem of the maximum likelihood estimation of the parameters for this process is also considered. Since a complex system of equations appears, with a solution that cannot be guaranteed by classical numerical procedures, we suggest the use of metaheuristic optimization algorithms such as simulated annealing and variable neighborhood search. Some strategies are suggested for bounding the space of solutions, and a description is provided for the application of the algorithms selected. In the case of the variable neighborhood search algorithm, a hybrid method is proposed in which it is combined with simulated annealing. In order to validate the theory developed here, we also carry out some studies based on simulated data and consider 2 real crude oil production scenarios from Norway and Kazakhstan.
This groundbreaking research extracted DNA from petroleum using nanoparticle affinity bead technology, yielding 3,159,020 petroleum DNA (pDNA) sequences, primarily environmental DNA. While most original in situ DNA (oriDNA) was lost, ancient DNA (aDNA) from petroleum offers an important source of ecological and evolutionary information, surpassing traditional fossils. This study reveals that oil, mainly sourced from algae and lower aquatic plants, now serves as a new type of fossil, providing detailed insights into Earth's hidden history, including unclassified species and ancient events, revolutionizing petroleum geology and paleontology.
Pranav Kulkarni, Adway Kanhere, Harshita Kukreja
et al.
Generative Adversarial Network (GAN)-based synthesis of fat suppressed (FS) MRIs from non-FS proton density sequences has the potential to accelerate acquisition of knee MRIs. However, GANs trained on single-site data have poor generalizability to external data. We show that federated learning can improve multi-center generalizability of GANs for synthesizing FS MRIs, while facilitating privacy-preserving multi-institutional collaborations.
Carbon dioxide storage and utilization has become an inevitable trend and choice for sustainable development under the background of global climate change and carbon neutrality. Carbon industry which is dominated by CO2 capture, utilization and storage/ CO2 capture and storage (CCUS/CCS) is becoming a new strategic industry under the goal of carbon neutrality. The sustainable development of carbon industry needs to learn from the experiences of global oil and gas industry development. There are three types of “carbon” in the earth system. Black carbon is the CO2 that has not been sequestered or used and remains in the atmosphere for a long time; grey carbon is the CO2 that has been fixed or permanently sequestered in the geological body, and blue carbon is the CO2 that could be converted into products for human use through biological, physical, chemical and other ways. The carbon industry system covers carbon generation, carbon capture, carbon transportation, carbon utilization, carbon sequestration, carbon products, carbon finance, and other businesses. It is a revolutionary industrial field to completely eliminate “black carbon”. The development of carbon industry technical system takes carbon emission reduction, zero carbon, negative carbon and carbon economy as the connotation, and the construction of a low-cost and energy-efficient carbon industry system based on CCUS/CCS are strategic measures to achieve the goal of carbon neutrality and clean energy utilization globally. This will promote the “four 80%s” transformation of China's energy supply, namely, to 2060, the percentage of zero-carbon new energy in the energy consumption will be over 80% and the CO2 emission will be decreased by 80% to ensure the carbon emission reduction of total 80×108 t from the percentage of carbon-based fossil energy in the energy consumption of over 80%, and the percentage of CO2 emission from energy of over 80% in 2021. The carbon industry in China is facing three challenges, large CO2 emissions, high percentage of coal in energy consumption, and poor innovative system. Three strategic measures are proposed accordingly, including: (1) unswervingly develop carbon industrial system and ensure the achievement of carbon neutrality as scheduled by 2060; (2) vigorously develop new energy sources and promote a revolutionary transformation of China's energy production and consumption structure; (3) accelerate the establishment of scientific and technological innovation system of the whole CO2 industry. It is of great significance for continuously optimization of ecological environment and construction of green earth and ecological earth to develop the carbon industry system, utilize clean energy, and achieve the strategic goal of global carbon neutrality.
高锦鸿1,2，卢奎1,3，孙强2，芦鑫2，马宇翔1，杜蘅1，杨红艳1 GAO Jinhong1,2，LU Kui1,3, SUN Qiang2, LU Xin2, MA Yuxiang1, DU Heng1, YANG Hongyan1
旨在为芝麻素酚在食品加工中的应用提供参考，以过氧化值和茴香胺值为指标，通过正交实验研究了不同高温加热条件下芝麻素酚对花生油的抗氧化作用，并对比了芝麻素酚与α-生育酚、二丁基羟基甲苯（BHT）、抗坏血酸棕榈酸酯的抗氧化效果及协同增效作用。结果表明：高温加热下花生油会发生显著氧化；在芝麻素酚添加量0.06%、加热温度150 ℃、加热时间15 h条件下，芝麻素酚对花生油过氧化值和茴香胺值的抑制率最高，且优于抗坏血酸棕榈酸酯、α-生育酚及BHT的抗氧化能力；芝麻素酚与α-生育酚及BHT均具有一定的协同增效作用。综上，芝麻素酚可以作为天然抗氧化剂应用于延缓高温加热油脂的氧化。To provide reference for the application of sesaminol on food processing, the antioxidation effect of sesaminol on peanut oil at heating temperature was investigated by orthogonal test using peroxide value and anisidine value as indexes. The antioxidant effects of α-tocopherol, dibutylhydroxytoluene (BHT) and ascorbyl palmitate were compared with that of sesaminol, and their synergistic effects were also investigated. The results showed that peanut oil underwent significant oxidation at high temperature. Under the conditions of heating temperature 150 ℃，heating time 15 h, and sesaminol addition 0.06%, sesaminol showed the highest inhibition rates of peroxide value and anisidine value, and had stronger antioxidant activity than ascorbyl palmitate, α-tocopherol and BHT, meanwhile, the sesaminol had synergistic effects with α-tocopherol and BHT. In conclusion, sesaminol can be used as a natural antioxidant for retarding the oxidation of edible oil at high temperature.
Rusul J. Alsaedi, Joachim Gudmundsson, André van Renssen
Given a set of $n\geq 1$ autonomous, anonymous, indistinguishable, silent, and possibly disoriented mobile unit disk (i.e., fat) robots operating following Look-Compute-Move cycles in the Euclidean plane, we consider the Pattern Formation problem: from arbitrary starting positions, the robots must reposition themselves to form a given target pattern. This problem arises under obstructed visibility, where a robot cannot see another robot if there is a third robot on the straight line segment between the two robots. We assume that a robot's movement cannot be interrupted by an adversary and that robots have a small $O(1)$-sized memory that they can use to store information, but that cannot be communicated to the other robots. To solve this problem, we present an algorithm that works in three steps. First it establishes mutual visibility, then it elects one robot to be the leader, and finally it forms the required pattern. The whole algorithm runs in $O(n) + O(q \log n)$ rounds with probability at least $1 - n^{-q}$. The algorithms are collision-free and do not require the knowledge of the number of robots.
欧锦强1，宋志华2，邵琳雅2，黄健花2，王兴国2，唐俊军3 OU Jinqiang1, SONG Zhihua2, SHAO Linya2, HUANG Jianhua2, WANG Xingguo2, TANG Junjun3
为了定量分析亚麻籽油中α-亚麻酸（ALA）顺反异构体，采用SLB-IL111色谱柱，对气相色谱法测定反式亚麻酸（TALA）的分析条件（载气流速和升温程序）进行优化，并对所建立的方法进行评价。结果表明：载气流速选择0.3 mL/min；升温程序为60 ℃保持5 min，以20 ℃/min升温至175 ℃保持15 min，以1 ℃/min升温至180 ℃保持28 min，以0.2 ℃/min升温至185 ℃保持40 min；该方法可实现9c,12t,15t-C18∶ 3和9t,12c,15t-C18∶ 3的较好分离，分离度为0.9，其余各TALA的分离度在1.6~8.7之间；所建立的方法重复性好，测定加热亚麻籽油样时各异构体日间相对标准偏差（RSD）为3.16%~5.65%，日内RSD为2.00%~4.13%，检出限为0.678 2~0.879 0 mg/L，定量限为2.260 6~2.930 1 mg/L；各ALA异构体含量与峰面积之间线性关系良好，相关系数均为0.999 8。综上，所建立的方法可用于亚麻籽油中8种不同构型ALA顺反异构体的定量分析。
In order to quantitatively analyze the cis- and trans- isomers of α-linolenic acid (ALA) in linseed oil, the analytical conditions (carrier gas flow rate and temperature program) for the determination of trans-linolenic acid (TALA) by gas chromatography were optimized on the SLB-IL111 column,and the established method was evaluated. The results showed that the optimal carrier gas flow rate was 0.3 mL/min, and the chromatographic column was initially heated at 60 ℃ for 5 min, then ramped at 20 ℃/min to 175 ℃ and held for 15 min, ramped at 1 ℃/min to 180 ℃ and held for 28 min, finally ramped at 0.2 ℃/min to 185 ℃ and held for 40 min. The method established had effective separation of 9c,12t,15t-C18∶ 3 and 9t,12c,15t-C18∶ 3 with a separation degree of 0.9. The separation degree of the other TALA ranged from 1.6 to 8.7. The method was reproducible and the relative standard deviation (RSD) of inter-day repetition of each isomer in heated lineseed oil was 316%-565%, and the RSD of intra-day was 200%-413%.The detection limit was 0.678 2-0.879 0 mg/L, and the limit of quantification was 2.260 6-2.930 1 mg/L. The linear relationship between the content of each ALA isomer and the peak area was good, and all the correlation coefficient was 0.999 8. In conclusion, the established method can be used for the quantitative analysis of 8 different isomer of ALA in linseed oil.
Gagnon Yancie, Mhemdi Houcine, Delbecq Frederic
et al.
The vegetable oil extraction process from seeds and nuts depends on mechanical and solvent (usually n-hexane) extractions. Despite the efficiency of n-hexane, its use is nowadays questioned due to health, environmental, and technological issues. As an alternative to hexane extraction, several greener solvents and extraction techniques have been developed and tested during the last decades. Among these alternatives, the Surfactant-Aqueous Extraction Process (SAEP) appears as a promising method. Initially developed for the petroleum sector, this method was then tested and optimized for vegetable oil extraction. Successful implementations at the laboratory scale led to slightly more than 90% oil yield, mainly by using so-called “extended surfactants”. Compare to conventional surfactants, these surfactants can efficiently solubilize a large amount of vegetable oil in water, despite the structural diversity and the bulkiness of vegetable oil molecules. The present review is devoted to extended surfactant applications to SAEP. This review summarizes and discusses the main findings related to the extended surfactant structures and properties, as well as the main experimental results on the SAEP, and the advantages and the current limitations towards a scaling-up of this promising process.
The dynamic cavity method provides the most efficient way to evaluate probabilities of dynamic trajectories in systems of stochastic units with unidirectional sparse interactions. It is closely related to sum-product algorithms widely used to compute marginal functions from complicated global functions of many variables, with applications in disordered systems, combinatorial optimization and computer science. However, the complexity of the cavity approach grows exponentially with the in-degrees of the interacting units, which creates a de-facto barrier for the successful analysis of systems with fat-tailed in-degree distributions. In this manuscript, we present a dynamic programming algorithm that overcomes this barrier by reducing the computational complexity in the in-degrees from exponential to quadratic, whenever couplings are chosen randomly from (or can be approximated in terms of) discrete, possibly unit-dependent, sets of equidistant values. As a case study, we analyse the dynamics of a random Boolean network with a fat-tailed degree distribution and fully asymmetric binary $\pm J$ couplings, and we use the power of the algorithm to unlock the noise dependent heterogeneity of stationary node activation patterns in such a system.
In this paper, the main chemical properties of Perilla seed oil (PO) obtained by mechanical pressing of Perilla seeds are reported. The analysis of fatty acid composition has highlighted a very high amount of (n-3) α-linolenic acid (ALA), more than 60%, higher than other ALA rich oils such as linseed and sacha inchi oils and similar to chia oil. PO has an important sterol (higher than 3000 mg/kg) and vitamin E (approx. 870 mg/kg) content, while biophenols are in quite low concentration. The analyzed sample showed a very low acid and peroxide value and this is the demonstration that, following the proper technological procedures, it is possible to obtain high quality oils even in presence of high α-linolenic acid concentrations.
Natural gas leakage and explosion have become the main extreme disasters of buried gas pipelines. TNT equivalent method is selected as the estimation model for small hole leakage explosion of gas pipeline. The material constitutive model is appropriately selected. Arbitrary Lagrange-Euler method (ALE) is used to verify the accuracy of the model. A three-dimensional solid model of buried gas pipeline-soil-explosive coupling is established. The relationship between dynamic response (equivalent stress, displacement) of buried gas pipeline and explosion center distance and explosive quantity is analyzed. The results show that the ALE method can well describe the propagation law of spherical explosion waves in the soil, and can visually show the macro-evolution process of the explosion cavity and the surface mound. Under the explosion center distance of 205 cm, the stress increases with the explosive charge. The anti-explosion limit charge of the pipe body is 58.4 kg. Both the stress and displacement increase with the explosive charge or the decrease of the explosion center distance. The <i>X</i>-direction displacement, the <i>Y</i>-direction displacement of front explosion point and rear explosion point increases first and then decreases with time. The <i>Y</i>-direction displacement of the top of the pipe presents a second shock that increases first and then decreases. The displacement of the bottom of the pipe changes approximately sinusoidally. There is a deviation between the strain-based and stress-based failure evaluation results. The strain-based results are relatively conservative. The study can provide a reference for the safety assessment of parallel gas pipelines.
Chemical engineering, Petroleum refining. Petroleum products
Abstract Saybolt color determination is one of the techniques used to evaluate the quality of petroleum products as an indicator of the degree of refinement. As color is a property readily observed by operators, conventional procedures require operators to determine Saybolt color either through direct visual observation or through Saybolt chromometers. These methods are subjective due to the variability in perception of colors across different observers and may be influenced by external factors such as the level of illuminance. Digital oil color analyzers, on the other hand, cost almost four times as much as Saybolt chromometers. An alternative approach to color measurement is to develop a correlation model between Saybolt color with the physical and chemical properties of condensates and light crude oils from Malaysian oil and gas fields. This work applies several multiple linear regression techniques (such as stepwise regression) performed both manually and using the R software (version 3.6.1) to obtain statistically significant results. The step, regsubsets and glmulti functions from R are explored to develop the correlation model which predicts Saybolt color using only identified key properties, overcoming the possible drawbacks associated with conventional laboratory analysis. The models developed through these different techniques are analyzed and compared based on criteria indicated through the coefficient of multiple determination, R 2 and F-tests to infer on suitable regression approaches. Results obtained from these regression methods for models with and without interaction terms report deviations of less than 5% for 75% of the samples used for validation.
Ussama Zahid, Giulio Gambardella, Nicholas J. Fraser
et al.
Deep neural networks (DNNs) are state-of-the-art algorithms for multiple applications, spanning from image classification to speech recognition. While providing excellent accuracy, they often have enormous compute and memory requirements. As a result of this, quantized neural networks (QNNs) are increasingly being adopted and deployed especially on embedded devices, thanks to their high accuracy, but also since they have significantly lower compute and memory requirements compared to their floating point equivalents. QNN deployment is also being evaluated for safety-critical applications, such as automotive, avionics, medical or industrial. These systems require functional safety, guaranteeing failure-free behaviour even in the presence of hardware faults. In general fault tolerance can be achieved by adding redundancy to the system, which further exacerbates the overall computational demands and makes it difficult to meet the power and performance requirements. In order to decrease the hardware cost for achieving functional safety, it is vital to explore domain-specific solutions which can exploit the inherent features of DNNs. In this work we present a novel methodology called fault-aware training (FAT), which includes error modeling during neural network (NN) training, to make QNNs resilient to specific fault models on the device. Our experiments show that by injecting faults in the convolutional layers during training, highly accurate convolutional neural networks (CNNs) can be trained which exhibits much better error tolerance compared to the original. Furthermore, we show that redundant systems which are built from QNNs trained with FAT achieve higher worse-case accuracy at lower hardware cost. This has been validated for numerous classification tasks including CIFAR10, GTSRB, SVHN and ImageNet.