In this study, the influence of total sn-2 palmitic triacylglycerols (TAGs) and ratio of 1-oleoyl-2-palmitoyl-3-linoleoylglycerol (OPL) to 1,3-dioleoyl-2-palmitoylglycerol (OPO) in human milk fat substitute (HMFS) on the metabolic changes were investigated in Sprague–Dawley rats. Metabolomics and lipidomics profiling analysis indicated that increasing the total sn-2 palmitic TAGs and OPL to OPO ratio in HMFS could significantly influence glycine, serine and threonine metabolism, glycerophospholipid metabolism, glycerolipid metabolism, sphingolipid metabolism, bile acid biosynthesis, and taurine and hypotaurine metabolism pathways in rats after 4 weeks of feeding, which were mainly related to lipid, bile acid and energy metabolism. Meanwhile, the up-regulation of taurine, L-tryptophan, and L-cysteine, and down-regulations of lysoPC (18:0) and hypoxanthine would contribute to the reduction in inflammatory response and oxidative stress, and improvement of immunity function in rats. In addition, analysis of targeted biochemical factors also revealed that HMFS-fed rats had significantly increased levels of anti-inflammatory factor (IL-4), immunoglobulin A (IgA), superoxide dismutase (SOD), and glutathione peroxidase (GSH-px), and decreased levels of pro-inflammatory factors (IL-6 and TNF-α) and malondialdehyde (MDA), compared with those of the control fat-fed rats. Collectively, these observations present new in vivo nutritional evidence for the metabolic regulatory effects of the TAG structure and composition of human milk fat substitutes on the host.

In this study, arabinoxylans (AXs) with different molecular structures were first extracted from corn brans using different concentrations of NaOH (0.25, 0.5, 0.75 and 1 M). AX extracted with 0.25 M NaOH (AX-0.25 M) had the highest ferulic acid content of 5.11 mg/g and AX extracted with 1 M NaOH (AX-1 M) had the lowest arabinose to xylose ratio of 0.47. Meanwhile, the molecular structure of AX and its addition amount both had an influence on the microstructure and physicochemical properties of dual-induced casein/arabinoxylan composite hydrogels (CAS/AX hydrogels) by laccase and glucono-delta-lactone. With CAS/AX mass ratio of 16:1, CAS/AX-0.25 M and CAS/AX-1 M hydrogels exhibited stronger gel strength, better water holding capacity, higher hardness, gumminess and epigallocatechin gallate (EGCG) loading capacity, which was possibly due to their stronger covalent and non-covalent interactions, and denser network microstructures. Besides, the EGCG-loaded CAS/AX-0.25 M hydrogel with CAS/AX mass ratio of 16:1 showed the lowest EGCG cumulative release percentage in 2 h of simulated gastric fluid (31.56%) and 4 h of simulated intestinal fluid (52.94%), while almost all the remaining EGCG (46.57%) was fast released in simulated colonic fluid, showing the most optimum colon-targeted release behavior and protection of EGCG at low pH and protease degradation conditions.

In this study, dual-induced casein and arabinoxylan composite gels (casein/AX gels) were successfully prepared by laccase and glucono-6-lactone (GDL). The influence of AX amount on the microstructure and properties of casein/AX gels was also evaluated. Compared to that of the single casein gel, the rheological and textural properties of casein/AX gels were improved, which might be attributed to the co-existence of covalent and non -covalent interactions of AX and casein molecules, resulting in the interconnected network gel microstructure observed by scanning electron microscopy (SEM). Besides, with the increased amount of AX, casein/AX gels showed an increased water holding capacity and a decreased water mobility significantly (P < 0.05). Meanwhile, epigallocatechin gallate (EGCG) was loaded into casein/AX gels more efficiently and released in a pH-responsive pattern. Only 10-22% of the loaded EGCG was released from casein/AX gels in the simulated gastric fluid (SGF). In contrast, most EGCG was released in the simulated intestinal fluid (SIF), showing great potential to protect EGCG from degradation at low pH conditions.

In this study, the impact of sn-2 palmitic triacyclglycerols (TAGs) in combination with their ratio of two major TAGs (1-oleoyl-2-palmitoyl-3-linoleoylglycerol (OPL) to 1,3-dioleoyl-2-palmitoylglycerol (OPO)) in human milk fat substitute (HMFS) on bile acid (BA) metabolism and intestinal microbiota composition was investigated in newly-weaned Sprague–Dawley rats after four weeks of high-fat feeding. Compared to those of control group rats, HMFS-fed rats had significantly increased contents of six hepatic primary BAs (CDCA, αMCA, βMCA, TCDCA, TαMCA and TβMCA), four ileal primary BAs (UDCA, TCA, TCDCA and TUDCA) and three secondary BAs (DCA, LCA and ωMCA), especially for the HMFS with the highest sn-2 palmitic acid TAGs of 57.9% and OPL to OPO ratio of 1.4. Meanwhile, the inhibition of ileal FXR-FGF15 and activation of TGR5-GLP-1 signaling pathways in HMFS-fed rats were accompanied by the increased levels of enzymes involved in BA synthesis (CYP7A1, CYP27A1 and CYP7B1) in the liver and two key thermogenic proteins (PGC1α and UCP1) in perirenal adipose tissue, respectively. Moreover, increasing sn-2 palmitic TAGs and OPL to OPO ratio in HMFS also altered the microbiota composition both on the phylum and genus level in rats, predominantly microbes associated with bile-salt hydrolase activity, short-chain fatty acid production and reduced obesity risk, which suggested a beneficial effect on host microbial ecosystem. These observations provided important nutritional evidence for developing new HMFS products for infants.

In this study, the effect of sn-2 palmitic triacylglycerols (sn-2 palmitic TAGs) and the ratio between the two major sn-2 palmitic TAGs (OPL to OPO ratio) in a human milk fat substitute (HMFS) on growth, fatty acid and calcium absorptions, and lipid and bile acid metabolic alterations was investigated in Sprague-Dawley rats. After 4 weeks of high-fat feeding, rats fed with the HMFS containing a sn-2 palmitic acid content of 57.87% and an OPL to OPO ratio of 1.4 showed the lowest TAG accumulation in their livers and hypertrophy of perirenal adipocytes, compared to the groups fed with fats containing a lower sn-2 palmitic acid content or a lower OPL to OPO ratio. Meanwhile, synergistically improved absorption of fatty acids and calcium and increased levels of total bile acids (BAs), especially for the tauro-conjugated BAs (TCDCA, TUDCA, TαMCA, TβMCA, TDCA and TωMCA), were observed in rats by both increasing the sn-2 palmitic acid content and the OPL to OPO ratio in HMFS. In addition, the levels of total BAs and tauro-conjugated BAs were negatively correlated with serum TAG, TC, and LDL-c levels and positively correlated with HDL-c levels according to Spearman's correlation analysis (P < 0.05). Collectively, these findings present new nutritional evidence for the potential effects of the TAG structure and composition of a human milk fat substitute on the growth and lipid and bile acid metabolism of the host in infancy.

A novel nonparametric method based on manifold learning is proposed for industrial process monitoring. In conventional algorithms, to preserve the global and local structure information of data, heat kernels containing two auxiliary parameters are introduced to define the global and local weight matrices, respectively. However, it is difficult to identify and choose these two parameters empirically. The inadequate selection of parameters can lead to one-sided and inappropriate global and local feature extractions, resulting in an inadequate fault detection performance. To resolve the above problems, a nonparametric strategy is used in this study to generate two nonparametric weight matrices to replace the heat kernel-based weight matrices. Consequently, the proposed method requires no auxiliary parameters in defining the weight matrices, making it more practical. Moreover, it automatically determines a good trade-off between global and local feature extractions. A process monitoring model based on the proposed method was developed. The feasibility and effectiveness of the new nonparametric method are evaluated using a synthetic example and the Tennessee Eastman chemical process.

Graph Neural Networks (GNNs) have recently received a surge of popularity due to their superiority in modeling realistic complex systems into graphs. Better yet, many approaches have made extraordinary contributions to this topic under the context of dynamic graph. However, current dynamic graph models merely focused on the design of the model but didn’t lay enough emphasis on the dynamism of data, which weakened the expressiveness of the output. Hence, we propose the Adaptive Sampling Temporal Graph Network (ASTGN), a Continuous-Time Dynamic Graph (CTDG) algorithm which casts the sampling strategy as a contextual bandit problem. To capture the dynamic information of graphs, we use a change-detection mechanism to keep the sampling strategy up with the times. Besides, we propose an additional constraint to keep our sampling path causal. We experimentally demonstrate the effectiveness of our approaches with six benchmarks and show the superiority over state-of-the-art baselines.

To capture higher-order structural features, most GNN-based algorithms learn node representations incorporating k-hop neighbors' information. Due to the high time complexity of querying k-hop neighbors, most graph algorithms cannot be deployed in a giant dense temporal network to execute millisecond-level inference. This problem dramatically limits the potential of applying graph algorithms in certain areas, especially financial fraud detection. Therefore, we propose Asynchronous Propagation Attention Network, an asynchronous continuous time dynamic graph algorithm for real-time temporal graph embedding. Traditional graph models usually execute two serial operations: first graph querying and then model inference. Different from previous graph algorithms, we decouple model inference and graph computation to alleviate the damage of the heavy graph query operation to the speed of model inference. Extensive experiments demonstrate that the proposed method can achieve competitive performance while greatly improving the inference speed. The source code is published at a Github repository.

Nowadays, the application of dynamic graphs in the modeling of complex systems has made a great achievement, which has aroused people's attention to anomaly detection of dynamic graphs. As an unsupervised learning task, anomaly detection is target at identifying the abnormal data that is different from the majority. One-class support vector machine, one of the classic anomaly detection algorithms, has been widely applied to find the outliers for it's stability, robustness and convenience. However, traditional anomaly detection algorithms always lose their effectiveness when applied to dynamic graph anomaly detection task. In order to solve the above problem, we design one-class temporal graph attention neural network (OCTGAT) for anomaly detection on dynamic graph. OCTGAT aims to integrate the powerful representation capabilities off temporal graph neural networks and the classical one-class objective. Compared with the given benchmarks, OGTGAT achieves significant improvements in the experiments.

激光焊接技术凭借其高效、高精度的特点已广泛应用在汽车制造产业,在涡轮增压器的生产中,激光焊接也逐步被应用于实现废气阀门与阀杆的连接.虽然当前对于激光焊接的工艺研究很多,但对工艺参数的优化也一直缺少一个科学有效的方法.本文以熔接深度、熔接宽度和内部缺陷作为响应变量,选取焊接功率、焊接速度作为因子,运用响应曲面法(RSM)设计实验方案,通过对响应结果进行分析,研究焊接参数对焊接质量的影响,拟合能够预测最优参数的数学模型,最终将获得的最优参数运用到生产中进行验证.

Nowadays, graph-structured data are increasingly used to model complex systems. Meanwhile, detecting anomalies from graph has become a vital research problem of pressing societal concerns. Anomaly detection is an unsupervised learning task of identifying rare data that differ from the majority. As one of the dominant anomaly detection algorithms, one-class support vector machine has been widely used to detect outliers. However, those traditional anomaly detection methods lost their effectiveness in graph data. Since traditional anomaly detection methods are stable, robust and easy to use, it is vitally important to generalize them to graph data. In this work, we propose one-class graph neural network (OCGNN), a one-class classification framework for graph anomaly detection. OCGNN is designed to combine the powerful representation ability of graph neural networks along with the classical one-class objective. Compared with other baselines, OCGNN achieves significant improvements in extensive experiments.

3-Monochloropropane 1,2-diol (3-MCPD) esters are toxicants formed during food thermal processing, and their testicular toxicities were widely reported. In this 90 day in vivo study, Sprague-Dawley rats were treated with 3-MCPD 1-monooleate at 10 and 100 mg/kg body weight (bw)/day or 1-monostearate at 15 and 150 mg/kg bw/day. Histological results indicated that testicular impairment was observed, and the level of serum testosterone was decreased dose dependently, while the levels of serum transforming growth factor beta and interferon-γ in rats' serum were increased dose dependently. To address the molecular mechanisms leading to testicular toxicities of 3-MCPD esters, testes samples were investigated with a mass spectrometry proteomic approach. The deregulated proteins affected by 3-MCPD esters include many enzymes related with the inflammatory necrosis pathways. While verifying the results in cellular level, 3-MCPD 1-monooleate and 3-MCPD 1-monostearate showed almost similar testicular cytotoxicity, and they could activate RIPK1 and MLKL pathways at the cellular level. All of these results showed the possible mechanisms about the toxicity of 3-MCPD esters in rats' testes and play a vital role in understanding the toxic effects of 3-MCPD esters both in vivo and in vitro.

Fatty acid esters of 3-monochloropropane 1,2-diol (3-MCPD esters) are processing-induced food toxicants, with the kidney as their major target organ. For the first time, this study treated Sprague Dawley (SD) rats with 3-MCPD 1-monooleate at 10 and 100 mg/kg BW/day and 1-monostearate at 15 and 150 mg/kg BW/day for 90 days and examined for their potential semi-long-term nephrotoxicity and the associated molecular mechanisms. No bodyweight difference was observed between groups during the study. Both 3-MCPD 1-monooleate and 1-monostearate resulted in a dose-dependent increase of serum urea creatinine, uric acid and urea nitrogen levels, and histological renal impairment. The proteomic analysis of the kidney samples showed that the 3-MCPD esters deregulated proteins involved in the pathways for ion transportation, apoptosis, the metabolism of xenobiotics, and enzymes related to endogenous biological metabolisms of carbohydrates, amino acids, nitrogen, lipids, fatty acids, and the tricarboxylic acid (TCA) cycle, providing partial explanation for the nephrotoxicity of 3-MCPD esters.

Feature extraction plays a key role in the data-driven process monitoring. Recently, manifold learning approaches have shown good effectiveness of preserving manifold structure features of process data. However, conventional algorithms only focus on low-order statistics while ignore high-order statistics, which makes them unable to extract non-Gaussian features of data effectively. Moreover, process data may not strictly follow Gaussian distribution in the complex modern industrial processes. In this paper, to address the above issues, a new manifold learning method named statistics local and nonlocal embedding (SLNLE) is proposed for non-Gaussian process monitoring. Firstly, both low-order and high-order process statistics are conducted by statistics pattern analysis (SPA) algorithm under given time window. Then, local and nonlocal embedding (LNLE) is adopted to preserve manifold structure information of them. Compared with locality preserving projections (LPP) algorithm only retaining local structure features, LNLE implements global and local feature extraction simultaneously, which can preserve non-Gaussian features more effectively. Lastly, a SLNLE-based monitoring model is developed and kernel density estimation (KDE) is applied to obtain more accurate control limits for better process monitoring performance. The feasibility and superiority of the proposed approach are tested on the Tennessee Eastman (TE) process.

Recently, deep generative models have become increasingly popular in unsupervised anomaly detection. However, deep generative models aim at recovering the data distribution rather than detecting anomalies. Moreover, deep generative models have the risk of overfitting training samples, which has disastrous effects on anomaly detection performance. To solve the above two problems, we propose a self-adversarial variational autoencoder (adVAE) with a Gaussian anomaly prior assumption. We assume that both the anomalous and the normal prior distribution are Gaussian and have overlaps in the latent space. Therefore, a Gaussian transformer net T is trained to synthesize anomalous but near-normal latent variables. Keeping the original training objective of a variational autoencoder, a generator G tries to distinguish between the normal latent variables encoded by E and the anomalous latent variables synthesized by T, and the encoder E is trained to discriminate whether the output of G is real. These new objectives we added not only give both G and E the ability to discriminate, but also become an additional regularization mechanism to prevent overfitting. Compared with other competitive methods, the proposed model achieves significant improvements in extensive experiments. The employed datasets and our model are available in a Github repository.

Fault monitoring and diagnosis system are very important in detecting system failures and keeping the stability of production line for the modern industrial system. This paper studies the data-driven fault diagnosis and tolerant control integrated technique focused on Solid Oxide Fuel Cell (SOFC) system, especially for the stack degradation fault. The multivariable statistical approach Support Vector Machine (SVM) as well as Principal Component Analysis (PCA) are studied for the multi-fault classification and diagnosis purpose. Then based on the diagnosis results, the decision-making part is designed to select appropriate fault reconfigurable control strategy which can handle five types of stack faults and recover the thermal and electrical parameters to normal operating condition. The core of integrated data-based fault monitoring and control approach is to take full advantage of available SOFC measurements data aiming to acquire the useful fault condition information and choosing adequate fault recovery control method. The results are that the proposed strategy can keep the system power near the normal operating state and make the SOFC temperature steady near the enactment value when working in faulty conditions, which may result in both lifetime and durability improvement for the SOFC system.

Attention-based encoder–decoder framework has greatly improved image caption generation tasks. The attention mechanism plays a transitional role by transforming static image features into sequential captions. To generate reasonable captions, it is of great significance to detect spatial characteristics of images. In this paper, we propose a spatial relational attention approach to consider spatial positions and attributes. Image features are firstly weighted by the attention mechanism. Then they are concatenated with contextual features to form a spatial–visual tensor. The tensor is feature extracted by a fully convolutional network to produce visual concepts for the decoder network. The fully convolutional layers maintain spatial topology of images. Experiments conducted on the three benchmark datasets, namely Flickr8k, Flickr30k and MSCOCO, demonstrate the effectiveness of our proposed approach. Captions generated by the spatial relational attention method precisely capture spatial relations of objects.

深度流形表示学习对于自动学习系统的本质特征有着重要的作用.论文提出了一种基于深度流形表示学习的多故障识别方法.所提的多故障识别方法可以分为三个阶段:第一,将故障识别问题转化为分类问题,定义正常和故障状态,以及预处理原始数据;第二,利用深度流形表示学习对深度神经网络进行预训练;第三,利用故障标签数据全局训练深度网络.所提出的方法被应用于由一种典型的工业系统生成的两个不同尺寸以及多个故障类型的数据集.测试结果表明,所提方法能够准确预测故障类型,优于其他两种分类方法.此外,由于所提出的方法仅需要数据,因此很容易迁移到其他的工业系统.