Explore the vast range of titles available.

Although Gaussian process regression (GPR) is a powerful Bayesian nonparametric regression model for engineering problems, its predictive performance is highly dependent on a kernel for covariance function of GPR. However, choosing a proper kernel is still challenging even for experts. To choose proper kernel automatically, this study proposes a compositional kernel (CPK) learning using tree-based genetic programming (GEP). The optimal structure of the kernel is defined as a compositional representation based on sums and products of eight base-kernels. The CPK can be encoded as a tree-structure, so that tree-based GEP is employed to discover an optimal tree-structure of the CPK. To avoid overly complex solution in GEP, the proposed method introduced a dynamic maximum tree-depth technique. The novelty of the proposed method is to utilize more flexible and efficient learning capability to learn the relationship between input and output than existing methods. To evaluate the learning capability of the proposed method, seven test functions were firstly investigated with various noise levels, and its predictive accuracy was compared with existing methods. Reliability problems in both parallel and series systems were introduced to evaluate the performance of the proposed method for efficient reliability assessment. The results show that the proposed method generally outperforms or performs similarly to the best one among existing methods. In addition, it is also shown that proper kernel function can significantly improve the performance of GPR as the training data increases. Stated differently, the proposed method can learn the function of being fitted efficiently with less training samples than existing methods. In this context, the proposed method can make powerful and automatic predictive modeling based on GPR in engineering problems.

By virtue of the advances in sensing techniques, finite element (FE) model updating (FEMU) using static and dynamic data has been recently employed to improve identification on updating parameters. Using heterogeneous data can provide useful information to improve parameter identifiability in FEMU. It is worth noting that the useful information from the heterogeneous data may be diluted in the conventional FEM framework. The conventional FEMU framework in previous studies have used heterogeneous data at once to compute residuals in the objective function, and they are condensed to be a scalar. In this implementation, it should be careful to formulate the objective function with proper weighting factors to consider the scale of measurement and relative significances. Otherwise, the information from heterogeneous data cannot be efficiently utilized. For FEMU of the bridge, parameter compensation may exist due to mutual dependence among updating parameters. This aggravates the parameter identifiability to make the results of the FEMU worse. To address the limitation of the conventional FEMU method, this study proposes a sequential framework for the FEMU of existing bridges. The proposed FEMU method uses two steps to utilize static and dynamic data in a sequential manner. By using them separately, the influence of the parameter compensation can be suppressed. The proposed FEMU method is verified through numerical and experimental study. Through these verifications, the limitation of the conventional FEMU method is investigated in terms of parameter identifiability and predictive performance. The proposed FEMU method shows much smaller variabilities in the updating parameters than the conventional one by providing the better predictions than those of the conventional one in calibration and validation data. Based on numerical and experimental study, the proposed FEMU method can improve the parameter identifiability using the heterogeneous data and it seems to be promising and efficient framework for FEMU of the existing bridge.

In probabilistic risk assessment (PRA), two main methods exist for quantifying fault trees: theoretical and empirical (sampling). The efficiency of PRA quantification varies depending on the sampling method used. This study evaluated the feasibility of using quasi-Monte Carlo simulation (Quasi-MCS) and progressive Latin hypercube sampling (P-LHS) for PRA quantification. Eight risk outcomes were derived through PRA for internal and external events in four cases. The PRA convergence, variability, and error rates of each sampling method were compared and analyzed. The comparison analysis revealed that all sampling methods had an error rate of approximately 2% with 9,000 total samples. P-LHS exhibited the best convergence and variability among the methods, followed by Quasi-MCS and LHS. Although Quasi-MCS showed more significant variability than LHS as the number of events increased, its error rate remained within 2% with 9,000 samples. Therefore, both P-LHS and Quasi-MCS are feasible for PRA quantification.

In this paper, a new finite element model updating (FEMU) method is proposed based on mixed-integer nonlinear programming (MINLP) to deal with model-form uncertainty in FE models. Depending on modelers’ preference and experience, various FE models can be constructed for a specific structure in practice. However, no one can guarantee that a specific model representation is always best (Model-form uncertainty). Conventional method should perform model updating for each FE model independently and select a best one among them, so that it becomes computationally intensive with many candidate FE models. To handle model-form uncertainty, this study formulates FEMU as the MINLP problem. The proposed method assigns an integer variable for model choice, while continuous real variables are used for the updating parameters. With this formulation, the optimization algorithm can explore both model and parameter space simultaneously to deal with the model-form uncertainty in FE models. Firstly, three numerical experiments were explored to evaluate the performance of the proposed method by considering possible situations in reality as follows: (1) a true FE model exists in model space with an admissible FE model; (2) only admissible FE model exists in model space; and (3) no true and admissible FE models exist in model space. Then, the proposed method was experimentally validated through a real bridge. The results show that the proposed method can find a best FE model with optimal estimates of the updating parameters with much less computational efforts against the conventional FEMU.

This paper presents a multi-functional strand capable of introducing prestressing force in prestressed concrete (PSC) girders and sensing their static and dynamic behavior as well. This innovative strand is developed by replacing the core steel wire of the strand used in PSC structures with a carbon fiber-reinforced polymer (CFRP) wire with a built-in optical Fiber Bragg Grating (FBG) sensor. A full-scale girder specimen was fabricated by applying this multi-function strand to check the possibility of tracking the change of prestressing force at each construction stage. Moreover, dynamic data could be secured during dynamic loading tests without installing accelerometers and made it possible to obtain the natural frequencies of the structure. The results verified the capability to effectively manage the prestressing force in the PSC bridge structure by applying the PC strand with a built-in optical sensor known for its outstanding practicability and durability.

Computer experiments are widely used to evaluate the performance and reliability of engineering systems with the lowest possible time and cost. Sometimes, a high-fidelity model is required to ensure predictive accuracy; this becomes computationally intensive when many computational analyses are required (for example, inverse analysis or uncertainty analysis). In this context, a surrogate model can play a valuable role in addressing computational issues. Surrogate models are fast approximations of high-fidelity models. One efficient way for surrogate modeling is the sequential sampling (SS) method. The SS method sequentially adds samples to refine the surrogate model. This paper proposes a multiple-update-infill sampling method using a minimum energy design to improve the global quality of the surrogate model. The minimum energy design was recently developed for global optimization to find multiple optima. The proposed method was evaluated with other multiple-update-infill sampling methods in terms of convergence, accuracy, sampling efficiency, and computational cost.

Background Perineural invasion (PNI) has emerged as an important factor related to colorectal cancer spread; however, the impact of neoadjuvant chemoradiotherapy (nCRT) on PNI remains unclear. Herein, we investigated the prognostic value of PNI, along with lymphovascular invasion (LVI), in rectal cancer patients treated with nCRT. Methods This single-center observational study of pathologic variables, including PNI and LVI, analyzed 1411 invasive rectal cancer patients (965 and 446 patients treated with primary resection and nCRT, respectively). Results The overall detection rates of LVI and PNI were 16.7 and 28.8%, respectively. The incidence of LVI was significantly lower in patients treated with nCRT (8.1 vs. 20.6%, P  < .001); this was confirmed by multivariate analysis. However, PNI was not affected by nCRT (with nCRT 28.3% vs. without nCRT 29.1%, P  = .786). In the 446 patients with nCRT, multivariate analysis revealed that PNI was an independent prognostic factor for both disease-free survival (DFS) and overall survival (OS). For the prediction of both 5-year DFS and OS, the C-index for the combinations of T-stage with the PNI (TPNI) system showed favorable result, especially in patients with a total number of harvested lymph nodes <8. Conclusion PNI is a meaningful prognostic factor for rectal cancer patients treated with nCRT, especially when <8 lymph nodes are harvested. The lack of influence of nCRT on the PNI incidence suggests that residual tumor cells with PNI are more radioresistant or biologically aggressive than those without.

High quality and nutritional benefits are ultimately the desirable features that influence the commercial value and market share of broad bean (Vicia faba L.). Different cultivars vary greatly in taste, flavor, and nutrition. However, the molecular basis of these traits remains largely unknown. Here, the grain metabolites of the superior Chinese landrace Cixidabaican (CX) were detected by a widely targeted metabolomics approach and compared with the main cultivar Lingxiyicun (LX) from Japan. The analyses of global metabolic variations revealed a total of 149 differentially abundant metabolites (DAMs) were identified between these two genotypes. Among them, 84 and 65 were up- and down-regulated in CX compared with LX. Most of the DAMs were closely related to healthy eating substances known for their antioxidant and anti-cancer properties, and some others were involved in the taste formation. The KEGG-based classification further revealed that these DAMs were significantly enriched in 21 metabolic pathways, particularly in flavone and flavonol biosynthesis. The differences in key secondary metabolites, including flavonoids, terpenoids, amino acid derivates, and alkaloids, may lead to more nutritional value in a healthy diet and better adaptability for the seed germination of CX. The present results provide important insights into the taste/quality-forming mechanisms and contributes to the conservation and utilization of germplasm resources for breeding broad bean with superior eating quality.

PurposePrevious studies have reported paradoxical survival prognoses for some node-negative and node-positive colon cancer patients. However, current guidelines recommend adjuvant chemotherapy (CT) only for node-positive patients. This study investigated the efficacy of adjuvant CT for patients who underwent radical surgery for colon cancer with solitary lymph node (LN) metastasis.MethodsThis study included 281 patients treated between 2004 and 2015. Patients were classified into no-CT (n = 39) and CT (n = 242) groups, and the survival outcomes and recurrence-related follow-up data were analyzed.ResultsThe groups exhibited similarities in tumor sidedness, tumor differentiation, and pathologic stage. However, the age, ASA class, and preoperative CEA level were relatively lower in the CT group. Although the CT group had a higher 5-year overall survival (OS) rate than the no-CT group (88.4% vs. 65.3%, p < 0.001), the groups did not differ in terms of 5-year disease-free survival (DFS) (CT, 84.1% vs. no-CT, 83.3%, p = 0.490). A multivariate analysis identified adjuvant CT as an independent factor for OS but not for DFS. A highly examined LN count (≥ 12) was associated with improved DFS improvement. However, D3 LN dissection was not associated with DFS or OS. For DFS, intermediate/apical positive LNs received a high hazard ratio relative to pericolic/epicolic LNs (2.080, 95% confidence interval: 0.979–4.416), but this was not significant (p = 0.057).ConclusionsAdjuvant chemotherapy did not provide clear advantages for colon cancer with solitary LN metastasis. Further large studies that analyze several prognostic factors are needed to establish tailored adjuvant CT administration guidelines.