A self‐correction technique for real‐time simulation of regulated rivers

An automatic self‐correction scheme is presented for online adjustment of stream‐flow forecasts during simulation of river systems with regulating structures, when actual flow data for interior points are available in real time. Another scheme is presented which allows for the modelling of channel losses due to physical effects such as evapotranspiration, bank storage and diversions. A previously developed river simulation/reservoir management model has been modified to include both schemes. The model is particularly useful for real‐time, short‐term flood forecasting and reservoir operation. The model application is made to the Lower Colorado River in Texas.     

Aromatic products of 340 °C supercritical-toluene extraction of two Turkish lignites: an n.m.r. study

Fractions of Elbistan and Seyitomer (Turkish) lignites, extracted with supercritical toluene at 340 °C and 8 MPa, have been separated by solvent extraction and silica-gel chromatography. Analyses by n.m.r. and i.r. spectroscopies and other methods have been combined in structural-analysis schemes to yield information about the average molecule in aromatic extracts. Carbon aromaticities, f_a, derived from 22.63 MHz ¹H-decoupled pulse Fourier-transform (PFT) ¹³C-n.m.r. are more widely spread for Elbistan (0.34–0.56) than for Seyitomer (0.40–0.43), and are lower than for supercritical-gas (SCG) products from bituminous coals. ¹³C-n.m.r. also reveals the presence of aromatic ether-O in polar fractions. Narrow aromatic signals in 100 MHz ¹H-n.m.r. spectra suggest the presence of single-aromatic-ring average structures. In the hexane-soluble aromatics, 27% (Elbistan) and 29% (Seyitomer) of the available sites are substituted by alkyI groups, some of which are at least eight carbon atoms long; the hexane-soluble polar and asphaltene/asphaltol fractions contain fewer such groups.     

Sr2+–Mg2+-Doped SiAlON Ceramics

In the present work, multi-cation-doped (Sr²⁺–Mg²⁺) SiAlON ceramics were investigated. MgO and SrO were used in 100:0 and 50:50 molar ratios. The mixture was sintered at 1800° and 1830°C for 1 h in a gas pressure-sintering furnace. The results showed that sintered samples were composed of mainly α- and β-SiAlON phases and small amounts of some Sr-containing phases and SiAlON polytypes. According to Rietveld analysis of X-ray diffraction patterns, Mg is incorporated into the α-SiAlON structure. However, the incorporation of Sr is limited.     

A two-phase variable neighbourhood search algorithm for assembly line worker assignment and balancing problem type-II: an industrial case study

The assembly line worker assignment and balancing problem type-II (ALWABP-2) occurs when workers and tasks (where task times depend on workers’ skills) are to be simultaneously assigned to a fixed number of workstations with the goal of minimising the cycle time. In this study, a two-phase variable neighbourhood search (VNS) algorithm is proposed to solve the ALWABP-2 due to the NP-hard nature of this problem. In the first phase of the algorithm, a VNS approach is applied to assign tasks to workstations with the aim of minimising the cycle time while in the second phase, a variable neighbourhood descent method is applied to assign workers to workstations. The performance of the proposed algorithm is tested on well-known benchmark instances. In addition, the proposed algorithm has been used to solve a real case study from a consumer electronics company that manufactures LCD TVs. The results show that the algorithm is superior to the methods reported in the literature in terms of its higher efficiency and robustness. Furthermore, the algorithm is easy to implement and significantly improves the performance of the final assembly line for the investigated LCD TV real case study.     

Eigenclassifiers for combining correlated classifiers

In practice, classifiers in an ensemble are not independent. This paper is the continuation of our previous work on ensemble subset selection [A. Ula?, M. Semerci, O.T. Y?ld?z, E. Alpayd?n, Incremental construction of classifier and discriminant ensembles, Information Sciences, 179 (9) (2009) 1298–1318] and has two parts: first, we investigate the effect of four factors on correlation: (i) algorithms used for training, (ii) hyperparameters of the algorithms, (iii) resampled training sets, (iv) input feature subsets. Simulations using 14 classifiers on 38 data sets indicate that hyperparameters and overlapping training sets have higher effect on positive correlation than features and algorithms. Second, we propose postprocessing before fusing using principal component analysis (PCA) to form uncorrelated eigenclassifiers from a set of correlated experts. Combining the information from all classifiers may be better than subset selection where some base classifiers are pruned before combination, because using all allows redundancy.     

Cost-conscious comparison of supervised learning algorithms over multiple data sets

In the literature, there exist statistical tests to compare supervised learning algorithms on multiple data sets in terms of accuracy but they do not always generate an ordering. We propose Multi²Test, a generalization of our previous work, for ordering multiple learning algorithms on multiple data sets from “best” to “worst” where our goodness measure is composed of a prior cost term additional to generalization error. Our simulations show that Multi²Test generates orderings using pairwise tests on error and different types of cost using time and space complexity of the learning algorithms.     

Output-associative RVM regression for dimensional and continuous emotion prediction

Many problems in machine learning and computer vision consist of predicting multi-dimensional output vectors given a specific set of input features. In many of these problems, there exist inherent temporal and spatial dependencies between the output vectors, as well as repeating output patterns and input–output associations, that can provide more robust and accurate predictors when modeled properly. With this intrinsic motivation, we propose a novel Output-Associative Relevance Vector Machine (OA-RVM) regression framework that augments the traditional RVM regression by being able to learn non-linear input and output dependencies. Instead of depending solely on the input patterns, OA-RVM models output covariances within a predefined temporal window, thus capturing past, current and future context. As a result, output patterns manifested in the training data are captured within a formal probabilistic framework, and subsequently used during inference. As a proof of concept, we target the highly challenging problem of dimensional and continuous prediction of emotions, and evaluate the proposed framework by focusing on the case of multiple nonverbal cues, namely facial expressions, shoulder movements and audio cues. We demonstrate the advantages of the proposed OA-RVM regression by performing subject-independent evaluation using the SAL database that constitutes naturalistic conversational interactions. The experimental results show that OA-RVM regression outperforms the traditional RVM and SVM regression approaches in terms of accuracy of the prediction (evaluated using the Root Mean Squared Error) and structure of the prediction (evaluated using the correlation coefficient), generating more accurate and robust prediction models.     

Prioritization methodology of dangerous substances for water quality monitoring with scarce data

Monitoring and control of dangerous substances discharged into receiving waters have attracted more attention lately. Since it is not possible to analyze every single substance, a prioritization methodology is needed for the selection of those to be monitored. Existing well-developed models require significant amount of data for reliable outcomes. This paper presents a methodology to prioritize the dangerous substances having adverse effects on freshwaters in Turkey, where data are scarce. Such a methodology will also serve as a solid model for other countries with limited background data. The adopted methodology enabled the elimination of chemicals to generate a candidate list composed of 608 substances among more than 5000 substances. Further screening and prioritization were conducted using different assessment methods (i.e., Total Hazard Value, Total Impact Value, Combined Monitoring-based, and Modelling-based Priority Setting) to obtain a proposed Final Candidate Specific Pollutants List of 150 dangerous substances. The proposed Candidate National Pollutant List of Turkey was established by combining 45 priority pollutants of the European Union with a list of candidate specific pollutants. According to the outcomes of this study, monitoring and controlling of 195 dangerous substances in freshwaters are recommended. Further detailed studies should be conducted in order to observe the actual levels of these dangerous substances in freshwaters followed by a review of the monitoring list accordingly. Moreover, further revisions might be required in the proposed list due to some possible versatile conditions in terms of sampling points (i.e., change in the location of industries).     

Effect of Deep Cryogenic Treatment on Wear Resistance of AISI 52100 Bearing Steel

In this study, the effects of deep cryogenic treatment (DCT) on the wear resistance of AISI 52100 bearing steel were investigated. For this purpose, a number of bearing steel samples were held for different times (12, 24, 36, 48, 60 h) at deep cryogenic temperatures (?145 °C). The wear experiments were carried out in a ball–disk arrangement, by applying loads of 10 and 20 N and a sliding velocity of 0.15 m/s. After conducting the experimental studies, 36 h was found to be the optimal holding time. At this holding time, the wear rate and friction coefficient were decreased, while the hardness reached to maximum values. It was observed that DCT led to significant microstructural changes, which resulted in improved tribological properties.