Regularized nonnegative shared subspace learning

Joint modeling of related data sources has the potential to improve various data mining tasks such as transfer learning, multitask clustering, information retrieval etc. However, diversity among various data sources might outweigh the advantages of the joint modeling, and thus may result in performance degradations. To this end, we propose a regularized shared subspace learning framework, which can exploit the mutual strengths of related data sources while being immune to the effects of the variabilities of each source. This is achieved by further imposing a mutual orthogonality constraint on the constituent subspaces which segregates the common patterns from the source specific patterns, and thus, avoids performance degradations. Our approach is rooted in nonnegative matrix factorization and extends it further to enable joint analysis of related data sources. Experiments performed using three real world data sets for both retrieval and clustering applications demonstrate the benefits of regularization and validate the effectiveness of the model. Our proposed solution provides a formal framework appropriate for jointly analyzing related data sources and therefore, it is applicable to a wider context in data mining.     

Synthesis of highly concentrated suspension of chemically converted graphene in organic solvents: Effect of temperature on the extent of reduction and dispersibility

We report the effect of temperature on the extent of graphene oxide reduction by hydrazine and the dispersibility of the resulting chemically converted graphene (CCG) in polar organic solvents. The extent of graphene oxide reduction at high temperatures was only slightly higher than at low temperatures (30–50 °C), while the dispersibility of the resulting CCG in organic solvents decreased markedly with increasing temperature. The low dispersibility of CCGs prepared at high temperatures was greatly affected by reduction and influenced by the formation of an irreversible agglomerate of CCG at high temperatures. The reduction of graphene oxide at low temperatures is necessary to prepare highly dispersible CCG in organic solvents. CCG prepared at 30 °C is dispersible in N-methyl-2-pyrrolidone concentrations as high as 0.71 mg/mL. The free-standing paper made of this CCG possessed an electrical conductivity of more than 22,000 S/m, one of the highest values ever reported.     

A novel inverse modeling control for piezo positioning stage

Journal of Mechanical Science and Technology - This paper focuses on the application of a novel inverse nonlinear autoregressive with exogenous input (NARX) structure and a fuzzy inference system...     

Modeling of microstructure effects on the mechanical behavior of ultrafine-grained nickels processed by hot isostatic pressing

Bulk ultrafine-grained nickel specimens having grain sizes in the range of 0.25-5 μm were consolidated by hot isostatic pressing technique. The resulting microstructures were characterized by transmission electron microscopy and X-ray diffraction analysis. Compression tests were carried out at room temperature and at strain rate of 1.6×10⁻⁴ s⁻¹. It was found that the measured yield strength does not follow the Hall-Petch law as a consequence of the presence of oxide phase. Therefore, the use of micromechanics based model, which takes into account only the Hall-Petch relationship at grain level for predicting the grain sized effects on mechanical behavior of this kind of materials, is not accurate yet. In this study, a modification made to the generalized self-consistent model was proposed for studying both grain size and oxide phase dependence of ultrafine-grained materials behavior. Because of the novel modification, an optimization procedure with two steps was required to identify the parameters of micromechanical model. An acceptable agreement between experimental and numerical results was achieved. Moreover, the influence of texture on the yield strength and the application of the proposed model to the spark plasma sintering processed materials were also discussed.     

Neurofuzzy control of a low‐temperature desulphurization plant

Low‐temperature desulphurization (LTD) is a low‐cost alternative to conventional wet scrubbing for removing sulphur dioxide from flue gas produced by power generating plants. A problem in the design of conventional controllers to achieve and maintain process conditions for optimal desulphurisation is the lack of mathematical models to characterise the complex desulphurization process and unexpected environ mental disturbances. The aim of the work reported in this paper was to replace a skilled human operator who could successfully regulate the LTD plant through manipulation of low‐level controllers with a competitive neurofuzzy system, which possesses both the learning ability of neural networks and the structural transparency of fuzzy logic systems. A hierarchical control structure was adopted whereby the competitive neurofuzzy method was used at the top level for co‐ordinating actions of low‐level conventional controllers. This approach would considerably simplify the task of designing the control system and has been shown to yield high‐level controllers with performances at least equalling that of expert operators, as demonstrated in this paper. Copyright © 2001 John Wiley & Sons, Ltd.     

Enhanced removal of 1,2-dichloroethane by anodophilic microbial consortia

1,2-Dichloroethane (1,2-DCA) is a well-known recalcitrant groundwater contaminant. New environment-friendly approaches for the removal of 1,2-DCA that does not bring about volatilization of the compound are required. In this study, different anodophilic consortia enriched in microbial fuel cells (MFCs) operated under airtight conditions were shown to effectively degrade 1,2-DCA (up to 102 mg per liter reactor volume per day), while concomitantly generating a current. An anodophilic consortium previously enriched with acetate as the electron donor changed its composition at the rate of 48% per week and increased its richness (Rr) 3-fold, upon adapting to 1,2-DCA as the new electron donor. After being stable, during 1 month of operation, it removed up to 95% of the 1,2-DCA amount in the medium in the first 2 weeks, while converting 43 ± 4% of electrons available from the removal to electricity. A natural consortium from a 1,2-DCA contaminated site changed its composition at the rate of 9% per week and increased its Rr 2-fold, upon adapting to the MFC anode conditions with 1,2-DCA as the electron donor. After being stable, during 1 month of operation, it removed up to 85% of the 1,2-DCA amount in the medium in the first 2 weeks and the coulombic efficiency was 25 ± 4%. The operation of the MFCs under closed circuit conditions resulted in higher 1,2-DCA removal rates than the operation under open circuit conditions, indicating that bioelectrochemical activities enhanced the removal of 1,2-DCA in the MFC anode. The production of ethylene glycol, acetate and carbon dioxide indicated that the anodophilic bacteria oxidatively metabolized 1,2-DCA, probably by means of a hydrolysis-based pathway. The results show that MFCs can be potentially used as a practically convenient technology for the biological removal of 1,2-DCA.     

THE VALIDATION OF THE ROBUST INPUT ESTIMATION APPROACH TO TWO-DIMENSIONAL INVERSE HEAT CONDUCTION PROBLEMS

A novel adaptive and robust input estimation inverse methodology of estimating the time-varying unknown heat flux, named as the input, on the two active boundaries of a 2-D inverse heat conduction problem is presented. The algorithm includes using the Kalman filter to propose a regression model between the residual innovation and the two thermal unknown boundaries flux through given 2-D heat conduction state-space models and noisy measurement sequence. Based on this regression equation, a recursive least-square estimator (RLSE) weighted by the forgetting factor is proposed to on-line estimate these unknowns. The adaptive and robust weighting technique is essential since unknowns input are time-varied and have unpredictable changing status. In this article, we provide the bandwidth analysis together with bias and variance tests to construct an efficient and robust forgetting factor as the ratio between the standard deviation of measurement and observable bias innovation at each time step. Herein, the unknowns are robustly and adaptively estimated under the system involving measurement noise, process error, and unpredictable change status of time-varying unknowns. The capabilities of the proposed algorithm are demonstrated through the comparison with the conventional input estimation algorithm and validated by two benchmark performance tests in 2-D cases. Results show that the proposed algorithm not only exhibits superior robust capability but also enhances the estimation performance and highly facilitates practical implementation.     

复合热稳定剂对绝缘纸协同抗老化效应的分子模拟研究

为了从微观层面探究热稳定剂的复合添加对绝缘纸协同抗热老化特性的影响机理,以双氰胺、三聚氰胺为热稳定剂研究对象,对老化过程中的热稳定剂、β-D-吡喃葡萄糖分子、水分子和小分子酸等进行了量子化学计算,利用前线轨道理论考察了不同分子之间的化学反应活性。此外,采用分子动力学模拟软件,搭建了末添加、单一添加和复合添加热稳定剂的四种纤维素含水含酸模型,从纤维素与水分子、蚁酸分子的氢键结合数、相互作用能这两个方面考察了不同改性体系的差异及抗老化性能差异的机理。结果表明:热稳定剂分子与水分子、小分子酸的化学反应活性要强于β-D-吡喃葡萄糖分子与这些物质的反应活性;两种热稳定剂分子分别与水分子、小分子酸的反应活性存在差异,复合添加可以集中二者的优势,起到协同延缓绝缘纸老化的效果;双氰胺与三聚氰胺的复合添加在降低纤维素分子对水分子、蚁酸分子的吸附能力上存在协同作用。