Search

The paper proposes an ontology construction approach that combines Fuzzy Formal Concept Analysis, Wikipedia and WordNet in a process that constructs multiple concept lattices for sub-domains. Those sub-domains are divided from the target domain. The multiple concept lattices approach can mine concepts and determine relations between concepts automatically, and construct domain ontology accordingly. This approach is suitable for the large domain or complex domain which contains obvious sub-domains.

In order to solve the problem of ignoring items’ attributes in single criteria collaborative filtering recommendation algorithm, a collaborative filtering recommendation algorithm based on multi-criteria decision making was proposed. Multi-criteria decision making was used during the computing of users’ similarity in the algorithm. By averaging all the similarities of sub evaluation with their weight, the algorithm chose the nearest neighbors. Finally, the algorithm used several nearest neighbors to get the recommendation results. Experiments show that the algorithm can find the nearest neighbors more similar after detailing evaluation, and make better predictions for the target users.

According to the different specifications of capacitor frame structure similar, designed by different designers, it will cause unnecessary duplication of labor, long design cycle, high cost, and the probability of error is relatively large, probably the same problem recurring. It causes not only the waste of manpower and resources, but also the existence of manufactured parts cannot be assembling a series of follow-up questions, it needs to be modified to normal use, causing great resistance to the research and development of new products. So that the competitiveness of enterprises in the same industries was at a disadvantage, based on the secondary development of Pro/E to solve the above problem.

Clostridium acetobutylicum parent strain was treated by ultraviolet irradiation at different minutes. The UV treated strains were cultivated in cornmeal medium with butanol added to the medium prior to cultivation. By increasing butanol content gradually in the medium, high butanol tolerance mutants were screened. Results show that mutants by UV treated for 4 minutes display best butanol tolerance and high butanol productivity. With 12g/L butanol added in the medium, the mutants still grow rapidly and accumulated 18.0g/L butanol at the end of 69 hr cultivation. Using sweet corn stalk juice as raw material, the maximum butanol produced by fermentation with the mutants reached 15.3g/L and 18.3g/L in batch and fed-batch fermentation respectively.

A series of Al-Cu-Si alloys were melted and spun into ribbons of about 100m thickness. The alloy compositions were selected to be suitable for brazing below 580C as filler metals. In the as-quenched state the foils were relatively brittle due to the occurrence of metal metastable phases. After appropriate annealing treatments between 400~450C the metastable phases were transformed into fine (Cu2Al) particles within the -Al matrix, and the mechanical properties of the Al-Cu-Si alloy ribbons were considerably improved.

Fractionation of softwood BCTMP pulp fiber was carried out with a cylindrical hydrocyclone. Pulp fiber length characterization in different streams have been examined using FQA. The fiber length fraction density function f is calculated and the separation index H(L) is used to evaluate fiber fractionation performance for different streams, and its specific expression was formed, so the new index may be widely used in the study of the evaluation of the separation performance of the hydrocyclone.

Aluminum substitution in CuInSe2could have important implications for photovoltaic applications. To better understand the Al doping effects, we have performed density functional calculations on the CuInSe2chalcopyrite as well as on Al-doped derivative compounds with different concentrations using the generalized gradient approximation. The structural and electronic properties of the pure and Al-doped CuInSe2have been calculated. We find that the substitution of In by Al creates structural deformation, and the band gap of CuIn1-xAlxSe2broadens as Al content increases. These are further discussed with the analysis of lattice parameters, bond lengths and angles, and electronic structures changes.

Based on density functional theory (DFT) within the framework of the generalized gradient approximation method, we have studied the Cd1-xZnxS sphalerite crystal structure and optical properties. We calculated electronic and optical properties of Cd1-xZnxS at the doping concentration x = 0, 0.25, 0.50, 0.75, 1.0. Optical properties (reflectivity, absorption coefficient, refractive index, dielectric function) and the electrical properties (band structure, electron density, etc.) are obtained including Zn-doing effects on the crystal structure, optical properties and electrical properties. With the increase of doping concentration x, the lattice parameter reduces from 0.5910 nm to 0.5409 nm; as a direct wide band gap semiconductor, its band gap increased from 1.15 to 2.22. Optical absorption coefficient increases with the increase of doping concentration, but the rate of increase is relatively small.

Secondary ion mass spectroscopy (SIMS) has been adopted to study the diffusion of Al and Si in Cu44.25Ag14.75Zr36Ti5bulk metallic glass (BMG). It has been found that around the transition temperature of metallic glass, the relation between its diffusion coefficient and the temperature satisfy the same Arrhenius relation, which means the metallic transition has not caused change to the diffusion mechanism. In addition, the radius of Al atom is close to that of Si atom, but under the same temperature and time condition, the diffusion coefficient of Si atom in bulk metallic glass (BMG) is twice that of the Al atom, while there is not a big difference in diffusion activation energy. This is because as non-metallic element, the radius of Si atom has a strong binding force with the metal atoms in the base material, which also has a bigger diffusion coefficient.

Partial occlusion and non-rigid variation are challenging problems in object tracking. To address this problem, robust gradient part-based models are proposed for object tracking in this paper. Our models constructed multiple well-chosen parts based on the gradient energy map of the object. And the local optimization algorithm, mean shift, is used to search the best locations of the multiple parts, which can be used to rectify the location of the tracked object by weighted feedback. Meanwhile, the models of the root and parts are updated online, which can improve tracking accuracy and robustness. Further, our models are easy to be embedded into different tracking algorithms and we implement the mean shift based on gradient part-based models. Experiments results show that our gradient part-based models are robust enough for object tracking, even though the objects are non-rigid or occluded.

To detect the targets in single frame infrared image, a target detection method based on multi-mixture filters model were proposed. The multi-mixture filters model involves four modules. Firstly, smooth filter is used to restrain and eliminate random noise. Secondly background Gaussian filter is used to construct a rough background model. Thirdly target Gaussian filter is used to construct an accurate target model. Finally morphologic filter is used to optimize the final target shape. Experiments show that this method can achieve a higher detection efficiency and accuracy.

In this paper, a novel approach is proposed for learning visual object classifiers with only positive images. Both positive and negative examples are collected from the positive images, and the false negative examples are filtered out by the Ramp Loss function, which has a strong ability of suppressing the influence of outliers. Meanwhile, the Latent Structural Support Vector Machines are adopted to estimate the best bounding boxes due to the unavoidable annotation errors. In learning the sub-problems, an extremely fast learning algorithm, CCCP-SGD, is proposed to optimize the Ramp Loss-based SVMs based on the Stochastic Gradient Descent algorithm. Experiments demonstrate that the CCCP-SGD algorithm can reduce the computing time both in the training and predicting phases, especially with a significant speedup in training, whilst it can also yield robust generalization performance when dataset is noisy. Besides, the visual object classifiers learnt by our approach with only positive images almost have the same performance as the classifiers learned with both positive and negative examples in object detection tasks.

Foldable structures are rapid to erect and easy to collapse for reuse. Four basically geometric design parameters of self-locking foldable latticed shell unit were defined. The recommended ranges of the geometric parameters and cross-section sizes were given by investigating foldable latticed shell units using ANSYS, with the consideration of element stability and the requirements of elastic design.

Engineering application of the flat tensegrity structures was studied based on the genetic algorithm. Optimal design of the tensegrity structure considering both the size of cross-section and prestress variables is presented and the related optimization problems are studied. Mutual influence and constraints of prestressed state and components were considered.

Using the Fiber Quantity Analyzer (FQA) for analysis of the paper pulp fiber moving in the Cylindrical Hydrocyclone, making use of the way of Limited Element Analysis and Calculation Fluid Dynamic for the numerical simulation of the velocity distribution of moving process. Based on the mathematic model of it, point out the idea of design on the geometric parameters of the cylindrical hydrocyclone. The result show, the suit length of pole is benefit to improve the machine handle ability and separation efficiency.

The optimal control strategy is introduced to improve the motion accuracy of planar 3-DOF parallel mechanism. The optimal control of error performance of mechanism is achieved by structuring and quantizing the quadratic form index of the error performance, and taking into account both the system response and energy control in state space, so we study the problem which can be boil down to solve two-point boundary value problem of differential equation. The simulation results show that the actual motion of mechanism keep the better tracking accuracy with the global motion while dynamics parameters is changing, and the controller is ease to accomplishment in the industrial due to the simply structure and the stable change of control energy.

Integrating and querying data from heterogeneous sources is a hot research topic in database research field. The emergence of the Semantic Web brings new paradigm shift of computing in data integration research where data is heterogeneous and distributed. To solve the problem this paper proposes a semantic-based approach. A Semantic-based Heterogeneous Relational Data Integration System, called SHRDIS, is presented. In this system, ontology is used as the mediated schema for the representation of the data source semantics. SPARQL queries over global schema are rewritten into local SQL queries that can be executed on heterogeneous relational databases. The architecture and implementation of SHRDIS is illustrated in detail. The experiment results show that the SHRDIS system has nice performance and scalability.