Applied Mechanics and Materials Vols. 543-547

Abstract: Based on the structure analysis of certain mortal-howitzer breech mechanism, virtual prototype is established by using Pro/E software an ADAMS. Dynamics models of breech mechanisms breechblock opening plate, closing lever and closing spring are established by simulating breech mechanisms working process when mortal-howitzer is during a shoot and manual. Simulate breech mechanisms working process when mortal-howitzer is during a shoot and manual. Such as, model of driving force during a shoot, the simulation of breech mechanism movement characters, and Dynamics simulation of breech mechanism in the course of contact. These models are carried out simulation experiment. The result shows simulation result is consistent with actual working condition of breech mechanism and proves validity of designed models.

Abstract: Vortex detection for surface flow field is a challenging issue in feature visualization. This paper presents a new vortex detection technique for surface flow field based on rotation index. Rotation index is the statistical quantity of vectors around a point, and indicates the rotation character of the sample region around the point. Then, an ELIC-based IBFVS method is applied to visualize the result flow field. A discussion of results is included to demonstrate our algorithm which leads to a more reasonable visualization results.

Abstract: Texture adaptation is a challenging issue in texture-based feature visualization. In order to visualize as more information as we can, this paper presents a texture adaptation technique based on extended information entropy. A new definition of information entropy for 3D vector field is proposed to quantitatively represent the information carried by them. A discussion of results is included to demonstrate our algorithm which leads to a more reasonable visualization results based on texture adaption with information entropy.

Abstract: The complexity of optimization problems encountered in various modeling algorithms makes a selection of a proper optimization vehicle crucial. Developments in particle swarm algorithm since its origin along with its benefits and drawbacks are mainly discussed as particle swarm optimization provides a simple realization mechanism and high convergence speed. We discuss several developments for single-objective case problem and multi-objective case problem.

Abstract: According to students' employment problem, employment data mining model of university graduates is presented. The decision tree is very effective means for classification, which is proposed according to the characteristics of employment data and C4.5 algorithm. The C4.5 algorithm is improved from ID3 algorithm that is the core algorithm in the decision tree. The C4.5 algorithm is suitable for its simple construction, high processing speed and easy implementation. The model includes preprocess of the data of employment selection of decision attributes, implementation of mining algorithm, and obtainment of rules from the decision tree. The rules point out which decision attributes decide the classification of employers. Case study shows that the decision tree algorithm applied to employment information data mining, can classify data of employment correctly with simple structure and faster speed, and find some valuable results for analysis and decision. so the proposed algorithm in this paper is effective.

Abstract: With the advent of the era of big data, people increasingly strong demand for intelligent information. How to make news industry provide more accurate and efficient service for people has become a significant issue. This paper describes how to build a predictable model about Chinese news classification using Bayesian algorithms, and discusses the performance of Naive Bayesian model in different dataset environments, where contains unbalanced data and equilibrium data.

Abstract: In order to solve the problem that the speed of augmenting in NEAT algorithm is obviously not fast enough. We bring in the hybridization scheme. The scheme mainly gets new progeny chromosomes by combining parents ones. But it will probably bring in abnormal genes. As we record the time stamp for every gene, we can easily ensure that those genes are the same, so that we can simply get those completely reasonable hybrid chromosomes. We mainly use the time stamp of every gene to record information and make progeny chromosomes.

Abstract: Consumer reviews have been considered as a golden mine which can be used to analysis consumers preferences, attitudes and opinions about products or services or online stores. How to exploit this mine effectively has attracted researchers attention. In this paper, a new classification method based on fuzzy cognitive maps is proposal. The experience results showed the approach is more effective way of classifying Chinese reviews based on calculating the precision rate and the recall rate.

Abstract: In this paper, we compare the estimation performances of 7 different kernels (i.e., Uniform, Triangular, Epanechnikov, Biweight, Triweight, Cosine and Gaussian) when using them to conduct the probability density estimation with Parzen window method. We firstly analyze the efficiencies of these 7 kernels and then compare their estimation errors measured by mean squared error (MSE). The theoretical analysis and the experimental comparisons show that the mostly-used Gaussian kernel is not the best choice for the probability density estimation, of which the efficiency is low and estimation error is high. The derived conclusions give some guidelines for the selection of kernel in the practical application of probability density estimation.

Abstract: The kernel was the key technology of SVM; the kernel affected the learning ability and generalization ability of support vector machine. Aiming at the specific application of harmful text information recognition, combining traditional kernel function the paper structured a new combination kernel, modeling for the independent harmful vocabulary and co-occur vocabularies, and then evaluation the linear kernel, homogeneous polynomial kernel, non homogeneous polynomial kernel and combination kernel function in the sample experiment. The experimental results of combination kernel function showed that the effect has increased greatly than other kernel functions for the application of harmful text information filtering. Especially the Rcall value achieved satisfactory results.