Applied Mechanics and Materials Vols. 198-199

Abstract: AR model is widely used which based on stationary time series used for short-term prediction. However, in fact the time series we got is often non-stationary, and there is little literature researching the smooth processing, modeling and forecasting and then restoring the results in system. In view of this, this paper provides a method, that is, differential autoregressive of cycle prediction. First, explain the basic principles and give the calculation steps of smooth processing, modeling and forecasting and restoring the results. Then, applied the prediction method in the short-term forecast of coal arrive of a provincial. Model implementation is based on java programming. We get high prediction accuracy, the system easily integrated, can be widely used, and can achieve rolling forecast.

Abstract: When solving large scale combinatorial optimization problems, Max-Min Ant System requires long computation time. MPI-based Parallel Max-Min Ant System described in this paper can ensure the quality of the solution, as well as reduce the computation time. Numerical experiments on the multi-node cluster system show that when solving the traveling salesman problem, MPI-based Parallel Max-Min Ant System can get better computational efficiency.

Abstract: The technology industry development project is a carrier of technology industry. Scientific and fair projects optimization method can improve the efficiency of decision-making, the effectiveness of industry resource allocation and investment efficiency, the market competitiveness of technology products. With expert advice and relevant literature, we can establish a comprehensive evaluation index system of science and technology industrialization projects. Base on Tianjin city in the three same type of science and technology Industrialization Projects in 2010, by using Fuzzy AHP comprehensive evaluation, the projects were sequenced to determine the priority of the items, and then which provide a reference for policy makers.

Abstract: Gaussian processes (GPs) is a very promising technology that has been applied both in the regression problem and the classification problem. In recent years, models based on Gaussian process priors have attracted much attention in the machine learning. Binary (or two-class, C=2) classification using Gaussian process is a very well-developed method. In this paper, a Multi-classification (C>2) method is illustrated, which is based on Binary GPs classification. A good accuracy can be obtained through this method. Meanwhile, a comparison about decision time and accuracy between this method and Support Vector Machine (SVM) is made during the experiments.

Abstract: In multi-objective particle swarm optimization (MOPSO), the selection of global guides for all partials is vital to improve the convergence and diversity of solutions. In this paper, the related work of global guides searching in MOPSO is introduced, and a new Pareto–based selecting strategy is proposed. Basing on the analysis of the structure and mapping relation of the particle swarm and the nondominated solutions archive, considering the density information, the global guides selecting frequency and other factors, a new gbest selecting strategy for each particle in the swam is presented. Experimental results of contrasting experiments of two typical MOPSO functions demonstrate that the proposed strategy is satisfying.

Abstract: The rapid growth of biological texts promotes the study of text mining which focuses on mining biological knowledge in various unstructured documents. Meanwhile, most biological text mining efforts are based on identifying biological terms such as gene and protein names. Therefore, how to identify biological terms effectively from text has become one of the important problems in bioinformatics. Conditional random fields (CRFs), an important machine learning algorithm, are graphical models for modeling the probability of labels given the observations. They have traditionally been trained with using a set of observation and label pairs. Here we use CRFs in a class of temporal learning algorithms, reinforcement learning. Consequently the labels are actions that update the environment and affect the next observation. As a result, from the view of reinforcement learning, CRFs provide a way to model joint actions in a decentralized Markov decision process, which define how agents can communicate with each other to choose the optimal joint action. We use GENIA corpus to carry on training and testing the proposed approach. The result showed the system could find out biological terms from texts effectively. We get average precision rate=90.8%, average recall rate=90.6%, and average F1 rate=90.6% on six classes of biological terms. The results are pretty better than many other biological named entity recognition systems.

Abstract: In this paper, a new method of the recurrence analysis pitch detection of nonlinear dynamical characteristics for speech signals is designed，which calculated firstly the pitch by recurrence quantification，and then distinguished accurately voiced/unvoiced by the product of the recurrence degree and the pitch, and modified the fluctuating pitch. The results show that the new method performance is better than the conventional autocorrelation algorithm and cepstrum method，especially in the part that the surd and the sonant are not evident, and get a high robustness in noisy environment.

Abstract: The research of joint replenishment problem in multi-item spare part inventory is an important orientation of inventory control, and it is difficult to solve. A model for continuity joint replenishment in stochastic demand is advanced. And the arithmetic is designed. Firstly, this arithmetic reserves reorder point of items according to the constraints of spare part supportability. Secondly, the cost and parameters of policy is computed. Finally, a solution is advanced for multi-item inventory cost. At last, the results indicate that the heuristic algorithm can effectively solve the problem.

Abstract: X-ray radiography has more and more application value and X-ray simulation system has practical significance for improving the quality of X-ray imaging and replacing the expensive devices but it’s a time-consuming work. This paper proposes an improved fast ray-tracing method extended from Siddon’s and Zhao’s methods. For a 2D array, we compute level by level which has natural modes, by comparing the incident x-coordinate with the characteristic constant. In the case of 3D models, the volume data are back projected into 1D linked list and compute indices and lengths similarly like the case of 2D. Compared with Zhao’s method, the new method avoids skipping parametric planes and computing several starting voxels and ending voxels in one level. The time consumed in the new algorithm has reduced by 5/6 regarding the conventional Siddon’s algorithm.

Abstract: The classical rough set can not show the fuzziness and the importance of objects in decision procedure because it uses definite form to express each object. In order to solve this problem, this paper firstly introduces a special decision table in which each object has a membership degree to show its fuzziness and has been assigned a weight to show its importance in decision procedure. And then, the special decision table is studied and the relevant rough set model is provided. In the meantime, relevant definitions and theorems are proposed. On the above basis, an attribute reduction algorithm is presented. Finally, feasibility of the relevant rough set model and the presented attribute reduction algorithm are verified by an example.