Applied Mechanics and Materials Vols. 397-400

Abstract: The research as identity recognition using photographs evoked EEG signal, results show that it is easy to identify subjects, but it is in the subjects with the experimental condition, when the subjects don't fit or deliberately wrong, generate camouflage signal, what impact will the identification, existing research rarely the subjects involved, camouflage case EEG and under the condition of normal EEG signal difference, carries on the analysis from the brainwave components, in order to find the similarities and differences between them, to provide the reference for the identification .

Abstract: This paper mainly description image enhance based on wavelet transform. The author gave the theory of wavelet analysis, the application of wavelet transform in image enhancement. The impression drawing is given by means of MATLAB simulation. Through tone up to disperse wavelet transform, rebuild coefficient, the siginal of image is more boost up in low contrast section than high contrast section, the margin is distinct.

Abstract: In this paper, a data acquisition subsystem for aluminum reduction cells under reconstruction is designed on memory mapped file and MODBUS, has realized the sampling procedure and the main control program of real-time data transmission through the memory mapped file, has realized the MODBUS communication between the sampling procedure and the prestage using MSComm control, the system can work stably at present.

Abstract: The cased hole resistivity logging technology is one of high-tech logging technologies being studied in China. The signal source requires high voltage, high current, ultra-low frequency, high stability, high resolution, etc. Based on the characteristics, proposed were the ideas that the direct digital synthesis technique should be applied to produce the required sine logging exciting signals, and high-power FET were used to produce the signals with high power and high voltage. The results show that the output frequency of signal source is 0~ 10 Hz, frequency resolution 0. 002 Hz, output voltage range -300 V~+300V, output current less than 6 A. The design has been successfully applied in the research project of extremely weak nV-level electric signal acquisition theory and experimental study on formation resistivity measurement through metal casing.

Abstract: Gas insulated switchgear (GIS) has been widely used in global power grids currently. However, there are several disadvantages of GIS which will cause partial discharge (PD) in the insulation and then make GIS shut down. In this paper, the application of the ultra high frequency (UHF)method for partial discharge monitoring is discussed in detail, including the placement of the UHF sensors , detection system and the different characteristics of the different defects in the PD signals diagram are analyzed as well. Finally, some relevant conclusions are drawn based on Finite-Diffrenee Time-Domain algorithm (FDTD) for simulation of the electromagnetic waves by UHF PD detection technology.

Abstract: One of the major drawbacks of orthogonal frequency division multiplexing (OFDM) signals is the high peak to average power ratio (PAPR) of the transmitted signal. In this paper, a new method based on hexagonal constellation is proposed to reduce PAPR and to eliminate side information for OFDM signals. In order to get better PAPR reduction performance, we propose a new PAPR reduction algorithm based on parametric minimum cross entropy (PMCE). Simulation results show that PMCE algorithm is an efficient method to achieve significant PAPR reduction.

Abstract: When the mass data information for drug contains no real or noise effects of information, the general data mining analysis method will have a great advantage, the traditional medicine association behavior is used in the analysis of a categories data mining analysis method, ignore drug data analysis results and the practical next behavior prediction contact. In order to solve the problem put forward the k-medoids and c5.0 joint constraints drug data mining methods, and in the first step of clustering analysis of fully considering the effect of noise and isolated points, with the first step clustering results were late to the classification of the decision tree as data sample, and the appropriate data pretreatment, so it can guarantee the accuracy of the calculation model. Experiments prove this drug data analysis models are available, the mining efficiency is higher.

Abstract: In current, there are complex relationship between the assets of information security product. According to this characteristic, we propose a new asset recognition algorithm (ART) on the improvement of the C4.5 decision tree algorithm, and analyze the computational complexity and space complexity of the proposed algorithm. Finally, we demonstrate that our algorithm is more precise than C4.5 algorithm in asset recognition by an application example whose result verifies the availability of our algorithm. Keywordsdecision tree, information security product, asset recognition, C4.5

Abstract: This paper is concerned with the problem of license plate recognition of vehicles. A recognition algorithm based on dynamic sliding window to binarize license plate characters is proposed. While a connected domain approach is presented to cope with the degradation characters. There are three steps to recognize the characters. First, the characters are classified by their features. Then, based on such classification a grid method is used to construct the feature vector. Finally, least square support vector machine is employed to recognize these characters. The test results show the high recognition rate and also illustrate the effectiveness of the proposed algorithm.

Abstract: With the era of big data approaching, data becomes more and more important. Faced with such massive amounts of data space, how to quickly identify the contents of a field that the users are interest in and extract them out, is an urgent problem to be solved. To identify the content that users are interested in, we can use NLPIR Chinese word segmentation framework for speech segmentation, and identify named entity according to part of speech tagging. For extraction, using Hadoop, parallel cluster platform based on a big data MapReduce framework, using the Hadoop Distributed File System (HDFS) for efficient data access and starting Map and Reduce tasks to extract the information of named entity. This task extracts the required information from the interactive encyclopedia and then stores them in the knowledge base. It implements the task of extracting the information data of parallelization of named entity based on Hadoop platform.