Table of Content

    20 June 2014
    Volume 44 Issue 3
    The weighted maximum vector angular margin core vector machinefor imbalanced data classification
    LU Shuxia, LI Limin
    JOURNAL OF SHANDONG UNIVERSITY (ENGINEERING SCIENCE). 2014, 44(3):  1-7.  doi:10.6040/j.issn.1672-3961.2.2013.376
    Abstract ( 710 )   PDF (906KB) ( 385 )   Save
    Related Articles | Metrics
    A new classification approach was proposed in order to deal with large size and imbalanced datasets classification problem. The proposed method was based on the maximum vector angular margin core vector machine to implement classification for large datasets. For the imbalanced datasets classification problem, each sample was assigned by different weights, which could improve the classification performance of algorithm. The proposed approach could effectively solve the imbalanced datasets classification problem and implement fast training on large datasets.
    Recommendation algorithm based on hierarchical item similarity
    SUN Yuanshuai, CHEN Yao, LIU Xiangrong, CHEN Ke, LIN Chen
    JOURNAL OF SHANDONG UNIVERSITY (ENGINEERING SCIENCE). 2014, 44(3):  8-14.  doi:10.6040/j.issn.1672-3961.1.2013.073
    Abstract ( 704 )   PDF (1491KB) ( 575 )   Save
    Related Articles | Metrics
    To solve the problem that CF(Collaborative Filtering) recommendation highly depends on the accurate similarity measurement, a novel recommendation algorithm based on item hierarchy similarity was proposed, which was named REHIS(Recommendation Hierarchical Similarity). The framework of REHIS was described as follows. First, the mining association rules and KNN (K Nearest Neighbor) algorithm were used to complement the hierarchy structure. Afterwards, the TopK method was employed to compute the similarity between items. Finally, scores were predicted by using the framework of itembased CF algorithm. On the other hand, to solve the CF poor scalability problem, the TopK algorithm were further extended to the cosine distance and Pearson correlation coefficient, both of which were commonly used similarity measurement methods. Experimental results showed that, compared with existing algorithms, REHIS could achieve a better recommendation in term of root mean square error, and TopK could reduce the time cost for searching the most similar items, too.
    Overfitting phenomenon  of  the series of single value decomposition algorithms in rating prediction
    CHEN Dawei, YAN Zhao*, LIU Haoyan
    JOURNAL OF SHANDONG UNIVERSITY (ENGINEERING SCIENCE). 2014, 44(3):  15-21.  doi:10.6040/j.issn.1672-3961.1.2013.380
    Abstract ( 948 )   PDF (1322KB) ( 2397 )   Save
    Related Articles | Metrics
    The collaborative filtering algorithm was improved to prevent overfitting in training rating prediction model. Experiments were put forth for  overfitting phenomenon  of the series of single value decomposition algorithms in rating prediction, and a method of adjusting parameters and iteration count to avoid overfitting phenomenon was proposed. The experimental results showed that this method could  find better rating prediction and avoid overfitting at the same time.
    A semi-supervised image retrieval algorithm based onfeature fusion of the region of interest
    KONG Chao1,2, ZHANG Huaxiang1,2*, LIU Li1,2
    JOURNAL OF SHANDONG UNIVERSITY (ENGINEERING SCIENCE). 2014, 44(3):  22-28.  doi:10.6040/j.issn.1672-3961.2.2013.281
    Abstract ( 703 )   PDF (2082KB) ( 691 )   Save
    Related Articles | Metrics
    A method of image retrieval based on the feature fusion of region of interest was proposed to realize the semantic correlation of images content. First, the regions of interest were divided and the integrated underlying characteristics of image were extracted. Second, the characteristics were used as training data to classify the images by semisupervised learning, then the mapping between images and categories of semantic was established. Finally, the quadratic distance and the improved Canberra distance were respectively used for measuring lowlevel features, and the cluster centers of images in the feature space were updated iteratively through positive feedback. The experiments compared with other algorithms showed that the proposed image retrieval algorithm had higher accuracy and performed more effectively than traditional algorithms.
    Semantic similarity computation of Baidu encyclopedia entries based on SimRank
    YIN Kun, YIN Hongfeng*, YANG Yan, JIA Zhen
    JOURNAL OF SHANDONG UNIVERSITY (ENGINEERING SCIENCE). 2014, 44(3):  29-35.  doi:10.6040/j.issn.1672-3961.2.2013.282
    Abstract ( 1017 )   PDF (1462KB) ( 2868 )   Save
    Related Articles | Metrics
    The measurement of the semantic similarity using semistructured data on Baidu encyclopedia was proposed. The encyclopedia entries and related entries were considered as two nodes of a directed graph, of which there was a link between two nodes. Then SimRank algorithm was used to calculate the semantic similarity of encyclopedia entries. Experimental results showed that the proposed measure significantly outperformed the traditional similarity measures, and might accurately reflect  the semantic relationship between words.
    A method of building Chinese microblog sentiment lexicon
    ZHOU Yongmei1, YANG Aimin1, LIN Jianghao2
    JOURNAL OF SHANDONG UNIVERSITY (ENGINEERING SCIENCE). 2014, 44(3):  36-40.  doi:10.6040/j.issn.1672-3961.1.2013.103
    Abstract ( 1210 )   PDF (1192KB) ( 1447 )   Save
    Related Articles | Metrics

    A method of building Chinese microblog sentiment lexicon was proposed,which adopted the discovery strategies of context entropy for network language, acquired network languages from the secondary filtration by TF-IDF and computed the sentiment weights of network language by SO-PMI algorithm in the labeled corpus. The built lexicon was applied into the analysis experiments of micro-blog sentiment,which was compared with that of naive bayesian classifier. Experiment results showed that the efficacy of classification by the built micro-blog sentimental lexicon was better than that by naive bayesian classifier,and was simple and rapid in the classification process.

    Transfer function modeling algorithm  for digital background calibration in Pipeline ADC
    GONG Yuehong, LUO Min*, YU Mingyan, JIN Jie
    JOURNAL OF SHANDONG UNIVERSITY (ENGINEERING SCIENCE). 2014, 44(3):  41-47.  doi:10.6040/j.issn.1672-3961.0.2013.113
    Abstract ( 875 )   PDF (1903KB) ( 540 )   Save
    Related Articles | Metrics
    In the background calibration, the residual amplifier transfer function modeling was applied to estimate the errors for improving the performance of pipeline analog to digital converter (ADC). To find a compromise among calibration resolution, hardware consumption, power consumption and the convergence time, an interpolation algorithm was needed to be chosen. Using a 12 bits 40 M sample rate pipeline ADC as prototype, the piecewise linear and cubic polynomial were respectively used to model the first stage inter-stage amplifier transfer function. The model was described by verilog, and the analog-digital mixed simulation was put forth by the analog circuit. The code was synthesized to estimate the overhead of the two algorithms. Simulation results showed that piecewise linear interpolation consumered less hardware and power consumption between these two algorithms, while the cubic polynomial interpolation was more precise and convergences faster.
    Welding defect detection method based on feature extraction and extreme searching
    LIANG Wei1,2, TAO Liang3, ZHANG Guangxian1, LI Zhenhua1
    JOURNAL OF SHANDONG UNIVERSITY (ENGINEERING SCIENCE). 2014, 44(3):  48-51.  doi:10.6040/j.issn.1672-3961.0.2014.013
    Abstract ( 680 )   PDF (1342KB) ( 468 )   Save
    Related Articles | Metrics
    A location detecting algorithm of the welding defect based on the grayscale distribution curve feature was proposed. At first, the welding line X-ray digital image was acquired by the flat panel detector. Secondly, because the problem of large noise and low dynamic range is ubiquitous, the time domain image frame sequence integration method was adopted to reduce the image noise and enhance the defect detail. Then, the optimal integration time was calculated. Thirdly, the grayscale value curve was drawn parallel to the welding line direction. The welding line image was segmented on the basis of its second derivative curve. At last, extreme point was searched in the welding line image. The size and location were decided if obvious maximum value was existed. Experimental results proved the high accuracy.
    Dynamic hand gesture recognition algorithm based on depth information
    YU Jing1, TIAN Guohui1*, YIN Jianqin2
    JOURNAL OF SHANDONG UNIVERSITY (ENGINEERING SCIENCE). 2014, 44(3):  52-56.  doi:10.6040/j.issn.1672-3961.0.2013.378
    Abstract ( 875 )   PDF (2177KB) ( 2314 )   Save
    Related Articles | Metrics
    To solve the complex calculation and the unreliable feature extraction of gesture recognition, a fast algorithm of dynamic gesture recognition based on the depth information of Kinect was proposed. Firstly,the depth camera of Kinect was used to get the depth image. Then, the method of threshold segmentation was used for image preprocessing, using OpenCV library and depth information to extract the foreground. Finally, Dynamic time warping algorithm calculated the similarity between test behavior template and reference behavior template for classifying the samples. The algorithm was realized under VS2010 by integrating OpenNI and OpenCV. Compared with other algorithms, this algorithm improved the extraction method of the dynamic gesture characteristics and the classification trajectory. Experimental results showed that the proposed method had a high recognition rate for the dynamic hand gestures with characteristics of time and space, and it had robustness under different illumination and complex background.
    Visual servo grasping of household objects for NAO robot
    YUAN Li, TIAN Guohui*, LI Guodong
    JOURNAL OF SHANDONG UNIVERSITY (ENGINEERING SCIENCE). 2014, 44(3):  57-63.  doi:10.6040/j.issn.1672-3961.0.2013.359
    Abstract ( 1036 )   PDF (2315KB) ( 2123 )   Save
    Related Articles | Metrics
    An improved positionbased visual servo (PBVS) grasping algorithm was proposed for the service robot in home environment. First, the rapid identification of object was achieved by using Naomark, and the position and pose information of the object was estimated through the world homography matrix decomposition. Secondly, the kinematic model of the NAO robot′s arm was established, and PBVS control law was designed for either unimanual or bimanual grasping. Finally, for purpose of improving the stability and robustness of the grasping operation, path planning of the endeffector was added to the original PBVS control law. Experimental results showed that the proposed method could grasp the objects rapidly and stably.
    Design of pre-amplifier circuit on blood alcohol detection  based on NIR
    LU Liqian, LI Zengyong*, CUI Ruofei, ZHOU Weiwei
    JOURNAL OF SHANDONG UNIVERSITY (ENGINEERING SCIENCE). 2014, 44(3):  64-68.  doi:10.6040/j.issn.1672-3961.0.2013.168
    Abstract ( 849 )   PDF (1818KB) ( 1025 )   Save
    Related Articles | Metrics
    A blood alcohol detection method was investigated based on NIR(near-infrared) spectroscopy.And a lownoise, low-drift and high-precision preamplifier was designed to detect the blood alcohol volume fraction accurately. In the current-voltage conversion section, the simulation of drift and error of two kinds of circuits were carried out by Multisim software.AD620 was selected in the preamplifier part which could improve the CMRR(common mode rejection ratio) of circuits. The results of simulation showed that the current-voltage conversion circuit of symmetrical compensational low-temperature drift can suppressed the drift and reduced error better. Compared with the OP07 current-voltage conversion circuit, the error could be reduced by 2%, while the draft could be reduced by 97%, which could support weak signal analyzing.
    Tolerance design of automotive body sheet metal parts based on a compliant assembly variation model
    YU Kuigang1,2, YANG Zhihong1,2*
    JOURNAL OF SHANDONG UNIVERSITY (ENGINEERING SCIENCE). 2014, 44(3):  69-74.  doi:10.6040/j.issn.1672-3961.0.2013.288
    Abstract ( 840 )   PDF (1095KB) ( 614 )   Save
    Related Articles | Metrics
     An optimal tolerance design method of compliant sheet metal parts was proposed based on an assembly variation model of compliant sheet metal parts. The average variance model of all measurement points obtained from the determinate models of measurement points was adopted as the objective function of the tolerance design, and the process capacity index and the variance ratio of parts were the constraint functions. The rear floor assembly was used to validate the tolerance design method. The results showed that tolerance designs of parts were synchronously done in the lap welding structure, and independently done in the butt welding structure while in the corner welding structure the tolerance design of the butt part was firstly done and then that of the lap part was done. When there existed a very large rigidness difference between lap parts, the less rigidness part could adopt a bigger tolerance value.
    Detection  method of karst caves in city subway based on  the cross-hole resistivity tomography
    ZHANG Wenjun, LI Shucai, SU Maoxin*, XUE Yiguo, QIU Daohong
    JOURNAL OF SHANDONG UNIVERSITY (ENGINEERING SCIENCE). 2014, 44(3):  75-82.  doi:10.6040/j.issn.1672-3961.0.2013.343
    Abstract ( 894 )   PDF (5213KB) ( 639 )   Save
    Related Articles | Metrics
    Karst caves were serious potential threats to the construction of subway. Four kinds of karst caves models were built and the numerical simulation were put forth by using the finite element and iteration method. The typical response characteristics were obtained to cross-hole resistivity tomography method of the models, and the response features were analyzed and summarized. Project cases in Dalian subway verified its favourable application effect in practice. Research results proved that cross-hole resistivity tomography method was an effective technology to karst cave detection and could help to ensure the safety of subway construction. Finally, based on numerical simulation and project cases, a few recommendations were given both in working method and data processing when using cross-hole resistivity tomography method to detect karst caves.
    The spatial model of vehicle-pavement coupling vibration and its dynamic responses analysis
    LIU Bo, WANG Youzhi*, AN Junjiang, WANG Yilin, YUAN Quan
    JOURNAL OF SHANDONG UNIVERSITY (ENGINEERING SCIENCE). 2014, 44(3):  83-89.  doi:10.6040/j.issn.1672-3961.0.2013.207
    Abstract ( 819 )   PDF (3377KB) ( 703 )   Save
    Related Articles | Metrics
    A fine spatial vehicle-pavement coupling vibration model was established to simulate the dynamic responses of rigid pavement under moving vehicle, in which the mass-spring-damper model was adopted for the vehicle and the finite element model based on the Winkler foundation was adopted for the concrete pavement. The improved harmonic superposition method was used to consider the 3-D distribution of road surface roughness, and the displacement coordination equation was used for the simultaneous solution of the vehicle and pavement vibration. Example analysis showed that the established vehiclepavement coupling vibration model could truly reflect the space geometric and mechanical coupling relationship between vehicle and pavement. With the stimulus of threedimensional road roughness, the dynamic response of the vehicle and pavement showed significant spatial distribution characteristics. Comparing with the results of the transient dynamic analysis method, the deflection and stress of the pavement significantly increased.
    Superplastic behavior of 1420Al-Li alloy
    ZHENG Lu1, ZHANG Yanling2, WANG Yaoqi2, HOU Hongliang2, FAN Tongxiang1*
    JOURNAL OF SHANDONG UNIVERSITY (ENGINEERING SCIENCE). 2014, 44(3):  90-94.  doi:10.6040/j.issn.1672-3961.0.2013.354
    Abstract ( 731 )   PDF (1947KB) ( 541 )   Save
    Related Articles | Metrics
    1420Al-Li alloy was used to explore the best deformation results including high elongation, uniform, and fine-grain structure by changing deformation temperature and strain rate. The results showed that, on the condition of deflection temperature T=480℃ and strain rate  =3×10-4s-1, 1420 Al-Li alloy showed good superplasticity, of which the maximum elongation δ=550% and the minimum flow stress σ=1.9MPa. Furthermore, the grain microstructure was isometric and fine after superplastic deformation through metallurgical analysis and TEM microstructure observation. Besides, the re-crystallization grains appeared which could promote the super-plastic deformation process during the deformation.