15.07.2014 Views

Conference Program of WCICA 2012

Conference Program of WCICA 2012

Conference Program of WCICA 2012

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

<strong>WCICA</strong> <strong>2012</strong><br />

Book <strong>of</strong> Abstracts: Friday Sessions<br />

Xie, Nan<br />

Tongji Univ.<br />

According to production tasks, making manufacturing resources reconstruction<br />

via machine tool selection is one <strong>of</strong> the key issues to achieve<br />

manufacturing system reorganization. At the time when machine tool<br />

is selected, qualitative factors and quantitative factors must be considered.<br />

In order to solve the problem well, the evaluation index system<br />

with a hierarchical structure was presented. A multi-objective optimization<br />

model was established, and then the model was analyzed by using<br />

hierarchical weighting methods. Finally, a practical application was provided<br />

to illustrate the scientific validity <strong>of</strong> the model and the effectiveness<br />

<strong>of</strong> the methodology.<br />

◁ PFrA-07<br />

A Hybrid Method for Short-term Load Forecasting in Power System,<br />

pp.696–699<br />

Zhu, Xianghe<br />

Qi, Huan<br />

Huang, Xuncheng<br />

Sun, Suqin<br />

Huazhong Univ.<strong>of</strong> Sci. & Tech.Wuchang Branch<br />

huazhong Univ. <strong>of</strong> Sci. & Tech.<br />

Electric Power <strong>of</strong> HeNan<br />

Electric Power <strong>of</strong> HeNan<br />

In order to improve the accuracy <strong>of</strong> power load forecasting, this paper<br />

proposes a hybrid model based on Ensemble Empirical Mode Decomposition<br />

(EEMD), least square-support vector machine (SVM) and BP<br />

nature network as a short-term load forecasting model. At first, the actual<br />

power load series is decomposed into different new series based<br />

on EEMD. Then the right parameters and kernel functions are chosen<br />

to build different LS-SVM model respectively, to forecast each intrinsic<br />

mode functions, due to the change regulation <strong>of</strong> each <strong>of</strong> all resulted<br />

intrinsic mode functions. Finally, we use the BP network to reconstruct<br />

the forecasted signals <strong>of</strong> the components and obtain the ultimate forecasting<br />

results. Simulation results show that the proposed forecasting<br />

method possesses accuracy.<br />

◁ PFrA-08<br />

Group Decision-Making Based Case Retrieval and Its Application,<br />

pp.773–778<br />

Zhang, Chun-xiao<br />

YAN, Aijun<br />

Zhao, Hui<br />

Wang, Pu<br />

Beijing Univ. <strong>of</strong> Tech.<br />

Beijing Univ. <strong>of</strong> Tech.<br />

Beijing Univ. <strong>of</strong> Tech.<br />

Beijing Univ. <strong>of</strong> Tech., China<br />

The distribution <strong>of</strong> the case feature attribute weights directly affects the<br />

case retrieval result. Aim at improving the retrieval precision, a retrieval<br />

method is proposed based on group decision-making for optimizing<br />

the case feature attributes weights. Firstly, multiple groups <strong>of</strong> initial<br />

weights are obtained by genetic algorithm. Then, the multiple sets <strong>of</strong><br />

retrieval results produced by these weights are optimized through group<br />

decision-making method, and the weights can be dynamic adjusted<br />

through the deviations between individual decision results and group<br />

result. The simulation results indicate that the proposed approach can<br />

fully excavate the potential knowledge <strong>of</strong> attribute weights and thus result<br />

in higher retrieval accuracy in a case-based reasoning system. The<br />

PID adjusting comparison experiment <strong>of</strong> typical two order delay system<br />

verifies the effectiveness <strong>of</strong> the new method.<br />

◁ PFrA-09<br />

Attribute Reduction Method Using Water-Filling Principle for Case-<br />

Based Reasoning, pp.779–782<br />

Zhao, Hui<br />

YAN, Aijun<br />

Zhang, Chun-xiao<br />

Wang, Pu<br />

Beijing Univ. <strong>of</strong> Tech.<br />

Beijing Univ. <strong>of</strong> Tech.<br />

Beijing Univ. <strong>of</strong> Tech.<br />

Beijing Univ. <strong>of</strong> Tech., China<br />

As the large number <strong>of</strong> feature attributes in Case-based reasoning system<br />

(CBR) brings a huge information redundancy which reduces the<br />

retrieval efficiency, a novel reduction method based on Water-Filling is<br />

proposed to remove those unnecessary attributes. In the method, the<br />

importance <strong>of</strong> each attribute could be calculated by utilizing the ratio<br />

<strong>of</strong> the standard deviation and the mean value <strong>of</strong> each attribute data as<br />

evaluation parameter, and the impotance result <strong>of</strong> each attribute is then<br />

used to guide the reduction process. The experiments on glass identification<br />

showed that the new method could get a better retrieval accuracy<br />

as well as a greater efficiency compared with the methods which do not<br />

conduct the reduction process.<br />

◁ PFrA-10<br />

Application <strong>of</strong> Self-organizing Feature Map Neural Network Based on<br />

Data Clustering, pp.797–802<br />

Hu, Xiang<br />

Yang, Yun<br />

Zhang, Lihong<br />

Xiang, Tao<br />

Hong, Chengqiu<br />

Zheng, Xiaotong<br />

Tsinghua Univ.<br />

Tsinghua Univ.<br />

Tinghua Univ.<br />

Tsinghua Univ.<br />

Tsinghua Univ.<br />

Tsinghua Univ.<br />

Outlier detection is <strong>of</strong> much importance in preprocessing <strong>of</strong> data collected<br />

from complex industry system, for the data has strong nonlinearity<br />

and poor stability, involving much noise. Outlier detection based<br />

on clustering, rejects abnormal data points which have significant difference<br />

from others according to the definition <strong>of</strong> similarity. Self-organizing<br />

Feature Map (SOM) Neural Network algorithm has the self-study and<br />

adaptive functions <strong>of</strong> neural networks, so as to be a hot research in<br />

clustering analysis recently. This paper first introduces Self-organizing<br />

Feature Map algorithm based on artificial neural network, and then improves<br />

the algorithm by using weighted Euclidean distance, finally uses<br />

the s<strong>of</strong>tware <strong>of</strong> MATLAB to analyze some actual data <strong>of</strong> electrical power.<br />

The result shows that SOM algorithm achieves a very good effect in<br />

clustering, and the MATLAB toolbox shows favorable visual effects.<br />

◁ PFrA-11<br />

Flow Rate Control and Resource Allocation Policy with Security Requirements<br />

in OFDMA Networks, pp.1020–1025<br />

Zhu, Xingzheng<br />

Yue, Jianting<br />

Yang, Bo<br />

Guan, Xinping<br />

SHANGHAI JIAOTONG Univ.<br />

SHANGHAI JIAOTONG Univ.<br />

Shanghai Jiao Tong Univ.<br />

Shanghai Jiao Tong Univ.<br />

OFDMA-based network performs well in maximizing the overall<br />

throughput, and meanwhile, problems with security requirements are<br />

focused to satisfy the increasing needs <strong>of</strong> confidential data transmissions.<br />

According to information theory, however, secure transmission<br />

undesirably antagonizes larger flow rate, thus a trade<strong>of</strong>f is expected to<br />

balance these two matters. In this paper, we consider a downlink situation<br />

in OFDMA network, where all users request secure transmission<br />

from the base station yet the resources including power and subcarriers<br />

are limited such that an appropriate policy on resource allocation can<br />

play an instrumental role. Dynamic queueing and Lyapunov optimization<br />

are two highlights in this process. We utilize dynamic queueing<br />

so that merely instantaneous state <strong>of</strong> the network is enough, which is<br />

advantageous for easier implementations. On the other hand, by the<br />

means <strong>of</strong> Lyapunov optimization, we also prove our algorithm can obtain<br />

a performance <strong>of</strong> flow rate admission extremely close to the optimal.<br />

Applying this policy can reduce computational complexity significantly<br />

but still perform close to optimality.<br />

◁ PFrA-12<br />

Augmented Dimension Algorithm Based on Sequential Detection for<br />

Maneuvering Target Tracking, pp.1323–1327<br />

Pan, Baogui<br />

Peng, Dongliang<br />

Shao, Genfu<br />

Inst. <strong>of</strong> Information & Control<br />

Hangzhou Dianzi Univ.<br />

Inst. <strong>of</strong> Information & Control<br />

In order to solve the problem that target tracking algorithm based on<br />

single model has poor tracking performance when the target occurs<br />

high maneuver and that IMM algorithm has low accuracy in tracking a<br />

constant velocity target, an augmented dimension algorithm based on<br />

sequential detection for maneuvering target tracking is proposed. First,<br />

the KF-UKF joint filtering is proposed. The Kalman filter based on the<br />

CV model is used to estimate the state <strong>of</strong> a constant velocity target.<br />

When the target maneuver is detected, the dimension <strong>of</strong> the CV model<br />

is augmented, and the unscented Kalman filter is used to estimate<br />

the state. Second, a fading memory sequential detection algorithm is<br />

111

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!