ISSN 0253-2778

CN 34-1054/N

2015 Vol. 45, No. 4

Display Method:
Original Paper
A new estimate of DoA for saturated systems and its applications
SHANG Weike, ZUO Bin, SUN Yuanfeng, ZHANG Jinyuan
2015, 45(4): 259-267. doi: 10.3969/j.issn.0253-2778.2015.04.001
A new method for estimating the domain of attraction(DoA) for saturated systems was presented. Compared with the existing results, the advantage of the new result is mainly twofold: ① It does not include any product by the system matrix and the Lyapunov matrix; ② It does not result in heavy computing cost. It will be seen that these features are essentially important in system analysis. For comparison, the new result was extended to uncertain saturated systems, which shows that it leads to less conservativeness. Numerical examples verify the correctness of the conclusion.
Hybrid controller design and analysis for experimental greenhouse temperature system
CHU Zhudong, QIN Linlin, LU Linjian, MA Guoqi, WU Gang
2015, 45(4): 268-274. doi: 10.3969/j.issn.0253-2778.2015.04.002
Due to the interaction between discrete on-off controls and continuous environmental factors, greenhouse temperature control systems can be regarded as a class of hybrid system. Most previous greenhouse control algorithms rely on an exquisite system model and classical or modern control theories, and fail to consider the actual conditions of greenhouses in China in their design of a controller and their system anaysis does not include the hybrid properties of the greenhouse. Based directly on the hybrid automata theory, a hybrid controller was designed for controlling the temperatures of experimental greenhouse in summer. The controller was shown to be non-blocking and deterministic in hybrid automata theory framework. Experiments were performed in the spring and summer of 2014, and the controller behaved reasonably and timely when events triggered state transitions. Further more, a controller framework containing two typical hybrid controller modes was established for winter and summer to meet the needs of continuous control all year round. And the referential time points for mode transitions were obtained by analyzing daily lowest temperaturs from 2013 to 2014.
Corn-counting method based on photoelectric signal
CHANG Li, MA Chengxue, GAO Lifu
2015, 45(4): 275-279. doi: 10.3969/j.issn.0253-2778.2015.04.003
1000-grain weight is the single crucial criterion for breeding high quality corn. In order to measure the 1000-grain weight, the problem of counting the corn grains must first be solved. Photodiode was used to obtain the image information of corn grains when free falling, and an identification algorithm was designed for calculating the number of corn grains. The method and the algorithm are capable of high accuracy and a high speed, and have very good reference value to the general grain counting.
A three dimensional robust guidance law design based on RBF neural network gain adjustment
CHEN Bo, YANG Kaihong, JI Haibo
2015, 45(4): 280-285. doi: 10.3969/j.issn.0253-2778.2015.04.004
By adopting the three dimensional nonlinear model for the relative motion of missiles and targets,a scheme of guidance law was presented. The theoretical basis of the guidance law includes input-to-state stability (ISS) as well as the dynamic adjustment and self-study ability of the radial basis function (RBF) neural network. The control law is capable of dynamically adjusting the gain of nonlinear guidance law with the angular rate change of LOS (line of sight). The guidance law can avoid the undershoot augment caused by gain fixation and large-scale target-maneuvering, and also effectively trace as well as intercept the target making a variety of maneuvers. The numerical simulation results demonstrate the adaptivity and easy implementation of the control law.
The research of speaker diarization based on BIC and G_PLDA
LI Rui, ZHUO Zhu, LI Hui
2015, 45(4): 286-293. doi: 10.3969/j.issn.0253-2778.2015.04.005
The traditional technology for speaker diarization(SD), which exploits the Bayesian information criterion(BIC) as the similarity metric, can obtain good results in the short dialogue task, but with the length of the dialogue increasing , single Gaussian model of BIC is insufficient to describe the information distribution of different speakers. Moveover, it is difficult to delineate the threshold between the same speakers and different speakers when using hierarchical clustering (HAC). To solve this problem, a fusion method between BIC and G_PLDA was proposed, so as to make full use of the reliability of BIC in short-term clustering and the excellent discriminating power of G_PLDA in long utterancs. A set of experiments based on NIST 08 Summed shows that this new fusion method reduces the diariazation error rate (DER) from 2.34% of BIC baseline system to 1.54%, improving performance of speaker diarization by 34.2%.
Cooperative scheduling and power allocation scheme based on user clustering in LTE uplink
ZOU Guoqi, XU Jing, ZHU Yuanping
2015, 45(4): 294-301. doi: 10.3969/j.issn.0253-2778.2015.04.006
Inter-cell interference is serious in LTE co-channel networks, which becomes the bottleneck of outage performance enhancement in LTE networks. However, inter-cell interference is not considered when applying conventional single-cell scheduling algorithms in LTE uplink. A cooperative scheduling and power allocation scheme based on user clustering was proposed in order to mitigate inter-cell interference and enhance system performance, in which users from different cells who share the same resource blocks were clustered. The cooperative scheduling algorithm based on the proportional fair criterion was developed to allocate resourcs among user clusters. The cooperative power allocation was achieved by solving a set of power optimization problems formulated with interference among users in the same user cluster. The simulation result shows that the cooperative scheduling and power allocation scheme based on user clustering is efficient in terms of outage throughput, while greatly enhancing system throughput.
Application of adaptive cross approximation combined with compressedsensing to fast solution of electromagnetic scatteringproblems of electrically large objects over wide angles
CAO Xinyuan, CHEN Mingsheng, KONG Meng, ZHANG Liang, CHENG Liangliang, CHEN Bingbing, QI Qi
2015, 45(4): 302-307. doi: 10.3969/j.issn.0253-2778.2015.04.007
Fast analysis of electromagnetic scattering properties of various objects, especially electrically large objects over a wide angle, is always a difficult problem in computational electromagnetics. A new solution using compressed sensing in conjunction with adaptive cross approximation was proposed, and a new incident source including different angle information was constructed based on compressed sensing theory, which could reduce the number of computation times for method of moments. Meanwhile, adaptive cross approximation technique was also introduced to method of moments to form a low rank decomposition of the impedance matrix. Thus a new scheme was finally formed to rapidly analyze electromagnetic scattering problems for electrically large objects over a wide angle. Numerical results show that this solution can reduce operation time effectively while retaining the accuracy of calculation results.
Research on passive human activity recognition using WiFi ambient signals
GU Yu, QUAN Lianghu, CHEN Mengni, REN Fuji
2015, 45(4): 308-313. doi: 10.3969/j.issn.0253-2778.2015.04.008
Although traditional k-nearest neighbor(K-NN) and Bagging can recognize effectively less human activities using WiFi ambient signal, recognition by either alone of the seven states, namely, empty, walking, sitting, standing, sleeping, falling and running, is not ideal. To improve recognition rates, a new algorithm, fusion algorithm, was designed. It significantly outperforms K-NN and Bagging in terms of recognition ratios in both single-subject and multi-subject experiments.
A latent semantic analysis classification technique based on optimized categorization information
JI Duo, BI Chen, CAI Dongfeng
2015, 45(4): 314-320. doi: 10.3969/j.issn.0253-2778.2015.04.009
As an effective method in the way of dimensionality reduction, latent semantic analysis( LSA) has been widely applied to many text learning missions, such as information retrieval and text categorization. Based on professional literature text classification tasks, features of text from same and different categories were analyzed under a strict classification system, patent documents classification was taken as an example, an optimized LSA classification technique was purposed based on categorization information. Utilizing features information from different category text, the technique divided original documents into a variety of fake documents, strengthens occurrence frequency of exclusive features from different categories, thus building optimized latent semantic space and improving the performance of the classification model. The experimental result shows that the method effectively improves categorization precision when applied to text categorization.
Recognition of ancient Chinese characters based on hybrid kernel WLS-SVR
HU Gensheng, SUN Yingying, XU Lingying, LIANG Dong, SUN Xiaoqi
2015, 45(4): 321-328. doi: 10.3969/j.issn.0253-2778.2015.04.010
The shapes of ancient Chinese characters are often uncertain, which reduces the accuracy of recognition by many classifiers. To solve this problem, a new recognition algorithm combining adaptive weighted least squares support vector regression(WLS-SVR) with hybrid kernel function was proposed to recognize ancient Chinese characters. The weight coefficients of WLS-SVR decayed at a rate of the exponential function of prediction errors. The hybrid kernel was constructed using the wavelet kernel function with local properties and RBF kernel function with global properties. For feature extraction, global point density and component structure are fused with local features of pseudo 2D elastic mesh and local point density. Experiment results show the good robustness and high recognition accuracy of the proposed method.
Positive domain reduction in intuitionistic fuzzy objective information systems
BAO Zhongkui, YANG Shanlin
2015, 45(4): 329-336. doi: 10.3969/j.issn.0253-2778.2015.04.011
The classical rough set theory can not be directly used to reduce knowledge for intuitionistic fuzzy objective information systems. To solve this problem, dominance relation was firstly introduced to intuitionistic fuzzy objective information systems, and intuitionistic fuzzy rough set based on dominance relation was defined. Then, the notion of the relative positive domain and the significance of attributes in classical rough set theory were generalized to intuitionistic fuzzy objective information systems, while the monotone property of the relative positive domain was investigated. According to the different characteristics of attributes and the definition of positive domain reduction, the judgment theorem for positive domain reduction was given, the positive domain reduction algorithm using attribute significance as heuristic information was presented, and the complexity analysis of the algorithm was given. Finally, the effectiveness of the proposed algorithm was illustrated with comparative experiments.
Entropy-based image noise variance estimation
YANG Tao, FANG Shuai, CHENG WenJuan
2015, 45(4): 337-344. doi: 10.3969/j.issn.0253-2778.2015.04.012
In the de-noising and segmentation algorithm used to deal with the noise image, it is necessary to know the distribution model and the statistical parameters of noise. A novel noise estimation algorithm was thus proposed. First, the combined value of the input noise image variance and local entropy of each image block was calculated. Then all the comprehensive values were arranged in a descending order, and de-noising was calculated using the corresponding standards deviations in that order. Finally, final noise estimates were selected using the image quality evaluation algorithm. The proposed algorithm does not need pre-processing such as complex filtering, wavelet transform, etc., and can obtain the variance of noise by directly processing a series of input image data. It is simple and easy to implement, has high computational efficiency, and enable BM3D and similar de-noising algorithm to denoise adaptively.