Loading...

Current Issue

21 February 2008, Volume 19 Issue 1
ELECTRONICS TECHNOLOGY
On the improvement of the mutual coupling compensation in DOA estimation
Wu Yujiang & Nie Zaiping
2008, 19(1):  1-6. 
Abstract ( )   PDF (922KB) ( )  
Related Articles | Metrics

A new and exact calculation method for the mutual impedance matrix of receiving arrays is proposed. The mutual impedance matrix is derived from electromagnetic boundary conditions and can be used to relate the coupling free open-circuit voltages, instead of the conventional ones, to the measured voltages. A remarkable improvement on compensation for the coupling effects is shown in the direction finding applications, while a simple relationship between measured terminal voltages and the coupling free voltages is remained.

Modified unscented particle filter for nonlinear Bayesian tracking
Zhan Ronghui, Xin Qin & Wan Jianwei
2008, 19(1):  7-14. 
Abstract ( )   PDF (1185KB) ( )  
Related Articles | Metrics

A modified unscented particle filtering scheme for nonlinear tracking is proposed, in view of the potential drawbacks (such as, particle impoverishment and numerical sensitivity in calculating the prior) of the conventional unscented particle filter (UPF) confronted in practice. Specifically, a different derivation of the importance weight is presented in detail. The proposed method can avoid the calculation of the prior and reduce the effects of the impoverishment problem caused by sampling from the proposal distribution. Simulations have been performed using two illustrative examples and results have been provided to demonstrate the validity of the modified UPF as well as its improved performance over the conventional one.

Wentzel-Kramer-Brillouin and finite-difference time-domain analysis of Terahertz band electromagnetic characteristics of target coated with unmagnetized plasma
Liu Shaobin, Zhou Tao, Liu Meilin & Hong Wei
2008, 19(1):  15-20. 
Abstract ( )   PDF (1491KB) ( )  
Related Articles | Metrics

We investigate computationally the attenuation and reflection of Terahertz (THz) wave using targets coated with plasmas. The simulators are the Wentzel-Kramer-Brillouin (WKB) method and finite-difference timedomain (FDTD) method. The relation between the frequency of the incident electromagnetic (EM) wave and the attenuation caused by unmagnitized plasma is analyzed. The results demonstrate that the amount of absorbed power is a decreasing function of the EM wave frequency and the plasma collision frequency. For THz band incident wave, the attenuation that is caused by plasma is small when the plasma has common density and the collision frequency. This conclusion has fine applying foreground for plasma anti stealth.

Channel capacity of multiple-input multiple-output systems with transmit and receive correlation
Wang Jun, Zhu Shihua, Wang Lei & Liu Fang
2008, 19(1):  21-26. 
Abstract ( )   PDF (869KB) ( )  
Related Articles | Metrics

In order to investigate the impact of channel model parameters on the channel capacity of a multipleinput multiple-output (MIMO) system, a novel method is proposed to explore the channel capacity under Rayleigh flat fading with correlated transmit and receive antennas. The optimal transmitting direction which can achieve maximum channel capacity is derived using random matrices theory. In addition, the closed-form expression for the channel capacity of MIMO systems is given by utilizing the properties of Wishart distribution when SNR is high. Computer simulation results show that the channel capacity is maximized when the antenna spacing increases to a certain point, and furthermore, the larger the scattering angle is, the more quickly the channel capacity converges to its maximum. At high SNR (>12 dB), the estimation of capacity is close to its true value. And, when the same array configuration is adopted both at the transmitter and the receiver, the UCA yields higher channel capacity than ULA.

Posterior Cramer-Rao lower bounds for bearing-only tracking
Guo Lei, Tang Bin & Liu Gang
2008, 19(1):  27-32. 
Abstract ( )   PDF (838KB) ( )  
Related Articles | Metrics

In the state estimation of passive tracking systems, the traditional approximate expression for the Cramer-Rao lower bound (CRLB) does not take two factors into consideration, that is, measurement origin uncertainty and state noise. Such treatment is only valid in ideal situation but it is not feasible in actual situation. In this article, considering the two factors, the posterior Cramer-Rao lower bound (PCRLB) recursion expression for the error of bearing-only tracking is derived. Then, further analysis is carried out on the PCRLB. According to the final result, there are four main parameters that play a role in the performance of the PCRLB, that is, measurement noise, detection probability, state noise and clutter density, amongst which the first two have greater impact on the performance of the PCRLB than the others.

Optimal training sequences for MIMO systems under correlated fading
Pang Jiyong, Li Jiandong, Lu Zhuo, Zhao Linjing & Chen Liang
2008, 19(1):  33-38. 
Abstract ( )   PDF (926KB) ( )  
Related Articles | Metrics

The optimal design of training sequences for channel estimation in multiple-input multiple-output (MIMO) systems under spatially correlated fading is considered. The channel is assumed to be a block-fading model with spatial correlation known at both the transmitter and the receiver. To minimize the channel estimation error, optimal training sequences are designed to exploit full information of the spatial correlation under the criterion of minimum mean square error (MMSE). It is investigated that the spatial correlation is helpful to decrease the estimation error and the proposed training sequences have good performance via simulations.

Study of adaptive modulation and LDPC coding in multicarrier systems
Huo Yongqing, Peng Qicong & Shao Huaizong
2008, 19(1):  39-45. 
Abstract ( )   PDF (1080KB) ( )  
Related Articles | Metrics

An adaptive modulation (AM) algorithm is proposed and the application of the adapting algorithm together with low-density parity-check (LDPC) codes in multicarrier systems is investigated. The AM algorithm is based on minimizing the average bit error rate (BER) of systems, the combination of AM algorithm and LDPC codes with different code rates (half and three-fourths) are studied. The proposed AM algorithm with that of Fischer et al is compared. Simulation results show that the performance of the proposed AM algorithm is better than that of the Fischer’s algorithm. The results also show that application of the proposed AM algorithm together with LDPC codes can greatly improve the performance of multicarrier systems. Results also show that the performance of the proposed algorithm is degraded with an increase in code rate when code length is the same.

Heuristic based data scheduling algorithm for OFDMA wireless network
Guo Kunqi, Sun Lixin, Jia Shilou & Yu Xiaoyang
2008, 19(1):  46-51. 
Abstract ( )   PDF (1777KB) ( )  
Related Articles | Metrics

A system model based on joint layer mechanism is formulated for optimal data scheduling over fixed point-to-point links in OFDMA ad-hoc wireless networks. A distributed scheduling algorithm (DSA) for system model optimization is proposed that combines the randomly chosen subcarrier according to the channel condition of local subcarriers with link power control to limit interference caused by the reuse of subcarrier among links. For the global fairness improvement of algorithms, a global power control scheduling algorithm (GPCSA) based on the proposed DSA is presented and dynamically allocates global power according to difference between average carrier-noise-ratio of selected local links and system link protection ratio. Simulation results demonstrate that the proposed algorithms achieve better efficiency and fairness compared with other existing algorithms.

Investigation on full distribution CNC system based on SERCOS bus
Guo Wei, Chen Zongyu & Li Congxin
2008, 19(1):  52-57. 
Abstract ( )   PDF (1738KB) ( )  
Related Articles | Metrics

A full distribution CNC system based on SERCOS bus is studied in accordance with the limitations of traditional PC-based motion card. The conventional PC-based motion control card is dispersed into several autonomous intelligent servo-control units with the function of servo driver. The autonomous intelligent servocontrol units realize the loop control of position, velocity and current. Interpolation computation is completed in PC and the computational results are transferred to every autonomous intelligent servo-control unit by high speed SERCOS bus. Software or hardware synchronization technology is used to ensure all servomotors are successive and synchronously running. The communication and synchronization technology of SERCOS are also researched and the autonomous intelligent servo-control card is developed byself. Finally, the experiment of circle contour process on a prototype system proves the feasibility.

Multi-constraint quality of service routing algorithm for dynamic topology networks
Wang Ping, Chen Bingcai, Gu Xuemai & Liu Gongliang
2008, 19(1):  58-64. 
Abstract ( )   PDF (1874KB) ( )  
Related Articles | Metrics

An adaptive multi-QoS routing algorithm called AMQRA is proposed for dynamic topology networks, such as satellite networks and Ad-hoc networks. The AMQRA is a distributed and mobile-agents-based routing algorithm, which combines ant quantity system (AQS) with ant colony optimization (ACO) that is used in AntNet routing algorithm. In dynamic topology networks, the AMQRA achieves timely optimization for concave metric QoS constraint and fast convergence. The proposed routing algorithm is simulated in Iridium satellite constellation on OPNET. The results show that AMQRA not only outperforms the AntNet in convergence rate in dynamic topology networks but also can optimize concave metric QoS constraint and reasonably allot bandwidth to the load to avoid networks congestion.

Local prediction of the chaotic fh-code based on LS-SVM
Wang Yi & Guo Wei
2008, 19(1):  65-70. 
Abstract ( )   PDF (1883KB) ( )  
Related Articles | Metrics

Support vector machine (SVM) is powerful to solve some problems such as nonlinear classification, function estimation and density estimation. To consider the chaotic fh (frequency hopping)-code’s characters in chaotic dynamic system, the forecasting model of the support vector machine in combination with Takens’ delay coordinate phase reconstruction of chaotic times is established and the least squares model for large-scale problems is used in local training for this model. Finally, a fh-code series generated by Logistic-Kent mapping is applied to verify the local prediction model. Simulation results show that the high accuracy and fault tolerant SVM model has an excellent performance in predicting the fh code, with a very low mean square error and a high relative coefficient.

DEFENCE ELECTRONICS TECHNOLOGY
Elevation estimation for low-angle target based on reflection paths suppression
Hou Yuguan, Shen Yiying & Zhang Zhongzhao
2008, 19(1):  71-75. 
Abstract ( )   PDF (1174KB) ( )  
Related Articles | Metrics

In the signal processing for metrewave radar, the reflection paths of target echoes can cause severe error in the elevation estimation for the low-angle target tracking. The exact angles of the reflection paths are unknown beforehand, and therefore, the reflection paths can not be suppressed easily. Therefore, in this article, an improved reflection paths suppression approach is presented. A block matrix aggregate is constructed based on the possible angles of the reflection paths. Combined with the beamforming-like processing, a generalized maximum likelihood estimation is derived to optimize the estimation. Moreover, the noise reduction method based on the Toeplitz covariance matrix is used for better performance. This approach is applied to the real data collected by the low-angle tracking radar with 8-channel vertical array. The experiment results show that the reflection effects are reduced and the accuracy of the elevation estimate is improved.

Multiple targets vector miss distance measurement accuracy based on 2-D assignment algorithms
Fang Bingyi & Wu Siliang
2008, 19(1):  76-80. 
Abstract ( )   PDF (505KB) ( )  
Related Articles | Metrics

An extension of 2-D assignment approach is proposed for measurement-to-target association for improving multiple targets vector miss distance measurement accuracy. When the multiple targets move so closely, the measurements can not be fully resolved due to finite resolution. The proposed method adopts an auction algorithm to compute the feasible measurement-to-target assignment with unresolved measurements for solving this 2-D assignment problem. Computer simulation results demonstrate the effectiveness and feasibility of this method.

Efficient analysis of dielectric radomes using multilevel fast multipole algorithm with CRWG basis
Que Xiaofeng, Nie Zaiping & Hu Jun
2008, 19(1):  81-87. 
Abstract ( )   PDF (1708KB) ( )  
Related Articles | Metrics

A full-wave analysis of the electromagnetic problem of a three-dimensional (3-D) antenna radiating through a 3-D dielectric radome is preserued. The problem is formulated using the Poggio-Miller-Chang-Harrington-Wu(PMCHW) approach for homogeneous dielectric objects and the electric field integral equation for conducting objects. The integral equations are discretized by the method of moment (MoM), in which the conducting and dielectric surface/interfaces are represented by curvilinear triangular patches and the unknown equivalent electric and magnetic currents are expanded using curvilinear RWG basis functions. The resultant matrix equation is then solved by the multilevel fast multipole algorithm (MLFMA) and fast far-field approximation (FAFFA) is used to further accelerate the computation. The radiation patterns of dipole arrays in the presence of radomes are presented. The numerical results demonstrate the accuracy and versatility of this method.

Effect of beam-pointing errors on bistatic SAR imaging
Zhang Yabiao, Yuan Bingcheng, Tang Ziyue & Zhu Zhenbo
2008, 19(1):  88-93. 
Abstract ( )   PDF (2426KB) ( )  
Related Articles | Metrics

The purpose is to conduct a research in the energy variation of echo wave and the imaging effect caused by the aero bistatic SAR pointing errors. Based on the moving geometry configuration of aero bistatic SAR, a model of beam pointing errors is built. Based on this, the azimuth Doppler frequency center estimation caused by these errors and the limitation to the beam pointing synchronization error are studied, and then the imaging result of different errors are analyzed. The computer’s simulations are provided to prove the validity of the above analysis.

SYSTEMS ENGINEERING
Transformation and entropy for fuzzy rough sets
Zhang Chengyi, Li Dongya, Fu Haiyan & Chen Guohui
2008, 19(1):  94-98. 
Abstract ( )   PDF (216KB) ( )  
Related Articles | Metrics

A new method for translating a fuzzy rough set to a fuzzy set is introduced and the fuzzy approximation of a fuzzy rough set is given. The properties of the fuzzy approximation of a fuzzy rough set are studied and a fuzzy entropy measure for fuzzy rough sets is proposed. This measure is consistent with similar considerations for ordinary fuzzy sets and is the result of the fuzzy approximation of fuzzy rough sets.

Combined heuristics for determining order quantity under time-varying demands
Tang Jiafu, Pan Zhendong, Gong Jun & Liu Shixin
2008, 19(1):  99-111. 
Abstract ( )   PDF (564KB) ( )  
Related Articles | Metrics

The time-varying demands for a certain period are often assumed to be less than the basic economic order quantity (EOQ) so that total replenishment quantity rather than economic order quantity is normally considered by most of the heuristics. This acticle focuses on a combined heuristics method for determining order quantity under generalized time-varying demands. The independent policy (IP), abnormal independent policy (AIP) and dependent policies are studied and compared. Using the concepts of normal/abnormal periods and the properties of dependent policies, a dependent policy-based heuristics (DPH) is proposed for solving the order quantity problems with a kind of time-varying demands pattern under which the first period is normal. By merging the Silver-Meal (S-M) heuristics and the dependent policy-based heuristics (DPH), a combined heuristics (DPH/S-M) is developed for solving order quantity problems with generalized time-varying demands. The experimentation shows that (1) for the problem with one normal period, no matter which position the normal period stands, the DPH/S-M could not guarantee better than the S-M heuristics, however it is superior to the S-M heuristics in the case that the demands in the abnormal periods are in descending order, and (2) The DPH/S-M is superior to the S-M heuristics for problems with more than one normal period, and the more the number of normal periods, the greater the improvements.

Analysis of effect factors-based stochastic network planning model
Chu Chunchao
2008, 19(1):  112-118. 
Abstract ( )   PDF (1880KB) ( )  
Related Articles | Metrics

Looking at all the indeterminate factors as a whole and regarding activity durations as independent random variables, the traditional stochastic network planning models ignore the inevitable relationship and dependence among activity durations when more than one activity is possibly affected by the same indeterminate factors. On this basis of analysis of indeterminate effect factors of durations, the effect factors-based stochastic network planning (EFBSNP) model is proposed, which emphasizes on the effects of not only logistic and organizational relationships, but also the dependent relationships, due to indeterminate factors among activity durations on the project period. By virtue of indeterminate factor analysis the model extracts and describes the quantitatively indeterminate effect factors, and then takes into account the indeterminate factors effect schedule by using the Monte Carlo simulation technique. The method is flexible enough to deal with effect factors and is coincident with practice. A software has been developed to simplify the model-based calculation, in VisualStudio.NET language. Finally, a case study is included to demonstrate the applicability of the proposed model and comparison is made with some advantages over the existing models.

Self-organizing fuzzy clustering neural network and application to electronic countermeasures effectiveness evaluation
Li Zhisheng, Li Junshan, Feng Fan & Zhao Xin
2008, 19(1):  119-124. 
Abstract ( )   PDF (369KB) ( )  
Related Articles | Metrics

A self-organizing fuzzy clustering neural network by combining the self-organizing Kohonen clustering network with the fuzzy theory is proposed. This network model is designed for the effectiveness evaluation of electronic countermeasures, which not only exerts the advantages of the fuzzy theory, but also has a good ability in machine learning and data analysis. The subjective value of sample versus class is computed by the fuzzy computing theory, and the classified results obtained by self-organizing learning of Kohonen neural network are represented on output layer. Meanwhile, the fuzzy competition learning algorithm keeps the similar information between samples and overcomes the disadvantages of neural network which has fewer samples. The simulation result indicates that the proposed algorithm is feasible and effective.

CONTROL THEORY AND APPLICATION
Unpower aerocraft augmented state feedback tracking guaranteed cost control
Xu Jiansheng, Wu Hao & Wang Yongji
2008, 19(1):  125-130. 
Abstract ( )   PDF (719KB) ( )  
Related Articles | Metrics

Aimed at designing the unpower aerocraft attitude control system in a simple and practical way, the guaranteed cost control is adopted. To eliminate the steady-error, a novel tracking control approach—augmented state feedback tracking guaranteed cost control is proposed. Firstly, the unpower aerocraft is modeled as a linear system with norm bounded parameter uncertain, then the linear matrix inequality based state feedback guaranteed cost control law is combined with the augmented state feedback tracking control from a new point of view. The sufficient condition of the existence of the augmented state feedback tracking guaranteed cost control is derived and converted to the feasible problem of the linear matrix inequality. Finally, the proposed approach is applied to a specified unpower aerocraft. The six dimensions of freedom simulation results show that the proposed approach is effective and feasible.

Stabilization of discrete nonlinear systems based on control Lyapunov functions
Cai Xiushan, Wang Xiaodong & Lv Ganyun
2008, 19(1):  131-133. 
Abstract ( )   PDF (346KB) ( )  
Related Articles | Metrics

The stabilization of discrete nonlinear systems is studied. Based on control Lyapunov functions, a sufficient and necessary condition for a quadratic function to be a control Lyapunov function is given. From this condition, a continuous state feedback law is constructed explicitly. It can globally asymptotically stabilize the equilibrium of the closed-loop system. A simulation example shows the effectiveness of the proposed method.

Novel active fault-tolerant control scheme and its application to a double inverted pendulum system
Cui Ping, Weng Zhengxin, Patton Ron
2008, 19(1):  134-140. 
Abstract ( )   PDF (886KB) ( )  
Related Articles | Metrics

On the basis of the gain-scheduled H∞ design strategy, a novel active fault-tolerant control scheme is proposed. Under the assumption that the effects of faults on the state-space matrices of systems can be of affine parameter dependence, a reconfigurable robust H∞ linear parameter varying controller is developed. The designed controller is a function of the fault effect factors that can be derived online by using a well-trained neural network. To demonstrate the effectiveness of the proposed method, a double inverted pendulum system, with a fault in the motor tachometer loop, is considered.

Neural network-based H∞ filtering for nonlinear systems with time-delays
Luan Xiaoli & Liu Fei
2008, 19(1):  141-147. 
Abstract ( )   PDF (731KB) ( )  
Related Articles | Metrics

A novel H∞ design methodology for a neural network-based nonlinear filtering scheme is addressed. Firstly, neural networks are employed to approximate the nonlinearities. Next, the nonlinear dynamic system is represented by the mode-dependent linear difference inclusion (LDI). Finally, based on the LDI model, a neural network-based nonlinear filter (NNBNF) is developed to minimize the upper bound of H∞ gain index of the estimation error under some linear matrix inequality (LMI) constraints. Compared with the existing nonlinear filters, NNBNF is time-invariant and numerically tractable. The validity and applicability of the proposed approach are successfully demonstrated in an illustrative example.

Novel backstepping design for blended aero and reaction-jet missile autopilot
Liu Zhong & Jia Xiaohong
2008, 19(1):  148-153. 
Abstract ( )   PDF (1215KB) ( )  
Related Articles | Metrics

The advanced missile uses blended control of aero-fin and reaction-jet to improve missile maneuverability. The blended control design, which is multi-inputs and multi-outputs (MIMO), severe nonlinear, and model uncertain, is much more complex than conventional aero-fin control. A novel nonlinear backstepping control approach is proposed to design the blended autopilot. Missile model is reformed to a new one by state reconstruction technique so that it is easy to be handled by the backstepping method. Then a Lyapunov function is chosen to avoid oscillation caused in normal backstepping way when control parameters are mismatched. In distribution of both inputs, optimal energy logic is proposed. In addition, a fuzzy cerebellar model articulation controller (FCMAC) neural network is used to guarantee controller robustness to uncertainties. Finally, simulation results demonstrate the efficiency and advantages of the proposed method.

New adaptive quasi-sliding mode control for nonlinear discrete-time systems
Wang Weihong & Hou Zhongsheng
2008, 19(1):  154-160. 
Abstract ( )   PDF (1048KB) ( )  
Related Articles | Metrics

A new adaptive quasi-sliding mode control algorithm is developed for a class of nonlinear discrete-time systems, which is especially useful for nonlinear systems with vaguely known dynamics. This design is model-free, and is based directly on pseudo-partial-derivatives derived on-line from the input and output information of the system using an improved recursive projection type of identification algorithm. The theoretical analysis and simulation results show that the adaptive quasi-sliding mode control system is stable and convergent.

Predictive control for satellite formation keeping
He Donglei & Cao Xibin
2008, 19(1):  161-166. 
Abstract ( )   PDF (2638KB) ( )  
Related Articles | Metrics

Based on a Hill equation and a nonlinear equation describing the desired and real dynamics of relative motion separately, a predictive controller is brought forward, which makes the real state track the desired ones to keep satellite formation. The stability and robustness of the controller are analyzed. Finally, comparing the simulation results of the proposed controller with that of the traditional, proportional-differential controller shows that the former one is capable of keeping the satellite formation more favorably, considering the disturbances such as the J2 perturbations.

SOFTWARE ALGORITHM AND SIMULATION
Learning algorithm and application of quantum BP neural networks based on universal quantum gates
Li Panchi & Li Shiyong
2008, 19(1):  167-174. 
Abstract ( )   PDF (1466KB) ( )  
Related Articles | Metrics

A quantum BP neural networks model with learning algorithm is proposed. First, based on the universality of single qubit rotation gate and two-qubit controlled-NOT gate, a quantum neuron model is constructed, which is composed of input, phase rotation, aggregation, reversal rotation and output. In this model, the input is described by qubits, and the output is given by the probability of the state in which |1 is observed. The phase rotation and the reversal rotation are performed by the universal quantum gates. Secondly, the quantum BP neural networks model is constructed, in which the output layer and the hide layer are quantum neurons. With the application of the gradient descent algorithm, a learning algorithm of the model is proposed, and the continuity of the model is proved. It is shown that this model and algorithm are superior to the conventional BP networks in three aspects: convergence speed, convergence rate and robustness, by two application examples of pattern recognition and function approximation.

RKP based secure tracking in wireless sensor networks
Wang Jiahao, Qin Zhiguang, Geng Ji & Wang Shengkun
2008, 19(1):  175-183. 
Abstract ( )   PDF (2213KB) ( )  
Related Articles | Metrics

To enhancing the wireless sensor network’s security in target tracking and locating application, this article proposes a tracking cluster based mobile cluster distributed group rekeying protocol (MCDGR). Based on the given sensitivity, sensors can locate the moving object in the monitored area and form a tracking cluster around it. This tracking cluster can follow the target logically, process data on the target and report to the sink node, and thus achieve the tracking function. We introduce a multi-path reinforcement scheme, q-composition scheme and one-way cryptographic hash function based random key predistribution algorithm (RKP), which can guarantee a high accuracy and security and a low energy consumption on the same time in large-scale sensor networks.

Support vector classifier based on principal component analysis
Zheng Chunhong, Jiao Licheng & Li Yongzhao
2008, 19(1):  184-190. 
Abstract ( )   PDF (369KB) ( )  
Related Articles | Metrics

Support vector classifier (SVC) has the superior advantages for small sample learning problems with high dimensions, with especially better generalization ability. However there is some redundancy among the high dimensions of the original samples and the main features of the samples may be picked up first to improve the performance of SVC. A principal component analysis (PCA) is employed to reduce the feature dimensions of the original samples and the pre-selected main features efficiently, and an SVC is constructed in the selected feature space to improve the learning speed and identification rate of SVC. Furthermore, a heuristic genetic algorithm-based automatic model selection is proposed to determine the hyperparameters of SVC to evaluate the performance of the learning machines. Experiments performed on the Heart and Adult benchmark data sets demonstrate that the proposed PCA-based SVC not only reduces the test time drastically, but also improves the identify rates effectively.

Parameter selection of support vector machine for function approximation based on chaos optimization
Yuan Xiaofang & Wang Yaonan
2008, 19(1):  191-197. 
Abstract ( )   PDF (2826KB) ( )  
Related Articles | Metrics

The support vector machine (SVM) is a novel machine learning method, which has the ability to approximate nonlinear functions with arbitrary accuracy. Setting parameters well is very crucial for SVM learning results and generalization ability, and now there is no systematic, general method for parameter selection. In this article, the SVM parameter selection for function approximation is regarded as a compound optimization problem and a mutative scale chaos optimization algorithm is employed to search for optimal parameter values. The chaos optimization algorithm is an effective way for global optimal and the mutative scale chaos algorithm could improve the search efficiency and accuracy. Several simulation examples show the sensitivity of the SVM parameters and demonstrate the superiority of this proposed method for nonlinear function approximation.

Compression method based on training dataset of SVM
Ban Xiaojuan, Shen Qilong, Chen Hao & Tu Xuyan
2008, 19(1):  198-201. 
Abstract ( )   PDF (690KB) ( )  
Related Articles | Metrics

The method to compress the training dataset of Support Vector Machine (SVM) based on the character of the Support Vector Machine is proposed. First, the distance between the unit in two training datasets, and then the samples that keep away from hyper-plane are discarded in order to compress the training dataset. The time spent in training SVM with the training dataset compressed by the method is shortened obviously. The result of the experiment shows that the algorithm is effective.