A kind of improved genetic algorithm foridentifying transfer function of thermal process in powerplant is introduced. In the algorithm, floating-point coding,rank-based selection, elitist reservation and groupingmetho...
详细信息
A kind of improved genetic algorithm foridentifying transfer function of thermal process in powerplant is introduced. In the algorithm, floating-point coding,rank-based selection, elitist reservation and groupingmethod are used, the premature convergence is restrained,the global and local searching ability is improved. Thegenetic algorithm-based model identification MATLABprogram is designed, the transfer functions of thermalprocess can be got with it according to the operating datalog files. The identification results to topical thermalprocess is given. It is shown by simulation research thataccurate identification results can be got no matter whatkind of input signal is used, such as step signal, randomoperating signal, even if there is a strong noise in the inputsignal.
The use of new types of actuators such as bimetal stripsand McKibben muscles in robotics has become *** use of these actuators can have the disadvantage ofneeding to design controllers for non-linearity's and time...
详细信息
The use of new types of actuators such as bimetal stripsand McKibben muscles in robotics has become *** use of these actuators can have the disadvantage ofneeding to design controllers for non-linearity's and timedependent parameters, which can arise in such *** problems that occur in robotics are the increasedproblems associated with systems having large degrees offreedom. This paper presents a method of designing acontroller that has the potential to overcome theseproblems. A nonlinear controller is evolved using anevolutionary algorithm for a two-link system and isshown to be robust and capable of controlling timevarying parameter systems. The controller has thepotential of being able to be extended to much largersystems with higher degrees of freedom.
This paper proposes a text classification methodbased on TAN model. Naive Bayesian classifier is the mosteffective and popular text classification method, but its attributeindependence assumption makes it unable to ex...
详细信息
This paper proposes a text classification methodbased on TAN model. Naive Bayesian classifier is the mosteffective and popular text classification method, but its attributeindependence assumption makes it unable to express thedependence among text terms. TAN (Tree Angmented NaiveBayes) combines the simplicity of Naive Bayesian with theability to express the dependence among attributes in Bayesiannetwork. This paper reviews some existing text methods,introduces TAN model, and applies TAN model to textclassification. Naive Bayesian and TAN classifiers are alsocompared by our experiments. Experimental results show TANclassifier has better performance.
WDM optical networks represent the direc-tion to the future high capacity wide-area network applica-tions. By creating the optical paths between several nodesin the core networks, logical topology can be created overt...
详细信息
WDM optical networks represent the direc-tion to the future high capacity wide-area network applica-tions. By creating the optical paths between several nodesin the core networks, logical topology can be created overthe physical topology. In this paper we consider themethod of designing logical topology in wavelengthrouted all optical networks. The exact formulation of thelogical topology problem is usually given as a Mixed In-teger Linear Programming, but it grows intractable withincreasing size of network. Heuristic approaches havebeen proposed, here we also proposed a different heuristicapproach to determine the logical topology, and compare itwith other methods.
In this paper, a new adaptive watermarkingalgorithm is presented for video image coding. Withoperating compressed video data-stream directly, we embedwatermarking information into motion vectors. According towatermark...
详细信息
In this paper, a new adaptive watermarkingalgorithm is presented for video image coding. Withoperating compressed video data-stream directly, we embedwatermarking information into motion vectors. According towatermarking information and the feature vectors of motionvectors, the proposed algorithm can adaptively selectembedding scheme with the aim to get good watermarkedimage quality. Watermark is extracted blindly withoutoriginal images and experimental results show that theproposed algorithm has a good compatibility with thepresent compression standards, and it is very simple andefficient.
Maneuver detection is a key issue in themultiple-model target tracking applications. In Kalmanfilter, the conventional maneuver detection algorithmfails when the maneuver is almost buried in noise. Themultiresolutiona...
详细信息
Maneuver detection is a key issue in themultiple-model target tracking applications. In Kalmanfilter, the conventional maneuver detection algorithmfails when the maneuver is almost buried in noise. Themultiresolutional approach behavior well for maneuverdetection, but the time delay limits its application. Inthis paper, we suggest a new algorithm for maneuverdetection based on the third-order cumulant of themeasurements. As cumulants are blind to Gaussiannoise, the maneuver is easy to be detected in thecumulant domain. Taking the sliding windowprocessing, the maneuver detection algorithm isreal-time without any time delay. Simulation resultsdemonstrate the effectiveness of the algorithm over theconventional one, especially at low SNR.
The third-order common-base transistor circuitreported firstly in [2] is a kind of chaotic signalgenerator, which can be easily realized in anylaboratory. The principal purpose of this paper isstudying the dynamics an...
详细信息
The third-order common-base transistor circuitreported firstly in [2] is a kind of chaotic signalgenerator, which can be easily realized in anylaboratory. The principal purpose of this paper isstudying the dynamics and great variety ofone-parameter bifurcation phenomena in it. Computersimulation and laboratory experiment are used toanalyze its characteristics. Several strange attractors,the intermittent attractor, periodic windows, cascades ofperiod doubling, period plus two and period minus onephenomena are observed by adjusting the value of aninductor. The computer simulation results agree withthe experiment results quite nicely.
Based on the universal approximation capa-bility of the Takagi-Seguno (TS) fuzzy systems, we presenta solution to the model reference control of continuous-time nonlinear systems problem. Using the assumptionthat a fu...
详细信息
Based on the universal approximation capa-bility of the Takagi-Seguno (TS) fuzzy systems, we presenta solution to the model reference control of continuous-time nonlinear systems problem. Using the assumptionthat a fuzzy model exists for the considered nonlinear sys-tems class, a direct TS fuzzy adaptive controller is designedto achieve the tracking objective. It is not required to es-timate the fuzzy model, only its existence assumption isrequired. The proposed adaptive scheme is asymptoticallystable and the tracking error converges to zero. Simulationresults illustrate the proposed algorithm performance.
Fast and efficient disparity estimation(DE)isa key technology in stereoscopic image sequenceprocessing. On the basis of analyzing the correlationbetween disparity vectors and motion vectors. a new fastdisparity estima...
详细信息
Fast and efficient disparity estimation(DE)isa key technology in stereoscopic image sequenceprocessing. On the basis of analyzing the correlationbetween disparity vectors and motion vectors. a new fastdisparity estimation algorithm for stereo video coding ispresented in this paper. Experimental results show that,compared with the full search algorithm, the proposedalgorithm can significantly reduce DE complexity whilethe reconstructed image quality is quite close to that offull search method.
The World Wide Web tody has become amajor force in driving the economy, and thetechnology. In spite of the abundance of informationfreely available on the web, the information is notaccessible to everyone. There are a...
详细信息
The World Wide Web tody has become amajor force in driving the economy, and thetechnology. In spite of the abundance of informationfreely available on the web, the information is notaccessible to everyone. There are a couple of reasonsfor this trend. Firstly, most of this information hasprimarily been used by and restricted to persons whoconnect and use the web via a traditional *** access to web information, for an increasingnumber of mobile device users, is very limited at thispoint. Secondly, most of the contents on the web are ina format unsuitable for disabled people. Contens areeither visually rich or highly structured. A goodsolution for this is to access web by voice, that is,voice browsing. In this paper we describe researchwork people have done on voice browsing and itsfuture development.
暂无评论