Wavelet And Date Processing

At present, the wavelet transform signal processing has become a hot research field, in contrast to having an essential way of progress, it is to overcome the the Fourier analysis can not be to local analysis veyywell. the wavelet analysis is a landmark development results, at present, not only the domestic but also the foreign, the wavelet transform signal processing has become a hot area of research to many scholars to study the wavelet results,which in the theory as applied to image compression, feature extraction, signal filtering and data fusion and other aspects, and applications of wavelet transform is also continuously evolving. But in a class of important signal processing functions-functions with limited research is still in early stages, this is done in this regard some preliminary results.As the application of wavelet approximation method, we consider the use of wavelet to processing the signal, in practice, we find that it is difficult to achieve good wavelet alone treatment effect, combined with support vector machines to do double data processing in practical applications will not to show the general effectiveness of support vector machine (SVM),which is based on statistical learning theory based on a new machine learning method, which allows the structure to minimize risk and solve the small sample learning problem. Used SVW as a predictor, combined with genetic algorithm to solve one of the optimal control signal sequence, we designed controller has good land quality, to better track and the reference output. Therefore, we support vector machines in the study had some work done.In this paper, the completion of two parts:first,In this paper, we use the wavelet as an approximation tool for entire functions that the exponential type were reconstructed, and calculate the error bound, given the wavelet reconstruction of entire functions of exponential type form.second, We make some studies based on statistical learning theory,which based on support vector machine (SVM) learning methods to solve the small sample case study, by using support vector machines for nonlinear model identification, design a control structure, and use the genetic algorithm to predict the control law.