Research And Implementation Of Harmonic Analysis Based On Support Vector Machine

Long-term since, the traditional way of fossil energy utilization is characterized by high consumption and heavy pollution, which has appeared serious crisis. On one hand, we are facing the depletion of conventional resource. On the other hand, the environmental issue are becoming more and more urgent. Therefore, providing a reliable, stable and efficient electrical power supply is very significant. Many factors influence the level of power quality, among which harmonic is particularly prominent in these factors. The effective control and governance of harmonic expansion is needed inevitably. Obviously, it must be based on the premise of accurate harmonic detection and analysis.Electric power harmonic detection and analysis based on the embedded DSP microprocessor has received extensive attention of the industry because of the high precision, small volume, low power consumption, and fast response, etc. This paper takes electric power harmonic as the research object. The application of the Support Vector Machine algorithm after parameter selection and optimization of solution process, in harmonic detection and analysis was studied and designed, and a platform of DSP microprocessor was set up for the implement and verification of the designed algorithm.First, the statistics learning basic theory was expounded and analyzed.The support vector machine algorithm based on this theory was introduced, and the algorithm was analyzed and studied in detail. The concrete steps of Support Vector Machine(SVM) analysis regression were summarized. In this paper, the mathematical model of the harmonic signal in the theory of Support Vector Machine was analyzed and expounded, and the deficiency of the existing algorithm was pointed out.Secondly, the genetic algorithm was introduced to research and optimize the relevant control parameters of the Support Vector Machine. The mathematical model was used to analyze and research the algorithm based on the theory of the algorithm. The algorithm analysis of the diversity maintained of samples and its convergence problem were analyzed deeply. The specific steps of how the parameters of Support Vector Machine were optimized by Genetic Algorithm were clarified.Furthermore, the function relation between the special approximation error and produce cost was chosen. Combining with analysis theory of the power system signals based on Support Vector Machine, mathematical model of the analysis of harmonic signal was reestablished. The iteratively reweighted least squares analysis of the model with steps and processes was concluded. With the help of MATLAB simulation environment, related characteristics of hybrid algorithm optimized were comparatively analyzed.Finally, platform of harmonic monitoring and analysis, which can be used for the designed algorithm, based on the TMS320F28335 microprocessor was designed and build. And Support Vector Machine algorithm based on genetic algorithm optimization program was written, then a harmonic analysis experiment was also carried on with the platform. The experimental results show that the proposed algorithm has high accuracy in harmonic analysis.