Application Study On Electrical Load Forecasting And Warning Based On Support Vector Machine

Human entered since the 1990s, accompanied by economic globalization and the development of economy and science and technology, energy is also a global, ecological, pluralism, security, science and technology-intensive trend of development. Against this background paper on electricity energy forecast early warning system for the Construction of the, to resolve the increasingly tense conflict between energy development and provide a way.Among the paper focus and center of the forecast, it has a direct impact on the success or failure of early warning. Machine learning applied to return to the field more and get the attention of experts at home and abroad, based on statistical learning theory of support vector machines (SVM) Yijuntuqi, rapid development in all walks of life of the related applications. This paper is mainly based on the forecast subsystem SVM, comprehensive evaluation of the use of the traditional model of thinking and reasoning model into the forecast system, in order to predict the results of scientific and reasonable. In addition, examples of this paper, data on the return of grey forecasting methods and SVM predicted a comparative analysis, the aim is for future research work to find suitable research tools and research materials. Another important part of this paper is the design of early warning systems, in support of theories has done some originality of the work - that matched genetic algorithm theorem and the preparation of the Java programming. This choice of SVM machine learning methods can also be used in the field of pattern recognition of their years of research experience on improving early warning systems. Finally, use the above theory, the power load forecasting system of early warning systems in the framework of a preliminary study, using systems thinking modules made by the four sub-systems: the noise of subsystems, forecast analysis system, early warning systems and match Secondary defense systems. Noise analysis of defense systems and subsystems with secondary supporting role, but essential, and these two systems of innovation in this paper also points.This paper is organized in such arrangements; first of all, in statistical learning theory on the basis of the study is described SVM classification and regression in the basic theory. Secondly, the paper discussed the genetic algorithm, given its Java programming instructions of the match and proved Theorem. Finally, given the power load forecasting the basic flow chart of early warning systems. Papers in Figure 24, Table 6, 45 references.