The Situation Assessment Model And Algorithm Of Voltage Situation Picture Of Distribution Network Center Of Substation

With the installation of special transformer terminal and public transformer terminal,as well as the promotion of distribution automation and the opening and closing substation,the technical conditions of flow calculation,the situational awareness,the situation visualization of distribution network increasingly mature.Machine learning algorithms are active in various fields of research in recent years,which are data processing,inductive reasoning,calculus,simulation and other application scenarios.With its extraordinary slice data analysis effect and precise data classification result,it is favored by researchers in various fields.SVG(Vector Graphic Scale)image format is a unified graphics format in global power system.With the application and popularization of the visualization technology of power system,a lot of time stamp images of SVG format have been accumulated.If SVG format image information can be extracted in the form of data and be modeled successfully,graphics information may be able to be more effectively and fully used.The work and innovation of this paper are as follows:Research on the theoretical aspect of the key technology of the SVM algorithm for the voltage situation map of the distribution network in the substation.The part analyze the development situation of power system visualization in domestic and foreign,and point out that the application for power grid real-time graphics and data is mainly used in power system visualization,reliability analysis,fault diagnosis and so on.Scientific computing aim to insight,rather than get data.If appropriate mathematical model isn’t established,it is difficult to transform data into information,but itself requires a lot of cost of storage.Point out that the prediction and assessment of the SVM algorithm,and summarizes the application case of the method.Put forward some methods of feature vector extraction from the image data that SVM algorithm needed.Discuss the process of SVM algorithm,and study RBF kernel SVM algorithm deeply.Research on feature vector extraction model and its dimension reduction method based on the situation of voltage contour line in substation-center distribution network.This part is the basic data preparation stage of graphic situation assessment,and it is an important link in the establishment of situation model.Extracting appropriate feature vectors has a very important influence on the establishment of SVM model,and it is also one of the key factors to evaluate the accuracy of the model.This part first gives the situation of distribution network graphic SVG format,then puts forward the model for extracting feature vector graphics based on SVG format to provide the SVM model for the analysis and provide with data support,after that,puts forward the extraction method of the model based on Java programming language.Research on prediction model of voltage qualification rate based on clustering analysis and SVM algorithm.The part describes the definition of the qualified rate of the voltage value firstly,and its significance on the reliability and decision of power grid,then put forward the qualified rate of the voltage value calculation model,and then introduces the method of building SVM model according to the algorithm of RBF kernel function based on feature vector extraction voltage qualification rate of the graphic,and then set up two SVM prediction model,respectively,before and after clustering model,the parameters of SVM algorithm optimization using genetic algorithm,finally compare the results analysis.Study on the accurate value of voltage qualification rate of the SVM model based on the SVM super plane segmentation method.The first part presents the voltage qualification rate model based on the SVM algorithm of hyper plane separation method,and introduces the mapping theory of low dimensional space to high dimensional space of SVM algorithm RBF kernel function,and then put forward the exact value of voltage qualification rate-feature vector SVM model,finally use large sample and small sample to test the accuracy of the model.