Research On Feature Compression And Extraction Method Of Electrical Equipment Condition Monitoring Data

With the continuous advancement of smart grid construction and the increasingly stable operation of power system,the breadth and depth of electrical equipment monitoring are expanding.The state monitoring data and the source is large.There may be multiple variables describing the same state,and there is redundancy between the partial state attributes.It is possible to effectively reduce the difficulty of equipment state analysis and improve the efficiency of monitoring data processing by extracting key parameter which can respond to the state of the equipment from many status monitoring parameter and deleting the useless and repeated status parameters.In this paper,the present situation and characteristics of electrical equipment condition monitoring are analyzed in detail,and the research methods of data feature extraction are summarized.Then,the application of three feature extraction methods in electrical equipment monitoring data are described in detail.Based on the problems of the type of electrical equipment,such as the type of miscellaneous,the number of parameters and the unclear relationship between the parameters,the key parameter is proposed to improve the principal component analysis.The quantitative parameters of the state parameters are constructed by the fault of the electrical equipment,the major defects and the general defects.The principal component analysis is used to reduce the state parameters to the coordinate system with the comprehensive evaluation as the axis.According to the weight of each state parameters,the weight of the size of the key parameters as a basis for selection,removal of equipment and the state is not relevant sate parameters.Based on the analysis of the relationship between the state parameter and the state of the equipment,proposes a method based on an association rule feature extraction method,which uses the association rules to quantify the relationship between the state of each equipment and the state parameters and calculate the support degree and the confidence degree,the support degree and the confidence threshold as the basis for the feature extraction,select the key parameters of electrical equipment state;Attribute reduction is a feature extraction method which can find out the optimal subset of attribute attributes to replace the original feature set of the original parameters attribute.By reducing the state monitoring parameters of electrical equipment,it can effectively reduce the dimension of equipment condition monitoring information decision.It is possible to improve the efficiency of the state of the electric electrical equipment more quickly and accurately by analyzing the reduced attribute parameters of the equipment state evaluation parameter and the state monitoring data storage space.Finally,Taking the transformer status monitoring data as an example,the application of the above three methods in the state feature extraction of the transformer is described in detail,which proves the feasibility of the method and provides the idea for the analysis and processing of the state data monitoring data.