Research On State Evaluation Method Of Smart Electricity Meter Based On Uncertainty Information Fusion

Smart electricity meter is entering into each family gradually,realized the complete coverage of information collection of electricity,it plays more important role in the construction of smart grid.The life of smart electricity meter is limited,and the process of recycling is cumbersome,which will cause great waste of resources.If the computer technology can be used to predict the state of smart electricity meter and obtain the status information of smart electricity meter,it can be used to maximize the use of smart electricity meter.Through years of accumulation,the State Grid company has accumulated a large number of it can be used to maximize the use of smart electricity meter quality data,which covers the evaluation of smart electricity meter status of the required information basically,but it have not been fully exploited and used.At home and abroad,there is few research on the state evaluation of smart electricity meter,especially in the domestic,the research of smart electricity meter is still in the initial stage.Information fusion also called multi-source information fusion.It refers to merging and combining information and data from multiple sources to form a unified result,which is more precise and complete than single information source.The process of information fusion mainly includes multi-sensor data acquisition,data preprocessing,data fusion processing(feature extraction,data fusion Computation)and output process.In order to solve the problem of resource waste caused by smart electricity meter,the state evaluation of smart electricity meter is studied.A lot of data collected by the State Grid are taken into account,although it covers the information needed for state evaluation,there are many interference data in it.In order to solve this problem,a multi information fusion method is adopted in this paper.The main research work of this paper is as follows:First,this paper introduces the development of smart electricity meter based on the research of smart grid and other scholars.From two aspects of civil and military,this paper introduces the application of information fusion technology,expounds the flow of information fusion technology,and gives a brief description of the methods and techniques involved.Second,in view of the large amount of data given by the National Electric Net Ltd,this paper proposes an state evaluation method of smart electricity meter based on the maximum margin Bayesian network classifier.Second,in view of the characteristics of a large number of data given by the State Grid,this paper presents an smart electricity meter state evaluation method based on the maximum margin Bayesian network classifier.In this method,the Maximum margin Bayesian network classifiers classifier(MMBN)is used to classify the state of the smart electricity meter.Since most of the data are time information and have continuous properties,a continuous attribute discretization method based on information entropy is used in data preprocessing,and the data is discrete in equal distance.Then,genetic algorithm is applied to feature selection of preprocessed data.Finally,MMBN method is used to complete state evaluation.Compared with the SVM classifier,the experimental results show that the method achieves better state evaluation effect.Third,in view of the poor generalization ability of single classifier and the problem that multiple classifiers may cause the waste of inefficient resources,this thesis proposes a smart electricity meter state evaluation method based on selective integration.This method trains the same classification algorithm many times,uses the K-means algorithm as the choice strategy,and uses the coincidence error of the two classifiers as the standard of measuring distance,and divides the classifiers that produce the same error into the same group.The number of selected classifiers is determined by experiments,and finally the voting results of the selected classifier are polled.Experimental results show that the selective ensemble method improves the accuracy of classification.