Study On Phase Identification And Transformer Area Identification Algorithm Of Single-phase Smart Meter

With the increasing penetration of distributed energy resources in the power distribution network,new energy system represented by distributed energy resources and renewable energy resources play an increasingly important role in the economy and society.Accurate power distribution network connectivity model is the key to ensure the safe and stable operation of energy management systems,power outage management systems and power distribution management systems in these new energy systems.Transformer area and phase connection information are an important part of the power distribution network connectivity model.However,due to the line optimization and rectification,the construction of the new community,etc.,the phase and station information is often inconsistent with the actual information.The station area and phase identification technology have become the key technologies in the automatic meter reading system.With the development of new energy system and smart grid,the power line carrier automatic meter reading system cannot meet the requirements of power companies gradually.There are many new automatic meter reading technologies which are not power line carriers and whose communication channels are separated from power lines.This thesis focuses on the phase and station identification of non-power line carrier automatic meter reading system,and makes full use of the relevant ideas of clustering to further study the new phase and station identification technology.The main work of this thesis includes the following:Research on phase identification based on spectral clustering.There are relatively few phase recognition algorithms for non-power line carrier automatic meter reading system,and most of them have cost problems based on hardware,while most of the phase recognition algorithms based on data have low identification accuracy and efficiency.Therefore,this thesis proposes a spectral clustering phase identification algorithm based on voltage data.No additional equipment is required.The algorithm divides the data set into several clusters using normalized spectral clustering,and then determines the phase label of a cluster based on the feeder measurement and part of the accurate phase information.Experiments were carried out using the IEEE European Low Voltage Test Feeder datasets.The effects of random removal of some meters and the total time and voltage time resolution of different samples were studied in the case of no noise and noise situation respectively.Compared with the traditional k-means phase recognition algorithm.Simulation and comparison experiments show that our proposed algorithm has higher accuracy,stability and reliability.Research on transformer area identification based on k-means.The existing transformer area identification methods are mostly manually identified by a station identifier that combines the power line carrier and the pulse current.The recognition algorithm based on smart meter data is rare,and there are problems such as slow recognition speed and low precision.Therefore,this thesis makes full use of the voltage data of smart meters and clustering method,proposes a k-means based transformer area identification algorithm.The solution of transformer area identification by customer and concentrator is proposed,including the algorithm and system design of platform area identification.The algorithm first determines the cluster number of the cluster by the silhouette score method,then uses k-means to cluster the voltage data of these smart meters,and finally determines the station attribution of the meter according to the feeder measurement and part of the accurate transformer area information.The validity of the algorithm is proved by the validation of the public data set.