Distribution Station Monitoring System Based On Deep Learning And Knowledge Base

With the rapid development of renewable energy and distributed generation technologies,the customers connected to distribution stations are also becoming energy producers.Emerging technologies such as renewable energy and the Internet are facilitating the interconnection of multiple complex systems and promoting sustainable economic and social development.In the future,the monitoring and management model for distribution stations will also change dramatically..At present,the intelligent degree of state perception of distribution station is not high.New forms of knowledge representation are needed to express the entities and data sets in the knowledge base of distribution stations.It is urgent to improve the informatization and automation level of distribution station.The specific work in this paper is as follows.(1)The theory of knowledge graphs and deep learning are analyzed in depth.This paper discusses the basic structure of knowledge graph and its construction principle.The basic theory of convolutional neural network is discussed and the structure of each layer of the network is discussed in detail.Distributed deep learning effectively increases the amount of data in the training neural network and improves the complexity of neural network model.Through the combination of deep learning and knowledge graph,the ability to extract candidate units from the knowledge base is improved to complete the representation,organization and storage of knowledge.(2)Research on distribution station monitoring based on deep learning and knowledge graph.The equipment model and time series model of knowledge graph of distribution station are constructed.The knowledge of candidate units is extracted from the heterogeneous knowledge base of the distribution station to construct the knowledge map of the distribution station.The convolutional neural network based on TransR model is used to evaluate the overall state of power distribution.The knowledge inference engine is used to realize the coordination of various services.The unstructured data of the knowledge base of distribution station contains a large number of state parameters of distribution station.In view of this characteristic,knowledge extraction of distribution station based on improved convolutional neural network and distributed deep learning is proposed.In this paper,the application of densely connected neural network in state feature recognition of monitoring images of distribution stations is deeply studied,and distributed deep learning is used to further improve the accuracy of model training.The effectiveness of the proposed method is verified by simulation analysis.(3)The framework of distribution station monitoring system based on deep learning and knowledge graph is constructed.The system is composed of the "end" layer,"edge"layer,"tube" layer,"cloud" layer and "intelligence" layer.Realize data acquisition,transmission,management,value creation.The efficient operation of the distribution station is realized through the cooperation of the five-layer structure.The distribution station monitoring and control system is developed by using graph database,Web server,REST link library,B/S mode and other key technologies,and the basic functions of the system are realized.To sum up,the equipment model and the time series model of the knowledge graph of the distribution station are constructed in this paper.A knowledge extraction method based on improved convolutional neural network and distributed deep learning is proposed,and the effectiveness of the proposed method is verified by simulation.The framework of distribution station monitoring system based on deep learning and knowledge base is constructed.The monitoring system of distribution station is developed.The system has friendly interface and easy operation.