Community Gas Risk Prediction Technology Based On Temporal Dynamic Knowledge Graph

With the rapid popularization of city gas,there are also potential risks while improving the community residents’ quality of life.In order to improve the safety of community residents’ gas use,it is necessary to predict the community gas risk.However,the existing gas risk assessment methods are mainly based on static risk assessment,which is difficult to fully reflect the dynamic characteristics of community gas risk.In addition,these methods often ignore the interaction among community gas risk factors,and cannot effectively predict and prevent potential coupling risks.Therefore,the thesis proposes a community gas dynamic risk prediction method based on temporal dynamic knowledge graph,which aims to realize the dynamic prediction of risk level in community gas system by using the temporal and dynamic characteristics of temporal dynamic knowledge graph.The research content of the thesis is as follows:(1)The thesis analyzes the risk factors of the community gas system,locates the risk sources and risk problems of the community gas system,and summarizes them.It concludes five potential causes of gas system safety incidents,and establishes a community gas risk assessment index system.The thesis further analyzes its weight and determines the risk level of community gas,and processes the community gas data into a temporal dynamic knowledge graph of community gas according to the index system and risk level.(2)According to the temporal dynamic knowledge graph to predict community gas risk,the thesis mainly proposes two community gas risk prediction methods,and verifies the two methods respectively.The first method is a community gas risk prediction method based on a relational graph convolutional network,which is divided into node information aggregation module,temporal information aggregation module and risk prediction module.The node information aggregation module uses the relational graph convolutional network to complete the aggregation of node information in the graph space.The temporal information aggregation module uses the long short-term memory network to complete the aggregation of the graph temporal information.The risk prediction module obtains the conditional distribution of gas risk prediction through encoding and Multilayer Perceptron(MLP)decoding,and obtains the gas risk level.The second method is based on temporal dynamic knowledge graph for prediction,which is divided into historical information learning module,current information learning module and risk prediction module.The historical information learning module mainly uses the MLP network to learn historical knowledge graph information,generates index vectors and estimates the probability of gas risk entities in the historical vocabulary through the softmax function.The current information learning module uses the MLP network and the gated recurrent unit network to generate index vectors respectively,and synthesizes the above index vectors through the softmax function to obtain the probability of the gas risk entity.The risk prediction module performs a weighted summation based on the probabilities of the gas risk entities of the first two modules to obtain the final prediction result.(3)The thesis realizes the community gas risk prediction technology platform,which is specifically divided into data collection and transmission module,community gas risk prediction algorithm module and community gas risk prediction visualization module.The data collection and transmission module is mainly based on6 Lo WPAN to realize the data acquisition of the community gas system,and upload it to the Io T cloud platform(Things Board)through the MQTT protocol.The community gas risk prediction algorithm module mainly completes the code implementation of the prediction algorithm.The community gas risk prediction visualization module receives data from the Io T cloud platform,and uses the community gas risk prediction algorithm to process the above data.The front-end of the module uses Vue.js technology,and the back-end uses the Spring Boot framework to visualize the community gas risk prediction results exhibit.In the thesis,the community gas dynamic risk prediction algorithm based on temporal dynamic knowledge graph improves the real-time performance of community gas risk prediction and the accuracy of gas risk prediction,realizes real-time community gas data monitoring and gas risk prediction results through the community gas risk prediction platform.The visual display has certain reference significance for community gas risk monitoring and prevention.