Research On Waterway Intelligent Service System Based On Tensorflow

In recent years,with the advancement of the construction of "digital waterway",the main inland rivers,including the Yangtze River,Xijiang and Heilongjiang waterway management departments,have mastered and accumulated a large number of navigation waterway sensing data results in various aspects such as electronic waterway maps,navigation marks and water levels,and Provide waterway information services to the public through websites,APPs and WeChat.However,most of the waterway information currently served is only real-time dynamic or historical data that has been simply analyzed and processed.The mining of massive data perceived by digital waterways is not deep enough,and it is difficult to provide intelligence that can better reflect the relationship and evolution of waterway elements.service.In view of this problem,this paper is in line with the development trend of "smart waterway",using the latest artificial intelligence technology to research the intelligent service of inland navigation waterway information,and constructs the waterway intelligent service system architecture based on Tensorflow,and implements and demonstrates the application.The main research contents of this paper are as follows:(1)Analyze and explore the intelligent service mode and technology of inland waterway,and determine the technical solutions centered on Tensorflow,Tensorflow Serving and Docker.Guided by the intelligent service layering idea,the overall architecture of the inland waterway intelligent service including intelligent layer,transport layer and interaction layer is designed.(2)Taking the water level prediction as the application scenario,the intelligent water level prediction model based on LSTM and its implementation in Tensorflow are studied.The deployment architecture of the inland waterway intelligent service system was designed and implemented with Tensorflow serving and Docker as the technology core.Thereby,the complete process of the waterway intelligent service model from construction,training,deployment to service is realized.(3)Based on gRPC and RESTful API,the application mode of waterway intelligent service in different clients such as APP and Web browser is studied.The integration scheme between the intelligent service system and the existing service system is designed,and the interaction and application of the traditional service and the intelligent service are realized for the intelligent route planning scenario in the ship navigation APP.With Tensorflow,Tensorflow Serving and Docker as the core of the technology,this paper deeply studies the architecture of the inland navigation waterway intelligent system and the training,deployment and application of the model,and carries out the demonstration application.It is a new benefit to the inland waterway intelligent service.