Research On Urban Land Use Function Identification And Design Of Online Visual Microservice Computing Engine

Urban land use function data is an important basic data in the field of urban computing.Accurate classification of urban land use functions can help experts to carry out detailed planning.However,there are still many difficulties in the research of urban land use function identification.The data source suitable for urban land use functions classification is relatively single and there are huge differences between different cities.Moreover,the research in urban computing lacks a fully functional online visual analysis software engine.This thesis aims to accurately identify the urban land use functions of Shenzhen and develop an online visual analysis software engine.This thesis proposes a functional classification model of urban land use based on multisource data.In terms of feature engineering,this thesis uses the TF-IDF algorithm to preprocess the POI business data,and solves the problem of uneven classification of POI data;by using mobile phone base stations as discrete points of Tyson polygons,this thesis assigns the flow characteristics of people to At the architectural level;for the architectural form feature data,this thesis has carried out feature expansion.In terms of models,this thesis proposes a XGBoost-based urban building function classification model,and conducts experiments with Shenzhen as an example.The experimental results show that the model proposed in this research has obtained better results in terms of the number of classifications and classification accuracy.This thesis has developed an online interactive analysis and interactive visualization software engine based on WebGL.The server uses the Spring Cloud technology architecture.The front-end uses React component technology,and the data storage module uses MySQL relational database,Redis cache database,and Nginx static resource server for separate storage according to the different characteristics of the data.By storing the trained machine learning model as a PMML format file,this software server program provides users with online calling services of a variety of machine learning models.The different modules on the server side of this software are deployed using Docker containerized virtualization technology,and the container group is arranged using K8 s.By encapsulating the business code of other research in the urban computing field into the Pod container group of K8 s,the software engine also realizes the rapid horizontal expansion of software functions.