| In the "Industry 4.0" era,modern information technology is developing in the direction of digitization,visualization,and intelligence.Its essence is to enable the physical world and the information world to better interact and integrate.A large number of electronic devices are stored in the data center computer room,and stable operation can only be achieved if the environmental parameters are kept within the normal range.The current environmental monitoring of the computer room is mainly based on human inspection records,which is not only inefficient,but also unable to realize the monitoring of the overall condition,and the environmental data is collected at fixed time intervals,which is real-time.The current monitoring methods cannot predict and warn of abnormalities in advance based on real-time data.Digital Twin is a synchronous digital mirror image of the physical world,which can feed back optimization information to the physical world by analyzing and predicting derived data.Aiming at the deficiencies of the computer room environment monitoring and combining the concept of digital twins,this thesis realizes a 3D computer room environment monitoring and early warning system based on digital twins,which effectively solves the existing problems.The main content of this thesis is as follows:(1)A digital virtual mapping of the overall environment of the computer room was carried out,and the overall outline of the computer room and a three-dimensional model of the main equipment were constructed in the virtual space.First,use the modeling software3 D Max to build a virtual model of each part of the computer room with the actual proportions,build a virtual model of the overall outline of the computer room through the virtual engine Unity3 D,and import the equipment model into Unity3 D to form a complete3 D computer room scene,and then Further optimize the scene,add light effects to enhance the visual reality,give the model physical attributes in reality,and realize the user perspective roaming function in the 3D computer room scene.(2)Perform predictive analysis on the twin data between the computer room entity and the virtual model,and obtain the simulation prediction data to optimize and adjust the entity.Aiming at the characteristics of the computer room environment data meeting the time series,a non-linear combined time series prediction model based on RBF neural network is proposed.The LSTM and Fprophet models are used to model the computer room environment data,and then the two prediction models are used to perform non-linear predictions.Linear combination is used to obtain the optimal weight to predict the simulation data at the next time point.The combined prediction model proposed in this thesis is compared with the traditional prediction model through temperature data,which proves the effectiveness of the combined prediction model and can realize early warning of abnormal environmental data.(3)With the support of the virtual model of the computer room and environmental prediction data,a 3D computer room environment monitoring and early warning system was further designed.The system includes a data storage layer,a business logic layer,and a user layer.The data storage layer includes the environment collection data,user interaction data,and analysis and prediction data of the computer room.The data-driven model at the business logic layer realizes the logic functions of each scene interaction.Finally,a human-computer interaction interface was designed to integrate all scene resources into the login module,monitoring module and alarm module of the interface,realizing the overall real-time environmental monitoring,querying the information of each device in the roaming scene,reminding of the alarm and early warning information.Compared with traditional monitoring methods,the functions of positioning and displaying abnormal information on the model have the advantages of real-time visualization,accurate positioning of abnormalities,and early warning. |
PDF Full Text Request