Design And Implementation Of A Real-time Acquisition And Processing Platform For Internet Of Things

With the proposal and rapid development of technologies such as 5G and artificial intelligence,the Internet of Things has led another wave of information revolution.More and more traditional enterprises are transforming and developing their own Internet of Things platforms.Facing the increasing complexity of application scenarios and the increasing number of devices,it has become a hot issue to be solved for the current Internet of Things platform.Under the above background,this paper designs and implements an Internet of Things real-time acquisition and processing platform.The platform completes the processes of data acquisition and access,data transmission,data processing and data storage through eventdriven network programming framework and stream processing framework.It realizes the mapping and management between equipment and platform and provides the corresponding data service capabilities.The main work of this paper is as follows:(1)Zoo Keeper is designed to realize distributed Netty cluster to ensure high concurrency and high availability of the platform for service stability.The platform supports service registration and discovery,adapts to a variety of protocol parsing capabilities and opens up secondary development interfaces to achieve custom data parsing and improve the overall scalability of the platform.The middleware Pulsar is used to improve the overall stability and reliability of the platform.(2)Flink is used for real-time logical process and data push for the real-time requirement of the platform.Flink SQL CDC is designed to provide data synchronization.(3)Realize the alarm prediction function for massive time series alarm logs generated by the Internet of Things platform.Train the algorithm model based on CBOW-LSTM.According to the existing alarm log,the test has achieved good results and improves the intelligent level of the platform.(4)Design the system architecture and database table structure.Build the model and complete the code writing.Class diagram and sequence diagram are used to explain the functions of each module in detail.After testing and practical use of the platform,the functions and performance of the platform meet the requirements.The CBOW-LSTM algorithm also has high accuracy in the alarm prediction function.