Research And Implementation Of Monitoring System For Provincial Meteorological Information Based On Data Mining

With the rapid development of Chinese meteorological cause and the improvement of meteorological information construction level,the meteorological department requires data transmission more accurate, timely and effective. Provincial meteorological information center as the master node of data transmission and storage is mainly responsible for the protection of transmission and data including meteorological observation data and product information. However, the monitoring system for provincial meteorological information transmission currently is unable to meet the more rigorous requirements for transmission monitoring because of its distracted business, its low degree automation and its weak monitoring capability and fault analysis capability.In order to solve these problems, a new meteorological information monitoring system based on data mining is designed and developed in the paper. This monitoring system can inspect the system status, make intelligent analysis on alarm information and find the fault source. In one word, this monitoring system as the intellectualized monitoring system can provides effective opinions for the staff. The main contributions are as follows:Firstly, according to the actual situation in Provincial Meteorological Information Center, alarm information of transmission equipment including automatic observation station, area observation station, atmospheric detection and radar data are all integrated into a uniform monitoring system. The situations such as the system in disorder, the transaction not timely and the process not in place are improved, and the work efficiency is increased.Secondly, adding alarm reminding service and duty management functional modules to the monitoring system for recording shift turnover, querying information and so on. At the same time, the personal information, work information and equipment information can be managed in this system. After that, the work process is optimized.Finally,in order to design and complete a meteorological information alarm association model based on association rules mining theory, data mining technology is studied deeply. This model consists of three modules: real-time alarm recognition module based on matching degree, the association rule mining module based on Apriori algorithm and the threshold optimization module based on neural network. The history alarm information is used to create and update the knowledge base about the alarm information and fault source. Then the system takes the real-time alarm message to match the rules and outputs the alarm message with the highest matching degree. After that, parameters are optimized by the threshold optimization module and the monitoring level is improved.