基于数据挖掘的灾害预警云计算平台设计

传统的基于无线传感器网络的灾害预警平台通过ZigBee无线传感器网络及时采集环境数据,数据由GPRS网关节点传输到远程监控中心,并进行数据分析与预警,没有对数据进行汇聚处理,无法解决数据间的规约问题,预警效率和准确率等性能均较低。设计基于数据挖掘的灾害预警云计算平台,平台整体结构包括感知层、网络层和应用层;平台通过数据汇集调度模块对实时环境信息数据进行汇聚,解决不同数据间的规约情况;利用灾害事件触发模块根据各环境信息监测站获取的实施环境信息对比标准信息得到的灾害情况,将灾害预警等级划分为五级。平台软件中设计灾害事件触发检验流程,并通过数据挖掘算法并行化技术缩短数据挖掘时间,提升灾害预警时效性。实验证明,所设计平台在保障有效进行灾害预警的同时,平均传输速度为7.97 byte/s,平均传输时延为13.13 ms;CPU占用率为28%～34%,内存占用率为9%～19%,且准确率较高、稳定性佳。

Design of Disaster Early Warning Cloud Computing Platform Based on Data Mining

The traditional disaster early warning platform based on wireless sensor network collects environmental data in a timely manner through the ZigBee wireless sensor network.The data is transmitted from the GPRSP network clearance node to the remote monitoring and control center,and data analysis and early warning are conducted.Without data aggregation processing,the protocol problem between data can not be solved,early warning efficiency and accuracy can not be achieved.Rate and other properties are low.Design disaster warning cloud computing platform based on data mining.The overall structure of the platform includes perceptual layer,network layer and application layer.The platform gathers real-time environmental information data through the data collection scheduling module to resolve the statutes between different data;Based on the disaster information obtained by each environmental information monitoring station,the disaster event trigger module is used to classify the disaster early warning level into five levels.In the platform software,the disaster event trigger inspection process is designed,and the data mining algorithm concurrent technology is used to shorten the data mining time and improve the timeliness of disaster warning.Experiments show that while the designed platform guarantees effective disaster warning,the average transmission speed is 7.97 byte/S,and the average transmission time is 13.13 MS;The CPU occupancy rate is 28%-34%,the memory occupancy rate is 9%? 19%,and the accuracy is high and the stability is good.