Research On The Core Issues Of Precaution System In Cloud

Disaster Precaution System(DPS) is useful for disaster avoiding. Usually, a DPS need to process a large volume of data and react to the disaster just-in-time. However,because the traditional DPSs are usually built on a single computing node, it is hard to work well in scenarios above. Concerning the benefits of Cloud computing and Big Data processing environments, it is a good choice to built and run a DPS in Cloud.(1)The thesis proposed a kind of support environment of DPSs in Cloud, which can provide computing and storage abilities on demand. A Hadoop cluster is built on the platform of Openstack Cloud. Then, to enhance its efficiency, a MapReduce model for distributed data storage and access is proposed, which can reduce the data transportation costs.(2)Focusing on the drawbacks of traditional BP algorithms in the large volume of data processing, a distributed and parallel BP algorithm is designed for DPSs to improve its training efficiency. Experiments and simulation show that the algorithm can improve the efficiency and accurate of DPSs.(3)Based on the research results above, a distributed DPS is designed and developed.