| In the era of big data, the value of electric power data get increasing attention. Along with the electric powerinformatization proopulsion, the power data presents an explosive growth. The use frequency, concurrent access and statistical calculation of power information systems are growing at a geometric level. The completion of the State Grid centralized deployment-based information system brings adavantages to the management of power information system.At the same time it leads to the disappointing performance of some high concurrency application. To solve the contradiction between the increasing data qunitity and the backward computation ability in the powerinformation system, high performance computing method needed to be developed.The foundation of the information technology era is based on the "calculation".Currently, distributed computing MapReduce, memory computing HANA and Exadata Oracle are commonly used to deal with big data calculation. These methods can be used to improve the performance of the power information system in a certain extent. However, there also exist some problems such as:the bottleneck of the disk I/O, high cost, renovation workload and so on.Meanwhile, part of the core technology is confidential, which makes the system’s opration and maintenance difficult. Based on the existing method of high performance computing, this paper proposes an Object of Parallel Computing Architecture(OPCA), which combined the technique of objectification, parallel computing and memory computing. OPCA has been used in the State Grid asset quality supervision, Equipment lean management system and so on since January,2015. The experiment and applicationresults show that:On the one hand OPCA geatly improving the reliability of the system, on the other hand improve the system efficiency to hundreds of times.Its calculation efficiency, reliability and system renovation workload are better than similar products at home and abroad. |
PDF Full Text Request