The Research On Online Adaptive Settings

With the development of communication technology and means of information collection, this paper presents on-line adaptive setting technology, which is distinctly better than off-line on selectivity, rapidness and sensitivity. Using WAMS, SCADA and EMS to attain the grid operation mode, protection value refresh itself on real-time according to adaptive setting algorithm. Considering the large amount of value computation, and the low real-time performance, the paper designs a large power grid partitioning strategy of large change in sensitivity to decomposition the grid. The sub-region updates the setting value based on disturbance protection area, and accelerates the calculation speed by cluster technology and multi-core parallel computing techniques.Large-scale network is division into several sub-networks, which means the character properties coefficient matrix of power system network is divided into several sub-matrices with the interconnection of coordinating parts. Each sub-matrix can be solved independently, which can develop the parallelism of trends and short-circuit. This method reduces the order of network, and improves computing speed and efficiency by parallel computation. This paper introduces the technology of large changes in sensitivity to deal with the grid mode, which proposes a new equivalent decomposition idea. The detailed analysis of the solution process and the communication process is done. What's more, this paper decomposition the standard IEEE 13 nodes model and verify the equivalence.Power system calculation rapidly and real-time is the premise of on-line checking and on-line setting of large-scale grids. But the traditional method of parallel computing can not take the advantage of widely used multi-core CPU. This paper presents multi-core parallel computing technology which is suitable to power system analysis computation, and shows the basic process on how to use it to deal with matrix decomposition and short circuit calculation. Verified by simulation of the proposed method, the results are compared before the algorithm and after. On the other hand, cloud computation as a new parallel technology is researched on the background of power system calculation. This paper analyzes the development to power system real-time computation by cloud technology with combined traditional cluster and new multi-core parallel technology.Finally, considering only the current operation mode, we test the sensitivity of over-current protection setting value on EPRI 7 nodes mode, whose result proved adaptive setting value is more sensitive than traditional. What's more, we test the fault calculation on Central China Power Grid with the multi-core parallel computation, and compare the performance before the method and after. The results show that the proposed multi-core parallel computing method is correct and effective. What'more, we test the short circuit calculation speed before and after the decomposition of network equivalent based on the Big Change Sensitivity technology, combined with parallel techniques. The results show that the network decomposition technique can be adaptive to parallel technology, and do well in the speed and performance of on-line protection settings.