Study On The Warning Model Of Cascading Failure In Power System

In recent years, several large-scale blackouts happened in the world caused huge losses of social economy and people’s life. Generally, the large-scale blackouts are always caused by cascading failures. Therefore, further research still needs to be done to study the mechanism of cascading failure and the characteristics of critical state. We can take effective measures such as early warning and prevention to the system before large-scale failures occur. It has important significance to improve the grid’s reliability.However, online early-warning analysis of cascading failures are having trouble in the computational ability and the precision of calculation. Therefore, this paper studied the early-warning model of cascading failures on the basis of summing up the present studies of power system cascading failure and reliability assessment. Moreover, two aspects of small probability event simulation and parallel simulation are studied to speed up the running speed of the model.Firstly, the early-warning model of cascading failure is built based on the critical characteristic of complex system, including two layer circulations of inside and outside, using the optimal power flow to calculate the reliability index of system. By increasing the load level gradually, to simulate the evolution of system operation pressure until the system reaches the critical state, and study the effect of the margin of transmission line on the grid’s criticality. Secondly, splitting technology is applied to improve the efficiency of simulation of large-scale cascading failure which has small probability.. Splitting is a technique of variance reduction. The large-scale fault is divided into some readily grasped small events and simulated respectively, finally we can obtain the probability of large-scale events for they are the conditional probability of these small events. Compared with the traditional Monte Carlo simulation, splitting technique can capture the fault chain in far less simulation, and the results can be used as the of prediction basis of large-scale fault chain. Finally, by using the technology of parallel simulation, allocating the tasks to different sub-processor and running at the same time. Parallel simulation can further reduce the simulation time, meanwhile maintain the accuracy of simulation results.