Power System Time Domain Simulation Under Uncertainty

Power system time domain simulation is an undispensable tool for planning, design, operation, research etc. As the approximate of model parameters and the error in the measurement, conversion and transmission, the parameters may have significant uncertainty; Due to finite accuracy in computers, accumulation of rounding error exists in the simulation process. The mainly method in time domain simulation is deterministic simulation techniques, thus the results may be unreliable. Therefore, taking uncertainty into time domain simulation is significantly important.In this paper, we propose power system time domain simulation under uncertainty based on Monte Carlo method, interval arithmetic and interval Taylor model arithmetic respectively on the basis of Taylor series expansion. The proposed methods have been verified on some test systems.Monte Carlo method is an effective technique for time domain simulation under uncertainty, which can accurately analyze the impacts of uncertainty of mechanical power inputs, inertia constants and initial values on simulation results. However, this method suffers a large computational burden.Interval arithmetic is an important tool for reliable computing. In this paper, the uncertainty of model parameters is described by intervals and each variable in differential equations is replaced with interval variable. Thus, time domain simulation based on interval arithmetic is presented. This method allows a rigorous estimation of the influence of either form of uncertainty and only needs one simulation. It is computationally very fast compared with Monte Carlo method.Interval Taylor model arithmetic is one of the productions by introducing symbolic techniques into naive interval arithmetic, which can record the correlation among variables to some extent, thus help to reduce the overestimation. Time domain simulation under interval Taylor model arithmetic is presented. This method can avoid the conservative results of naive interval arithmetic and computationally more efficient than Monte Carlo method without sacrificing accuracy.The results of WSCC 3-unit system and New England 10-unit system demonstrated the effectiveness and practical value of the methods.