Fast Estimation For Small Disturbance Stability In Power System N-1 Verification

The N-1 small disturbance stability verification is used for evaluating stability when any element outage in power systems because of fault or scheduling. Stability could be analyzed off-line for network planning, but also real-time scheduling is guided on-line by the method. If one element is disconnecting, recomputation of small disturbance stability is needed by traditional approach for N-1 verification. So critical eigenvalues are calculated repeatedly for large-scale power system and which is time-consumed. Obviously, the traditional approach can not be fast enough for on-line analysis. For accelerating computation, a fast estimated approach based on sensitivity analysis used for judging stability is proposed in this thesis.Eigenvalue sensitivity can quantificationally provides the effect of system parameters or transfer functions to eigenvalues on the effecting degree and tendency. As an important and effective tool, it has been effectively used for devising controller parameters, selecting locations for controllers and re-allocating generator power etc. Generally, computation of eigenvalue sensitivity is complicated and time-consumed. In this thesis, the PMT method is used for modeling the system, and then the state matrix is obtained conveniently. So the program has good transplant quality and expansibility, it make good use of sparse techniques and eigenvalue sensitivity with respect to system parameters could be calculated conveniently and quickly.In this thesis, critical eigenvalues describing system stability are analytically expressed as nonlinear functions of line parameters, then estimation equation is obtained from Taylor expansion with higher term omitted. Eigenvalues under line outage are estimated quickly by the equation and the system stability is judged. N sets of eigenvalues are obtained by one complete eigenvalue sensitivity calculation in proposed approach. While repetitive calculation for eigenvalues is replaced and computational time is largely reduced.The most time consuming part in the estimation equation is the calculation for the first and second order eigenvalue sensitivities with respect to branch parameters. The branch admittance coefficient is introduced as branch parameter, then eigenvalue sensitivity with respect to complex admittance is replaced by the real coefficient and the calculation is reducing.The model error in the proposed approach is the truncation error of the third and higher order terms'. So the proposed approach could be incorporated with traditional approach and it is only regarded as a selective critical branch subset method. The correcting values in eigenvalue estimation are regarded as an index, then the branch is in this critical subset when correcting values is larger because of its outage. The eigenvalues are calculated by traditional approach when a branch outage which belongs to the subset. The rapidity and effectiveness of the proposed approach is examined on two testing system. The superiority in computing time of proposed method becomes greater with system scale increasing.