Power Flow Computation Based On Tensor Method And Identification For Coherent Generators

Power flow calculation and recognition for coherent generators are the most basic issues in power system stability analysis. Ill-conditioned power flow will likely appear when the system structure is fragile under fault or when system at heavy-load situation to make the power flow calculation be difficult to converge.Thus the study of ill-conditioned power flow algorithm is very significant for planning and stability analysis of power system. Moreover, the recognition for coherent generators has attracted great attention for years because of its great significance for analysis and criterion of system stability. To focus on the deficits and difficulties of algorithm for ill-condition power flow and recognition for coherent generators, a thoroughly exploration and study have been done in this thesis and many satisfactory achievements are obtained. The main content and achievements of this thesis consist of two parts which are presented as follows.1,Two new approaches for power flow calculation based on tensor method are presented in the thesis.Method one used least squares optimization algorithm to obtain tensor increment when the real roots of tensor equation did not exist, which led to a better convergence. Method two adopted tensor direct calculation strategy in polar coordinates to achieve the quadratic of the increment by the direct calculation of the second derivative of power flow equations; therefore, its convergence and calculation speed were significantly improved; although its calculation speed was nearly the same as that of Newton method for light or normal load situation, it was obviously faster than that at the heavy load situation, sometimes even 30% faster. The results of several examples demonstrate that the two methods proposed are valid.2,Two new approaches for coherent generators identification are presented in the thesis.Method 1 is the algorithm for recognition of given generator swing curves. The effectiveness indicators is applied for to obtain the optimal generators clustering in method 1, with the given generator swing curves. Method 2 is the algorithm for recognition of coherent generators under given fault. In this method, CCT(critical clearing time) for the given fault is calculated first, and then the generator swing curves for the given fault cleared at CCT ,which is also called generator critical swing curves, could be achieved. Generator classification can be obtained by the fuzzy clustering process of generator critical swing curves, and the generators belonging to same classification are the coherent generators, because the generator critical swing curves include sufficient coherency information of generators. The results of several examples demonstrate that the two methods proposed are valid.Since approaches presented in this thesis have strong theoretical fundation, the softwares developed based on them have the protential to be applied in practical engineering.