| Two kinds of transient stability controls (TSC), namely preventive control (PC) and emergency control (EC), have been applied widely in modern power systems. PC usually employs continuous actions, influences pre-contingency operation and affects the system dynamics during and after any contingency. EC actions are usually discrete, contingency specific and free from influencing the normal operation. With relatively low cost, PC actions have to pay the cost for the whole action period whether the objective disturbances occur or not. This may be uneconomical or even infeasible in some circumstances. In contrast, EC actions usually cost dearly, and need restoring time if the objective disturbances actually occur. Obviously, it is great significant for assuring large scale power systems in security and economy operation states to optimize PC EC and their coordination.Currently most works on PC employ the sensitivities of the stability energy margin to the generator power injections, which exist some disadvantages and the cause of potential divergence in computation are analyzed. The amount of pre-assigned harmful contingencies in large power systems is a real chanllenge for PC. In this thesis, it is settled soundly through grouping these contingencies into some subsets according to each unstable mode (UM) provided by EEAC, and identifying the worst contingency among each subset by stability energy margin provided also by EEAC. Then a preventive control technique is proposed, which is according to that a generator belongs to the critical or non-critical cluster to instruct the generation allocation, and through an iterative procedure to approach the anticipant objectives. During the procedure, the index of selecting the generators to be controlled would be set flexibly with the objective, so it does not just provide a final optimization result, but an open scheme for different control aims. To stabilizethe system most efficiently, a index based on the kinetic energy at the dynamic saddle point in the controlling one-machine infinite-bus plane provided by EEAC, is built to select the worst affected generators only through one numerical integration. Simulations on a real large power system confirmed its effect.The optimization of EC is taken as a combination optimization problem, in this thesis, which is investigated from model analysis, algorithm design, and its application on the basement of computation complexity theory. Based on Tabu search, a local search technique with changeable local structure (algorithm 2) is proposed for this NP-Hard problem, which provides an approach to control the optimizing process flexibly. In solving NPC problems, the development of complete polynomial time approximation scheme for 0-1 knapsack problem is the most successful example, the thesis analyzes the common points between the scheme and algorithm1 (Being in practice), 2, then describes the key factors influencing the performances of the algorithm 1, 2, which provide a good approach to instruct the improvement and application of EC optimization.Due to the complex mutual-influences between PC and EC, the global optimal TSC is even much more difficult. Prof. Xue.Y(2002) proposed a general optimization framework for the coordination, where the task is formulated as a nonlinear hybrid-programming problem with both integer and continuous variables and with numerous stability constraints. The objective function is the sum of the daily cost for PC and the possibility-weighted cost for EC. In this thesis, a two-layer TSC optimization strategy is proposed. Unstable contingencies are firstly classified into subsets according to their UMs. During the optimal procedure, their UMs should be checked all along to identify the possible changes resulted from the actions. All contingencies in such a subset are of the same critical machines, thus a certain TSC action has the same qualitative effects on their stability. So the two-layerstrategy consists of (1) the lower layer, which coordinates the PC and EC actions for each...
|
PDF Full Text Request