Study On Ant Colony Optimization Algorithm And Its Application To Short Generation Scheduling Of Electric Power Systems

Optimal operation problems in electric power systems are typical non-linear programming problems, in addition, many non-linear, discrete, stochastic and undeterministic factors are involved. Conventional mathematical programming techniques are imperfect for the optimal operation of power systems due to inherent complexity. However, the optimization methods by simulated evolutions are appropriate for solving the non-linear, discrete problems and combination optimization problems which are difficult to be solved via conventional methods. A novel optimization method by simulated evolutions - Ant Colony Optimization algorithm and its application in the Unit Commitment problem (UC) are exploringly investigated in this thesis. The main achievements are as follows:According to the stochastic mechanism, a novel Ant Colony Optimization algorithm with Random Perturbation Behavior(RPACO) is presented in this paper. The new algorithm includes two important aspects: a amplify factor formulated by inverse exponent function is developed, which is used to avoid premature, on the other hand, corresponding transition strategy with random selection and perturbation behavior is designed, which is designed to prevent the algorithm from stagnating. Furthermore, selection of parameters in the new algorithm is analyzed, a selection method of parameters which possesses a generality is proposed based on many numerical simulations, and finally the best numerical range of each parameter is determined. Numerical simulations demonstrate that the new algorithm possesses more strong global optimization capability.The optimal unit commitment is solved via the Ant Colony Optimization algorithm in this paper. Concepts of a state and a decision are introduced, and the definition of a path is proposed. Thereby, the unit commitment problem is designed with a simulation to TSP. Because only the problem without constrains can be solved through ACO, two effective methods are developed to deal with all kinds of constrains of UC. First, the list is used to restrict states which don't satisfy the constrains, such as spinning reserve,minimum start up/shut down time. Second, the transmission line capacity limit is handled by an additive penalty item. Numerical results demonstrate the feasibility and availability of the algorithm.Considering the uncertainty of load forecast and likelihood of component failure, a certain generation reserve should be established. In this paper, the probabilistic technique is used to evaluate the spinning reserve requirements of the system, that is, the reliability constraint is considered in the model of unit commitment. The operating risk of the combination states in the system is evaluated via the security function method, and combinations, in which the security constraint doesn't be satisfied, will be eliminated. Moreover, the sensitivity of the desired security level to the optima is investigated.