Stochastic Optimal Power Flow Solution Research Based On Genetic Algorithm

Reliability and economy usually conflict with each other in complex power system. For one hand, the power grid of China, which also is the largest and most complicated network in the world, ask for high reliability of safe operation. On the other hand, due to the resource spent in economic development of China become more and more, it's high time that we should make the most efficient use of the available resources. So the power grid must play an important part in alleviating the conflict of reliability and economy, especially when different kinds of different uncertain factors must be considered. Stochastic Optimal Power Flow (SOPF) is proposed to study the kind of problem.SOPF is broadly applied in many aspects of power market, such as optimal dispatch, optimal ancillary control, bidding strategy, demand side management and so on. For this reason, the research and application of SOPF play an important role in smart grid building and development. However, the fitness and solution efficiency in present algorithm to SOPF problem don't agree with each other. The algorithm to SOPF problem applied in real-time computation is still under research. In order to this, this thesis makes an introduction of the kinds of SOPF models and the equivalence to power injected, puts forward an algorithm with high fitness, prospects the research and development of SOPF.From the beginning of the development of OPF, this thesis makes an introduction of SOPF, as well as the significance of the subject study at present. To the summary on the situation of SOPF research, this thesis conducts study on the significance of SOPF with Load Uncertainty. It also puts forward an algorithm with high fitness and the stochastic model considering the chance constraint of voltage and transmission limit applied in practice. Applying this model into IEEE 5-bus system and Heyuan power system, we can get the result of optimization. The SOPF problem uses the genetic algorithm to deal with the uncertainties and common PF method with Monte Carlo Simulation to fulfill chance constraints, which has high fitness. At last, by contrast between genetic algorithm and certainty algorithm, this thesis makes a conclusion that genetic algorithm is more adaptable to SOPF in real-time computation and much more of prospect.