Research On Computational Intelligence Method And Its Application Into Optimal Operation For Feeding Reservoir Group

Optimal dispatch of feeding reservoir group plays a very important role in water resources development, utilization and management. The applicability and validity of available optimization techniques sometimes maybe aren't very excellent in solving the problem of optimal dispatch of feeding reservoir group. It needs to be improved and perfected. Aiming at the research hotspots on Computational Intelligence's application in complex water resources system based on optimal dispatch of feeding reservoir group, in order to conduct some researches on forecasting and optimization, this paper takes optimal dispatch problem of feeding reservoir group in Bao'an District, Shenzhen City as an example, on the whole several researches have been conducted as follows:Firstly, the problem of runoff forecasting is researched for water supply reservoir group based on Phase Space Reconstruction Theory. After the consistency of runoff sequence is analyzed, the statistic method of BDS is applied to prove its non-linearity and the largest Lyapunov exponent is computed, which manifests that there is chaotic characteristics in the runoff sequence of reservoir group. Single-dimensional and multi-dimensional runoff forecast models are built and analyzed based on State Space Reconstruction Theory, Artificial Neural Network and Genetic Algorithm. Their performances in practice are compared and analyzed.Secondly, a hybrid intelligent algorithm called GAA is designed, which is based on Genetic Algorithm and Ant Algorithm and has the characteristic of random search and deterministic research. Penalty factor based on simulated annealing idea is introduced to deal with the optimal problem with constraint effectively. It is a relatively universal frame for optimal problems. This algorithm is compared with the unimproved algorithm through numerical experiment. Furthermore, in practice it embodies its relatively feasibility, applicability and versatility.Meanwhile, in order to overcome the problem of "Dimensionality Disaster" caused by many water supply districts, fuzzy overall evaluation model is used to aggregate many water supply districts into one aggregate district by assigning a relative weight for each district. Accordingly, a decomposition algorithm is designed. Considering the function of water conservation and water release of reservoirs, the notion of feasible interval of joint operation for feeding reservoir group is put forward and the according algorithm is designed, which is proved feasible in practice.Thirdly, Evolution Algorithm, which belongs to Computational Intelligence, is adapted to generate non-inferior solution for multi-objective reservoir group optimal operation and the non-inferior solutions are evaluated by fuzzy theory. After multiobjective evolution algorithm theory is read up and its critical technology is studied in detail, a evolutionary algorithm based on Pareto strength is designed and applied to solve a real case.After the non-inferior solutions are generated, a single-person fuzzy evaluation model and a multi-person fuzzy evaluation model are built based on classic similarity to ideal solution method, which are able to describe the subjective uncertainty of the decision makers in decision process.Lastly, fuzzy sets is used to describe the uncertainty in the reservoir inflow of reservoir groups and the fuzziness of the objective function. Degree of reliability of reservoir inflow and degree of satisfaction of object realization are put forward to evaluate the optimal operation scheme. The stochastic optimal operation model is translated into fuzzy programming model, which not only describes the uncertainty in reservoir inflow, but also simplifies the problem.