Longfengshan Reservoir Optimal Scheduling Algorithm Researching Of Swarm Intelligence

Reservoir operation, also known as the reservoir for use, defined as the runoff water live and hydrological forecasting, storage and discharge runoff storage in a planned way, the ability to adjust some of the most objective criteria as scheduling, make full use of the reservoir, on the inflow redistribution of space and time, and then get the most comprehensive benefits. Hennessy optimal scheduling is including water quality and storage capacity, etc., in order to achieve the greatest overall benefit.Based on the theory of sustainable development of water resources, in order to Longfengshan Reservoir as the research object, to fully understand the current situation in the reservoir operation of the premise, to seek optimal reservoir operation rules to get maximum benefit under the measure whether the reservoir optimal operation mode standards, according to the natural reservoir inflow over the years and in all sectors of water, the actual storage capacity integrated reservoir, reservoir inflow runoff will be described as deterministic, and establish a power plant generating capacity of the reservoir up to the best standards, using a variety of intelligent optimization algorithms theoretical methods and techniques developed Longfengshan reservoir operation control method for Longfengshan Reservoir to provide a reference to maximize the development of water resources. The main results are as follows:(1) According to G-P and Takens theory, calculate the minimum embedding dimension chaotic sequence of phase space reconstruction space required as a wavelet neural network input nodes, hidden layer nodes based on time-frequency analysis to determine wavelet neural network number, and ultimately determine the wavelet neural network topology is 12-24-1;(2) By increasing the momentum and Newton method utilization improved wavelet neural network to adjust the network parameters such that the convergence speed, increase learning efficiency, avoiding in multi-dimensional input, the network convergence speed greatly reduced this deficiency;(3) According to Longfengshan Reservoir 1994-2013 or 20 years of monthly runoff water storage materials, the use of trained improved wavelet neural network model to predict 2014 and 2015 Longfengshan Reservoir monthly runoff, reservoir operation to provide reasonable data.(4) The use of GA, PSO and AFSA improved algorithm to optimize the scheduling Longfengshan Reservoir, the analysis and forecast in 2014 and 2015 as the known conditions of storage of water in the lower reservoir to satisfy the constraints, scheduling, results can provide a reference for the reservoir in 2014 and 2015.