Study On Improved Particle Swarm Optimization And Its Application In Optimal Operation Of Reservior

Water is an important material basis for sustainable economic and social development,a serious shortage of water resources, comprehensive utilization rate is low, so the《NationLong-term Science and Technology Development Plan (2006-2020)》indicated the optimalallocation of water resources and comprehensive development and utilization will as apriority theme, proposed to focus on the research and development of key technologiesHydraulic governance regulation. With the construction of hydroelectric energy projectsforward, hydropower reservoir optimal scheduling has become a core issue in thelong-term economic operation of hydropower systems. By studying the relevant optimalscheduling technology to protect the security and stability operation, improve water useefficiency, comprehensive benefits play a reservoir of great significance.Affected by the hydrological profiles, power control and timing differences, thereservoir scheduling model presents a high-dimensional, non-convex, nonlinear and othercharacteristics, it is difficult to solve by traditional methods. Intelligent swarmoptimization algorithm to break the traditional computing model optimization problemaccurate model, suitable for those traditional methods to solve difficult problems. As anemerging meta-heuristic optimization techniques, particle swarm optimization throughsimulation of biological foraging behavior to solve optimization problems, with fastconvergence, strong optimization ability, etc. But the algorithm exists sensitivity topopulation initialization, easy algorithm precocious and other shortcomings, the paperstudies the rule of population evolution, then propose two improved algorithms and appliedto practical optimization problems. The main work and specific research contents are asfollows:(1) Aim to improve the particle swarm optimization (PSO) shortages,such as, fallsinto local minima easily and lack in population diversity, this paper proposes PSOalgorithm with improved mechanisms of composite chaos, introduced an ImprovedChaotic Particle Swarm Optimization (ICPSO), using the chaotic map’randomness andergodicity, mapped by using a composite consisting of different chaotic, dynamicoptimization of the initial population and the elite particle chaotic disturbance. numericalsimulation results show that the improved algorithm can effectively escape from localminima.(2) In order to improve the low convergence rate and poor population diversity ofstandard PSO, the Gaussian disturbance strategies and reverse particle learning strategies isintroduced to the algorithm named Adaptive Multi-strategy Particle Swarm Optimization (AMPSO). we use particle activity to judge the degree of convergence and select updatestrategy adaptively. Numerical simulation results show that the algorithm effectivelysolved the slow convergence, low accuracy problems. Finally, the algorithm is applied tominimize the error hyperspectral endmember extraction combinatorial optimizationproblems, simulations and real data experiments show that the algorithm can obtain a goodextraction results.(3) Long-term scheduling of reservoir for flood control and the factors to beconsidered in different periods generation, it is a set of multi-stage, multi-variable,multi-constraint in one of the complex nonlinear problems, so, using these improvedalgorithms to slove the reservoir optimal scheduling problem could be effective, numericalexample proved the accuracy of the algorithm can effectively solve the low convergenceand dimensionality disaster defects.