Font Size: a A A

Research On Oil Blending Optimization With Particle Swarm Optimization

Posted on:2008-09-21Degree:MasterType:Thesis
Country:ChinaCandidate:J H FengFull Text:PDF
GTID:2178360215994713Subject:Pattern Recognition and Intelligent Systems
Abstract/Summary:PDF Full Text Request
Gasoline blending is a key component that influences refinery economic profit, andits essence is to use the middle component oil of refinery to produce various products inaccordance with the appropriate mixed formula which meet the standards for gasolineblending. With the continuous rising of oil prices and the improvement of environmentalstandards, companies are requested urgently to improve their productivity and efficiency,produce clean gasoline which meets the environmental standards. As a result, moresevere requirements are proposed on oil blending.Blending problems can be generally abstracted as a complex nonlinear optimizationproblem. Particle Swarm Optimization (PSO) is a new swarm intelligent algorithmproposed by Russell Eberhart and James Kennedy in 1995. The algorithm is efficientparallel, simple, which can be used to solve large nonlinearity, non-peak and morecomplex optimization problems.The research target of this thesis is to put forward solutions for the complexoptimization problems of the gasoline blending process, through the analysis of PSOalgorithm, advanced PSO algorithms are proposed to help solving complex nonlinearoptimization problems. Formulation is done by using improved PSO algorithm to dealwith the gasoline blending examples.The main works of this thesis are as follows:(1) A two particle swarms optimization algorithm with mutation operator is investigated.The proposed algorithm constructs two swarms of particles with different velocityrestricts, introduces a communication mechanism and a special mutation operator,sequentially elevates the swarms to a higher ability of global convergence. The newmethod is illustrated with a blender optimization problem, and the feasibility andeffectiveness of the proposed algorithm is experimentally confirmed by thesimulation results.(2) Based on the simulation results, the influence of increased inertia weight and inertiaweight reduction on the PSO system is analysed. Two variable inertia weights PSOalgorithm is proposed, the simulation results prove the effectiveness of the improvedalgorithm. The two advanced algorithms are applied to the formulation of thegasoline blending problem, and the results are compared with other algorithms. The results showed that the advanced PSO algorithms can solve the blending optimizationproblem effectively.(3) Based on the one-to-one mapping thinking of monotone function, an approach isproposed for gasoline blending optimization problem under uncertainty. Themonotone relationship between the function and the argument are used, theprobability function constraints are changed to the uncertain parameter probabilityconstrains, so that it can the probability constraint can be represented with an explicitexpression. The chance constrains model is changed accordingly to the nonlinearoptimization model. Two paticle swarms optimization algorithm is also applied to thegasoline blending problem, the results show that the quality card edge of the refinedoil index can be achieved, and the settled probability target is also met perfectly.
Keywords/Search Tags:particle swarm optimization, mutation operator, inertia weight, blending recipe optimization, chance constraint
PDF Full Text Request
Related items