Font Size: a A A

Research On The Hazardous Chemical Release Source Characteristics Inversion Methods

Posted on:2017-10-30Degree:MasterType:Thesis
Country:ChinaCandidate:R ZhangFull Text:PDF
GTID:2348330536959032Subject:Power Engineering and Engineering Thermophysics
Abstract/Summary:PDF Full Text Request
Determination of emission sources characteristics is crucial to emergency decision involving hazardous chemical releases.A three stages search routine is proposed to deal with complicated sources characteristics inversion: emission sources characteristics preliminary estimation,emission sources position ranges determination,emission sources accurate positioning.The major contributions of the research can be concluded as follows:Firstly,a modified genetic algorithm(MGA)is proposed to deal with the sources characteristics preliminary estimation mission.The literature suggests that genetic algorithm is the most widely used direct optimization method for solving emission sources characteristics inversion problems.However,genetic algorithm has some drawbacks when it is used in engineering practice like premature convergence and high computation time cost.In the design of modified genetic algorithm,loser-gene is introduced to enhance population diversity and a strengthening local area search step is embedded to deal with genetic algorithm cataclysm problem.The simulation results showed that the modified genetic has better global optimum searching ability and is more robust to data error.Secondly,Markov Chain Monte Carlo sampling method coupled with adaptive Metropolis algorithm is proposed to deal with the emission sources position ranges determination mission.The preliminary result of MGA is used as sampling initial point for MCMC to accelerate sampling process.Take the logarithm of concentration data to balance influence of data across different orders of magnitude.“Partial Match and Cooperative Search” strategy is introduced to strengthen MCMC method robustness to data error.Thirdly,a modified guaranteed convergence Particle Swarm Optimization(MGC-PSO)is proposed to deal with the emission sources accurate positioning mission.In the design of MGC-PSO,“Requisition and Reset” strategy is introduced to accelerate search process and strengthen method robustness.A multiple sources simultaneous localization method is proposed to deal with complicated multiple emission sources problem based on MGC-PSO framework.Finally,emission sources characteristics inversion methodology frameworks for single emission source and multi-sources scene are proposed.MGA,MCMC and MGC-PSO method are used in series to complete single source inversion methodology framework.Primary component analysis is added to methodology framework to deal with the emission sources number estimation mission.“Divide and Rule” strategy is used in multi-sources inversion methodology framework to conduct sources characteristic inversion.
Keywords/Search Tags:source characteristic inversion, genetic algorithm, Markov Chain Monte Carlo sampling method, particle swarm optimization, source localization methodology framework
PDF Full Text Request
Related items