Font Size: a A A

Shuffled Frog Leaping Algorithm And Its Application In The Evaluation Of Water Safety

Posted on:2013-08-09Degree:MasterType:Thesis
Country:ChinaCandidate:C X YuFull Text:PDF
GTID:2268330425486704Subject:Science
Abstract/Summary:PDF Full Text Request
Comprehensive evaluation of water safety is the basis of the management anddecision-making of water environment and water resources, its purpose is toocomprehensive evaluate a country or a region’s water security, and to formulateappropriate measures to protect the country or the region’s social, economic andecological sustainable development based on the evaluation result of region watersecurity. How to effectively evaluate regional water security is important to watersustainable utilization and sustainable economic development. In order to explore thebetter comprehensive assessment model of water safety, and to reflect the actualsituation of regional water security more objective and more accurately, it is necessaryto explore comprehensive evaluation methods of water security.Under the support of the Nation Natural Science Fund (50739002,51179110), andthe base of other’ study on this area, this thesis systematically introduces three kind ofimproved frog leaping algorithm optimizing algorithm, Immune evolutionary shuffledfrog leaping algorithm, Immune evolutionary shuffled frog leaping algorithm based onparticle swarm optimization, Immune evolutionary shuffled frog leaping algorithmbased on chaotic mutation, to optimize the six index formulae(including the exponentialuniversal formula of logarithm type power function, the universal formula based onWeber-Fechner law, the universal semi-Г type distribution index formula, the pollutionhazard S-type function formula, the universal index with power function of weightedsum and the parameterization combination operator universal index formula), projectionpursuit regression model(PPR), radius basis function neural network model(RBF) andartificial neural network model (BP). Formulae and models optimized by the leap frogalgorithm were used in water safety evaluation about Jinan City, Dalian City,Xiangjiang River and other area, and the favorable results, which direct new way forregion water safety evaluation and provide foundation of decision making for environmental management department, show that the formulae and models proposed inthis paper not only universal and feasibility, but also practicality.To sum up, the principal findings in the research are as follows:(1) Just as normalized symmetry principle commonly existing in nature, the watersecurity information described the characteristics of factors and changes in laws arebound to meet the normalized symmetry. Thus, if a proper reference value andnormalized transformation form appears, the new models, new methods and newformulae of environmental information analysis based on normalized transformationwould become simple and universal.(2) Six index formulae(including the exponential universal formula of logarithmtype power function, the universal formula based on Weber-Fechner law, the universalsemi-Г type distribution index formula, the pollution hazard S-type function formula,the universal index with power function of weighted sum and the parameterizationcombination operator universal index formula), projection pursuit regressionmodel(PPR), radius basis function neural network model(RBF) and BP neural networkmodel with normalized indices values were established, which based on properreference values and transformed formulae were set for each water security indexrespectively in typical regional.(3) In order to avoid premature convergence, three kind of improved optimizingalgorithm were proposed, which are Immune evolutionary shuffled frog leapingalgorithm(IE-SFLA), Immune evolutionary shuffled frog leaping algorithm based onparticle swarm optimization(IEPSO-SFLA), Immune evolutionary shuffled frog leapingalgorithm based on chaotic mutation(IEC-SFLA). Selecting eight test functions to testperformance of the algorithm, results show that the improved algorithms can not onlyfree from trapping into local optimum, and close to the global optimal solution with thehigher precision, but also speed the convergence.(4) The results of evaluation examples show that the various universal indexformulae and models for water security evaluation have the characteristics of simplicityin form, simple for calculation, as well as universality and commonness.
Keywords/Search Tags:Water security evaluation, Shuffled frog leaping algorithm, Indexformula, Projection pursuit model, Radius basis function neural network model, BPneural network model
PDF Full Text Request
Related items