Evolutionary Algorithms-based Parallel Simulation-Optimization Framework for Solving Inverse Problems

Inverse Problems;Simulation-Optimization;Grid Computing;Parallel Computing;Evolutionary Algorithms

Mirghani, BahaEldin YousifAhmed ; Dr. G. Kumar Mahinthakumar, Committee Chair,Dr. S. Ranji Ranjithan, Committee Co-Chair,Prof. E. Downey Brill, Jr., Committee Member,Prof. John W. Baugh, Jr., Committee Member,Mirghani, BahaEldin YousifAhmed ; Dr. G. Kumar Mahinthakumar ; Committee Chair ; Dr. S. Ranji Ranjithan ; Committee Co-Chair ; Prof. E. Downey Brill ; Jr. ; Committee Member ; Prof. John W. Baugh ; Jr. ; Committee Member

University:North Carolina State University

关键词: Inverse Problems;    Simulation-Optimization;    Grid Computing;    Parallel Computing;    Evolutionary Algorithms;   

Others  :  https://repository.lib.ncsu.edu/bitstream/handle/1840.16/3276/etd.pdf?sequence=1&isAllowed=y

美国|英语

来源: null

【 摘 要 】

Inverse problems are computationally challenging to solve due to inherent ill-posedness and computational intractability. In this dissertation the use of a simulation-optimization approach that couples a numerical simulation model with evolutionary algorithms for solution of the inverse problem is adopted. In this approach, the simulation model is solved iteratively during the evolutionary search, which in general can be computationally intensive since a large number of forward model evaluations are typically required for solution. Parallel simulation models and surrogate models are investigated with the aim of improving the simulation model efficiency. Numerical search methods such as parallel hybrid methods and noisy genetic algorithms are investigated for optimization algorithm improvement. In addition, high performance computing and grid computing are explored as a means to facilitate computationally tractable solution of such problems. In this dissertation, the solution of a groundwater inverse problem is explored to validate and illustrate the methods, an array of illustrative groundwater inverse problem applications are demonstrated. The results tend to confirm the effectiveness of the parallel simulation and surrogate modeling for improving the simulation model executing time. Also the results support and illustrate the advantage of using the newly developed EA-based parallel hybrid and noisy genetic algorithms that enhance the efficiency of solving the inverse problem. Finally, the high performance computational experiments performed on the National Science Foundation's TeraGrid resources demonstrate the effectiveness of the grid-enabled simulation-optimization approach in improving solution time.