2026-05-20T18:56:56Zhttps://keep.lib.asu.edu/oai/request

oai:keep.lib.asu.edu:node-1992812024-12-23T18:01:48Zoai_pmh:alloai_pmh:repo_items

199281 https://hdl.handle.net/2286/R.2.N.199281 http://rightsstatements.org/vocab/InC/1.0/ All Rights Reserved 2024 91 pages Doctoral Dissertation Academic theses Text eng Su, Xin Pavlic, Theodore TP Pedrielli, Giulia G Yan, Hao H Berman, Spring S Arizona State University Partial requirement for: Ph.D., Arizona State University, 2024 Field of study: Industrial Engineering This dissertation explores an innovative approach to enhancing engineering design optimization, where mathematical optimization techniques generate designs that meet specified objectives within defined feasibility constraints. For single-objective problems, achieving a competitive solution requires searching the parameter space broadly to ensure global competitiveness. Multi-objective problems demand a diverse set of solutions representing balanced trade-offs among objectives. In both cases, optimizers face the challenge of balancing broad exploration of parameter space with intensive refinement of promising regions. In non-optimization contexts, consensus algorithms on networks can exhibit ``hyper-jump diffusion'' where the group exhibits unpredictable but ergodic jumps. Although this is typically undesirable for network consensus, this collective variation has potential value for population-based optimizers as a driver of diverse search patterns across parameter space. Inspired by this, this dissertation introduces Fading Consensus~(FC) dynamics into population-based optimization to harness these effects and improve both global exploration in single-objective problems and solution performance measures in multi-objective ones. Specifically, I apply FC to a population-based extension of the Newton--Raphson method and to Particle Swarm Optimization~(PSO), showing enhanced performance on benchmark problems. Further, I adapt FC to multi-objective optimization methods like Multi-Objective Particle Swarm Optimization~(MOPSO) and Non-dominated Sorting Genetic Algorithm~(NSGA-II), examining when and how beneficial diffusion can be predicted and used dynamically to improve search behavior. Overall, Fading Consensus networks emerge as a valuable tool for enhancing variation in population-based optimizations, promising significant gains in both solution quality and diversity. Industrial Engineering Dynamic Consensus Link Failure Local Active Information Storage Metaheuristics Optimization Algorithms Population-based Optimization Agreeing to Disagree: Incorporating Fading Consensus Dynamics for Better Exploration and Exploitation in Population-Based Numerical Optimization