Optimizing Gradient Methods for IoT Applications

Hosseini, Eghbal; Reinhardt, Line Blander; Rawat, Danda B.

doi:10.1109/jiot.2022.3142200

Cited by 3 publications

(1 citation statement)

References 29 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Numerous machine learning techniques, heuristics, and meta-heuristic algorithms have been proposed in the literature to address a wide array of optimization problems [50], [51], [52], [53], [54], [55], [56], [57], [58], and [59]. However, the inherent complexity of multi-objective optimization problems requires the creation of hybrid algorithms.…”

Section: B Hybrid Algorithm For Moopmentioning

confidence: 99%

Optimizing Hybrid Renewable Energy Systems:Novel Models and Heuristic Algorithms

Hosseini,

Al-Ghaili,

Deveci

et al. 2024

Preprint

Self Cite

View full text Add to dashboard Cite

In practical applications, real-world challenges ofteninvolve the integration of multiple renewable energy sources,including solar, wind, hydro, biomass, geothermal, and energystorage solutions like batteries. The primary objective ofthese integrated systems is to optimize energy production,enhance system reliability, and simultaneously minimize costssuch as those associated with fossil fuels and greenhouse gasemissions. Therefore, it is crucial to model these complexproblems as all objectives are covered. In this paper, wepropose multi-objective and multi-level mathematical modelsfor Hybrid Renewable Energy Systems (HRESs). These modelsallow us to simultaneously address various objectives andlevels of decision-making in the context of renewable energyintegration. Then, to effectively tackle these complex models,the paper introduces two efficient hybrid algorithms for bothmulti-level and multi-objective models. First algorithm is acombined smoothing approach to address multi-level problem.The algorithm uses Karush-Kuhn-Tucker (KKT) conditions,some mathematical principles and proofs, and a heuristicfunction to smooth the multi-level model. Finally, the algorithmemploys Taylor approximation to further refine the smoothedproblem. The second algorithm, focuses on multi-objective model,which involves two phases: first, a heuristic algorithm simplifiesobjective functions through interpolation; second, the populationis enhanced using the Laying Chicken Algorithm (LCA); and aneural network refines the best LCA generation to identify thePareto front in multi-objective problems.

show abstract