Path optimization model of smart elderly care service based on reinforcement learning and particle swarm optimization DOI
Yana Tang, Peng Fu,

Bo Zhang

и другие.

Journal of Computational Methods in Sciences and Engineering, Год журнала: 2025, Номер unknown

Опубликована: Апрель 25, 2025

To enhance the efficiency and adaptability of path planning in smart elderly care services, this paper proposes a hybrid optimization model integrating Deep Q-Network (DQN) Particle Swarm Optimization (PSO). Traditional approaches struggle with dynamic environmental adaptation multi-objective optimization, often leading to suboptimal routing. address this, proposed employs DQN establish an autonomous decision-making framework, leveraging reinforcement learning dynamically optimize selection based on variations reward mechanisms. Meanwhile, PSO enhances global search capability, adjusting particle positions velocities mitigate risk local optima improve overall efficiency. A framework is further introduced, incorporating weighted coefficients balance competing objectives ensure comprehensive optimization. Additionally, mechanism enables real-time data updates, allowing system swiftly respond sudden changes continuously refine selection. Experimental results demonstrate model’s superior performance, achieving average time 52 seconds service response 17 s. The approach maintains high accuracy, minimal distance deviation 1.35%, delivers optimized objective function value 0.84. These findings highlight effectiveness adaptive planning, offering robust intelligent solution for mobility

Язык: Английский

Adaptive Particle Swarm Optimization with Landscape Learning for Global Optimization and Feature Selection DOI Creative Commons

Khalil Abbal,

Mohammed El Amrani, Oussama Aoun

и другие.

Modelling—International Open Access Journal of Modelling in Engineering Science, Год журнала: 2025, Номер 6(1), С. 9 - 9

Опубликована: Янв. 20, 2025

Particle swarm optimization (PSO), an important solving method in the field of intelligence, is recognized as one most effective metaheuristics for addressing problems. Many adaptive strategies have been developed to improve performance PSO. Despite these advances, a key problem lies defining configuration criteria algorithm. This study presents variant PSO that relies on fitness landscape analysis, particularly via ruggedness factor estimation. Our approach involves adaptively updating cognitive and acceleration factors based estimation using machine learning-based deterministic way. We tested them global functions feature selection problem. The proposed gives encouraging results, outperforming native almost all instances remaining competitive with state-of-the-art methods.

Язык: Английский

Процитировано

0

Path optimization model of smart elderly care service based on reinforcement learning and particle swarm optimization DOI
Yana Tang, Peng Fu,

Bo Zhang

и другие.

Journal of Computational Methods in Sciences and Engineering, Год журнала: 2025, Номер unknown

Опубликована: Апрель 25, 2025

To enhance the efficiency and adaptability of path planning in smart elderly care services, this paper proposes a hybrid optimization model integrating Deep Q-Network (DQN) Particle Swarm Optimization (PSO). Traditional approaches struggle with dynamic environmental adaptation multi-objective optimization, often leading to suboptimal routing. address this, proposed employs DQN establish an autonomous decision-making framework, leveraging reinforcement learning dynamically optimize selection based on variations reward mechanisms. Meanwhile, PSO enhances global search capability, adjusting particle positions velocities mitigate risk local optima improve overall efficiency. A framework is further introduced, incorporating weighted coefficients balance competing objectives ensure comprehensive optimization. Additionally, mechanism enables real-time data updates, allowing system swiftly respond sudden changes continuously refine selection. Experimental results demonstrate model’s superior performance, achieving average time 52 seconds service response 17 s. The approach maintains high accuracy, minimal distance deviation 1.35%, delivers optimized objective function value 0.84. These findings highlight effectiveness adaptive planning, offering robust intelligent solution for mobility

Язык: Английский

Процитировано

0