An Improved Escape Algorithm with Reinforcement Learning and Multi-Strategy Synergy for Global Optimization and Real Problems
- 1Teachers College, Columbia University, New York, NY 10027, USA
- 2Department of Computer Science, Durham University, Stockton Road, Durham DH1 3LE, UK
- 3Polytechnic Institute, Zhejiang University, Hangzhou 310015, China
Abstract
To address the limitations of the standard Escape Algorithm (ESC), including its tendency to become trapped in local optima and its slow convergence speed in complex optimization scenarios, this paper proposes an Improved Escape Algorithm (IESC) for global optimization and robot path planning. First, a reinforcement learning–based decision-making mechanism using Q-learning is introduced to replace the fixed rule–driven population update strategy in ESC. By constructing a closed-loop framework of “individual–action–reward,” the proposed method enables adaptive strategy selection and effectively balances exploration and exploitation during different stages of the optimization process. Second, to enhance information interaction among individuals, a thermal conduction–enhanced search strategy is developed. Inspired by heat transfer principles, this strategy leverages global population information through the combined guidance of the population means and the global best solution, thereby improving local search directionality and convergence stability. Third, an elite-guided Lévy flight and quadratic interpolation fusion strategy is incorporated to strengthen global exploration and fine-grained local exploitation. Elite individuals guide long-range Lévy flights to avoid premature convergence, while quadratic interpolation refines solutions around high-quality regions to improve optimization precision. The performance of the proposed IESC is comprehensively evaluated on the CEC2017 benchmark suite, covering unimodal, multimodal, hybrid, and composite functions with multiple dimensional settings. Experimental results demonstrate that IESC consistently outperforms several state-of-the-art metaheuristic algorithms in terms of optimization accuracy, convergence speed, and robustness. Furthermore, IESC is applied to robot path planning, where a multi-objective model considering path length, smoothness, and feasibility is constructed under collision-avoidance constraints. Simulation results show that the paths generated by IESC are shorter, smoother, and more stable than those produced by comparative algorithms. These findings indicate that the proposed IESC provides an effective and reliable optimization framework for robot path planning and other complex optimization problems.
Related
- A novel deep learning prediction method for PM2.5 integrating improved MSTL decomposition and FATA optimization algorithm
- Optimization Strategy of PID-Inspired Metaheuristic Algorithm-Tuned Fractional-Order Controllers Applied to Nonlinear AVR System
- Stock Exchange Trading Optimization Algorithm Improved Based on Information Technology Investment Analysis for Numerical Optimization and Practical Problems
- A Power Regulation Strategy for Grid-Forming Wind Power Clusters Based on Deep Reinforcement Learning
- Dynamic Multi-Pair Trading Strategy in Cryptocurrency Markets with Deep Reinforcement Learning
BibTeX
@article{jin2026improved,
title = {{An Improved Escape Algorithm with Reinforcement Learning and Multi-Strategy Synergy for Global Optimization and Real Problems}},
author = {Bolai Jin and Haotian Zheng and Yangjian Yang},
journal = {Mathematics},
year = {2026},
month = oct,
doi = {10.3390/math14203830},
url = {https://www.mdpi.com/2227-7390/14/20/3830},
note = {CC BY 4.0},
}
RIS
TY - JOUR
TI - An Improved Escape Algorithm with Reinforcement Learning and Multi-Strategy Synergy for Global Optimization and Real Problems
AU - Bolai Jin
AU - Haotian Zheng
AU - Yangjian Yang
PY - 2026
DA - 2026/10/21
JO - Mathematics
DO - 10.3390/math14203830
UR - https://www.mdpi.com/2227-7390/14/20/3830
AB - To address the limitations of the standard Escape Algorithm (ESC), including its tendency to become trapped in local optima and its slow convergence speed in complex optimization scenarios, this paper proposes an Improved Escape Algorithm (IESC) for global optimization and robot path planning. First, a reinforcement learning–based decision-making mechanism using Q-learning is introduced to replace the fixed rule–driven population update strategy in ESC. By constructing a closed-loop framework of “individual–action–reward,” the proposed method enables adaptive strategy selection and effectively balances exploration and exploitation during different stages of the optimization process. Second, to enhance information interaction among individuals, a thermal conduction–enhanced search strategy is developed. Inspired by heat transfer principles, this strategy leverages global population information through the combined guidance of the population means and the global best solution, thereby improving local search directionality and convergence stability. Third, an elite-guided Lévy flight and quadratic interpolation fusion strategy is incorporated to strengthen global exploration and fine-grained local exploitation. Elite individuals guide long-range Lévy flights to avoid premature convergence, while quadratic interpolation refines solutions around high-quality regions to improve optimization precision. The performance of the proposed IESC is comprehensively evaluated on the CEC2017 benchmark suite, covering unimodal, multimodal, hybrid, and composite functions with multiple dimensional settings. Experimental results demonstrate that IESC consistently outperforms several state-of-the-art metaheuristic algorithms in terms of optimization accuracy, convergence speed, and robustness. Furthermore, IESC is applied to robot path planning, where a multi-objective model considering path length, smoothness, and feasibility is constructed under collision-avoidance constraints. Simulation results show that the paths generated by IESC are shorter, smoother, and more stable than those produced by comparative algorithms. These findings indicate that the proposed IESC provides an effective and reliable optimization framework for robot path planning and other complex optimization problems.
SN - 2227-7390
ER -