Automated surrogate-assisted particle swarm optimizer with an adaptive parental guidance strategy for expensive engineering optimization problems

Abstract Surrogate-assisted evolutionary algorithms have emerged as powerful tools for addressing complex engineering optimization problems, but their implementation often involves extensive manual intervention. To reduce the subjectivity of human design and lower computational costs, this paper proposes a generalized framework of automated surrogate-assisted optimization inspired by ensemble learning. Within this framework, we develop the AutoSAPSO algorithm, which integrates model and algorithm adaptation at the control layer. The algorithm employs an adaptive parental guidance strategy to dynamically adjust optimizer selection based on population diversity and convergence trends, an automated parameter control strategy to fine-tune key algorithm parameters during optimization, and an ensemble model dynamic selection strategy to enhance surrogate model accuracy by leveraging problem-specific characteristics. Additionally, an interactive cooperative optimization strategy is introduced to improve generalization ability and prediction accuracy by enabling dynamic interactions between the metamodel and objective function. The performance of AutoSAPSO is evaluated using 20 benchmark functions and verified with three real-world engineering optimization problems. Results demonstrate that AutoSAPSO significantly outperforms state-of-the-art approaches regarding convergence accuracy and robustness, confirming its potential as a general-purpose optimization tool for expensive engineering optimization problems.