A self-adaptive differential evolution algorithm with robust decision-making for school layout problem
Abstract
As urban populations expand and student distributions become increasingly uncertain, the optimization of school layout problems (SLP) faces growing challenges in solution robustness. However, robust decision-making has received limited attention in the SLP literature. This paper proposes a self-adaptive differential evolution algorithm with robust decision-making (SaDE-RD) to address SLP. A robustness evaluation model is first established using two budgeted uncertainty sets that capture fluctuations in student numbers and commuting distances, and is reformulated into a tractable linear form via robust optimization theory. To enhance search efficiency, SaDE-RD introduces a self-adaptive parameter control mechanism that dynamically adjusts the scaling factor and crossover rate based on the rate of best-solution updates. Moreover, a new stagnation process strategy is employed to keep high diversity by retaining suboptimal solutions. Extensive experiments on real-world datasets demonstrate that SaDE-RD outperforms baseline algorithms in overall quality.