A Multi-Strategy Crested Porcupine Optimizer for Work–Family Resource Allocation: Childcare Facility Siting and Family-Friendly Shift Scheduling

The time cost of care constrains female labour supply; two policy levers act on it: where childcare capacity is sited and how shifts are scheduled, both large combinatorial problems. The crested porcupine optimizer (CPO) initialises at random, guides exploitation by a single random or best individual and resizes its population on a fixed cycle. The proposed multi-strategy CPO (MSCPO) keeps CPO’s four defence operators and replaces those original components with Sobol quasi-opposition initialisation, elite-guided exploitation and monotone fitness-based population reduction. On the 29 CEC2017 functions at 10, 30, 50 and 100 dimensions (30 runs of 30,000 evaluations), MSCPO had the best Friedman mean rank among ten algorithms at every dimension (1.29 against 2.53 for CPO; χ2(9) = 878.5, p < 0.001), and pooled Holm tests separated it from all nine competitors (p ≤ 0.002). A factorial ablation attributed about 70% of the gain to elite-guided exploitation. The advantage over CPO vanished at the full budget of 104D evaluations. In childcare siting, MSCPO ranked first, leaving the dispersion of access 28–35% below the most equal siting rule with at least 99% of the best rule’s employment; in shift scheduling, it ranked fifth, with instrumentation suggesting a limitation of large steps under a rank-key decoder.