Multi-objective block form optimization for photovoltaic performance and heat island mitigation in plateau regions

Maximizing solar renewable energy utilization is crucial for advancing low-carbon urban construction. However, most existing studies focus solely on rooftop photovoltaic (PV) systems while overlooking the substantial power generation potential of building facade PV, particularly in solar-rich plateau regions. To address this gap, this study proposes a optimization framework integrated with an interpretable machine learning model, aiming to maximize the roof-facade solar energy generation indicator (SEGI), minimize PV economic payback period (EPP), and mitigate accumulated urban heat island intensity (AUHII). Applied to residential blocks in Lhasa (a typical plateau city), results show XGBoost outperforms other algorithms as a surrogate for traditional physical simulations, improving computational efficiency by over 50 times. SHapley Additive exPlanations (SHAP) method identified PV type, building orientation, and height as key drivers of SEGI and EPP, while building height, density, and sky view factor dominate AUHII mitigation. Sensitivity analysis further reveals significant differences in the impacts of morphological parameters on performance indicators. Pareto set features a spatial layout of low southeast and high northwest buildings, with an optimal orientation of 30° west of south. The optimal design scheme identified by entropy weighting method achieves a 66.17% increase in SEGI, 7.02% reduction in EPP, and 58.73 ℃·hr decrease in AUHII compared to the baseline scenario. These findings underscore the necessity of integrating urban morphology effects into plateau block design and demonstrate the proposed framework’s potential for balancing block performance. The revealed driving factors and critical thresholds provide morphology-tailored planning insights for PV-integrated sustainable block design in plateau regions.