Multi-level uncertainty optimisation on phase change materials integrated renewable systems with hybrid ventilations and active cooling

Multi-level uncertainty-based analysis and optimisation on renewable systems are important, as operating parameters are with multi-dimensional uncertainties during real application processes. However, current literature shows quite limited progress towards uncertainty-based performance, and methodologies for uncertainty-based optimisation are quite rare. In this study, a coordinated framework for multi-level uncertainty-based analysis and optimisation was proposed and presented, through a novel renewable system with active cooling, latent storage for radiative cooling and hybrid ventilations. An uncertainties-based optimisation methodology was proposed, with the integration of surrogate model in uncertainty-based optimisation function. The learnt optimisation function was thereafter flexibly integrated in a heuristic algorithm for the global optimisation. Comparative analysis between deterministic case, uncertainty case and uncertainty-based optimal case has been investigated to highlight the importance of uncertainty-based optimisations. Results showed that, compared to the deterministic scenario, the stochastic scenario can improve the peak power from 11.7 to 15.7 kW, and the electricity generation from 1776.9 to 2261.5 kWh, by 27.3%. Furthermore, by adopting the multi-level uncertainty-based optimisation methodology, the peak power and renewable generation can be further increased to 25 kW and 2340 kWh. A novel multi-level uncertainty-based optimisation methodology was proposed to promote green buildings with optimal design parameters. •A novel uncertainty-based optimisation method was proposed for multivariable design.•Supervised learning surrogate model was developed to improve prediction efficiency.•Uncertainty-based optimisation function was developed through stochastic scenarios.•A heuristic teaching-learning-based algorithm was employed for the optimal design.•Multi-level uncertainty-based optimisation was studied to improve system robustness.