Design and Optimization of Composite Thermal Insulation Layers for Mine Ecological Restoration Under Simulated Solar Radiation: Integrating Response Surface Screening with Gaussian Process Bayesian Optimization

Thermal regulation of surface soil is critical for vegetation establishment in degraded mining environments, yet the systematic design of insulation layers tailored to mine restoration remains underdeveloped. Fifteen candidate thermal insulation materials from six categories were evaluated under simulated solar radiation, and a two-stage optimization framework was established. The first stage employed response surface methodology (RSM) for preliminary screening; the second used Gaussian process regression with Bayesian optimization (GPR-BO) for mixture refinement. RSM identified hollow glass microspheres as the strongest positive contributor and wood chips as the most detrimental component. The GPR-BO framework yielded an optimal formulation achieving T90 = 12.80 °C, heating rate v = 0.0311 °C/min, and heat resistance efficiency η = 65.85%, with R2 exceeding 0.95 for all response variables. The observed thermal regulation arose from the synergy of three mechanisms: surface radiative heat suppression, internal conductive path interruption, and transient thermal buffering. These findings offer a practical design route for high-performance insulation layers in cold-region mine ecological restoration.