Abstract
Heliostats, with their large-area reflective surfaces, are inherently sensitive to wind loads. This study investigates wind load sensitivity of heliostats in open terrains through wind tunnel simulations encompassing 130 operational conditions, systematically analyzing probability characteristics and extreme value prediction of surface fluctuating wind pressures. The analytical framework establishes diagnostic criteria for pressure distribution patterns, and subsequently validated through the analysis of probability density histograms and pressure time-history traces. To address prevalent non-Gaussian characteristics, Block Maxima (BM), Peak Over Threshold (POT), and Bayesian methodologies were implemented for extreme pressure estimation. Rigorous threshold optimization and parametric refinement demonstrated these methods' superior predictive capability over conventional peak factor approaches. Results indicate Gaussian zones predominantly occupied central regions below 60°elevation, while peripheral areas exhibited non-Gaussian behavior due to vortex shedding. For extreme value prediction under non-Gaussian conditions, comparative analysis demonstrated that the Bayesian method achieved a narrower 95% confidence interval width of 18.86 compared to the baseline method, indicating substantially improved estimation precision with reduced uncertainty. Furthermore, chi-square goodness-of-fit tests confirmed the conformity between the empirical distribution of observed wind pressure data and the Bayesian-derived parametric model, with most interval-specific components below 2.0 and central probability regions consistently below 1.0. The proposed risk-operational identification mechanism based on azimuth-elevation combinations and Bayesian extreme value paradigm provides a theoretical basis for wind-resistant heliostat design.