In response to the shortcomings of traditional solutions in thermal energy management for intelligent sports facilities, this paper proposes an adaptive thermal energy dynamic management algorithm that incorporates spatiotemporal characteristics. The algorithm constructs a three-layer architecture of "spatial partitioning - time segmentation – intelligent adjustment", analyzes thermal energy demand through a spatiotemporal coupling model, and optimizes energy consumption and comfort through adaptive adjustment. The experiment utilises the main gymnasium of a city sports centre as the subject, and compares the new algorithm with traditional PID and fuzzy control algorithms. The spatiotemporal model integrates spatial thermal coupling (σ = 5 m, calibrated via venue lay-out) and time-dependent demand (β = 0.2, γ = 0.1 from six month data), unlike PID static control. Zoning adjusts air conditioning output by 30% in high density areas (e.g., auditorium front rows) using personnel density correction. This explains 0.52℃ SD vs. PID 1.23℃, as dynamic adjustments match real-time heat loads.