Accurately forecasting when and where electric vehicles will recharge on long-distance corridors is essential for sizing highway fast-charging infra-structure. This study presents a high-resolution demand-prediction frame-work that fuses anonymised electronic toll-collection travel-chain data with a Monte-Carlo state-of-charge depletion model and adaptive driver-behaviour rules. Two million six hundred and sixty thousand single-day trips on a major Chinese expressway are reconstructed and iterated 1 000 times to capture variability in initial SOC, vehicle mix and charging thresholds. Hourly load curves are produced for several representative service areas and validated against the province’s benchmark. Results show pronounced 16:00–17:00 peaks, with median peak powers of 300–475 kW and daily energies of 2.7–6.4 MWh; Monte-Carlo P5–P95 bands quantify a ± 10% uncertainty window around the crest. The mod-el outperforms a flat-traffic baseline by ≈25% at peak, and recommends transformer upgrades or much more storage for the busiest stations. By de-livering probabilistic, site-specific demand profiles, the proposed approach provides a decision-ready input for charging planning on China’s fast-growing motorway network.