Abstract: Large-scale EV integration increases grid uncertainty. V2G offers flexibility, yet its capacity depends on user willingness. This paper proposes a hosting capacity assessment method incorporating user willingness. A dynamic participation model is built via fuzzy inference using SOC, discharge revenue, and travel urgency, and the willingness is converted into power correction coefficients. Monte Carlo simulation generates EV travel behaviors for aggregation into nodal V2G loads, followed by probabilistic power flow. A multi-dimensional evaluation framework (security, flexibility, economy) with fuzzy AHP weights is established. IEEE 33-bus results show that the proposed method enables dynamic V2G capacity variations. Compared with fixed-participation V2G, the comprehensive index rises from 0.420 to 0.720, network loss drops 9.7%, and voltage compliance improves 12.2%, confirming its effectiveness.