With the mass adoption of electric vehicles (EVs), the profitable EV charging station serves as a flexible interface connecting the distribution network to these price-incentivized loads by submitting strategic energy bids. This paper proposes a bidding model for EV charging station in day-ahead electricity market using a stochastic bilevel approach, while the time uncertainty of driving pattern mostly ignored in previous studies is taken into consideration. In the upper level, the EV charging station will submit an energy bid to independent system operator (ISO) with the goal of maximizing its own benefit considering time uncertainty of EV driving patterns. While in the lower level, ISO will then conduct centralized market clearing to minimize the electricity procurement costs. By introducing Karush-Kuhn-Tucker conditions and linearization techniques, the bilevel model can be transformed into a convex single-level problem that is solved in a computationally tractable way. Numerical simulations indicate that the proposed bidding strategy not only allows EV charging station make arbitrage when facing time-varying settlement prices, but also fully exploits these potential flexible resources in smoothing daily system load profile and coordinately realizing peak-load shifting.