The hybrid game system suffers from the constantly time-varying environment and the heterogeneous behaviors of decision-making subjects (players) when the hybrid situation emerges, bringing significant challenges to situation representation and strategic reasoning. With the help of the powerful modeling ability of hybrid systems for situation representation and reasoning in network topology, we propose the six-tuple hybrid game model based on modified hybrid stochastic timed Petri nets (M-HSTPN) to reveal the internal cross-level operating mechanisms. Then, we carry out the nonlinear mapping relationship between the hybrid game elements and the symbols of M-HSTPN. Taking the intelligent drones combat problem as an example, we compare differential game, event game, and proposed hybrid game scheme in pursuit–evasion trajectories, action timing, bombing sequence, and formation. The proposed M-HSTPN-based hybrid game scheme can effectively describe the constantly changing pursuit–evasion trajectories and select the action timing to reduce the energy consumption of interception and the target-missing quantity. Furthermore, the proposed hybrid game scheme can reason the optimal bombing sequence and narrow the scope of optimal formation through enriching payoff evaluation indexes with the current hybrid situation compared with the event game scheme and differential game scheme.