Aiming at how panic and chaos affect pedestrian behavior and evacuation efficiency, we propose a panic crowd evacuation model based on entropy. Boltzmann entropy and Shannon entropy are used to simulate people's escape behavior in panic disorder. The simulation results suggest that the proposed model can effectively reflect the negative impact of chaos on evacuation, which leads to longer evacuation time. Compared with Shannon Entropy, Boltzmann Entropy is more sensitive to the number of persons and helpful to identify the evacuation bottleneck area. This work provides a new insight for understanding how crowds behave in an emergency evacuation.