Aiming at the need for simulation of attitude control system (ACS) of micro-nano satellite with pulsed plasma thruster as actuator, a semi-physical simulation of attitude control system of micro-nano satellite was carried out on a dual-rotor experimental platform controlled by pulsed signal. The control strategy of Elman neural network PID is used to adjust the parameters of PID on-line and adapt to the dynamic system. Through the comparative experiment of semi-physical simulation platform, the advantages of Elman neural network PID control system, such as strong adaptive ability and small overshoot, are verified. At the same time, the validity of dual-rotor experimental platform for semi-physical simulation of pulsed plasma thruster is verified.