Aiming at the control problem of cascade STATCOM to compensate the load reactive current and stabilize the grid voltage, this paper establishes a mathematical model of cascade STATCOM under the steady state, analysis the relationship of phase angle δ and reactive current on compensating capacitive load, deducts the establishment of dynamic model of the device, obtains the relationship between the two variables δ、M and current, this paper proposes a control method based on neural network inverse system, through the verification of reversibility of the system, construction of neural network and the design of control system, realizing the linearization and decoupling control of active and reactive current of cascade STATCOM. The simulation results show that the control strategy has favourable dynamic performance, so that the device has better performance of anti parameters change and load disturbance. It illustrates the effectiveness and the feasibility of the control strategy.