Abstract: According to the voltage and current time domain model in the synchronous rotating dq reference frame,a constrained adaptive PI control strategy is proposed to address the inverter voltage tracking problem with input and output constraints under load disturbance. Firstly,a state translation function is used to transform the system with output constraints into an equivalent unconstrained one. At the same time,an auxiliary dynamic system is introduced to tackle the effects of the input constraints. Secondly,the controller is designed by using back-stepping method. Wherein,neural network is utilized to approximate the nonlinear term including the time derivative of the virtual control input,and then“explosion of complexity”in the traditional backstepping method is avoided. Finally,the designed virtual control law and control law are both composed of PI term and feedforward term,and the gain of the PI controller is continuously adjusted along with the adaptive law. The Lyapunov stability theory is used to prove that all the error signals of the closed loop system are uniformly ultimately bounded under the proposed control strategy. Besides,simulation results in Matlab/Simulink platform demonstrate the feasibility and effectiveness of this method.