To solve the shortcomings of large dependence on the parameters and need accurate cancellation the dynamic when using exact linearization method to nonlinear system, this paper introduced the Kalman filter to linear decoupling control to observe the load torque change of induction motor. At the same time, use reactive power model based on MRAS to identification rotor time constant of induction motor. So, can effective tracking the dynamic parameters online of induction motor and achieved settlement the problem of the input-output linearization effected by the variation parameters of motor. Finally, the control system designed in this paper studied by computer simulation. Simulation results show that parameters adaptive identification based on Kalman filter and reactive power model in nonlinear decoupling control of induction motor is feasible and can get satisfactory results.