The invention discloses an on-line optimization design method based on an extended Kalman filtering state estimation value, and the method comprises the steps: adding a sampling point factor to the state estimation value obtained through an extended Kalman algorithm to update the state estimation value so that updated state estimation value is used to replace a state observation value; further, acquiring updated observation prediction value and compared with the observation value, acquiring a residual information matrix and using residual information to check the state estimation value that is most suitable for the current time after update, and using the state estimation value as the current time state. estimated value.. Therefore, the influence of model dynamic deviation on system tracking can be reduced to a certain extent. The problem of insufficient precision caused by model deviation of an extended Kalman filter in the system is solved, the extended Kalman filter can be well applied to the field of nonlinear systems, and certain robustness is achieved.