Method for establishing adaptive extended Kalman filter noise model

The invention discloses a method for establishing an adaptive extended Kalman filter noise model. The method comprises steps as follows: collecting sensor data, recording difference between an observation value and a predicted value in each extended Kalman filter calculation and a change value of a course angle of an attitude after filter calculation, and calculating a correlation coefficient between the difference of the observation value and the predicted value and the change value of the course angle of the attitude after filter calculation; establishing a noise observation module accordingto the correlation coefficient between the difference of the observation value and the predicted value and the change value of the course angle of the attitude after filter calculation; utilizing parameters required by the noise observation model as input of a neural network, utilizing a target value as a corresponding output of the neural network, and performing neural network model training toobtain the optimal paramet