An iterative cubature unscented Kalman filtering method, comprising the following steps: selecting sigma points of an iterative cubature unscented Kalman filtering algorithm; re-determining a weighting coefficient of the sigma points; providing the procedure of the cubature unscented Kalman filtering algorithm; and iteratively calculating the cubature unscented Kalman filtering algorithm. Said method is able to be effectively applied in a highly-free strong-nonlinearity system containing random noise, and solves the calculation amount problem, the nonlinear filtering divergence problem and the negative weight problem by means of collaborative processing, and is able to effectively improve the estimation accuracy and real-time performance of a state amount, without diverging a filtering result. The present solution can better fit the statistical characteristics of a nonlinear system function with respect to cubature Kalman filtering, and can avoid the non-positive definiteness of a sigma point weight with respect to unscented Kalman filtering.
展开▼