In view of the problem of increasing the accuracy of the computation of the deterministic sampling filter in the state estimation, a new algorithm is proposed and applied to the prediction of fatigue crack growth. First of all, the basic principle of determining the sampling filter is described. Secondly, from the point of view of multidimensional numerical integration, the mathematical expectation of deterministic sampling filter is analyzed, and the integral node value, the number of nodes and the weights are calculated according to the fully symmetrical integral formula. Finally, the improved model is applied to the fatigue crack propagation of the members, and the accuracy of the crack propagation prediction is improved by comparing with unscented Kalman filtering algorithm and cubature Kalman filtering algorithm. The feasibility and effectiveness of the proposed method are verified by simulation analysis.