Accurate measurement of ship deformation angle is of great significance for building a unified space attitude reference and improving the combat performance of the entire ship. However, many problems existing in the actual sailing process restrict its measurement accuracy severely, such as the time delay between measurement equipment, the unknown characteristics of measurement noise and the unknown model parameters of the dynamic deformation. This paper proposes an adaptive variable parameter multiple model (AVPMM) algorithm that can simultaneously solve these problems. Firstly, a time delay compensation algorithm is derived to suppress the influence of time delay on measurement accuracy; then an adaptive Cubature Kalman filter (ACKF) is designed based on the idea of adaptive filtering to suppress the influence of unknown characteristics of measurement noise; and finally an AVPMM algorithm is designed by combining the above ideas to reduce the impact of unknown model parameters of dynamic deformation. The results of simulation and turntable experiment show that this algorithm has good measurement accuracy and the error is less than 0.15 ′ .