本文首次提出了一种基于三元数代数的三维过程新模型。与纯四元数模型相比,三元数模型更加紧凑,计算量更小,同时在自适应线性滤波方面具有类似或者可比的性能。此外,三元数模型可以有效表征卡尔曼滤波中状态转移的一般性关系,而纯四元数模型则无法对此进行表征。基于实测风力数据和合成数据集的仿真实验验证了这一新建模方法的潜能。 This paper presents for the first time a new three-dimensional process model based on ternary algebra. Ternary models are more compact and less computationally expensive than pure quaternion models, and have similar or comparable performance in adaptive linear filtering. In addition, the ternary model can effectively characterize the general state transition of the Kalman filter, whereas the pure quaternion model can not characterize this. Simulation experiments based on measured wind data and synthetic data sets verify the potential of this new modeling approach.