<p>In view of the problem that data-driven health indicators rely on signal decomposition and recombination, which is difficult to reflect the physical characteristics of products, this paper considers the non-cooperative degradation characteristics of multi-performance indicators of motor spindle, and proposes a multi-dimensional data health state evaluation method based on transient dynamics. Firstly, considering the influence of stiffness on the service performance of the motor spindle, a dynamic model of the motor spindle is established based on the theory of Timoshenko beam element. Secondly, the dynamic stiffness was obtained from multi-dimensional data composed of key performance indexes, such as front vibration, post vibration, and radial runout at the tool shank connection, and the physical health index of the motor spindle was constructed by calculating the deviation degree of the dynamic stiffness compared with the factory level. Then, transient dynamic simulation is carried out to prove the accuracy of the dynamic model, and the accuracy and effectiveness of the method proposed in this paper are proved by the degradation data of the motor spindle under 3 sets of different test conditions obtained from the reliability test bench of the motor spindle. Finally, the superiority of the proposed method is further verified by comparing and analyzing the health status evaluation results of the motor spindle under physical and data-driven conditions.</p>