To improve the effect of ballet teaching, this study used the scientific computing model integrated with motion capture to analyze and evaluate the teaching of ballet dance postures, so as to improve the intelligence of modern ballet teaching. Moreover, this study employed the coordinate transformation and D-H method to model and analyze the forward and inverse kinematics of the ballet posture teaching model, and used the Monte Carlo method to verify the correctness of the exoskeleton motion space analysis. In addition, this study established a dynamic model, and used the Lagrangian equation method for a dynamic solution to obtain the relationship between the position, velocity and torque of each component. The data analysis indicated that the ballet posture teaching system, which is based on the scientific computing model and integrated with motion capture, can play an important role in ballet teaching.