According to the given keyframes and associated timing, how to simulate the human motion that conforms to Newton’s law by optimization is an important problem. The convergence property of current optimization models where the constraint of Newton’s law is regarded as the direct constraint condition is not satisfactory in practice. There are 2 reasons for that. The first is that the nonlinearity of Newton’s law is strong. The second is that the current optimization methods can only find local minimizers. By converting the constraint of Newton’s law into the objective function and adding the optimization of given timing, the convergence property of which is better than that of the current optimization models and is independent of the type of simulated motion, the error of mass parameters of human body and the error of the given timing. The simulated human motion can conform to Newton’s law as much as possible. The efficacy of the new model is validated by simulating seven types of somersaults on the trampoline. Furthermore, this model has been applied to the analysis of forces acting on human body and the design of new motions in sports.