The results obtain with this algorithm are quite good and we can see on the figure 4 the comparison for on subject between the initial datas and the output of our model. Some of the artefact present on the original data set are not visible on the simulation result but the essentials characteristics like the amplitude and the time respons are very close. In this way we can say that our model is satisfactory. On a other hand, it will be irrational to wait for a better coincidence between the experiment and the simulation. The real system is non-linear and we don't want to represent exactly the comportment of this muscular system reflex.
Figure 4: Comparison between the Model and the real data