6. Conclusion

The modeling of random loading is based on statistical approaches and their interactions with the dynamic behavior of the loads studied. Statistical approaches include global and local approaches. Global statistical modeling leads to a probability law that can be used to assess the risk of reaching (or exceeding) a given amplitude. Local statistical modeling focuses on the extreme events of the stress under study. Assimilating a random stress to a random trajectory leads to probabilistic modeling, which integrates information from frequency analysis and, more specifically, data concerning the speed and acceleration of the stress under study. Probabilistic models can now be used to perform fatigue design calculations, as well as to define laboratory test conditions based on real-life conditions