Long-term extreme value estimation method to determine maximum buffet design loads

Determination of the dynamic limit load for a vibrating structure plays an important role in the dynamic strength analysis of aircraft. Based on the order statistics and Gumbel extreme value distribution theory, a method was proposed to estimate maximum load encountered in a long operating time of aircraft by means of the statistics analysis of short duration experimental data. Using the buffet response data of an aircraft empennage structure as an example, the effects of different Gumbel distribution models, parameter estimation methods and number of samples on the estimation accuracy of a long-term extreme value were discussed. The comparison between predicted and test results shows that Gumbel III distribution model is more suitable for prediction of buffet extreme value than Gumbel I distribution model, and the relative estimation error is below ±5%. The maximum likelihood estimation method also shows a little better than the least square method for the parameter estimation of probability distribution models. The further research shows that the least sample number of test data must reach to the required number for a valid long-term extreme value prediction. Moreover, the investigation shows that the long-term extreme value theory is more reasonable than the traditional 3σ criterion to determine the dynamic limit loads for random vibration problems, and the limit load determined by 3σ criterion is partially dangerous.