Comparison And Evaluation Of Two Approaches Of Uncertainty Modeling In Dynamcal Systems.

Abstract

When mechanical systems are modeled, uncertainties should be taken into account for improving
the predictability of the model. In this work a two d.o.f. (degrees of freedom) dynamical system
is used to compare two strategies to model uncertainties in structural dynamics. Uncertainties are considered
present only on the spring stiffnesses. In the first approach, uncertainties are inserted into each
spring stiffness. A probabilistic model is constructed for each random variable associated to each spring
stiffness. In the second approach, uncertainties are considered in a global way, that is, a probability
model is constructed for the stiffness matrix. In both approaches, the probability density functions are
deduced from the Maximum Entropy Principle, using only the available information. The simple example
used is helpful to assure a better understanding of the two approaches. The event space generated by
each strategy will be shown and it will be discussed how good they are to predict data uncertainties and
model uncertainties.