Uncertainty Propagation in Structures, Systems and Processes

现代工程越来越多地依赖于计算机模拟和建模，用于产品和结构生命周期中的各种过程。这种建模从设计阶段一直延伸到退役。由于制造用于测试的原型的费用，建模方法提供了明显的经济优势。然而，这种建模的一个主要问题是如何处理过程中的不确定性。举一个例子，在最近对心脏瓣膜的操作进行建模的例子中，发现结果对心脏组织的材料规格非常敏感。事实上，对于材料参数的某些值，阀门没有打开。这在我们的理解中是一个关键问题，因为人体组织的特性会发生很大变化。该项目将提高我们对不确定性下建模的一些问题的理解。在第一种情况下，该项目将调查不确定性量化问题。有许多理论处理不确定性和该项目将确定哪些理论是适当的在给定的条件下，以及如何分配一个明确的措施或程度的不确定性在一个给定的理论。第二个问题要调查有关融合，即如果一个是提出了一个以上的措施，不确定性从不同的理论，他们如何结合起来。最后，动力学最重要的问题是不确定性的传播。如果结构或系统的模型在其输入中具有不确定性，则必须确定输出或预测的可靠性。

Modern engineering is relying increasingly on computer simulations and modelling for a wide range of processes over the life cycle of products and structures. This modelling extends from the design stage right through to decommisioning. Because of the expense of making prototypes for tesing, the modelling approach offers clear economic advantages. However, a major concern with such modelling is how uncertainties in the process are handled. To give an example, in a recent example of modelling the operation of heart valves, it was found that the results were very sensitive to the material specification of the heart tissue. In fact, for some values of the material parameters, the valves did not open. This is a critical issue in our understanding, as the properties of human tissue are subject to great variation. This project will improve our understanding of a number of issues concerning modelling under uncertainty. In the first case, the project will investigate issues of uncertainty quantification. There are many theories dealing with uncertainty and the project will determine which theories are appropriate under given conditions and how to assign a definite measure or extent of the uncertainty within a given theory. The second problem to be investigated concerns fusion i.e. if one is presented with more than one measure of uncertainty from different theories, how can they be combined. Finally, the most important problem for dynamics concerns propagation of uncertainty. If a model of a structure or system has uncertainty in its inputs, one must determine the reliability of the outputs or predictions.