Hybrid Model/Data-Based Robust Design Under Model Uncertainty

In this chapter, two hybrid model/data-based robust design methods, which integrate the advantages of the model-based robust design and the data-based modeling, are presented to achieve the robustness of the partially unknown nonlinear system. First, the system is formulated into a linear structure. Its sensitivity matrix, incorporating influences of model uncertainty and nonlinearity, is different from the common constant sensitivity matrix. With the help of this estimated sensitivity bound and the nominal model of the system, two model-based robust design methods, one in deterministic nature and another in probabilistic nature, are then developed to minimize the influence of parameter variation to performance. Finally, the presented methods are compared with several common robust design methods on the selected examples.