Figure 3.4 Hybrid modular structures:
(a) modular complementary structure
(b) modular competitive structure.
While complementary hybrid structures have been frequently reported in the literature
competitive structures are not so common. The main reason for this may be due to the
lack of theoretical results in this area. A central question is: how should the weighting
system be defined. Some examples of weighting methods have been reported in the
literature, from which the most important are:
- Weighting methods based on clustering techniques (e.g. Simutis et al., 1995; Leonard
et al., 1992) - Weighting methods based on expert systems (e.g. Schubert et al., 1994b)
- Weighting methods based on gating networks (Peres et al., 2000)
Common hybrid model structuresSeveral hybrid model structures have been reported in the literature. An important class
of these structures are the neural-fuzzy systems. The neural-fuzzy systems combine
artificial neural networks and fuzzy logic in one single model structure (e.g. Gupta and
Rao, 1994; Werbos 1992; Shi and Shimizu, 1992; Lin and George Lee, 1991).
Another important approach for designing hybrid model structures is based on the
division of the process in study in several modules according to the kind of knowledge
available in different parts of the process. The result is expressed by a diagram of
New methodologies for multiphase bioreactors 3 73