According to the available information guiding the search for solution, well-defined problems are those that lead to a unique and exhaustive solution through a finite number of defined steps (the algorithmic method).

On the other hand, ill-defined problems are more complex and do not supply all the information required for the solution. In these cases, heuristic methods apply techniques that make use of "rules of thumb" as opposed to invariant procedures. These methods do not always guarantee to solve the problem, but often solve it well enough for most uses, and often do so more quickly than a more complete solution would.

The traditional approach to structural problem solving is based on the application of general rules or mathematical equations, but, in the case of a complex structure, this approach is restrictive. In fact, a structure is a real physical object inserted in its environment where a variety of factors (quantitative and qualitative) should be taken into consideration. Most of the problems are ill-defined and do not have a single optimal solution [1].

If one evolves from the idea of a structure, as a simple device for channelling loads, to that of a structural system, as "a set of interrelated components working together toward a common purpose" [2], it becomes possible to apply the approach of systems engineering to structural problem solving.

In order to achieve an intelligent synthesis of quantitative and qualitative aspects, this approach permits the integration of the traditional algorithmic methods with heuristic methods. Simultaneously, the application of techniques coming from the artificial intelligence (AI) field permits a global synthesis of the results [3].

The AI techniques that are successfully applied in the structural field are fuzzy logic, genetic algorithms and artificial neural networks.

Artificial neural networks are parallel systems for information processing, inspired in the way in which biological neuron networks process information in the brain. Their main application are focused on solving problems which are too complex for conventional techniques, which do not have a specific algorithm for their solution, or whose algorithm is too complex to be found.

In this paper they have been applied to solve an inverse problem of structural identification related to the modelling of a bridge deck. The searched values are obtained with an error smaller than 10technique in the structural field.

1

S. Arangio, (In Italian) "Risoluzione di Problemi Strutturali Diretti ed Inversi mediante Tecniche Euristiche", Graduate Thesis in Civil Engineering, University of Rome "La Sapienza", Advisor Prof. F. Bontempi, 2004.

2

P.H. Sydenham, "Systems approach to Engineering Design", Artech House, Boston, 2004.

3

L. Sgambi, "Artificial Intelligence for the analysis and design of complex bridge structures", Proceedings on Ninth International Conference on Structural Safety and Reliability, (ICOSSAR05), Rome, Italy, 2005.

purchase the full-text of this paper (price £20)

go to the previous paper
go to the next paper
return to the table of contents
return to the book description
purchase this book (price £135 +P&P)