Sequence-based prediction in conceptual design of bridges

This paper explores the application of a machine learning technique in knowledge support systems in civil engineering design. It presents a sequence-based prediction method for engineering design and demonstrates its utility in the conceptual design of bridges. The basic idea of sequence-based prediction is that the most recent numbers of similar design cases are used in predicting the characteristics of the next design and more recent cases are given stronger influence on decision making in the new design situation than older ones. This paper develops a model of sequence-based prediction and carries out a number of experiments using it. It is then applied to a set of standard data and the results of using a sequence-based prediction method are compared with other methods. The empirical results show the potential applications of the method in engineering design.