Development of neural networks in civil engineering applications

Abstract

Artificial neural networks (NNs) have been widely studied for their ability to correctly learn the true distribution of the data from a sample. This ability is called generalization. NNs are a new generation of information progressing systems that are deliberately constructed to make use of some of the organization principles that characterize the human brain. They are parallel computational models comprised of densely interconnected adaptive processing units. The collective behavior of an NN, like a human brain, demonstrates the ability to learn, recall, and generalize from training patterns or data. The motivation behind combining a number of NNs is to improve upon a network's generalization performance. Recently, NNs have been intensively studied with impressive successes across a wide variety of applications. These include pattern classification, speech synthesis and recognition, function approximation, combinatorial optimization, and nonlinear system modeling and control (Fausett, 1994). Adeli and Yeh (1989) were the first to apply the NNs in civil and structure engineering. Since then, a large number of applications of NNs in civil engineering areas have been done. These applications include pattern recognition and machine learning in structural analysis and design, structure system identification, structural control, structural material characterization and modeling, construction scheduling and management, construction cost estimation, etc (Adeli, 2001).

Development of a Displacement Prediction System for Deep Excavation Using AI Technology

“…The most common approach to the prediction of excavation-related wall displacements involves supervised learning models. Backpropagation neural networks (BPNN) have proven highly effective in these prediction scenarios, as evidenced by a success rate from 77% to 83% [32,33]. Huang [26] studied ten cases of inverse construction using a relatively complete database of the Taipei Basin.…”