Prediction models for calculating bolted connections using data mining techniques and the finite element method

• Published in: Engineering structures Vol. 32; no. 10; pp. 3018 - 3027
• Main Authors: Fernández, J., Pernía, A., Martínez-de-Pisón, F.J., Lostado, R.
• Format: Journal Article
• Language: English
• Published: Kidlington Elsevier Ltd 01.10.2010; Elsevier
• Subjects: Applied sciences; Artificial intelligence; Bolted connection; Bolting; Bolts; Buildings; Buildings. Public works; Computation methods. Tables. Charts; Computer simulation; Data mining; Exact sciences and technology; External envelopes; Finite element method; Joints; Mathematical analysis; Mathematical models; Prediction models; Structural analysis. Stresses; Finite element method; Bolted connection; Prediction models; Data mining; Artificial intelligence; Bolted joint; Connection; Forecast model; Mesh method; Modeling; Result
• ISSN: 0141-0296; 1873-7323
• DOI: 10.1016/j.engstruct.2010.05.021

This paper describes a method based on a combination of Finite Element Method (FEM) and Data Mining (DM) techniques to set up prediction models that can be used to calculate bolted connections. Based on the results of a finite element (FE) model validated by tests, a number of FE simulations is developed, varying the most significant parameters (thickness, bolt diameter, friction, etc.). The results of these simulations are used to generate a database which can then be used to create prediction models. The process centres on selecting the best technique from a set of DM and artificial intelligence (AI) algorithms to find the models which provide the most generally applicable solutions to the problem. This method, combining FE models with prediction techniques, is highly useful for the specific case of bolted connections, because it enables results to be obtained almost in real time with only slight prediction errors. This makes it an excellent tool for optimising the design of such connections.