Modelling of dam behaviour based on neuro-fuzzy identification

• Published in: Engineering structures Vol. 35; pp. 107 - 113
• Main Authors: Ranković, Vesna, Grujović, Nenad, Divac, Dejan, Milivojević, Nikola, Novaković, Aleksandar
• Format: Journal Article
• Language: English
• Published: Kidlington Elsevier Ltd 01.02.2012; Elsevier
• Subjects: ANFIS; Applied sciences; Arch dam; Artificial neural networks; Buildings. Public works; Dam behaviour; Dams and subsidiary installations; Displacement; Exact sciences and technology; Fuzzy; Fuzzy logic; Hydraulic constructions; Identification; Mathematical models; Modelling; Nonlinearity; Radial displacement; Stresses. Safety; Structural analysis. Stresses; Serbia; Europe; Arch dam; Identification; Radial displacement; ANFIS; Dam behaviour; Structural design; Conformational dynamics; Measurement result; Forecast model; Modeling; Displacement; Case study; Simulation; Arch dams; Behavior
• ISSN: 0141-0296; 1873-7323
• DOI: 10.1016/j.engstruct.2011.11.011

► We construct the neuro-fuzzy models to predict the radial displacements of arch dam. ► The ANFIS models were developed and tested using experimental data. ► The hybrid learning algorithm was used for updating of ANFIS parameters. ► Measured values were compared with predicted values. ► The proposed model shows excellent agreement with measured data. The radial displacement of one or several points of the dam is an important time-varying behaviour indicator and it is a nonlinear function of hydrostatic pressure, temperature and other unexpected unknown causes. Nonlinear system identification is becoming an important tool which can be used to time-varying behaviour modelling of engineering structures. Identification and prediction of complex nonlinear structural behaviour are complex tasks for which non-parametric models are often used. The objective of this study is to develop a neuro-fuzzy identification model to predict the radial displacement of the arch dam. The ANFIS (adaptive network-based fuzzy inference system) models were developed and tested using experimental data collected during 11years. Comparing the values predicted by the ANFIS with the experimental data indicates that soft computing models provide accurate results. These models can be applied for prediction of displacement in further studies.