Development of a data-driven fuzzy system for evaluating ultimate tendon stress of externally prestressed beams

Lau, See Hung (2015) Development of a data-driven fuzzy system for evaluating ultimate tendon stress of externally prestressed beams. PhD thesis, Universiti Malaysia Sarawak, (UNIMAS).

PDF (Please get the password by email to repository@unimas.my , or call ext: 082-583914/3942/3933) Lau See.pdf Restricted to Registered users only Download (3MB)

Abstract

The research work presented herein aimed to investigate the applicability of data-driven harmony search (HS) based single input rule modules (SIRMs)-connected fuzzy inference system (FIS) model in predicting stress increase, f ps and ultimate stress, ps f for externally prestressed beams. The study consisted of three major parts: (i) to evaluate the parameters that affect the increase in stress for externally prestressed tendons; (ii) to formulate and validate a data-driven HS-based SIRMs-connected FIS model as an optimisation problem; and (iii) to investigate the applicability of the proposed model with application to the externally prestressed beams with consideration on the monotonicity preserving property. In this study, the proposed method exploits the uncertain nature of the system where the relationship between the inputs and outputs are not well understood to yield an approximate solution to it. A parametric study was carried out to study the effects of changing various nondimensionalised parameters on the behaviour of beams prestressed with external tendons. It was identified that non-dimensionalised parameters deviator spacing to tendon depth ratio, d ps0 S d ; shear span to effective beam span ratio, L L s ; tendon depth to beam height ratio, d h ps0 ; and combined reinforcement index, q have significant influence to ps f in externally prestressed tendons. The applicability of the proposed model was used in predicting ps f in externally prestressed tendons. Although the proposed model yields better coefficient of correlation for ps f in externally prestressed tendons, it was visualized that the monotonicity property of the model was not preserved. Hence, the monotonicity preserving SIRMsconnected FIS model is proposed to enhance the performance of the model in predicting the ps f in externally prestressed tendons. Though the proposed model with MI gives less coefficient of correlation compared with the proposed model without MI, it still performs better than the other prediction equations proposed in the literature. Besides, it was also iii identified that the model with the presentation of monotonicity preserving property is a better model as it can filter out the noisy and/or missing data presence during the evaluations which affects the overall output of the model. The proposed model leads to a novel data-driven HS based SIRMs-connected FIS model with an application in civil engineering.

Actions (For repository members only: login required)

View Item