Bridge construction risks in complex environments – a hybrid analytic hierarchy process and optimised neural network model

Abstract

Translate Article

Take Notes

ABSTRACT A model was developed to predict construction risks for bridges by coupling the analytic hierarchy process (AHP) with an optimised Extreme Learning Machine (ELM) neural network. Firstly, by using the AHP method, 22 factors were identified to comprehensively represent risks during bridge construction. These factors were formulated in a two-level hierarchical structure. Secondly, a risk assessment system was formulated. Thirdly, an ELM neural network model was built to automate the risk prediction process and minimise the subjectivity associated with the traditional expert assessment system. The ELM model was optimised by the Sparrow Search Algorithm (SSA). Finally, the AHP-SSA-ELM model was tested on 50 bridge construction cases, and showed a 96% agreement with the expert assessment. This means that the proposed model can be used confidently to assess risks during bridge construction in complex environments. The model is accurate, practical, and efficient. It will inform risk management to avoid social and economic losses during bridge construction.