Random forest-based algorithms for accurate evaluation of ultimate bending capacity of steel tubes

Self Cite

Bridges all over the world are vulnerable to severe deterioration agents meanwhile their maintenance budgets are being tightened. This state of affairs necessitates the establishment of an autonomous deterioration model to predict the performance condition of bridges. This research paper explores the implementation of a set of intelligent data driven models to analyze the future condition ratings of bridge decks. These models comprise support vector machines, Gaussian process regression, regression tree, back propagation artificial neural network, Elman recurrent neural network, cascade forward neural network, long short‐term memory network and deep convolutional neural network. The performance comparison is carried out using five evaluation metrics of root mean squared percentage error, mean absolute percentage error, root mean squared error mean absolute error and relative absolute error. The models herein are developed and validated using the structural deterioration information retrieved from the National Bridge Inventory (NBI). It can be argued that the developed deterioration model could be implemented by departments of transportation to analyze and monitor the performance condition behavior of bridge components over their useful lifetime.

Section: Particle Swarm Optimization (Pso)mentioning

confidence: 99%

Scrutinizing the Performances of Hybrid ANN Models for Forecasting Condition of Bridges

Elshaboury,

Ben Seghier,

Abdelkader

et al. 2023

Self Cite

“…Meta-heuristic, also known as nature-inspired, optimization algorithms are a new trend in optimization models that outperform traditional gradient-based optimization models in terms of capturing global optima with a low convergence rate. This class of optimization algorithms has earned the trust of the scientific community and has been used in a variety of fields, including structural reliability [12][13][14], multi-objective optimization [15], and even improving the efficiency of machine learning models [16,17].…”

Section: Original Article Abstractmentioning

confidence: 99%

A Novel Chaotic Optimization‐Oriented Model for Bridge Maintenance and Rehabilitation Planning

Abdelkader,

Ben Seghier,

Elshaboury

et al. 2023

Self Cite

The rise in percentage of deteriorated bridges coupled with the escalation of maintenance costs call for systematic bridge management systems. Therefore, this research paper introduces an exponential chaotic differential evolution (ECDE) model for optimizing bridge maintenance plans at both project and network levels. The developed optimization model is envisioned on addressing the performance, economic, social, and environmental aspects of bridge maintenance management. In addition, it exploits the use of chaotic sequences to circumvent the critical shortcomings of local minima entrapment and premature convergence encountered by classical optimization metaheuristics. The validity of the developed model is experimented using a comparative analysis of a wide set bridge elements and over a varying multi‐year maintenance plan. Comparison results reveal that ECDE‐based Sinusoidal algorithm improve the performance diagnostics of classical meta‐heuristics by values ranging from 49.2% to 73.1% over the multi‐year maintenance plans. It can be argued that the developed model could benefit transportation agencies in formulating flexible, sustainable, and cost‐effective maintenance strategies in addition to preserving the structural condition of designated bridges in the network.

“…This technique is a strong supervised ensemble learning strategy that employs the concept of classification and regression trees (CART). Random forest proved to be effective in tackling a wide range of complicated prediction issues [27,28]. To achieve the final output results, the random forest is produced in two stages.…”

Section: Random Forest (Rf)mentioning

confidence: 99%

Assessment of artificial intelligence‐ based techniques for the estimation of pile group scour depth

Jafari‐Asl,

Seghier,

Panagiotis

et al. 2023

The scour phenomenon around piles is regarded as one of the main causes of serious damages to the pile‐supported structures such as bridges, jetties, wind turbines, and offshore platforms threatening their stability and sustainability in the long term. Thus, accurate forecast of scouring is vital for the design and operation of these structures. In this paper, three artificial intelligence‐based techniques including support vector regression, artificial neural network and random forest were applied to predict the local scour depth around pile groups. An experimental dataset is collected and used to construct the machine learning‐based models. The sediment number, shields parameter spacing, Keulegan‐Carpenter number and pile Reynolds number were used as input variables for the model development. Results assessment indicate that the artificial neural network model anticipated the highest performance among the three machine learning based models, with coefficient of determination of 0.97, and root mean square error of 0.15.