A development of a bridge management system using the LIRF-DBiLSTM approach with the DSF-MPO algorithm
Year: 2026
Type: preprint
Abstract: Bridge infrastructure has greater economic, social, and cultural value. The majority of the infrastructural assets are in poor condition. Therefore, a timely and accurate inspection as well as a planning strategy is required. Prevailing works failed to cover different types of bridges with diverse maintenance costs. To solve these issues, a Linear Interpolated Ridge Function based Deep Bidirectional Long Short Term Memory-centric Bridge Management System (LIRF-DBiLSTM-centric BMS) is proposed in this research method. Primarily, the bridge inventory system data is obtained. Afterward, by utilizing the Delta Scored Density-Based Spatial Clustering of Applications with Noise (DS-DBSCAN) approach, the bridge types are clustered. The damage history and repair history of the particular bridge are extracted from the clustered bridge type. Then, from the clustered bridge types as well as damage and repair history, the attributes are extracted. Thereafter, for predicting the damaged part with location, the extracted attributes and the land seismic data are provided as input to the classifier. The prioritization process is performed by using the Bridge Condition Index (BCI) and Normalized Pi-Functional Membership Function-based Fuzzy Inference System (NPFMF-FIS) if damage occurs. Next, for the prioritized damage, the management factors are derived. After that, concerning minimum cost and user delay rate, the factors are optimized utilizing Dynamic Stepsize Function-based Marine Predator Optimization (DSF-MPO). According to experimental analysis, a minimum cost with a minimum delay rate is obtained by the proposed framework. Moreover, the proposed approach attains 98.86% in damage prediction.
Source: Research Square
Author: Satyam Shivam
Language: English
Subfield: Civil and Structural Engineering
Field: Engineering
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7236008653
