Geographic Information System-based road network accessibility modeling using heuristic assisted adaptive temporal convolution network

Due to the rapid development of the country's economy, the utilization of the road network has greatly increased. It improves the facility of accessing educational, health centers, markets, and other essential places. Multiple models are introduced for calculating the accessibility of the road network. However, traditional frameworks struggle to capture dynamic temporal shifts and fail to align selection outcomes with complex, real-world structures. To address these limitations, this work develops a deep learning-based road network accessibility framework to analyze the road network structure to provide sustainable transportation. At first, the required data is gathered from the online resources and subjected to the feature extraction phase. Here, the Aggregate Transportation Score (ATS), Mean Fractal Dimension (MFD), Mean Road Network Density (MRND), and Mean Weighted Accessibility Index (MWAI) features are derived using “connectivity, accessibility, spatial pattern, and network development.” Then, these features are fused and transferred to the detection phase. Here, the road network accessibility modeling is executed using the Adaptive Temporal Convolution Network (ATCN). The modeling performance of the ATCN is enhanced by tuning its parameter using the Random Revised Student Psychology Optimization (RRSPO). Finally, the proposed approach offers an accurate outcome. The experimentation is evaluated to confirm the designed approach’s efficacy. The developed framework achieved MEP of 3.7, and RMSE of 13.77. Thus, the result showed the recommended model had better transportation planning than other models.