Scholarlink

Road infrastructure is a cornerstone of socioeconomic progress in developed and developing nations. The significance of maintaining road networks, especially pavements, has grown exponentially in recent times. The ability to effectively monitor and upkeep pavements and ensure their optimal functionality has become paramount. The Pavement Condition Index (PCI) has the principal task of assessing the functional health of road infrastructure, aiding in prioritizing maintenance and reconstruction efforts. This study presents an innovative solution that combines sophisticated data-driven methodologies with the transformative capabilities of machine learning and optimization. The Particle Swarm Optimized Long Short-Term Memory (PSO-LSTM) algorithm emerges as a cutting-edge fusion, seamlessly integrating optimization techniques with sequential learning. Leveraging the strengths of the long short-term memory (LSTM) neural network and particle swarm optimization (PSO), the PSO-LSTM model addresses complex temporal relationships in data while optimizing parameters for superior accuracy. In this study, the predictive performance of the PSO-LSTM model in comparison to conventional models such as LSTM, artificial neural network (ANN), support vector regressor (SVR), and M5 is comprehensively discussed and also through quantitative metrics including mean squared error (MSE), root mean squared error (RMSE), mean absolute error (MAE), and coefficient of determination (R2), the superiority of the PSO-LSTM model is distinctly demonstrated. Specifically, the PSO-LSTM algorithm showcases remarkable accuracy with significantly lower errors and a robust correlation between predicted and actual Pavement Condition Index (PCI) values. Notably, the PSO-LSTM model achieves an MSE of 6.86, RMSE of 2.62, MAE of 1.40, and R2 of 0.97. These metrics underscore the model’s effectiveness in precise PCI prediction, which is fundamental for enhancing road infrastructure management strategies. This research emphasizes the vital role of technology, particularly advanced algorithms like PSO-LSTM, in enhancing road maintenance and management. This enables road authorities and infrastructure managers to make proactive decisions and allocate resources effectively, leading to improved road network efficiency and longevity.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Optimizing Pavement Condition Index Prediction Using PSO-LSTM Hybrid Algorithm: A Data-Driven Approach for Road Maintenance Strategies

  • 章节

  • 发表日期：

    2025年

查看全文
  • Sachin Gowda,
  • Pala Gireesh Kumar,
  • Kesana Naga Suneetha,
  • Gopalapurapu Kavya Sri,
  • Aakash Gupta,
  • Bishnu Kant Shukla

摘要

Road infrastructure is a cornerstone of socioeconomic progress in developed and developing nations. The significance of maintaining road networks, especially pavements, has grown exponentially in recent times. The ability to effectively monitor and upkeep pavements and ensure their optimal functionality has become paramount. The Pavement Condition Index (PCI) has the principal task of assessing the functional health of road infrastructure, aiding in prioritizing maintenance and reconstruction efforts. This study presents an innovative solution that combines sophisticated data-driven methodologies with the transformative capabilities of machine learning and optimization. The Particle Swarm Optimized Long Short-Term Memory (PSO-LSTM) algorithm emerges as a cutting-edge fusion, seamlessly integrating optimization techniques with sequential learning. Leveraging the strengths of the long short-term memory (LSTM) neural network and particle swarm optimization (PSO), the PSO-LSTM model addresses complex temporal relationships in data while optimizing parameters for superior accuracy. In this study, the predictive performance of the PSO-LSTM model in comparison to conventional models such as LSTM, artificial neural network (ANN), support vector regressor (SVR), and M5 is comprehensively discussed and also through quantitative metrics including mean squared error (MSE), root mean squared error (RMSE), mean absolute error (MAE), and coefficient of determination (R2), the superiority of the PSO-LSTM model is distinctly demonstrated. Specifically, the PSO-LSTM algorithm showcases remarkable accuracy with significantly lower errors and a robust correlation between predicted and actual Pavement Condition Index (PCI) values. Notably, the PSO-LSTM model achieves an MSE of 6.86, RMSE of 2.62, MAE of 1.40, and R2 of 0.97. These metrics underscore the model’s effectiveness in precise PCI prediction, which is fundamental for enhancing road infrastructure management strategies. This research emphasizes the vital role of technology, particularly advanced algorithms like PSO-LSTM, in enhancing road maintenance and management. This enables road authorities and infrastructure managers to make proactive decisions and allocate resources effectively, leading to improved road network efficiency and longevity.