Improved BP Neural Network Ensemble Model for Asphalt Pavement Performance Prediction

Area of Science:

• Civil Engineering
• Materials Science
• Artificial Intelligence

Background:

• Traditional asphalt pavement performance models struggle with complex operational environments and increasing traffic loads.
• Nonlinear degradation of pavement performance over time requires advanced prediction methods.

Purpose of the Study:

• To develop a novel approach for accurate asphalt pavement service performance forecasting.
• To improve the prediction accuracy of the Road Condition Index (RCI) and Pavement Quality Index (PQI).

Main Methods:

• Optimized a Backpropagation (BP) neural network using the Levenberg-Marquardt (LM) algorithm.
• Utilized seven key input parameters: road age, average daily traffic, annual temperature range, precipitation, relative humidity, pavement thickness, and compressive strength.

Main Results:

• The proposed LM-optimized BP neural network model demonstrated superior performance in predicting RCI compared to traditional BP networks.
• Achieved a Mean Absolute Error (MAE) of 0.395 for RCI prediction.
• The model accurately predicts road condition and PQI based on the selected factors.

Conclusions:

• The developed model provides a significant advancement in predicting asphalt pavement service performance.
• The LM-optimized BP neural network effectively captures the complex degradation processes of asphalt pavements.
• The model's effectiveness is validated for practical application in pavement management.