Developing Enhanced Performance Curves of ITD Asphalt Pavements by Mining the Historical Data

Details

• Creators:
  Deng, Yong ; Shi, Xianming
• Corporate Creators:
  Washington State University. Department of Civil and Environmental Engineering
• Corporate Contributors:
  Idaho Transportation Department ; United States. Department of Transportation. Federal Highway Administration
• Subject/TRT Terms:
  Asphalt Pavements Asset Management Calibration Data Mining Forecasting Pavement Performance Validation
• Publication/ Report Number:
  FHWA-ID-23-293
• Resource Type:
  Tech Report
• Geographical Coverage:
  United States
• Edition:
  Final Report 09/01/20020 – 03/31/2023
• Corporate Publisher:
  Idaho Transportation Department
• Abstract:
  Current ITD approach to predict asphalt pavement performance over-simplifies the dynamic and nonlinear nature of pavement deterioration and fails to account for complexities in the pavement condition. As such, this project aimed to develop reliable, realistic and enhanced performance curves for ITD asphalt pavements by mining the historical data. The project consisted of literature review, practitioner survey, assessment of the current ITD pavement performance curves, data processing and modeling for both new and rehabilitated asphalt pavements. Neural networks calibrated with particle swarm optimization achieved desirable prediction performance for asphalt pavement rutting in Idaho using the AASHTOWare Pavement ME Design™ (PMED) data. Gene expression programming models achieved better prediction performance than linear regression and mechanistic-empirical models for four typical distresses – rutting, longitudinal cracking, thermal cracking and roughness of asphalt pavements using PMED data. Deep learning models achieved better prediction performance than statistical models for short-term rutting development of a field asphalt pavement with ITD PMS data than piece-wise regression models for overall condition index of sampled field asphalt pavements. These models can extend the application in terms of the distress type, pavement type and areas of interest. To avoid overfitting and ensure basic rationality of predictive models, statistical methods are necessary to check the stability, robustness, sensitivity, etc. of constructed models before application.
• Format:
  PDF
• Collection(s):
  US Transportation Collection
• Main Document Checksum:
  urn:sha-512:a99140c495b26ac7b37ec55c7ae9c447223808d0487c0c9406608bc8777a5a5402d94b64bc7f479ba10ee4316ade60c212dbbf386e4cd03d8e1e38a0525a087a
• Download URL:
  https://rosap.ntl.bts.gov/view/dot/72653/dot_72653_DS1.pdf
• File Type:
  [PDF - 3.11 MB ]