Loading...

Development of Pavement Performance Prediction Models Based on the Assumptions of Availablity and Ubavailabilty of Accurate Data

Ziyadi, Mojtaba | 2011

683 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 42484 (09)
  4. University: Sharif Universiyt of Technology
  5. Department: Civil Engineering
  6. Advisor(s): Tabatabaei, Nader; Shafahi, Yusof
  7. Abstract:
  8. Accurate prediction of pavement performance is essential to a pavement infrastructure management system. Selection of the prediction model is based on the extent of available data, assumptions used in performance modeling, ease of use and management purposes. Therefore, two methods were proposed in this thesis based on the assumptions of availability and unavailibility of accurate data. The first method presents a two-stage model to classify and accurately predict the performance of a pavement infrastructure system. Sections with similar characteristics are classified into groups using a support vector classifier (SVC). Then, a recurrent neural network (RNN) is utilized to predict performance. A case study shows that the proposed model is a good classification decision support system, has better prediction results than the single stage RNN model, and captures underlying effects of the different variables. In the second method, a stochastic expert-Markovian framework is proposed, especially useful for developing countries where the system lacks sufficient data. The two key components of the framework are eliciting expert knowledge using analytic network process (ANP) technique to derive Markovain TPM and regularly updating the TPM using new measurements from network inspections. To evaluate applicability of the proposed methodology the result is compared to the homogenous and non-homogenous Markov Chains obtained using historical data. A special case of Bayesian Inference is used for updating the TPM regularly. The results obtained from knowledge-based model were very close to those of the data-intensive models and this is encouraging because it shows that the proposed model is well suited for environments with limitation of adequate data. Moreover, the results of the updating process reveales that the expert knowledge precision has great influence on probing the model. Therefore, eliciting expert knowledge and techniques used for this purpose should be dealt with much care
  9. Keywords:
  10. Pavements ; Forecasting ; Recurrent Neural Networks ; Markov Chain ; Bayesian Inference

 Digital Object List

 Bookmark

No TOC