Research On Life-cycle Cost Of Bridge Based On MDP

Large bridges are deteriorating continuously during their service lives, because of external environment, load and other factors. Necessary maintenance and inspection should be taken on bridge structures for ensuring safety and reliability during service. However, irrelevancy maintenance policies may lead to high life-cycle cost or structural failure, cause enormous economic loss. Life-cycle design of bridge structure has great research significance. This dissertation focuses on investigation of life-cycle cost and maintenance policy of bridge structure by using a Markov Decision Processes-based (MDP) modeling approach. The main contents are as follows:Firstly, a new method based on Kringing technology and Monte Carlo important sampling was proposed for structural reliability analysis. Kriging technology can be used to alleviate the computational burden as well as response surface method (RSM). However, surrogate model constructed by Kriging technology is more accurate than that constructed by RSM, especially when the limit state function is strong nonlinear. Monte Carlo important sampling was used to obtain the failure probability of this surrogate model. The results show that the proposed method has excellent accuracy and is more efficient than the direct Monte Carlo sampling and important sampling.Then, a life cycle design method based on MDP was proposed and value iteration method was used to identify optimal maintenance policy. In order to build the Markov degradation model of bridge structures and determine the optimal maintence policy, the methods used to determine the self-transition matrix, decision-effect matrix and joint-transition matrix were researched.Finally, holding the objective of minimization of expected life-cycle cost, the economic model of bridge life-cycle design was developed to satisfy the reliability index. A MDP modeling approach was used to identify optimal structural initial resistance and their associated maintenance policies. The effects of initial resistance, resistance deterioration, discount rate and failure loss on bridge life-cycle costs, reliability index and maintenance policy were discussed.