3 sonuçlar
Arama Sonuçları
Listeleniyor 1 - 3 / 3
Yayın Maintenance policy analysis of the regenerative air heater system using factored POMDPs(Elsevier Ltd, 2022-03) Kıvanç, İpek; Özgür Ünlüakın, Demet; Bilgiç, TanerMaintenance optimization of multi-component systems is a difficult problem. Partially Observable Markov Decision Processes (POMDPs) are powerful tools for such problems under uncertainty in stochastic environments. In this study, the main POMDP solution approaches and solvers are surveyed. Then, based on experimental models with different complexities in the size of the system space, selected POMDP solvers using different representation patterns for modeling and different procedures for updating the value function while solving are compared. Furthermore, to show that factored representations are advantageous in modeling and solving the maintenance problem of multi-component systems where there exist also stochastic dependencies among the components, the maintenance problem of the one-line regenerative air heater system available in thermal power plants is modeled and solved with factored POMDPs. In-depth sensitivity analyses are performed on the obtained policy. The results show that factored POMDPs enable compact modeling, efficient policy generation and practical policy analysis for the tackled problem. Furthermore, they also motivate the use of factored POMDPs in the generation and analysis of maintenance policies for similar multi-component systems.Yayın An effective maintenance policy for a multi-component dynamic system using factored POMDPs(Springer Verlag, 2019-09-20) Kıvanç, İpek; Özgür Ünlüakın, DemetWith the latest advances in technology, almost all systems are getting substantially more uncertain and complex. Since increased complexity costs more, it is challenging to cope with this situation. Maintenance optimization plays a critical role in ensuring effective decision-making on the correct maintenance actions in multi-component systems. A Partially Observable Markov Decision Process (POMDP) is an appropriate framework for such problems. Nevertheless, POMDPs are rarely used for tackling maintenance problems. This study aims to formulate and solve a factored POMDP model to tackle the problems that arise with maintenance planning of multi-component systems. An empirical model consisting of four partially observable components deteriorating in time is constructed. We resort to Symbolic Perseus solver, which includes an adapted variant of the point-based value iteration algorithm, to solve the empirical model. The obtained maintenance policy is simulated on the empirical model in a finite horizon for many replications and the results are compared to the other predefined maintenance policies. Drawing upon the policy results of the factored representation, we present how factored POMDPs offer an effective maintenance policy for the multi-component systems.Yayın Maintenance policy simulation for a factored partially observable system(The Society for Modeling and Simulation International, 2019-07) Özgür Ünlüakın, Demet; Kıvanç, İpekTaking maintenance decisions is one of the well-known stochastic sequential decision problems under uncertainty. Partially Observable Markov Decision Processes (POMDPs) are powerful tools for such problems. Nevertheless, POMDPs are rarely used for tackling maintenance problems of multi-component systems because their state spaces grow exponentially with the increasing number of components. Factored representations have been proposed for POMDPs taking advantage of the factored structure already available in the nature of the problem. Our aim in this study is to show how to formulate a factored POMDP model for the maintenance problem of a multi-component dynamic system and how to simulate and evaluate the obtained policy before implementing it in real life. The sensitivity of the methodology is analyzed under several cost values, and the methodology is compared to other predefined policies. The results show that the policies generated via the POMDP solver perform better than the predefined policies.
