TY - GEN
T1 - Condition-Based Maintenance for Partially Observable Systems Under Dynamic Environmental Conditions
AU - Ge, Huikai
AU - Zhang, Nan
N1 - Publisher Copyright:
© 2026 IEEE.
PY - 2026
Y1 - 2026
N2 - This paper studies maintenance optimization for partially observable systems under dynamic environmental conditions. An infinite-horizon partially observable Markov decision process framework is developed to jointly model incomplete system state information, environmental dynamics, and the maintenance process over time. In the proposed model, the hidden system state is represented by a belief state, while a service-stage variable is used to describe the access process after a maintenance request is issued under environment-dependent accessibility constraints. The framework is applied to an offshore wind turbine maintenance problem. Numerical results show that the optimal policy depends not only on the estimated system state but also on environmental accessibility. Compared with two classical maintenance policies, the proposed model achieves consistently lower expected discounted costs across all tests. The results indicate that dynamic environmental conditions play a critical role in developing maintenance policies and establishing budgetary requirements.
AB - This paper studies maintenance optimization for partially observable systems under dynamic environmental conditions. An infinite-horizon partially observable Markov decision process framework is developed to jointly model incomplete system state information, environmental dynamics, and the maintenance process over time. In the proposed model, the hidden system state is represented by a belief state, while a service-stage variable is used to describe the access process after a maintenance request is issued under environment-dependent accessibility constraints. The framework is applied to an offshore wind turbine maintenance problem. Numerical results show that the optimal policy depends not only on the estimated system state but also on environmental accessibility. Compared with two classical maintenance policies, the proposed model achieves consistently lower expected discounted costs across all tests. The results indicate that dynamic environmental conditions play a critical role in developing maintenance policies and establishing budgetary requirements.
KW - Condition-based maintenance
KW - dynamic environmental conditions
KW - offshore wind turbine
KW - partially observable Markov decision process
UR - https://www.scopus.com/pages/publications/105052951667
U2 - 10.1109/SRSE70977.2026.11694593
DO - 10.1109/SRSE70977.2026.11694593
M3 - Conference contribution
AN - SCOPUS:105052951667
T3 - 2026 the 8th International Conference on System Reliability and Safety Engineering, SRSE 2026
SP - 723
EP - 729
BT - 2026 the 8th International Conference on System Reliability and Safety Engineering, SRSE 2026
PB - Institute of Electrical and Electronics Engineers Inc.
T2 - 8th International Conference on System Reliability and Safety Engineering, SRSE 2026
Y2 - 7 August 2026 through 10 August 2026
ER -