機台維修議題存在諸多不確定性，諸如機台真實狀態、狀態轉移情形等等，在資訊不完全的狀況下，工廠管理者難以採取最佳動作，有效降低機台長期維修與運轉成本。更何況，機台設備十分昂貴，管理者難以承擔此固定資產所造成的損失；因此，維修議題一直是工廠管理者所需要面臨的重大議題。
本研究旨在探討不同資訊不完美的情境下，提供決策者有效的最佳政策，得以下降機台設備的長期期望成本。本篇論文與過去文獻不同之處在於機台的狀態轉移矩陣為未知，需要藉由統計學習更新決策者對於潛在轉移矩陣的認知，以此採取最佳動作。
本研究考量單一機台情境，決策者知道真實轉移矩陣有兩種可能性，但並不知道為哪一者，需要藉由統計學習更新其認知情況，藉此採取當前最佳動作。本研究建構馬可夫決策過程模型，由數學模型推論最佳解相關特性，並以動態規劃方法進行數值求解，求得決策者在不同狀態下各期的最佳動作，進一步建構政策門檻圖，以達最小化長期成本之目標。
最終，本研究經由數學模型推論最佳解相關特性，並且以數值分析作為驗證，得出當決策者可以藉由觀察機台狀態更新其對於潛在真實矩陣認知的時候，決策者願意延遲維修採取不維修動作得以觀察機台狀態轉移資訊。實務上，大多產業均無法得知其機台設備的狀態轉移情況，而難以判斷最佳維修時間點，本研究所提供的方法論可供為參考，協助降低長期下的成本。

There are many uncertainties in the issue of machine maintenance. Under the
condition of incomplete information, it is difficult for factory managers to take the best action to effectively reduce the long-term maintenance and operation cost. In this study, we aim to explore optimal policies for the decision-maker (DM) in different situations of imperfect information, so as to reduce the long-term expected cost. The difference between this study and the previous literature is that the state transition matrix of the machine is unknown, and it is necessary to update DM’s cognition of the potential transition matrix through statistical learning. This study constructs a Markov decision process model, and uses the dynamic programming method for numerical analysis to obtain the best actions for DM. It is concluded that when DM can update his cognition of the potential real matrix by observing the machine state, he is willing to delay the maintenance. In practice, the methodology provided can be used as a reference to help factory manager reduce the long-term cost.

Aven, T., & Castro, I. (2009). A delay-time model with safety constraint. Reliability
Engineering & System Safety, 94(2), 261–267.
Bensoussan, A., & Sethi, S. P. (2009). A note on “the censored newsvendor and the
optimal acquisition of information”. Operations Research, 57(3), 791–794.
Cheng, G., Xie, J., & Zheng, Z. (2019). Optimal stopping for medical treatment with
predictive information. SSRN Electronic Journal.
Ding, X., Puterman, M. L., & Bisi, A. (2002). The censored newsvendor and the optimal acquisition of information. Operations Research, 50(3), 517–527.
Ehrenfeld, S. (1976). On a sequential markovian decision procedure with incomplete information. Computers & Operations Research, 3(1), 39–48.
Grosfeld-Nir, A. (1996). A two-state partially observable markov decision process with uniformly distributed observations. Operations Research, 44(3), 458–463.
Howard, R. A. (1960). Dynamic programming and markov processes.
Kim, M. J., Jiang, R., Makis, V., & Lee, C.-G. (2011). Optimal bayesian fault prediction
scheme for a partially observable system subject to random failure. European Journal of Operational Research, 214(2), 331–339.
Kim, M. J., & Makis, V. (2013). Joint optimization of sampling and control of partially
observable failing systems. Operations Research, 61(3), 777–790.
Lariviere, M. A., & Porteus, E. L. (1999). Stalking information: Bayesian inventory
management with unobserved lost sales. Management Science, 45(3), 346–363.
Lu, X., Song, J.-S., & Zhu, K. (2005). On “the censored newsvendor and the optimal
acquisition of information”. Operations Research, 53(6), 1024–1026.
Maillart, L. M. (2006). Maintenance policies for systems with condition monitoring and obvious failures. IIE Transactions, 38(6), 463–475.
Maillart, L. M., & Zheltova, L. (2007). Structured maintenance policies on interior sample paths. Naval Research Logistics (NRL), 54(6), 645–655.
McCall, J. J. (1965). Maintenance policies for stochastically failing equipment: a survey. Management Science, 11(5), 493–524.
Papadimitriou, C. (1995). Computational complexity (addison wesley, san diego).
Rosenfield, D. (1976). Markovian deterioration with uncertain information. Operations Research, 24(1), 141–155.
Ross, S. M. (1971). Quality control under markovian deterioration. Management Science, 17(9), 587–596.
Song, H., Liu, C.-C., Lawarr´ee, J., & Dahlgren, R. W. (2000). Optimal electricity supply
bidding by markov decision process. IEEE Transactions on Power Systems, 15(2),
618–624.
Valdez-Flores, C., & Feldman, R. M. (1989). A survey of preventive maintenance models for stochastically deteriorating single-unit systems. Naval Research Logistics (NRL), 36(4), 419–446.
Van Oosterom, C., Peng, H., & van Houtum, G.-J. (2017). Maintenance optimization for a markovian deteriorating system with population heterogeneity. IISE Transactions, 49(1), 96–109.
Van Oosterom, C. D., Elwany, A. H., C¸ elebi, D., & Van Houtum, G.-J. (2014). Optimal
policies for a delay time model with postponed replacement. European Journal of
Operational Research, 232(1), 186–197.
White, C. C. (1978). Optimal inspection and repair of a production process subject to deterioration. Journal of the Operational Research Society, 29(3), 235–243.
Xiao, G., Akc¸ay, A., Maillart, L., & van Houtum, G.-J. (2020). Optimality of chain-based threshold policies for machine maintenance under imperfect predictions. Available at SSRN 3683983.
Zhang, H. Z.,Weihua (2017). Analytical solution to a partially observable machine maintenance problem by a dual method.