Remote Electrical Tilt (RET) is a method for configuring antenna downtilt in base stations to optimize mobile network performance. Reinforcement Learning (RL) is an approach to automating the process by letting an agent learn an optimal control strategy and adapt to the dynamic environment. Applying RL in real world comes with challenges, for the RET problem there are performance requirements and partial observability of the system through exogenous factors inducing noise in observations. This thesis proposes a solution method through modeling the problem by a Partially Observable Markov Decision Process (POMDP). The set of hidden states are modeled as a high- level representation of situations requiring one of the possible actions uptilt, downtilt, no change. From this model, a Bayesian Neural Network (BNN) is trained to predict an observation model, relating observed Key Performance Indicators (KPIs) to the hidden states. The observation model is used for estimating belief state probabilities of each hidden state, from which decision of control action is made through a restrictive threshold policy. Experiments comparing the method to a baseline Deep Q- network (DQN) agent shows the method able to reach the same average performance increase as the baseline while outperforming the baseline in two metrics important for robust and safe control behaviour, the worst- case minimum reward increase and the average reward increase per number of tilt actions. / Fjärrstyrning av Elektrisk Lutning (FEL) är en metod för att reglera lutningen av antenner i basstationer för att optimera presentandan i ett mobilnätverk. Förstärkande Inlärning (FI) används som metod för att automatisera processen genom att låta en agent lära sig en optimal strategi för reglering och anpassa sig till den dynamiska miljön. Att tillämpa FI i ett verkligt scenario innebär utmaningar, för FEL specifikt finns det krav på en viss nivå av prestanda samt endast en delvis observerbarhet av systemet på grund av externa faktorer som orsakar brus i observationerna. I detta arbete föreslås en metod för att hantera detta genom att modellera problemet som en Delvis Observerbar Markovprocess (DOM). De dolda tillstånden modelleras för att representera situationer där var och en av de möjliga aktionerna behövs, det vill säga att luta antennen upp, ner eller inte ändra på lutningen. Utifrån denna modellering så tränas ett Bayesiskt Neuralt Nätverk (BNN) för att estimera en observationsmodel som kopplar observerade nyckeltal till de dolda tillstånden. Denna observationsmodel används för att estimera sannolikheten att vardera dolt tillstånd är det rätta. Utifrån dessa sannolikheter så görs valet av aktion genom ett tröskelvärde på sannolikheterna. Genom experiment som jämför metoden med en standardimplementering av en agent baserad på ett Djupt Qnätverk (DQN) visas att metoden har samma prestation när det kommer till en medelnivå på prestandaökning i nätverket. Metoden överträffar dock standardmetoden i två andra mätvärden som är viktiga ur aspekten säker och robust reglering, minimumvärdet på prestandaökningen samt medelökningen av prestandan per antal up- och nerlutningar som används.

In today's competitive environment, the importance of continuous production, quality improvement, and fast delivery has forced production and delivery processes to become highly reliable. Keeping equipment in good condition through maintenance activities can ensure a more reliable system. However, maintenance leads to temporary reduction in capacity that could otherwise be utilized for production. Therefore, the coordination of maintenance and production is important to guarantee good system performance. The central thesis of this dissertation is that integrating maintenance and production decisions increases efficiency by ensuring high quality production, effective resource utilization, and on-time deliveries. Firstly, we study the problem of integrated maintenance and production planning where machines are preventively maintained in the context of a periodic review production system with uncertain yield. Our goal is to provide insight into the optimal maintenance policy, increasing the number of finished products. Specifically, we prove the conditions that guarantee the optimal maintenance policy has a threshold type. Secondly, we address the problem of integrated maintenance planning and production scheduling where machines are correctively maintained in the context of a dynamic aircraft repair shop. To solve the problem, we view the dynamic repair shop as successive static repair scheduling sub-problems over shorter periods. Our results show that the approach that uses logic-based Benders decomposition to solve the static sub-problems, schedules over longer horizon, and quickly adjusts the schedule increases the utilization of aircraft in the long term. Finally, we tackle the problem of integrated maintenance planning and production scheduling where machines are preventively maintained in the context of a multi-machine production system. Depending on the deterioration process of machines, we design decomposed techniques that deal with the stochastic and combinatorial challenges in different, coupled stages. Our results demonstrate that the integrated approaches decrease the total maintenance and lost production cost, maximizing the on-time deliveries. We also prove sufficient conditions that guarantee the monotonicity of the optimal maintenance policy in both machine state and the number of customer orders. Within these three contexts, this dissertation demonstrates that the integrated maintenance and production decision-making increases the process efficiency to produce high quality products in a timely manner.