Climate change is increasingly affecting the water cycle, with droughts and floods posing significant challenges for agriculture, hydropower production, and urban water resource management due to growing variability in the factors influencing the water cycle. Reinforcement learning (RL) has demonstrated promising potential in optimization and planning tasks, as it trains models on historical data or through simulations, allowing them to generate new data by interacting with the simulator. This systematic literature review examines the application of reinforcement learning (RL) in water resource management across various domains. A total of 40 articles were analyzed, revealing that RL is a viable approach for this field due to its capability to learn and optimize sequential decision-making processes. The results show that RL agents are primarily trained in simulated environments rather than directly on historical data. Among the algorithms, deep Q-networks are the most commonly employed. Future research should address the challenges of bridging the gap between simulation and real-world applications and focus on improving the explainability of the decision-making process. Future studies need to address the challenges of bridging the gap between simulation and real-world applications. Furthermore, future research should focus on the explainability behind the decision-making process of the agent, which is important due to the safety-critical nature of the application.

In our digitalized world, Data Centers (DCs) serve as crucial infrastructure. Within the DC sector, data processing operations, including processes such as process cooling, hold special significance when investigated from an energy efficiency perspective, as they account for a substantial portion of total energy end-use. Therefore, it is important to prioritize data processing operations in energy management. The objective of this research is to explore the application of AI-powered clustering techniques to identify cooling operational statuses. Additionally, this research offers valuable perspectives on using AI for visualizing and identifying cooling patterns that deviate, which can provide valuable insights into DC energy management. The study object consists of a DC room equipped with Liquid Cooling Packages (LCPs). The findings show that the cooling power density on average is 9.1 kW/m 2 . Through analysis of the elbow curve, the optimal number of clusters is identified to be three. One of the identified clusters, i.e., Cluster 3, is characterized by large time periods with no supplied cooling from the LCPs. When comparing Clusters 1 and 2, Cluster 1 has a higher temperature difference between the chilled water supply and return, but a lower LCP return temperature compared to Cluster 2. Moreover, the quantified cooling characteristics contribute to the understanding of the LCPs’ operational statuses and cooling performance, which is useful for implementing targeted improvements, e.g., adjusting PID parameters, in the cooling infrastructure.

Machine Learning (ML) is frequently utilized in prediction tasks; however, its applications in hydropower forecasting,particularly in forecasting hourly power production, has not been thoroughly investigated. In this paper, two Deep Learning (DL) models,namely an autoregressive neural network and Long Short-Term Memory, are compared to a seasonal autoregressive moving average(SARIMA) model to forecast the hourly power production at a hydropower station situated in Linköping, Sweden. Hyperparameteroptimization algorithms are used to identify suitable DL models and algorithms for automatic model identification of SARIMA modelsare utilized. The three models are evaluated using a rolling origin strategy on a test dataset that consists of 10 months (January – October2023) of hourly power production. The DL models provided similarly accurate forecasts as the SARIMA model according to meansquared error and mean absolute error. However, the DL models are poorly calibrated, resulting in lower coverage compared to theSARIMA model. Furthermore, the models are using a univariate time series (i.e., using historical power production to forecast futurepower production) and future studies need to explore additional variables that may be useful in providing a more accurate forecast.

Information Communication Technology (ICT) centers play a vital role as essential facilities within our digitalized society. Energy efficiency holds great significance in the ICT sector, driven by the rising energy costs and to reduce the environmental impact. Simultaneously, it is essential to ensure a sufficient cooling supply for servers. Artificial Intelligence (AI) can be used to analyze patterns in large datasets, facilitating valuable insights that are difficult for humans to analyze alone because of the complexity and size of the datasets. The aim of this research is to characterize cooling patterns and explore how AI-driven clustering algorithms can be used to identify cooling operational statuses. The research object is an ICT room situated in Linköping, Sweden, and operated by the global telecommunications company Ericsson AB. The ICT room has Liquid Cooling Packages (LCPs) for water-based cooling.The results show that the average cooling power density in the ICT room is 6.98 kW/m2, and the interquartile range is 8.26 kW/m2. The results also demonstrate the potentialities in using AI-based clustering algorithms, K-means in the presented research, to uncover insights related to cooling operational statuses. Furthermore, the results show that it is suitable to divide the data points into four clusters, providing a detailed description of the characteristics of the dataset. The identified clusters differ with regards to variables, among other, such as LCP return air temperature and temperature difference between chilled water supply and return. This is beneficial in identifying undesired operational statuses of LCPs, e.g., low temperature difference between chilled water supply and return, which is an indicator of a poor cooling performance.