Open Access

In industrial production, Explainable Artificial Intelligence offers the opportunity to better understand AI models, not only to understand their decisions but also to be able to improve processes, which is possible even without the necessary experience. There are several widely used methods to implement this for Artificial Intelligence models, but no uniform standards and definitions. There are also still some hurdles on a technical and legal level as well as trust on the human side. Ultimately, the application of such techniques is particularly important in this area, but this is often not possible without the well-functioning and widespread use of Artificial Intelligence in manufacturing industry. However, it is a further step towards intelligent manufacturing and the acceptance of Artificial Intelligence in this area. The publication not only shows the latest developments, but also some use cases of Explainable Artificial Intelligence in manufacturing. Not only legal aspects but also ethical aspects are highlighted. In conclusion, it can be stated that Explainable Artificial Intelligence methods per se are well established, but their implementation and use in practice is lagging behind. A major point here in the future will be the explanation of generative artificial intelligence and user-centred explanations for non-domain experts. A standardised definition of explainability and standards for evaluating the individual methodologies must also be created

Transportation planners are increasingly relying on AI to optimize logistics and solve persistent challenges. However, as AI advances rapidly, most software vendors are unable to evaluate and implement all new developments. This paper uses bibliometric methods to track and evaluate the emerging trend of neurosymbolic AI, which combines neural networks with symbolic AI to improve decision making. By analyzing literature and citation data, we gain insights into the development and impact of neurosymbolic AI. The results provide a scalable approach for practitioners to efficiently identify and evaluate AI trends to facilitate the strategic adoption of technologies and innovations in transportation planning.

Performance benchmarking is crucial for optimizing networks, including 5G Non-Public Networks (5G-NPN). Since one of the major advantages of 5G-NPN is to guarantee Quality of Service (QoS), ensuring optimal performance for their diverse applications is critical. This requires adjustments and testing of various radio-related parameters. Also, performance analysis and benchmarking need to be done based on the evaluation of relevant Key Parameter Indicators (KPIs) in order to identify the parameter set for the optimum performance for each application’s requirements. Many published results on performance benchmarking often lack transparency in their scoring methods. Additionally, QoS benchmarking evaluation for 5G-NPN use cases needs further steps due to the varying ranges of their KPIs. For example Block Error Rate (BLER) is mostly represented by percentage, while Reference Signal Received Power (RSRP) and Reference Signal Received Quality (RSRQ) are negative values.3GPP in TR 103.559 outlines practices for benchmarking network QoS, with a focus on Speech and multimedia Transmission Quality (STQ). This paper extends these outlines to evaluate 5G-NPN performance by defining a multi-objective function. We select a specific 5G-NPN use case as an example and apply four tests with varying network configurations. After each test, we collect most relevant 5G KPIs. To facilitate comparison, all KPIs are rescaled to a common scale. Additionally, we assign weights to each KPI based on its significance in the chosen use case. By combining rescaling and weight assignments, we propose a single metric that effectively characterizes the overall network performance for 5G-NPNs based on their specific use case requirements.

Variable refrigerant flow (VRF) systems are constantly prone to failures during their lifespan, causing breakdowns, high energy bills, and indoor discomfort. In addition to correctly identifying these defects, fault detection, and diagnostic studies should be able to anticipate and predict the anomalies before they occur for efficient maintenance. Therefore, this study introduces an efficient self-learning predictive maintenance system, CACMMS (Cloud Air Conditioning Monitoring & Management System), designed to anticipate refrigerant leaks in VRF systems. Unlike previous efforts, this system leverages advanced fault detection and diagnosis strategies in a real existing building to enhance prediction accuracy. The study employed three noise filtering models (Kalman filter, moving average, S-G smoothing) in the preprocessing phase. Ten features were selected for assessment, and four machine learning models (decision tree, random forest, K-nearest neighbor, support vector machine) were compared. The accuracy, precision, sensitivity, computation time as well and confusion matrix were used as performance indicators and metrics to evaluate and choose the best performant model. Results indicated that decision tree and random forest models achieved over 95 % accuracy with execution times between 0.70 s and 3.32 s, outperforming K-nearest neighbor and support vector machine models. These findings highlight the system’s potential to reduce downtime and energy costs through effective predictive maintenance.

Over the years, the Internet of Things has brought significant benefits to modern society, lives, and industries; however, the technology used has yet to mature sufficiently to provide secure devices and communication. Recently, the number of connected devices rapidly grows, thus adversaries have more opportunities to gain access to IoT devices and use them to launch what is called large-scale attacks. With the rapid proliferation of Internet of Things (IoT) devices, the need for efficient and effective Intrusion Detection System (IDS) tailored for IoT environments has become increasingly paramount. This paper explores various techniques employed in contemporary IoT IDS, including traditional signature-based approaches like Snort and Bro/Zeek, as well as emerging deep learning-based methods.