DWI – Leibniz Institute for Interactive Materials
Publikationsserver der RWTH Aachen UniversityNot a member yet
552398 research outputs found
Sort by
Control of dynamic maneuvers during launch of flying wings in airborne wind energy systems
No full textAirborne wind energy is an evolving technology that utilizes airborne systems operating in wind fields to harvest energy. To achieve high performance, employing airborne systems with favorable aerodynamic characteristics is promising. In addition, to enable launching and landing at various sites and eliminate elaborate ground infrastructure such as runways or catapult aid systems, the airborne systems must be able to take off and land vertically. A flying wing, span-wise equipped with propulsion units, allows such vertical operation with its nose pointing upward, improving the system's operational flexibility. In addition, it provides favorable aerodynamic characteristics for energy-harvesting flight, allowing enhanced power production. However, due to the specific flight characteristics of flying wings, the launch maneuver from vertical takeoff to energy-harvesting flight is challenging, particularly in strong winds. Moreover, a tether connecting the flying wing to the ground must also be considered. Consequently, controlling the flying wing's motion throughout the operation, especially during such dynamic launch maneuvers, is a significant challenge. This thesis presents a control concept for dynamic flight maneuvers of a flying wing operating in an airborne wind energy system. It focuses on the launch, including the dynamic transition maneuver from vertical takeoff to energy-harvesting flight, providing the potential for enhanced system performance with a flying wing configuration. This transition involves a wide range of attitude changes and high nonlinearities that must be considered in the control concept. Moreover, the specific flight characteristics of flying wings and the constraints imposed by the tether must be considered. The flight controller presented here is cascaded, comprising guidance, translational, and rotational controllers. This thesis primarily focuses on designing the translational controller using an incremental nonlinear dynamic inversion approach. This controller governs the translational motion of the flying wing across its entire operational envelope, including dynamic maneuvers. The control concept is implemented for an exemplary flying wing airborne wind energy system and thoroughly analyzed, with results from linear and nonlinear simulations and flight tests. The developed control concept proves effective for launch maneuvers, addressing the challenges of controlling flying wings during such operations
Measuring regenerative orientation: A new scale for firms' net-positive impact beyond sustainability
No full textFrequency generation for ADPLLs in automotive FMCW radar using nano-scale CMOS
No full textFrequency-modulated continuous-wave (FMCW) radars are extensively utilized in modern automobiles due to their capability to measure distance and velocity, thereby contributing to the reduction of traffic accidents. In these applications, generating the frequency-modulated (FM) signal commonly involves the use of a phase-locked loop (PLL). Traditional charge-pump analog PLLs, however, suffer from several drawbacks, including a large silicon footprint caused by the bulky passive loop filter and the significant contribution of the charge-pump to in-band phase noise.With advancements in technology, designing analog PLLs has become increasingly challenging due to reduced voltage headroom, lower quality factors of passive components, and increased flicker noise from active devices. To address these issues, the concept of the all-digital PLL (ADPLL) has been introduced. ADPLLs utilize a digital loop filter that can be synthesized with standard cells, eliminating the need for a large analog loop filter, thus reducing silicon area and costs. Additionally, ADPLLs offer the flexibility to adjust PLL parameters, such as loop bandwidth, post-fabrication, making them adaptable to various scenarios. They also leverage the continuous scaling-down of CMOS technology more effectively.Despite these advantages, ADPLLs place high demands on the performance of frequency generation circuits. The digitally controlled oscillator (DCO) plays a critical role, as it predominantly affects the out-of-band phase noise of the ADPLL and must have a wide tuning range to determine the chirp signal bandwidth and influence the distance measurement resolution of the FMCW radar. Furthermore, a fine frequency tuning resolution of the DCO is essential because a coarse resolution introduces quantization noise, which degrades the phase noise performance of the PLL. Moreover, the in-band behavior of the ADPLL is largely determined by the reference frequency, typically provided by a crystal oscillator. Reducing the start-up time of the crystal oscillator is also crucial, as it limits the system's response time.The aim of this dissertation is hence to explore architectures for frequency generation circuits, with a primary focus on the design and implementation of the DCO and the crystal oscillator, along with their auxiliary circuitry, such as low-noise power supplies, frequency dividers and buffers, to enhance the performance of ADPLLs for FMCW radar applications. The designs have been validated through measurement results from three 28-nm CMOS silicon prototypes
Multi-source energy harvesting for wireless IIoT sensors
No full textThe Industrial Internet of Things (IIoT) and Industry 4.0 are accelerating the spread of wireless sensors in production, which are used to digitize process, machine, and environmental data for self-optimization and automation. In addition to many positive aspects, such as flexibility,mobility, and low installation costs, the limited operating time due to battery operation prevents the widespread use and acceptance of wireless IIoT sensor technology. Multi-source energy harvesting (MSEH) is a possible solution to the energy supply problem of IIoT sensors. Byutilizing different energy sources simultaneously, the unpredictability and fluctuations in the availability and performance of ambient sources can be compensated for, and the energy harvesting system can be made more resilient. A multi-source energy harvesting system consistsof harvesters, a power management board, a rechargeable energy storage device, and the load to be supplied. However, a universal MSEH power management board customized for industry and application examples for industrial scenarios are still lacking. The MSEH power management board InduFlex developed in this thesis is based on a multiple input multiple output architecture with three independent inputs, each equipped with a powermanagement IC. The design of InduFlex considers the requirement categories of flexibility, performance, reliability, economy, and usability in industry and includes the analysis of the implementation framework and the operating conditions of solar, thermal, and vibration harvesters in the industrial environment. The InduFlex prototype was validated in eight different system context scenarios and three real industrial applications: a milling process, a metal forming process, and an environmental monitoring process in a biolaboratory. Light and vibrationwere the most promising energy sources in the three industrial applications. Based on the potential energy supply through multi-source energy harvesting in industrial applications, exemplary IIoT systems were selected, and the energy balance was calculated. The energy selfsufficient and supporting areas were shown for the required solar cell size depending on the sampling and transmission rate. Overall, the investigations show that the application of (multi-source) energy harvesting is significantly simplified by a flexible, universal power management board. However, the general use of energy harvesting is severely limited in terms of location and available energy due to the great dependence on the available environmental sources. As the application and, therefore, the IIoT sensor technology is the main focus, alternative energy supply solutions, such as larger energy storage units or wireless energy transfer, must be sought if the supply from MSEH is insufficient. Multi-source energy harvesting is not a universal solution to the energy supply problem of wireless IIoT sensor technology. Nevertheless, the potential of energy harvesting should also be evaluated when using wireless IIoT sensor technology, as there are specific areas of application and exceptional cases in which a ‘deploy and forget’ approach can be realized, or at least the operating time can be significantly extended
Pipeline for ontology-based data access and service-oriented integration of artificial intelligence in CAR-T cell production
No full textDespite the promising results of Chimeric Antigen Receptor (CAR)-T cell therapy for treating leukemia and lymphomas, several limitations for broad patient access remain. These include the complexity of the manufacturing process, the unpredictability of patient reactions, and the high manufacturing cost. The application of Artificial Intelligence (AI) has the potential to enhance the reliability and efficiency of the manufacturing process, thereby facilitating safe and effective CAR-T cell products. The integration of AI in CAR-T cell production faces significant challenges, as the complexity of the processes requires a deep data understanding. Additionally, barriers arise from the low level of digitalization and strict regulations in CAR-T cell production, which impedes the integration of AI in the operational environment. Ontology-based data access (OBDA) can provide a remedy by semantically enriching data to increase data understanding, while service-oriented architectures facilitate more efficient integration of AI applications. Therefore, this thesis answers the main research question: Can a pipeline for ontology-based data access and service-oriented AI integration (POBAI) be developed to meet the requirements for CAR-T cell production? POBAI is developed and validated by specific requirements derived from the state of the art and implemented using the research project AIDPATH (AI-driven Decentralized Production of Advanced Therapies in the Hospital) as a case study. For verification, POBAI integrates two AI applications (Digital Cell Twin and Reactive Online Process Control) from AIDPATH using a risk-based approach aligned with Good Automated Manufacturing Practice (GAMP). POBAI is comprised of four components addressing the specific requirements. The Ontology Editor enables OBDA through the creation of an ontology and its mapping to CAR-T cell production data. The AI Agent Modeler assists Data Scientists in developing AI applications, offering an overview of the CAR-T cell production system and assessing the model's alignment with POBAI requirements. The AI Agent Generator automates the creation of an AI agent utilizing FastAPI and Docker, enabling the AI application to function independently and self-contained within the IT infrastructure. The Deployment Manager ensures the AI agent operates as intended and establishes its connection to the operational environment. The main research question was affirmed, with eight out of the eleven specific requirements fully satisfied and the remaining three partially addressed. Implementation in a near-real environment demonstrated the system's functionality, and extensive testing verified its stability. POBAI's architecture was designed to be flexible and scalable, supporting the integration of diverse database types, AI applications, and input files
