This use case focuses on the classification of global Local Climate Zones (LCZs). Identifying urban structures and land use, in the form of LCZ classes, is essential for analyzing and understanding urban heat islands and spatio-temporal dynamics of cities worldwide. We will make use of the freely and publicly available Sentinel-1 and Sentinel-2 datasets of radar and optical images, where one year of global data (4 PB) is available.

The dataset is hand-labelled in a number of classes, e.g., “dense trees”, “heavy industry”, “water”, “open high-rise”, etc. The amount of available data allows for training of deep learning models. In the context of DAPHNE, we aim at streamlining the data preparation, classification, and analysis pipelines at scale, improving the classification accuracy, and developing means of temporal change detection and global data analysis.

An existing reliability test system performs accelerated stress tests and monitors test conditions, state of each device under test, as well as electrical signals of the stress pulse. Many devices are tested simultaneously and waveform data is stored. The stored data resembles a 3D time series (test cycle, time, amplitude), electrical measurements (voltage & current). Over the past years, KAI stored millions electrical signals per semiconductor technology.

Using physics-based models, the microscopic degradation in the thin metal layers are simulated and lifetime under different parameters of mechanical stress are estimated and serve as input for the subsequent reliability analysis and degradation modeling.

The current data analysis pipeline (which still includes manual effort) leverages only a subset of waveforms for electro-thermal simulations. In the context of this project, we aim to overcome the existing drawbacks by building an efficient and scalable integrated data analysis pipeline, which is easy to use and extend in order to explore different methods for analyzing the characteristics of waveforms over time and under different influencing parameters.

AVL is developing gas ejectors for fuel cell systems to compensate pressure losses in the anode recirculation path. The optimized ejector generates a high suction pressure, which is depending on geometric variables and operating conditions like fuel supply pressure.

AVL established a DOE workflow combining software and scripts and elaborated a continuously growing database which is getting more dense for standard fuel cell system applications.

The database consists of 600+ data sets, each is characterized by the operating conditions, 7 key and 17 side design parameters – and suction pressure.

Out of Daphne Framework a sophisticated optimizer component, trained on the available dataset, is expected with enhanced prediction quality for wider range of application considering a bigger number of design parameters. The desired benefit is a smaller number of numerically expensive CFD Simulations for prediction verification.

There are many stages within semiconductor manufacturing, one of which is implantation. Special equipment is built for this processing step. Inherent to implanters is beam tuning, which ensures that the current process step’s requirements are met under varying equipment conditions. A process step’s requirement is stated within a recipe.

The target is to predict which recipes are tunable with high probability and which recipes are not, given the equipment’s current state. A tuning fail decreases the OEE (Overall Equipment Efficiency) and generates the need to either schedule a batch (25 wafers) with a different recipe or a maintenance activity for the equipment. After repair, a wider variety of recipes are processable again. In order to reduce downtimes, it is key to predict tuning success before a decision is made on which batch to process next.

AVL List proposes the so-called Integrated and Open Development Platform (IODP) for the continuous verification and validation of vehicles under development. The IODP vision is to provide system maturity assessment competence at any point in time, i.e., checking whether the characteristic value CV of the current state of development can meet the target characteristic value TCV.

The figure shows the evolution of the exemplary characteristic value “fuel consumption” over time, i.e., along the vehicle development process VDP. A key IODP concept is to systematically link simulation and physical tests. Using this approach, highly structured data emerges that accurately represents the vehicle development process. Organizations want to make use of this data, apply data analytics and make inferences in order to optimize their vehicle development process. The development of these process mining solutions involves creating realistic training data and appropriate mining approaches. Therefore, large amounts of data need to be handled and Daphne project–related components are to be employed.