Case-Based-Reasoningzur Vorhersage von Produkteigenschaften

The presentation by B. Gerschütz, S. Goetz, M. Hörmann, and S. Wartzack focused on the application of Case-Based Reasoning (CBR) in predicting product properties. The team identified a key issue in product development: simulation departments often become bottlenecks, slowing down the process. Their solution was to alleviate this burden by reusing existing simulation data. They proposed a system of virtual engineering, which integrates databases filled with simulation data from both simulation and construction departments. This approach is designed to streamline the product development process by making it more efficient and less reliant on constant input from the simulation department. The core of their methodology involves the use of machine learning tools like sciKitLearn and Pandas for data analysis and result prediction. The process starts with importing existing simulation data into the database, which is then analyzed using these tools. The analysis includes local result predictions and the generation of outputs like 'results.json', which document the predictions. Their system is designed to be user-friendly, with interfaces that allow for easy data input, export, and import. This setup ensures that users can quickly understand and interact with the system, making the prediction of product properties more accessible. One of the key benefits of their approach is the significant workload reduction for simulation departments. By automating routine tasks and simplifying data analysis in the R&D process, the system allows for a more streamlined workflow. It also offers the flexibility to be expanded and integrated into different processes, making it a versatile tool in digital engineering.