This paper presents the current status of the open-source Modelica library SMArtInt (Simple Modelica Artificial Intelligence Interface), which offers users a straightforward and efficient approach of integrating artificial intelligence via neural networks directly into Modelica models. We provide a detailed overview of the library’s features, area of application, and development process. The primary focus is on the diverse use cases of SMArtInt. The new version 0.5.1 brings an exciting feature with the support of the ONNX format. ONNX (Open Neural Network Exchange) is an open, cross-platform format for the representation and exchange of deep learning models. This means that in addition to TensorFlow models (TFLite), many other models, such as PyTorch can now also run in SMArtInt via the ONNX format.

This paper describes the use of a Modelica system model to support the development of a heat management control system for a fossil-free sailing yacht. Due to tight project timelines, the control system was developed and tested virtually, avoiding delays associated with waiting for the physical system to become available. The system model covers key functions such as heat recovery and heat dumping, enabling automated testing of various operational scenarios. This approach not only accelerates development but also reveals early insights into interactions between the control logic and system dynamics. The model is designed to be seamlessly replaced by the real system once it is built. Future comparisons between simulated and real-world performance will guide refinements to improve model accuracy and support model-based tuning of the control system.