Artificial Intelligence in Hybrid Vehicle Transmission Control - Literature Review and Research Methodology

Abstract

The electrification of vehicles is making the development of control software for automatic transmissions increasingly complex. A new drive train behavior is created, as well as new driving functions and operating states. Furthermore, increasing customer requirements for comfort, dynamics and fuel consumption must be ensured over the entire service life. The authors analyze the state-of-art in transmission control showing that current development methods are reaching their limits. Artificial intelligence (AI) offers many advantages such as optimization, learning algorithms and databased modelling. AI thus opens up new possibilities for designing control algorithms. This paper includes a literature review showing that AI methods in vehicle powertrains has not yet been adequately researched, especially in the design of control algorithms in hybrid vehicle transmissions. From this, the authors derive the research hypothesis that AI methods increase the accuracy and robustness of control algorithms in hybrid vehicle transmissions. Research questions are: Which challenges exist in hybrid vehicle transmission control? Which AI methods are useful in this area? What are the boundary conditions and benefits of those methods? Finally, the paper describes a research methodology to answer those questions including the analysis of the state-of-art in transmission control, expert interviews, and validation experiments.