Porsche combines bayesian networks with modern approaches to optimize production planning
Aug 16, 2023
auto_awesome
Simon Dürr from Porsche and Prof. Dr. Marco Huber discuss the optimization of production planning using Bayesian networks. They delve into incorporating customer behavior and market variations, the creation and reinforcement learning of network structures, and the challenges of implementation in a large organization. They also highlight the importance of plant orders and the potential application in other industries.
The project aims to redesign the planning and ordering process at Porsche using AI and new technologies, addressing limitations of the current SAP-based system and accommodating global market demands.
The AI model developed utilizes bi-agent networks and reinforcement learning algorithms to generate virtual planned orders, improving transparency, flexibility, and stability in both the customer ordering and production planning processes.
Deep dives
Introduction and Background
Simon and Marco introduce themselves and their roles at Porsche and the University of Stuttgart, respectively. They explain the background of the project, which aims to redesign the planning and ordering process at Porsche using AI and new technologies.
Challenges and Goals
The main challenges identified were the limitations of the current SAP-based planning system and the need to incorporate new technologies and adapt to global market demands. The goal is to create a more flexible and efficient planning process that balances customer demand and production capabilities.
The AI Model and its Applications
The AI model developed is based on the concept of planned orders. It utilizes bi-agent networks and reinforcement learning algorithms to generate virtual planned orders that can accurately represent a wide range of customer configurations. The model also improves stability, transparency, and flexibility in both the customer ordering process and the production planning process.
Future Steps and Potential Applications
The next steps involve implementing and operationalizing the prototypes and concepts developed in the project. This will require collaboration with various departments within Porsche and aligning the organization with the new processes. The model has potential applications beyond Porsche, as it can be applied to other industries with highly customizable products and complex planning requirements.
We talk to Simon Dürr from Porsche and Prof. Dr. Marco Huber from Fraunhofer IPA. Simon was responsible for the project at Porsche and wrote his PhD thesis on it. Marco Huber was his supervisor and together they explain the approach.
Thanks for listening. We welcome suggestions for topics, criticism and a few stars on Apple, Spotify and Co.