Production Planning and Control - Dependable Thanks to AI
2023-02-17 Especially in mechanical and plant engineering, production processes are frequently complex. Often, lead times and delivery dates can only be estimated. Prof. Dr.-Ing. Matthias Schmidt seeks to apply artificial intelligence methods to ensure greater adherence to schedules.
If a company does not produce in series, but manufactures individual parts or small batches, delivery times are often not easy to plan: for example, a company receives an order to manufacture a complex production plant. Various working steps are necessary for the construction, such as metalworking processes in prefabrication and complex assembly processes in final assembly. Frequently, individual work systems are blocked by other orders during the order cycle; the external painter, for example, is busy handling a full order book himself, or the expert at the milling machine is sick. There are far more examples of how the workflow can be slowed down. For the client, who needs the production plant in his own company, it is then a matter of waiting. "In practice, planned lead times are often estimated with rough rules of thumb based on empirical values," says Dr.-Ing. Matthias Schmidt, Professor of Production Management.
The researcher seeks to apply artificial intelligence to predict planned lead times, i.e. the time from order to delivery, more reliably. Together with his team, he analyses data from six medium-sized companies and identifies patterns: What are the key parameters that have a significant impact on planned lead times? "Not all variables actually have an influence on the lead time," explains Matthias Schmidt. By means of AI methods such as neural networks, complex workflows are analysed, some of which contain more than 30 steps. At the end of the research project, companies will have a method at hand that is suitable for everyday use, with which they can predict planned lead times more accurately.