Fast AI-based Controller Tuners

Challenge

Manufacturers increasingly need to adapt their production processes to changing products, materials and operating conditions. Mass customisation, the growing use of recycled materials and variations in raw material properties make it more difficult to maintain consistent production quality.

Every new production batch or process restart can require adjustments to machine parameters and control settings. Today, these adjustments often rely on trial-and-error or the experience of a limited number of expert operators. This makes commissioning time-consuming, difficult to scale and highly dependent on individual expertise.

Although manufacturers collect valuable process data, historical production recipes, specifications and operator logs, this information is not always effectively reused when setting up new production runs. As a result, companies face longer start-up times, unnecessary material consumption and inconsistent product quality.

Project Goals

This project (fAIntune_IRVA) aims to develop an AI-supported commissioning and fine-tuning methodology that helps manufacturers identify suitable process settings faster, with less waste and reduced dependence on expert operators.

The project combines historical production knowledge, artificial intelligence and engineering expertise to improve how production recipes are selected, generated and adjusted.

The research focuses on two complementary developments.

AI-assisted recipe generation will use historical production data, machine measurements, specifications and operator logs to identify suitable settings for new production runs. Rather than relying exclusively on existing recipes, the project will explore AI-based methods to generate new parameter combinations adapted to changing production conditions. Simulation data and dedicated test runs may also support this process.

Safe and efficient process fine-tuning will help operators refine the proposed settings during production start-up. By combining engineering knowledge, physical models and controlled adjustment strategies, the project aims to minimise trial-and-error while keeping operators involved in critical decisions.

The ultimate ambition is to develop a reusable commissioning and fine-tuning toolset that supports faster process adjustments, improves knowledge retention and enables more consistent production outcomes.

For manufacturers, this could mean shorter commissioning cycles, fewer rejected batches, lower raw material consumption and improved first-time-right production performance. By making expert knowledge easier to capture and reuse, fAIntune also aims to help companies scale their production processes more efficiently while maintaining quality across different products, batches and operating conditions.