AI based machine optimization

AI-based process monitoring

The majority of manufacturing companies cannot afford machine downtime and scrap. These, however, are frequently the consequences of wear and machine overload. The Fraunhofer IPT is developing a sensor-aided assistance system which uses Artificial Intelligence (AI) to make reliable predictions about the production process.

Complex process condition and lack of monitoring impair production quality

Manufacturing companies depend on the expertise and experience of their machine operators for the production of high-quality, sophisticated, precision components. Complex phenomena such as tool wear, chatter or imbalance occur frequently during machining processes and have an adverse effect on the surface quality of the workpieces. In many cases, machine operators must also rely on their own perceptions to monitor and control manufacturing processes as the machine itself has little in the way of assistive equipment at its disposal.

Recognizing risks in the production process via Artificial Intelligence

The Fraunhofer IPT is developing an assistance system which will help to identify wear and disruptive factors in the production process at an early stage: Sensors positioned close to production activity provide information relating to the production process and the machine status. However, the sensor signals are affected by various external influences and often cannot be interpreted immediately by the user. Artificial Intelligence processes permit the relevant parameters to be extracted from these sensor signals and used to model process status. The data obtained is used by AI to learn about the prevailing process states and is then able to recognize early warning signs of any deviation from standard and predict risks in the production process.

The Fraunhofer IPT is developing assistance systems which will analyze sensor signals and detect emerging sub-optimal process states in time to allow the machine operator to respond. This means that even inexperienced machine operators can successfully monitor process control standing at the machine. A learned AI model can use sensor data to predict tool wear or chatter, for example, eliminating any need for the user to carry out additional measurements under a microscope. In the context of an ageing society and given regular staff turnover within a company, assistance systems like this can facilitate the operation of machine tools and even contribute to job security.