Intelligent quality assurance through AI-supported analysis of acoustic signals

In the context of Industry 4.0, more and more diverse inspection tasks are emerging, generating larger volumes of data that are often not easy to interpret. Artificial intelligence (AI) can assign meaning to these measured values and assess the overall situation. The Fraunhofer Institute for Ceramic Technologies and Systems IKTS in Dresden offers a comprehensive range of state-of-the-art artificial intelligence methods for the automatic evaluation of technical and non-technical processes.

Acoustic diagnosis methods can be used to detect faults in manufacturing and operating processes with little effort. Defective components as well as critical plant and operating processes have specific noise patterns that are used for quality assurance.

Automatic signal evaluation is learned

With the help of signal analysis, pattern recognition and machine learning methods, sensor signals can be automatically interpreted, and their meaning recognized, e.g. »test item good« or »component still has 20 percent remaining service life«. This automatic evaluation enables economical and reliable quality assurance.

AI-based measurement and inspection systems learn the principal relationship between sensor signals and their meaning for their individual inspection task from examples before commissioning. Later, they can be »taught« and corrected by humans and thus adapt and improve themselves during operation. This technology is much more flexible and powerful than traditional testing methods.

Wide range of applications

The underlying methods for artificial intelligence, machine learning and pattern recognition of acoustic signals are also of interest for other applications. They have already been successfully used in quality control of gears, determination of the remaining service life of solenoid valves, crack and impact detection in aircraft materials, condition monitoring of railroad wheels, softness testing of paper, pest detection in grain stores or monitoring of compressors.