How can we make industrial production processes more economical? This was the topic of a speech which was held by QASS-engineer Sören Barteldes in November 2016 at The Maintenance, Monitoring and Control Conference (AKIDA) of RWTH Aachen University. Central point of the speech: reducing false reject rate by signal evaluation and sensor building, using the example of cold forming.
Process anomalies produce false positives
Measuring acoustic emission inside of solid bodies during production of components is state of the art since years. In most cases, piezo-electric sensors are used. Frequently, the data from these sensors is displayed in a diagram with the axes time and amplitude. If an incidents exceeds certain threshold levels, the component (or process) is declared as faulty, although this measurement technology is prone to errors. For example: electrical influences, impulse-like incidents like cracks or pushes or other process anomalies deliver acoustic emission that may be wrongly interpreted as signals from a real crack or a real error.
Innovative signal analyses modernize industrial production processes
Fourier- and envelope-analysis help the user to transform the signal in order to gain more information, to isolate the source of a signal. In many cases, this is done post-process. Now, QASS defines this state of the art in a new way – with High-Frequency-Impulse-Measurement as an in-process method. With HFIM, signal analysis is realized by using Field-Programmable-Gate-Arrays (FPGA), in real-time, to generate a three dimensional waterfall diagram that visualizes the production process itself, from start to stop. HFIM makes pattern recognition possible, as well as envelope analysis and detection of statistical significances. With HFIM, it’s possible to evaluate many industrial production processes.
Analysing straightening processes
Another application for HFIM: straightening of heat-treted gears and shafts. During straightening, cracks can occur at the side of the component that suffers tensile load. These cracks show up as impulse-like signals with high amplitude and dampened signal. Unfortunately, noise is able to mimic these acoustic emission. Without HFIM, it’s hard to distinguish between signals that come from real cracks and noise that mimics these signals. This means that, in a worst case scenario, good components are recycled because of false-positives. For hardened components, QASS is confident to reduce false reject rate by 60 percent.
The Maintenance, Monitoring and Control Conference (AKIDA) of RWTH Aachen University took place at November 15th and 16th, 2016. Photo: Bastian Späth – with permission by Institute for Mineral Resources Machine Technology (RWTH Aachen)