A Cognitive Sensor System Architecture for the Monitoring of Flexible Machining Systems

The machining of complex free-form workpieces, e.g. in die- and mold-making, requires an exceptionally high level of quality. Even minimal deviations in dimensional accuracy or surface roughness due to tool wear or critical engagement conditions are unacceptable in applications such as injection molding or die-casting. To perform optimizations at an early stage, sensor-based in-process monitoring of machining is essential. However, applying existing sensor systems in this area is complicated due to the strong emphasis on flexible multi-axis machine tools as well as single-part and small batch production. Therefore, this paper proposes a new wireless sensor system architecture designed for permanent process monitoring in flexible machines and variable production scenarios at the hardware and information extraction levels. The approach uses sensitive MEMS accelerometers to detect process emissions in high resolution at components only indirectly involved in the process, e.g. the machine spindle. To evaluate the approach, the architecture is implemented based on an FPGA-System-on-Chip hardware platform and integrated into a five-axis CNC milling center. In a tool life experiment, we analyze the ability of the system to map tool wear and surface roughness despite the increased distance from the contact zone. Our methodology enables the mapping independent of the machine structure and thus offers the same capabilities as tool-integrated sensor system architectures while maintaining a higher flexibility and tool independence.