| Machining is a complex, dynamic operation which produces a variety of physical phenomena indicative of the process status. A machining process monitoring platform is designed to detect and isolate process defects from normal machining signal. The platform consists of multiple sensors in a distributed architecture to provide redundance and fault tolerance in signal monitoring. Dedicated digital filtering algorithms are used to isolate process defects from normal cutting signal. A knowledge-based event coordinator is used to prescribe control actions and reactive plans to implement these actions.; By interfacing to the CNC controller, this distributed system architecture and design provides the integrated multi-sensor feedback and the event coordination necessary to initiate corrective actions or recovery procedures in real-time. In the compensation of dynamic as well as quasi-static machine tool errors, this approach provides deterministic reasons for the control actions as opposed to symptomatic reasons. Examples of dynamic errors (time critical) are chatter, catastrophic tool failure, progressive edge chipping, and spindle bearing failure. Examples of quasi-static machine tool errors (non-time critical) are thermal distortion and tool wear during machining. |
