A new method for grinder dressing fault mitigation using real-time peak detection

To facilitate the implementation of machine monitoring algorithms on the shop floor, signal processing and decision-making strategies must be developed, which account for the difficulties associated with monitoring a machine in an industrial environment. Therefore, this paper focuses on introducing a new method for dresser contact detection, which takes into account sensor and data acquisition system costs, computational limitations of an embedded detection system, and noise robustness issues associated with shop floor implementation. This paper also discusses the way in which the monitoring algorithm directly interfaces with the machine control in real time in order to ensure that system faults are avoided. Furthermore, the new algorithm is compared with more traditional methods for change detection by applying each algorithm to the signal output from a dresser horsepower sensor. In an industrial application, the new algorithm is shown to provide zero missed detections, zero false alarms, and a sufficiently fast response time.