Abstract
Manufacturing processes have become increasingly sophisticated leading to greater usage of robotics. Sustaining successful manufacturing robotic operations requires a strategic maintenance program. Maintenance can be very costly, especially when some manufacturers unnecessarily spend resources (i.e., time, money) to maintain their equipment. To reduce maintenance costs, manufacturers are exploring how they can assess the health of their robot workcell operations to enhance their maintenance strategies. Effective health assessment relies upon capturing appropriate data and generating intelligence from the workcell. Multiple data streams relevant to a robot workcell may be available including robot controller data, a supervisory programmable logic controller data, maintenance logs, process/part quality data, and equipment/process fault and/or failure data. This data can be extremely informative, yet the extreme volume and complexity of this data can be both overwhelming, confusing, and paralyzing. Researchers at the National Institute of Standards and Technology have developed a test method and companion sensor to assess the health of robot workcells, which will yield an additional and unique data stream. The intent is that this data stream can either serve as a surrogate for larger data volumes to reduce the data collection and analysis burden on the manufacturer or add more intelligence to assessing robot workcell health. This article will present the immediate effort focused on verifying the companion sensor. Results of the verification test process are discussed along with preliminary results of the sensor’s performance during verification testing.
A Fuzzy Drift Correlation Matrix for Multiple Data Stream Regression
