China
Africa
Explore chapters and articles related to this topic
Variation in the Real World
Published in Mundwiller Stephen, Statistical Process Control, 2017
In the Journal of Quality Technology (October 1984), Lloyd Nelson1 published a set of rules for process control to determine out-of-control signals known as the Nelson Rules. These rules apply to the Xbar or variable values chart. 1.One point is more than three standard deviation units from the mean.2.Nine (or more) points in a row are on the same side of the mean.3.Six (or more) points in a row are continually increasing or decreasing.4.Fourteen (or more) points in a row alternate in a direction, increasing then decreasing.5.Two (or three) out of three points in a row are more than two standard deviations from the mean in the same direction.6.Four (or five) out of five points in a row are more than one standard deviation from the mean (process average) in the same direction.7.Fifteen points in a row are within one standard deviation of the mean (process average) on either side of the mean (process average).8.Eight points in a row exist, but none within one standard deviation of the mean (process average), and the points are in both directions from the mean (process average).
Sensor fault detection for urban drainage systems using redundant measurements
Published in Urban Water Journal, 2022
Martin Pleau, Diana Qing Tao, François Grondin, Shadab Shishegar, Olivier Fradet
The paper proposes an innovative and efficient real-time sensor fault detection approach based on the statistical properties of redundant data to improve the operation and maintenance of UDS. The approach is easy to implement and can be configured to detect in real time a large set of anomalies, including incipient failures. Assuming different failure signatures, the efficiency of the approach can be assessed a priori. This is a key feature when comes time to choose and optimize the validation rules. The novelty of the proposed approach deals mainly with the decision-making process where the residuals are characterized in real time in so-called ‘observations’. These observations can be obtained using various decision rules, such as the Western Electric Decision Rules (Western Electric Company 1956), the Nelson Rules (Nelson 1984) and the Westgard Rules (Harel et al. 2008). The choice of the decision rules is driven by the sensor failure signatures to be detected, the efficiency sought and the complexity of the algorithm to be implemented.