Quasi-stationary biases of change point and change magnitude estimation after sequential cusum test

Assuming the observation process is a Brownian motion with the drift parameter subject to sudden change, estimations of the change point and change magnitude after the sequential CUSUM test are proposed and investigated. By assuming that the change occures far away from 0, the biases of the estimations conditioning on that the change is detected are obtained as the control limit approaches infinity     

Fault detection and isolation analysis and design for solution copolymerization of MMA and VAc process

The problem of detecting and isolating distinguishable actuator and sensor faults in the solution copolymerization of methyl methacrylate and vinyl acetate monomers are considered in this work. To this end, first state estimates are generated using a bank of high‐gain observers, and nonlinear fault detection and isolation (FDI) residuals are defined. The process dynamics are further analyzed to categorize fault scenarios as distinguishable and indistinguishable, and the necessary and sufficient conditions for the classification are presented. Subsequently, filters are designed that enable FDI for the distinguishable fault scenarios, with the advantage of detecting and confining possible locations for indistinguishable faults. The FDI filters are implemented on the copolymerization process, and the results compared with a linear model based filter design. © 2015 American Institute of Chemical Engineers AIChE J, 62: 1054–1064, 2016