Multistate PCA for continuous processes

摘要

Multiway PCA and multiblock multiway PLS have significantly enhanced process monitoring of batch processes by providing expected trajectory profiles for process variables along the batch duration. However, continuous processes lack of an equivalent technology because of their flexible operations and undefined time duration. This work presents a novel methodology to define operating regions where continuous processes operate. Such operating regions are determined by special variables, named state variables. Transition trajectories between regions or states of operation are calculated to determine the most likely profile in terms of state variable changes. This methodology can be implemented in the context of empirical monitoring methods, named Multistate PCA. A case study that make use of CO2 capture process data shows how this methodology could enhance fault diagnostics and statistical monitoring for continuous processes.