Using orthogonal projection approach (OPA) for rank-deficient reaction processes

Orthogonal projection approach (OPA) has been used for monitoring batch processes. However, because of the mass balance constraints between the reactants and products in a chemical reaction, standard OPA will fail if rank-deficiency occurs. The effect of the rank-deficiency on using OPA for process reaction data sets is discussed, and a new automatic approach for resolution of rank-deficiency data sets by OPA is proposed for batch control. The method is demonstrated for an industrial batch data set.