Using clustering based logical equation set to decompose large scale chemical processes for parallel solving data reconciliation and parameter estimation problem

Abstract

Data reconciliation and parameter estimation (DRPE) are important to the performance improvement of real time optimization and process control. As the computational difficulty of nonlinear DRPE optimization problems increases significantly with the number of variables and the equations in large scale processes, solving several smaller DRPE problems iteratively can be more efficient than solving a single large one. A novel parallel processing strategy is proposed to address the optimal distributed DRPE sub-problems. The clustering based logical equation set decomposition (CLESD) is developed to decrease the size of DRPE sub-problems and minimize the information loss of the large-scale DRPE problem. The decomposition of a large DRPE problem requires a two-stage approach, including the equation clustering decomposition and the variable clustering decomposition. CLESD-DRPE is compared with the traditional DRPE method via two industrial applications. The results show that CLESD-DRPE outperforms the traditional DRPE in terms of the solution time and convergence efficiency.