Optimal decomposition for distributed optimization in nonlinear model predictive control through community detection

Distributed optimization, based on a decomposition of the entire optimization problem, has been applied to many complex decision making problems in process systems engineering, including nonlinear model predictive control. While decomposition techniques have been widely adopted, it remains an open problem how to optimally decompose an optimization problem into a distributed structure. In this work, we propose to use community detection in network representations of optimization problems as a systematic method of partitioning the optimization variables into groups, such that the variables in the same groups generally share more constraints than variables between different groups. The proposed method is applied to the decomposition of the optimal control problem involved in the nonlinear model predictive control of a reactor-separator process, and the quality of the resulting decomposition is examined by the resulting control performance and computational time. Our result suggests that community detection in network representations of the optimization problem generates decompositions with improvements in computational performance as well as a good optimality of the solution.