Abstract:

The constraints of output variables, input variables and intermediate variables exist generally in chemical process control.The inconsistency in different constraints cannot make all of constraints satisfied, and optimal controller does not have feasible solutions, which adversely affects production.Based on convex polyhedron geometry, the feasibility analysis of constrained optimal control in chemical processes was transformed into whether two convex polyhedra intersect or not, and the reasonable constraints adjustment was transformed into a series of linear programming or nonlinear programming when the constrained optimization problem was infeasible.An algorithm for feasibility analysis and automatic constraints adjustment of constrained optimal control independent of user intervention was proposed. Simulation results of typical control problem of a heavy oil fractionator proposed by Royal Dutch/Shell Group showed that when the constrained optimization problem was infeasible the proposed algorithm could make constraints adjustment automatically and effectively, with small overshoot, smooth change of control action, and a certain control overdesign.