A CO2 compression and purification unit (CPU) using the phase separation method is a promising option to obtain high-purity CO2 products from oxy-fuel combustion power plants. For such CPU systems, a self-optimizing control method is applied for the first time with the objectives of minimizing energy consumption and economic cost. Although the set of controlled variables is determined identically for self-optimizing control structures based on the targets of specific energy consumption (case 1) and specific economic cost (case 2), their control pairings are different. The optimized control structures are feasible for implementation of the operating targets with low energy and cost penalties. Case 1 would be the most suitable control strategy, because its absolute amplitude of variations for specific energy consumption and specific economic cost are only 14.02% and 10.29% of those for case 2, respectively. The results provide a new solution to achieve energy-efficient and cost-effective operations for oxy-fuel combustion systems.