Application of Constrained Multi-objective Evolutionary Algorithm in a Compressed-air Station Scheduling Problem

The compressor station is the utility that generates the compressed-air. This paper aims to present the formulation of the operation scheduling optimization problem for a compressor station, and apply a constrained multi-objective evolutionary algorithm, named CMOEA/D-CDP, to solve the problem. For this problem, it can be defined as one of the nonlinear 0–1 mixed integer multi-objective optimization problems due to the decision variables including continuous variables and 0–1 integer variables. In light of the relation between the unit commitment and rotational speed of a compressor station, a special encoding scheme which decreases the dimension of decision variables is proposed to change this nonlinear 0–1 mixed integer multi-objective optimization problem into a nonlinear continuous multi-objective optimization problem. The result obtained by CMOEA/D-CDP is a set of Pareto optimal solutions, which can help a decision maker select a specific preferred solution to guide the production under the different limits of objectives. Experimental results on two typical compressor stations indicate that the proposed encoding scheme shows apparent superiority compared to the normal encoding method.