A new coding for solving large-scale non-convex economic dispatch problems without a penalty factor

摘要

In some optimization problems, it is necessary that a sum of variables converges to an expected value during the optimization process. In solving this type of problem, a penalty factor is usually employed in the objective function. Thus, some disadvantages such as increased search space is unavoidable. This research proposes a new coding based on particle swarm optimization in order to solve economic dispatch problems without a penalty factor, which, in turn, controls the sum of all variables. The suggested coding could effectively escape from infeasible solutions; thereby, search efficiency and solution quality would be dramatically improved. In order to verify the effectiveness of the proposed coding, it is implemented on various economic dispatch problems, i.e. test systems with 6, 15 and 40 generating units as well as a typical Korean system with 140 generating units, all with non-convex and non-smooth cost functions. In addition, the numerical results are compared with those of other well-established and valid techniques.