A proposal for a cooperative cross-entropy method to tackle the unit commitment problem

Abstract

The paper presents a proposal about the Cooperative Cross-Entropy to solve Combinatorial Optimization Problems. The cooperative scheme consists of the definition of several instances of the Cross-Entropy method and the establishment of an exchange policy of solutions between the different instances participating in the model. Each instance consists of the implementation of a cross-entropy algorithm with particular parameters that originates a different probability density function for the generation of new solutions. The exchange policy is implemented synchronously between the instances, and the solutions for the exchange process are chosen within the elite solutions of each instance of the cooperative model. For the experimentation, we use a simplified version of the Unit Commitment Problem (UCP). The computational results show that for large instances of the UCP, the cooperative model is effective. Furthermore, the proposed model shows a better balance between intensification and exploration for a certain combination of parameters, and therefore, it seems to be a good approach to tackle other optimization problems.