A new ant colony system model for unit commitment problem

Abstract

In this paper, a new ant colony system (ACS) model has been proposed for solving practical unit commitment problem (UCP). ACS is more suitable for solving combinatorial optimization problem, so a new ACS model has been applied to the UCP which is hard combinatorial in nature. Minimum cost path for the scheduling of thermal units has been obtained based on the concept of artificial ants find the shortest distance to the food source from the nest. In the first step, ants are allowed to search (ant search space) all the possible combination of the states satisfying the load demand for the complete scheduling period. Then an ant search mechanism is adopted where the artificial ants explore the ant search space to find the overall minimum generation cost. The search mechanism incorporates a multi decision states that makes the ACS approach superior over traditional dynamic programming (DP) method. In this new ACS model, elistic ants are added with the existing simple ACS model making the new ACS based model slightly advantageous over the ACS models presently in use. The proposed new ACS model successfully solves the UCP and its performance has been tested for practical ten unit system and compared with the conventional dynamic programming and simple ACS. The proposed model may be extended to practical system with large number of generating units.