Application of soft computing in a discrete optimization problem

Holnicki, Piotr; Kałuszko, Andrzej

Download

Title: Application of soft computing in a discrete optimization problem

Abstract:

In the paper two methods of solving a discrete optimization problem are discussed. The problem itself is related to air quality protection on a regional scale. The approach refers to optimal allocation of financial means for emission reduction in a given set of power and heating plants. The implementation considered is sulfur-oriented. The problem is formally stated as cost-constrained minimization of environmental damage function by the optimal choice of desulfurization technologies, within the predefined set of the controlled plants. The receptor-oriented objective function utilizes air pollution forecast preprocessed by a regional scale dispersion model. An integer-type optimization problem is solved by two methods. The first method utilizes a heuristic algorithm designed for solving this specific problem, which directly finds discrete solution. Another approach is based on the classical gradient optimization algorithm and gives continuous, technologically not applicable solution. Then the continuous solution is transformed to the discrete form by enumeration of some discrete cases. Both algorithms has been implemented and tested on the real data for selected region. The case study relates to the set of major power planls in Silesia Region (Poland) and the basic desulfurization technologies, which are Io be allocated. The test calculations allows us to evaluate accuracy of the heuristic method as welI as applicability of both approaches for supporling decisions concerning optimal strategies of emission abatement on a regional scale.