Investigation the Success of Semidefinite Programming for the Estimating of Fuel Cost Curves in Thermal Power Plants

Abstract

Accurate estimation of fuel cost curve parameters in thermal power plants is of great importance because these parameters directly influence the economic dispatch calculations. In this paper, a semidefinite programming (SDP) approach was proposed for the estimation of fuel cost functions' parameters in thermal power plants. The parameter estimation problem was designed as a minimization problem, where the objective function was accepted as the total absolute error (TAE) in the study. Also, linear, quadratic, and cubic fuel cost functions were used to estimate the fuel cost parameters. Different fuel types such as coal, oil and gas were preferred for simulation studies. The results achieved from the semidefinite programming method were compared with that of particle swarm optimization (PSO), artificial bee colony (ABC), crow search algorithm (CSA) and least error square (LES) methods, respectively. The performance of the methods were compared according to the TAE parameter. Simulation results showed that SDP method is more successful than other methods considered in this paper. Clearly, the present paper showed that SDP has a higher potential to solve parameter estimation problems.