A genetic algorithm solution to a new fuzzy unit commitment model

This paper proposes a new fuzzy model for the unit commitment problem (UCP). A solution method for the proposed UCP model based on the genetic algorithms (GAs) is presented (FZGA). The model treats the uncertainties in the load demand and the spinning reserve constraints in a new fuzzy logic (FL) frame. The proposed FL model is used to determine a penalty factor that could be used to guide the search for more practical optimal solution. The implemented fuzzy logic system consists of two inputs: the error in forecasted load demand and the amount of spinning reserve, and two outputs: a fuzzy load demand and a penalty factor. The obtained fuzzy load demand is more realistic than the forecasted crisp one; hence the solution of the UCP will be more accurate. In the proposed FZGA algorithm, coding of the solution is based on mixing binary and decimal representation. The fitness function is taken as the reciprocal of the total operating cost of the UCP in addition to penalty terms resulted from the fuzzy membership functions for both load demand and spinning reserve. Results show that the fuzzy-based penalty factor is directly related to the amount of shortage in the committed reserve; hence will properly guide the search, when added to the objective function, in the solution algorithm of the UCP. Accordingly, acceptable level of reserve with better-cost savings was achieved in the obtained results. Moreover, the proposed FZGA algorithm was capable of handling practical issues such as the uncertainties in the UCP. Numerical results show the superiority of solutions obtained compared to methods with traditional UCP models.