Abstract
The resolution of the environmental/economic dispatch (EED) problem using the different methods which are proposed in literature consumes an important computing time. Thus, the present paper deals with a technique based on two steps to solve the EED problem of electric energy power in real-time for forecast load curve. The first step uses the NSGAII approach (Non-dominated Sorting Genetic Algorithm) to solve the multi-objective problem MOP for different levels of load by treating the two cases, problem without line constraints and with line constraints. To verify effectiveness of this approach, NSGAII is compared with other algorithms which are used in the literature. Such as, weighted sum method (WSM), NPGA (Niched Pareto Genetic Algorithm), NSGA and SPEA (Strength Pareto Evolutionary Algorithm). To exploit the results in real time for forecast load curve, second step uses a radial basis function neural network (RBFN) with 3 layers, input layer formed by the level of global load, hidden layer and output layer formed by the generations of the various machines. The validity and effectiveness of this technique are verified by an example of a load curve of a didactic electric network IEEE 30-bus system with 6-generating units.