A novel method based on modified shuffled frog leaping algorithm and artificial neural network for power system load prediction

Considering the necessity of accurate power load demand prediction, a sufficient method based on modified shuffled frog leaping algorithm (MSFLA) and artificial neural network (ANN) is proposed to enhance the degree of conformity of the predicted power demand to its actual value. The ability of ANN in finding any nonlinear mapping between different variables has resulted to a wide usage in prediction applications. On the other hand the ability of evolutionary algorithms in solving complex nonlinear problems and finding the global optimum solution has made them as one of the most important tools in electrical power systems. Therefore in this paper for the first time we have used the evolutionary MSFLA to adjust the weighting matrix of ANN and so reducing the degree of uncertainty existing in load demand prediction.