A Short-term Power Load Prediction Model Based on Morlet WNN

Abstract

Power load is the key to work of regional grid. The stability of the regional power grid will be improved if we can predict the power load in short term. So, this paper provides a new hybrid model to forecast electric power in short term based on generalized extremum learning machine (GELM) and Morlet wavelet neural net (WNN). Firstly, regarding power load probability as model output, this model will cut power load information into different parts according to frequency attributes using wavelet function, and then every part will be analyzed with similar scaler. Secondly, the WNN is trained quickly by the GELM. After randomly picking deviation between hidden layer and input coefficient, some output coefficient will be got by generalized inverse transformation of matrix. Then, after the displacement coefficient and scale coefficient of Morlet function being randomly selected, the hidden layer coefficients are gained with matrix inversion, and then the Morlet WNN is designed out. Finally, taking load data of some regional power grid as experimental data, the results show that the proposed model has high accuracy and reliability, which proves the high effectiveness of the algorithm.