A Comparison of Fuzzy Neural Networks for the Generation of Daily Average and Peak Load Profiles
P.K. Dash, A.C. Liew, S. Rahman
International Journal of Systems Science, vol. 26, no. 11, 1995, pp. 2091-2106
Two fuzzy neural network models, based on the multilayer perceptron and capable of fuzzy classification of patterns, are presented in this paper. The first type of fuzzy neural network uses the membership values of the linguistic properties of the past load and weather parameters, and the output of the network is defined as fuzzy class membership values of the forecasted load. The backpropagation algorithm is used to train the network. The second type of fuzzy neural network is developed based on the fact that any fuzzy expert system can be represented in the form of a feedforward neural network. This kind of fuzzy neural network is trained to develop fuzzy logic rules and to find optimal input/ output membership values. A hybrid learning algorithm, consisting of unsupervised and supervised learning phases, is used to train this network. Extensive tests have been performed on a two-year utility data for generation of peak and average load profiles in a 24-hours-ahead time frame, and results for two typical winter and summer months are given to confirm the effectiveness of the two models.