Artificial neural network based peak load forecasting using Levenberg-Marquardt and quasi-Newton methods

Daily electrical peak-load forecasting has been done using the feedforward neural network based on the Levenberg-Marquardt back-propagation algorithm, Broyden-Fletcher-Goldfarb-Shanno back-propagation algorithm and one-step secant backpropagation algorithm by incorporating the effect of eleven weather parameters, the type of day and the previous day peak load information. To avoid the trapping of the network into a state of local minima, the optimisation of user-defined parameters viz. learning rate and error goal has been performed. Training data set has been selected using a growing window concept and is reduced as per the nature of the day and the season for which the forecast is made. For redundancy removal in the input variables, reduction of the number of input variables has been done by the principal component analysis method of factor extraction. The resultant data set is used for the training of a three-layered neural network. To increase the learning speed, the weights and biases are initialised according to the Nguyen and Widrow method. To avoid over-fitting, an early stopping of training is done at the minimum validation error.

[1]  Rey-Chue Hwang,et al.  An adaptive modular artificial neural network hourly load forecaster and its implementation at electric utilities , 1995 .

[2]  John E. Dennis,et al.  Numerical methods for unconstrained optimization and nonlinear equations , 1983, Prentice Hall series in computational mathematics.

[3]  David A. Wismer,et al.  Introduction to nonlinear optimization , 1978 .

[4]  Bernard Widrow,et al.  Improving the learning speed of 2-layer neural networks by choosing initial values of the adaptive weights , 1990, 1990 IJCNN International Joint Conference on Neural Networks.

[5]  Mohammad Bagher Menhaj,et al.  Training feedforward networks with the Marquardt algorithm , 1994, IEEE Trans. Neural Networks.

[6]  Robert J. Marks,et al.  Electric load forecasting using an artificial neural network , 1991 .

[7]  Roberto Battiti,et al.  Learning with first, second, and no derivatives: A case study in high energy physics , 1994, Neurocomputing.

[8]  Roberto Battiti,et al.  BFGS Optimization for Faster and Automated Supervised Learning , 1990 .

[9]  Roberto Battiti,et al.  First- and Second-Order Methods for Learning: Between Steepest Descent and Newton's Method , 1992, Neural Computation.

[10]  G. Ramakrishna,et al.  Power-demand forecasting using a neural network with an adaptive learning algorithm , 1995 .

[11]  D. K. Ranaweera,et al.  Application of radial basis function neural network model for short-term load forecasting , 1995 .

[12]  Alireza Khotanzad,et al.  ANNSTLF-Artificial Neural Network Short-Term Load Forecaster- generation three , 1998 .

[13]  S. Vemuri,et al.  Neural network based short term load forecasting , 1993 .

[14]  Shangyou Hao,et al.  An implementation of a neural network based load forecasting model for the EMS , 1994 .

[15]  Magdi S. Mahmoud,et al.  Cascaded artificial neural networks for short-term load forecasting , 1997 .

[16]  H. Harman.  Modern factor analysis , 1961 .