Genetic algorithms and rough fuzzy neural network-based hybrid approach for short-term load forecasting

This article describes a way of designing a hybrid system for short-term load forecasting, integrating rough sets theory with fuzzy neural networks using a multi-objective genetic algorithm. The multi-objective genetic algorithm is used to automatically learn the knowledge of historical data and find the best factors that are relevant to electric loads. The concept of entropy is introduced to describe the uncertainty of decision rules with dependency factors, and the crude domain knowledge expressed by decision rules is applied to design the structure and weights of the neural network. Simulation results demonstrate that the rough fuzzy neural network has better precision and convergence than the traditional fuzzy neural network for its simple, transparent network structure and effective inputs