Performance of the novel rough fuzzy-neural network on short-term load forecasting

Abstract

A hybrid model Integrating with rough set theory and fuzzy neural network is presented for short-term load forecasting. Multi-objective genetic algorithm is used to learn automatically the knowledge of historical data set and find the best factors that are relevant to electric loads, and the crude domain knowledge extracted from the elementary data set 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. Moreover, it becomes easier to understand the transferring way of knowledge in neural network. Index Terms--Rough set, Fuzzy-neural network, Multiobjective genetic algorithm, Short-term load forecasting, Data miningThis work is supported by NSF of Zhejiang Province, China (No. 502053) Li Feng is a Ph.D. student in College of electrical engineering,

Survey of Rough and Fuzzy Hybridization

“…The resulting classification is also shown to be an improvement over the conventional approach. A similar improvement in performance with the use of rough-fuzzy neural networks over fuzzy neural networks was reported in [18] for prediction of short-term electricity load forecasting. They also used genetic algorithms for the selection of the best set of inputs for the prediction.…”

“…For example, [18] used GAs for selecting the best set of inputs for rough-fuzzy neurocomputing. Similarly, [49] used GAs in rough and fuzzy set based rule generation.…”