An electricity price forecasting model with fuzzy clustering preconditioned ANN

Satoshi Itaba, Hiroyuki Mori

Research output: Contribution to journalArticle


In this paper, a hybrid model of fuzzy clustering and ANN (Artificial Neural Network) is proposed for electricity price forecasting. Due to the complicated behavior of electricity price in power markets, markets players are interested in maximizing profits while minimizing risks. As a result, more accurate models are required to deal with electricity price forecasting. This paper proposes a new method that makes use of fuzzy clustering preconditioned GRBFN (Generalized Radial Basis Function Network) to provide more accurate predicted prices. Fuzzy clustering plays a key role to prevent the number of learning data from decreasing at each cluster. GRBFN is one of efficient ANNs to approximate nonlinear systems. Furthermore, a modified GRBFN model is developed to improve the performance of GRBFN with the use of DA (Deterministic Annealing) clustering for the parameters initialization and EPSO (Evolutionary Particle Swarm Optimization) for optimizing the parameters of GRBFN. The proposed method is successfully applied to real data of ISO New England, USA.

Original languageEnglish
Pages (from-to)90-98
Number of pages9
JournalIEEJ Transactions on Power and Energy
Issue number2
Publication statusPublished - 1 Jan 2018



  • Artificial neural network
  • Clustering
  • EPSO
  • Electricity price
  • Forecasting
  • Fuzzy logic
  • Optimization

Cite this