Int J Performability Eng ›› 2021, Vol. 17 ›› Issue (4): 364-370.doi: 10.23940/ijpe.21.04.p4.364370

• Original article • Previous Articles     Next Articles

Ameliorating Vertically Bundled Electricity Price Prediction Exclusively from ICMLP Network

S. Anbazhagan a and Bhuvaneswari Ramachandranb   

  1. a Department of Electrical and Electronics, Annamalai University, Annamalai Nagar, 608002, India Electrical and Computer Engineering, University of West Florida, Pensacola, 32514, USA


In this paper, an improved complex-valued multi-layer perceptron (ICMLP) network has been proposed with a view to forecast the fluctuating prices of the deregulated market. The existing literature focuses on a hybrid of hard and soft computing models that should be able to capture the nonlinearity associated with those electricity market prices. Apart from individual models, the hybrid algorithms are based on the integration of different computing paradigms, which in turn increases the time complexity. An individual model for estimating electricity prices has yet to be developed. This paper proposes an exclusive approach called ICMLP with logarithmic performance index for forecasting electricity prices. In ICMLP, a novel optimization criterion is used that takes into account minimizing both the errors by magnitude and phase. Efficacy of ICMLP is evaluated using benchmark data sets of Spanish electricity market. The results have been found to be promisingly better than the existing prediction-based models. Simulation results demonstrate that the ICMLP is superior in terms of accuracy, training speed, and structure compactness.


Key words: improved complex-valued multi-layer perceptron, deregulated electricity market, soft computing, electricity price forecasting