Title: Neural network storage unit parameters modelling

Authors: Adel El Shahat

Addresses: Engineering Science Department, Faculty of Petroleum and Mining Engineering, Suez University, Suez, Egypt

Abstract: This paper addresses a new method of electrical parameters modelling for lead-acid battery as a storage device based on performance characteristics using artificial neural network with suitable numbers of layers and neurons, with excellent regression constants. First, battery parameters are identified by curve fitting with improved Thevenin model, and validated with a 12 V, 4 Ah lead-acid battery. Second, ANN technique is used to model Thevenin electric model parameters. Models outputs are: discharging resistance, shunt resistance, back e.m.f. and charging resistance; each one is deduced with battery characteristics as inputs: charging/discharging rate, state of charge, time, voltage, and current. Finally, discharging and charging characteristics of the battery model are implemented for more visibility. These models easily identify parameters and characteristics for this battery type with capacity ranges 0.05, 0.1, 0.2, 0.4, 0.6, 1, 2 and 3 CA. Error and comparisons figures are adopted for validation purposes.

Keywords: parameter modelling; lead-acid batteries; ANNs; artificial neural networks; estimation; storage units; electrical parameters; battery parameters; curve fitting.

DOI: 10.1504/IJIED.2014.066215

International Journal of Industrial Electronics and Drives, 2014 Vol.1 No.4, pp.249 - 274

Received: 26 Feb 2014
Accepted: 20 Jun 2014
Published online: 20 Dec 2014 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article