Experimental and neural network approach to effective electrical conductivity of carbon nanotubes dispersed chiral nematic liquid crystals

Abstract

Single walled carbon nanotubes (SWCNT’s) doped cholesteric liquid crystal composite has been prepared and characterized for their electrical responses. Also theoretically, an artificial neural network (ANN) approach has been trained for predicting the effective electrical conductivity of these composites. The ANN models are based on a feedforward backpropagation (FFBP) network with such training functions as the adaptive learning rate (GDX), gradient descent with adaptive learning rate (GDA), gradient descent (GD), conjugates gradient with Powell-Beale restarts (CGB), one-step secant (OSS), and Levenberg–Marquardt (LM), and training algorithms run at the uniform threshold transfer functions-Tangent sigmoid (TANSIG) and pure linear (PURELIN) for 1000 epochs. Our modeling confirms that the expected effective electrical conductivity by different training functions of ANN is in higher agreement with the experimental results of SWCNT doped CLC composites.