A New Training Algorithm for Artificial Hydrocarbon Networks Using an Energy Model of Covalent Bonds

Abstract

Computational algorithms for modeling problems are widely used in control engineering systems. Several algorithms for modeling systems are proposed in literature. However, they have some drawbacks like stability of algorithms and absence of information inside the model acting as black boxes. Recently, a novel technique called artificial hydrocarbon networks (AHNs) was proposed to improve the latter drawbacks in computational algorithms; but there is no training algorithm that specifies how to build the structure and tune all parameters involved on it. Thus, this paper introduces a new training algorithm for AHNs using an energy model of covalent bonds inspired on organic chemistry observations. Results report that this training algorithm can be used for designing the structure of AHNs while obtaining the parameter values, both at the same time.