ABSTRACT
This chapter presents the alternatives under two categories: first is the type of connectivity, second is the scheme adopted for neuronal activation. In both models, the neural networks have single hidden layer with hyperbolic tangent-type neuronal nonlinearity and linear output neurons. The framework of artificial neural networks has established an elegant bridge between problems, which display uncertainties, impreciseness with noise and modeling mismatches, and the solutions requiring precision, robustness, adaptability, and data-centeredness. The discipline of control engineering with tools offered by artificial neural networks has stipulated a synergy the outcomes of which are distributed over an enormously wide range. Despite the presence of a number of learning algorithms, two of them have become standard methods in engineering applications, and are elaborated in the sequel. The desired path is followed at an admissible level of accuracy under the variation of battery voltage shown in the middle subplot.
