A Novel Identification-Based Convex Control Scheme via Recurrent High-Order Neural Networks: An Application to the Internal Combustion Engine

An Empirical Study on Application of Machine Learning and Neural Network in English Learning

“…In machine learning and related fields, the computational models of artificial neural networks are inspired by the central nervous system of animals and are used to estimate or can rely on a large number of inputs and general unknown approximate functions. Artificial neural networks are usually presented as interconnected "neurons" that can compute values from input and are capable of machine learning and pattern recognition due to their adaptive nature [3]. Artificial neural network also has the preliminary ability of self-adaptation and self-organization.…”

Exact Takagi-Sugeno descriptor models of recurrent high-order neural networks for control applications

“…The number of scalar decision variables and LMI rows in this example is N d = 644 and N l = 74, respectively; based on them, the computational burden is given by 10.2959 for solving a total of 10 LMI expressions. Comparatively, that of LMI conditions in (Armenta et al 2019, Theorem 3) is 11.0819 since N d = 515 and N l = 884 for a total of 442 LMIs; thus, the proposed methodology helps attenuating the computational burden while tackling the effects of the modeling error.…”

“…1 In prior literature, for simplicity, linear control techniques have been applied to RHONN models at the cost of losing important characteristics (Armenta et al 2017); nonlinear control, on the other hand, has been applied via model inversion (Sanchez and Bernal 2000) and sliding modes (Castañeda et al 2013), but these approaches cancel out the RHONN nonlinearities instead of using them appropriately. Recently, a solution for implementing nonlinear control to a RHONN in the form of parallel distributed compensation (PDC) (Wang et al 1996) has appeared in (Armenta et al 2019); a Takagi-Sugeno (TS) model allowing design conditions in the form of linear matrix inequalities (LMIs) (Boyd et al 1994) and sum-of-squares (SOS) (Prajna et al 2004) has been obtained via a suitable transformation of the RHONN. The LMI and SOS techniques belong to the convex optimization framework which allows solving conditions in polynomial time and straightforwardly incorporate performance specifications such as speed of convergence, input/output constraints, observation (Han et al 2018), and H ∞ disturbance attenuation (Chayaopas and Assawinchaichote 2018): this work pursues this path as H ∞ optimization is at the root of modern robust control methodologies.…”

“…Contribution A RHONN is rewritten as an exact TS descriptor model in a tensor-product form with a fixed number of nonlinearities despite its order (Armenta et al 2019), based on which quadratic (Taniguchi et al 2000) as well as semi-quadratic Lyapunov functions (Guerra et al 2007(Guerra et al , 2015 can be used to derive LMI conditions for H ∞ -optimal control for a prescribed output of interest y(t) (see Fig. 2).…”