Energy Efficiency in NARMAX Models for Reduced Carbon Footprint

Code efficiency has gained significant attention as an important mechanism to address the increasing energy consumption within the Information and Communication Technology sector. Despite advances in algorithm optimisation, there has been limited research on reducing the environmental impact of recursive functions, particularly in the context of nonlinear dynamical systems. To address this gap, this paper investigates the energy efficiency of NARMAX (Nonlinear Autoregressive Moving Average with Exogenous inputs) models, focusing on reducing energy consumption and CO2 emissions while maintaining computational accuracy. Using Horner's method on NARMAX models, the research optimises the performance of recursive algorithms and examines their energy consumption in nonlinear dynamical systems. The method is applied in four case studies, which demonstrate reductions of up to 27.9% in CO2 emissions and 28.85% in energy consumption.